Table of Contents I Cost of Illness Handbook I USEPA
http://www.epa.gov/oppt/coi/pubs/toc.html
742B91001
Cost of Illness Handbook
http://www.epa.gov/oppt/coi/pubs/toc.html
Last updated on Thursday, October llth, 2007.
You are here: EPA Home Prevention, Pesticides & Toxic Substances Pollution Prevention and Toxics
Cost of Illness Handbook Table of Contents
Table of Contents
PLEASE NOTE: Some of the documents mentioned in
this Section are in Adobe's Portable Document Format
(PDF). To view or print them you will need to have the
Adobe Acrobat Reader program installed on your
computer. The Reader can be downloaded and used
with no charge; check here for more information on the
Adobe Acrobat Reader.
Executive Summary [PDF, 19 KB]
Part I: Introduction to the Cost of Illness Handbook [PDF,
149 KB]
Part II: Cancer
• Chapter II. 1.
KB]
Chapter II.2.
Introduction to the Costs of Cancers [PDF, 87
Cost of Stomach Cancer [PDF, 143 KB]
Chapter II.3. Cost of Breast Cancer (pending)
Chapter II.4. Cost of Kidney Cancer [PDF, 36 KB]
Chapter II.5. Cost of Lung Cancer [PDF, 115 KB]
Chapter II.6. Cost of Skin Cancer (pending)
Chapter II.7. Cost of Colorectal Cancer [PDF, 61 KB]
Chapter II.8. Cost of Bladder Cancer [PDF, 125 KB]
Part III: Adverse Developmental Effects
Chapter III.1.
Illnesses and
Chapter III.2.
Chapter III.3.
Chapter III.4.
Chapter III.5.
Chapter III.6.
Chapter III.7.
Chapter III.8.
Chapter III.9.
Children [PDF
Introduction to the Costs of Developmental
Disabilities [PDF, 59 KB]
Cost of Low Birth Weight [PDF, 51 KB]
Cost of Cleft Lip and Palate [PDF, 54 KB]
Cost of Limb Reductions [PDF, 32 KB]
Cost of Cardiac Abnormalities [PDF, 46 KB]
Cost of Spina Bifida [PDF, 32 KB]
Cost of Cerebral Palsy [PDF, 35 KB]
Cost of Down Syndrome [PDF, 33 KB]
Cost of Reducing High Blood Lead Levels in
32 KB]
Part IV: Respiratory Illnesses
• Chapter IV. 1. Introduction to Respiratory Diseases [PDF, 9
KB]
• Chapter IV.2. Cost of Asthma [PDF, 221 KB]
• Chapter IV.3. Cost of Acute Respiratory Diseases:
Hypersensitivity Pneumonitis, Humidifier Fever, and
Legionnaires' Disease [PDF, 30 KB]
• Chapter IV.4. Middle Ear Infections (Otitis Media)
(pending)
Part V: Symptoms
• Chapter V.I. Symptoms [PDF, 6KB]
• Chapter V.2. Symptom Groups [PDF, 76 KB]
Iof2
12/4/2007 9:55 AM
-------
Table of Contents I Cost of Illness Handbook I USEPA
http://www.epa.gov/oppt/coi/pubs/toc.html
Appendices
• Appendix A: Inflation and Discounting Factors [PDF, 28
KB]
Glossary [PDF, 7 KB]
References [PDF, 39 KB]
2 of 2
12/4/2007 9:55 AM
-------
EXECUTIVE SUMMARY
The societal benefits of environmental regulations and programs are
typically manifested by the reduction in adverse health effects. These
reductions are associated with decreased exposure to environmental
agents. Ideally, valuation of these human health benefits would include all
costs to society associated with the benefits, including medical costs,
work-related costs, educational costs, the cost of support services required
by medical conditions, and the willingness of individuals to pay to avoid the
health risks. These factors can be referred to in aggregate as society's
total willingness to pay to avoid an illness. Many of these categories of
information are difficult to obtain. In particular, obtaining an accurate
measure of a society's total willingness-to-pay to avoid illnesses is often
not possible. Consequently, analysts often use alternative measures of the
costs saved when illnesses are avoided. Direct medical costs, which
measure non-subjective aspects of an illness — the expenditures on medical
care — are often used as a lower-bound estimate of avoiding an illness.
This handbook, developed by the U.S. Environmental Protection Agency,
provides direct per capita incremental medical costs of illnesses associated
with environmental pollutants.1 The handbook was developed in response
to the Agency's desire to provide information on the benefits associated
with disease avoidance resulting from environmental programs or
regulations. The data can be used in economic analyses, policy
development or evaluation, and various decision-making activities.
Improvements in human health frequently constitute a major portion of the
benefits resulting from environmental regulations. While there are a variety
of approaches to estimating the value of these benefits, one of the more
straightforward approaches is to calculate the medical and related costs
avoided. The medical costs in this handbook provide a relatively simple
and efficient lower-bound estimate of the costs of illnesses.
The cost of illness data provided in this handbook include some, but not all
components of the total benefit of avoiding a disease. Those outside the
scope of this analysis are direct wow-medical costs, the opportunity costs
of patients, family members or other unpaid caregivers, and what the
patient and others would be willing to pay to avoid the anxiety, pain, and
suffering associated with the illness. Due to the seriousness of most
illnesses in this handbook, these components may be substantial. The
1 This handbook was developed by the Office of Pollution Prevention and Toxics under the
direction of Dr. Nicolaas Bouwes (EPA WAM) by Abt Associates, Cambridge, Massachusetts (Dr. K.
Cunningham, Project Manager).
Cost of Illness ES-1 Executive Summary
-------
values reported in this handbook must therefore be viewed as partial
estimates of the economic costs, and are useful primarily as lower-bound
estimates of cost.
EPA selected diseases for inclusion in this handbook based on the known
or anticipated need for disease cost estimates for regulatory or policy
activities and a review of the environmental health literature. Estimates of
medical costs are provided for the following illnesses (click on the illness
name to link to the relevant chapter; click on the section numbers to reach
the introductory chapters):
Section II: Cancers2
stomach cancer
breast cancer
kidney cancer
lung cancer
skin cancer
colorectal cancer
bladder cancer
Section III: Developmental Illnesses and Disabilities
low birth weight
cleft lip and palette
limb reductions
cardiac abnormalities
spina bifida
cerebral palsy
Down syndrome
high blood lead levels
Section IV: Respiratory Diseases
asthma
acute respiratory illnesses
middle ear infections
Section V: Symptoms
symptom groups
The diseases are organized into handbook sections with similar types of
diseases, as shown in the list above. Each chapter contains background
information in the illness, method used to estimate medical costs, and
present value cost estimates discounted at zero, three, five, and seven
percent over the duration of the disease. Costs are provided which were
current in the year in which the chapter was written or revised (1996 and
2Bone and liver cancer costs are also briefly discussed in the introductory cancer chapter (Chapter
II.l).
Cost of Illness ES-2 Executive Summary
-------
Link to Appendix A
forward). The costs can be updated to the current year using the
Consumer Price Index (CPI) Medical Services inflation data provided in
Appendix A: Inflation and Discounting Factors.
Matrices with preliminary information on environmental agents that may be
associated with cancer and birth defects are provided in the chapters that
introduce each of those disease categories. Each chapter also discusses
causality and especially susceptible subgroups of the population.
A core of information is provided in each chapter on the methodology,
costs, sources of uncertainty, and background information on the illness.
The handbook was developed over many years, and the cost estimates for
each illness were developed to address specific program requirements
within the Agency. Consequently, the type of information provided and
level of detail involved in the analyses vary among the illnesses.
The direct medical costs incurred as the result of an illness were estimated
for the duration of the illness, i.e., from diagnosis to cure or patient death.
Expected costs are estimated for each year post-diagnosis until cure or
death, incorporating information on the likelihood and timing of receiving
specific treatments, as well as survival data, information on the age of onset
of the disease, and life expectancy data. Medical cost estimates are subject
to advances in medical practice and changes in the costs of both services
and materials. Most cost estimates are based on recent evaluations of
medical practice; the handbook provides dates when cost and treatment
data were obtained and descriptive information regarding disease definition
and treatment. The user should consider changes in practice over time,
however, if recent advances or changes in treatment have been reported.
The goal of the handbook is to provide cost estimates that are generalizable
to any area of the United States. To obtain cost data representative of the
nation as a whole, standard disease treatment methods, using generally
acceptable practices, and the average patient experience regarding
prognosis and survival (e.g., life expectancy) were used in cost estimates.
As noted above, the costs provided in this handbook do not include many
non-medical costs, which may be substantial and should be included in a
comprehensive benefit evaluation. Although non-medical costs may be an
important component of overall benefit, direct medical costs are likely to
comprise a substantial portion of the cost to society for the diseases
included in this handbook. Thus, the medical cost estimates provided in
the handbook offer reasonable lower-bound estimates for many illnesses of
environmental concern.
Cost of Illness ES-3 Executive Summary
-------
It is anticipated that the contents will continue to be supplemented with
new illnesses, and with revisions to illnesses currently included in the
Handbook. In addition, links will be made to other sources of information
on this topic. EPA welcomes the submission of new data, comments, and
recommendations from users of this Handbook.
Cost of Illness ES-4 Executive Summary
-------
CHAPTER 1.1. INTRODUCTION TO THE COST OF ILLNESS HANDBOOK
Clicking on the sections below will take you to the relevant text.
I.I. A Overview
1.1 .B Willingness to Pay and Cost Components
1.1 .B. 1. Definition of Willingness to Pay
1.1 .B.2 Components of Willingness to Pay
1.1 .B.3 Approaches to Measuring Willingness to Pay
I.1.B.4 Conculsions
1.1 .C Organization of Handbook.
1.1 .C. 1 Illnesses Covered in the Handbook
I.l.C.2 Chapter Format
I.l.C.3 Section Contents
I.l.C.4 Selection of Illnesses
1.1 .C.5 Linking Diseases to Agents
1.1 .D Methods used to Estimate Direct Medical Costs
1.1 .D. 1 Overview of Method
1.1. D. 2 Survivors and Nonsurvivors
I.1.D.3 Data Sources
1.1 .E Susceptible Subpopulations
1.1 .F Limitations
1.1 .F. 1 Uncertainties Regarding the Market Value of Medical Goods and
Services
1.1 .F.2 Differences in Critical Patient Characteristics.
1.1 .F.3 Geographic Differences in Medical Practices and Services
1.1 .F.4 Uncertainty Regarding the Application And/or Accuracy of Input
Data
Appendix 1.1-A Equations Describing the Expected Present Discounted Value of the Per
Capita Lifetime Stream of Costs Associated with a Given Illness
Chapter 1.1 1.1-1 Introduction to the Cost of Illness Handbook
-------
CHAPTER 1.1. INTRODUCTION TO THE COST OF ILLNESS HANDBOOK
This handbook is provided through EPA's website. All chapters in the
handbook have links to this chapter because it contains basic information
on objectives, content of the handbook, analytical methodology,
limitations, and results. It is anticipated that the web site will be updated
continuously and this chapter will be modified as new information becomes
available. As shown in the example below, you can use sidebars appearing
at the left to link to resources that may be useful while reading and using
the Chapters. These include the table of contents (which can be used to
link to any chapter), a glossary with abbreviation definitions, inflation and
discounting factors, an executive summary, and other useful websites (e.g.,
OPPT).
Sidebar example:
Click here to link to the Table of Contents
The cost of illness is an estimate of the incremental direct medical costs
associated with medical diagnosis, treatment, and follow-up care. This
includes various cost elements, such as physician visits, hospitalization, and
Pharmaceuticals. This Handbook does not estimate the costs in lost time
or wages that may be incurred by either a patient or his or her unpaid
caregiver. The costs also do not include pain and suffering, which may be
substantial. Rather, this Handbook provides information on medical
treatments and their costs, usually aggregated over the lifetime of the
patient. These costs are inflated to the current year, and summarized at
various discount rates.1 A discussion of willingness-to-pay, which is
presented later in this chapter, outlines in more detail the cost elements that
are and are not included in the Handbook.
A normative approach is used to estimate costs, whereby the average age
at diagnosis, the average life expectancy, and other average or mean values
are used to estimate costs. The text notes situations which may arise that
would lead to higher or lower costs than those estimated in this Handbook.
Due to the variability in medical costs geographically and over time, there
is considerable uncertainty associated with estimating direct medical costs.
An estimate of costs is often useful, however, when considering planning,
decision-making, and regulatory development. As such, it can provide an
efficient lower-bound estimate of the benefits of avoiding an illness.
1 The chapters were developed over many years and the costs are presented in the current dollar
value for the year the chapters were written. Inflation factors based on the Consumer Price Index can be
accessed by clicking on the lefthand sidebar and used to inflate the costs to any year up to the present.
Link to inflation factors: Appendix A
Chapter 1.1 1.1-2 Introduction to the Cost of Illness Handbook
-------
1.1.A Overview
Improvements in human health in the form of avoiding adverse health
effects frequently constitute a major portion of the benefits resulting from
environmental regulations. There are a variety of approaches to estimating
the value of these benefits, but one of the simplest and more
straightforward approaches is to calculate the medical and related costs
avoided because of the health improvements.
The purpose of this handbook is to present information on the direct
medical costs resulting from illnesses that are associated with exposure to
environmental agents.2 These cost data can be used for policy and
regulatory development and evaluation, benefits assessments (e.g., RIAs),
and other applications where there is interest in either medical costs
avoided due to pollution prevention or costs incurred due to a lack of
pollution control. Direct medical costs represent only a portion of the total
benefits associated with pollution prevention/reduction, but in many cases
these lower-bound estimates may be sufficient for decision-making
purposes.
This handbook has been developed over many years, beginning in 1991.
The level of sophistication and complexity in the field of health economics
has evolved and the approaches taken in the chapters has likewise evolved.
In addition, the needs of the Agency and the requirements of benefit cost
analysis have changed with the advancing field of economics. To address
specific Agency needs, the chapters have been tailored to address program
requirements. For example, some chapters (e.g., stomach cancer) provide
direct medical cost information for survivors and nonsurvivors separately,
while other chapters dealing with other cancers contain medical costs
averaged over all patients with the disease. In each case, the approach was
designed to meet the specific requirements of the Office within the Agency
for whom the cost estimates were prepared. Consequently, although all
chapters contain information on direct medical costs, the approaches and
level of detail vary, depending on when the chapter was written and the
specific requirements that the analyses were designed to address.
Many offices within EPA have funded the analyses of medical costs
discussed in this handbook. These include Office of Pollution Prevention
and Toxics (the primary funding office), the Indoor Environment Division
within the Office of Air and Radiation, the Office of Policy Planning and
Evaluation, and the Office of Water.
2 For simplicity and brevity's sake, illnesses in the Handbook refer to diseases, birth defects, and
other acute and chronic conditions requiring medical attention.
Chapter 1.1 1.1-3 Introduction to the Cost of Illness Handbook
-------
I.1.B Willingness to Pay and Cost Components
When calculating the value of human health benefits, the ideal approach
would be to estimate the value of these improvements in health to everyone
affected by an illness (e.g., the patient, family, friends, community).
Economists measure this value in terms of how much they are willing to
pay for it. Obtaining the detailed information necessary to comprehensively
estimate willingness to pay (WTP), however, is complex and expensive. In
addition, some components of WTP, such as the value of avoiding pain and
suffering, are very difficult to estimate with accuracy. As an alternative to
estimating WTP, the direct medical costs of treating diseases provides a
lower-bound estimate of the benefits of reducing exposure to harmful
pollutants.
I.1.B.1. Definition of Willingness to Pay
WTP is a measure of value based on the premise, central to economic
theory, that the value of a good is simply what it is worth to those who
consume it or benefit from it. The amount an individual is willing to pay
for a particular good may be higher, or lower, than the cost of that good.
WTP for a good will vary from one individual to another and may decline
with how much of the good an individual already has. In the case of
market goods, the comparison between the price of the good and the
individual's WTP for it determines whether or not he or she buys the good.
If the price is lower than his WTP, he will buy the good at less than he
would have been willing to pay for it, receiving what economists call
"consumer surplus." If the price exactly equals his WTP, then the
individual will be equally happy whether he keeps the money and forgoes
the good or pays the money and gets the good. If the price exceeds the
individual's WTP, he will not buy the good.
In the context of environmental regulations and policy, economists define
the value of a reduction in health risks as the sum of all individuals' WTP
for it. Most people would be willing to pay something for a reduction in
risk to themselves, but many people would also be willing to pay for a
reduction in risk to others. Most parents, for example, would probably be
willing to pay for a reduction in the risk of their children incurring a serious
illness. These altruistic components of WTP may be insignificant in many
cases, but they may be substantial in the case of serious diseases or
disabilities. The total value of a risk reduction, then, is the sum of all
WTPs for it.
Environmental contaminants generally cannot be linked with certainty to
specific health effects experienced by specific individuals. Instead, the
contaminants increase the likelihood, for all exposed individuals, of
contracting specific diseases. Rather than summing the WTPs for a given
Chapter 1.1 1.1-4 Introduction to the Cost of Illness Handbook
-------
risk reduction over all those who enjoy the risk reduction, however, it is
often easier to think in terms of the value of an adverse health effect
avoided. Some people who would have contracted the illness will now
avoid contracting it. The total value of an avoided illness is what the
otherwise-afflicted individual would be willing to pay to avoid it plus what
others would be willing to pay for him or her to avoid it. The sum of these
WTPs is the total value of the avoided case of illness, referred to here as
total WTP. In practice, average WTPs are used to value adverse health
effects avoided because WTP will vary from one individual to another.
The crosswalk between valuing risk reductions and valuing a case avoided
can be made by valuing a statistical case avoided. For example, suppose
that a regulation is passed that reduces the risk of contracting pneumonia
by a factor of 0.001. That means that one fewer individual out of every
1,000 people whom the regulation affects would be expected to contract
pneumonia. Suppose each person has some positive WTP for this risk
reduction of 0.001, and that the average WTP is $5. The total willingness
to pay to avoid the one case of pneumonia that would otherwise be
expected to occur per 1,000 people is $5,000 ($5 per person x 1,000
people). That is, the value of a statistical case of pneumonia avoided
would be $5,000. Regulations typically affect cities of substantially greater
size than 1,000 people, however, so there are typically many cases avoided.
For example, if a regulation reduced the risk of contracting pneumonia by
0.001 in a city with 3 million people, there would be 3,000 (0.001 x
3,000,000) fewer cases of pneumonia expected to occur in the city as a
result of the regulation. If the value of a statistical case of pneumonia
avoided in that city is $5,000, the pneumonia-related benefit of the
regulation in that city would be 3,000 statistical cases avoided x $5,000 per
statistical case avoided = $15 million.
In theory, nonmarket goods should be valued in exactly the same way as
market goods — in terms of what people would be willing to pay for them,
(i.e., their willingness to pay). Unlike most market goods, many nonmarket
goods are public goods, from which many people benefit simultaneously.
A reduction in the risk of an adverse health effect is such a public good,
because all the exposed individuals will experience a decrease in the
likelihood of contracting the disease.
I.1.B.2 Components of Willingness to Pay
WTP to avoid an illness contains several components. Illness imposes both
direct and indirect costs that would not be borne if the illness was avoided.
Direct costs result from the increased resource utilization caused by the
illness, and may be medical or non-medical. For example, the cost of an
ambulance used to transport a person to the hospital is a direct medical
cost, while child care and housekeeping expenses required due to illness are
non-medical direct costs. In addition to the direct costs, there are
Chapter 1.1 1.1-5 Introduction to the Cost of Illness Handbook
-------
opportunity costs (the value of productive and leisure time lost) to the
patient and possibly to others.3 Finally, illness causes anxiety, pain, and
suffering, the cost of which, although difficult to measure, is very real and
may be very large. Most people would be willing to pay something to
avoid the pain and suffering that comes with illness, as well as to see loved
ones avoid pain and suffering. There is also a perceived value to most
individuals of maintaining public health (i.e., most people would place
some value on reducing the number of children with asthma, the number of
people with cancer, the incidence of birth defects, and the occurrence of
most illnesses).
Finally, in some cases people may take precautionary actions to avoid
contracting environmentally-related illnesses. People may buy bottled
water, for example, if their water supply is contaminated or if they believe
it may be. In these cases, there are not only costs associated with the
occurrence of the illness, but costs incurred in efforts to prevent the illness.
These costs would be avoided or reduced if the risk of the illness were
reduced. The components of total WTP are shown in Figure 1.1-1.
I.1.B.3 Approaches to Measuring Willingness to Pay
The challenge confronting the analyst is to measure the total value
associated with avoiding an illness. This is a difficult challenge due to the
variability in human perceptions, responses, and the complexities involved
in measuring individual attitudes and extrapolating to a larger group.
Economists have developed several ways to measure the value of morbidity
avoided; each method has advantages and disadvantages.
3 Opportunity cost is the cost associated with forgone opportunities. Time spent in the hospital,
for example, is time that would otherwise have been spent in productive and/or leisure activities. The
opportunity cost of a hospital stay is the value of the productive and/or leisure time lost during the hospital
stay.
Chapter 1.1 1.1-6 Introduction to the Cost of Illness Handbook
-------
Figure 1.1-1. Elements of Willingness to Pay
Total Willingness to Pay (WTP)a
WTP to Avoid
Pain and Suffering13
Value of
Lost Productive
Timec
family and
community
patient
Value of
Lost Leisure
Time
patient
caregiver
(unpaid) "
for
Medical Services
for
Special Servicesd
paid directly by
patient
paid indirectly by
patient and rest of
public through
insurance
premiums, taxes,
etc. (see text for
additional detail)
paid by hospital or
MD
a. See text for a discussion of cost elements. The cost components above are associated with contracting a disease. People who avoid disease by employing averting
behavior may incur other costs (e.g., the cost of buying bottled water). Both the cost components listed above, and those associated with risk avoidance would be
reduced or eliminated if the risks were reduced or eliminated.
b. Heightened morbidity or other adverse effects associated with a lack of treatment (e.g., due to insufficient resources) may increase pain and suffering. This indirect
cost category is very difficult to measure.
c. Lost time includes a partial or complete loss of the ability to carry out activities (paid or unpaid).
d. Includes special education (children); worker retraining (adults); workers' disability; and/or specialized equipment, transportation, and other services required due to
the illness.
Chapter 1.1
1.1-7
Introduction to the Cost of Illness Handbook
-------
I.1.B.3.1 Averting Behavior.
One approach to valuing WTP relies on the averting behavior of people.
This "averting behavior" approach provides estimates of WTP based on
actual behavior in markets. The major drawbacks of this method, however,
are that (1) it is limited to situations in which averting behavior is possible
(i.e., not all contaminants can be avoided), and (2) it is difficult to isolate
WTP for improved health from WTP for other aspects of the averting
behavior. For example, while use of an air conditioner may reduce
exposure to ambient air pollutants, it also cools the house.
Evaluation of averting behavior may be complex because pollution
avoidance costs are situation dependent. Using the air conditioning
example, community factors that influence air conditioner use include the
extent of public notification about pollution problems, ambient
temperatures, etc. Individual decision factors include a subjective rating of
the pollutant's health risks.
I.1.B.3.2 Contingent Valuation
A second method, contingent valuation, is to simply ask people how much
they are willing to pay for a good or service that is not traded on the
market. The valuation is contingent upon establishing the market. This
method involves designing surveys that present people with hypothetical
situations in which they are queried about how much they would be willing
to pay for a specified nonmarket good (such as to avoid a case of
pneumonia). The advantage of the contingent valuation method is that it
attempts to estimate the right thing — individuals' WTP. In addition, in
contrast to the averting behavior approach, it can be applied to any risk
reduction or adverse health effect. It is a controversial method because it
of necessity elicits responses to a hypothetical situation. The reliability of
the estimates obtained through contingent valuation methods is questioned
by many economists and by others. The approach is resource intensive and
costly. It requires careful questionnaire design and interpretation of
responses. In spite of its drawbacks, it may be a useful tool for obtaining
valuation data, and has the potential for contributing to a variety of
planning, evaluation, and regulatory activities.
I.1.B.4 Conclusions
The cost of illness method used in this handbook estimates the direct
medical costs associated with an illness. This method has several
advantages. It is straightforward to implement and easy to understand. In
addition, it is likely to result in relatively accurate estimates of the
components of total WTP that it attempts to measure, the medical cost
component. The major drawback of the cost of illness method is that it
omits several components of total WTP, most notably the WTP of the
patient and of others to avoid the anxiety, pain, and suffering associated
with the illness. These components may be substantial, especially for
Chapter 1.1 1.1-8 Introduction to the Cost of Illness Handbook
-------
serious illnesses. In addition, this handbook does not include direct
non-medical costs, the opportunity costs of family members or other
unpaid caregivers, or time lost for the patient. Consequently, the values
reported here are only a partial estimation of the cost of illness, which, as
described above, is itself an underestimate of the total economic costs
associated with the diseases considered. Because it omits these
components, the cost of illness method provides an underestimate of WTP
to avoid the disease.4
I.1.C Organization of Handbook.
This handbook is organized into sections based on common features of
illnesses and the type of illnesses discussed, such as cancer, developmental
illnesses and disabilities, diseases of specific organ systems (i.e.,
respiratory), and acute illnesses. They were organized in this manner
because the diseases contained in these categories are similar in important
aspects, including: cost calculation methods; biomedical data on disease
definition, causality, susceptible subgroups, and treatment; survival
patterns; and the types of medical services required and their costs. For
example, cancers frequently require similar types of medical intervention,
share similar characteristics regarding survival data, and have many
causative agents in common. Developmental effects also share many
characteristics; they manifest early in childhood, involve protracted
treatment, occur in clusters, require both medical and other professional
intervention, and may have similar causative agents. For both cancer and
developmental effects, toxicological data are often not organ-specific, and
providing general information regarding chemical associations in an
introductory chapter was most appropriate. Most sections begin with a
chapter that discusses the common characteristics of a disease group with
respect to background medical data, cost, and causality.
4 Some researchers (Crocker and Agee, 1995) have suggested that the cost of illness approach
may not always underestimate costs (e.g., when treatments are painful and consequences are limited,
patients or their caregivers may have ambivalent attitudes). If this is true, however, it is likely to be
expected to be the exception rather than the norm under circumstances of fully informed medical
information, because most individuals place a higher value on regaining their health (in the case of
non-terminal diseases) than on avoiding medical procedures.
Chapter 1.1 1.1-9 Introduction to the Cost of Illness Handbook
-------
I.1.C.1 Illnesses Covered in the Handbook
Medical costs are provided for the following illnesses:
Cancers^
breast cancer
kidney cancer
lung cancer
skin cancer
stomach cancer
colorectal cancer
bladder cancer
Developmental Illnesses and Disabilities
low birth weight
cleft lip and palette
limb reductions
cardiac abnormalities
spina bifida
cerebral palsy
Down syndrome
high blood lead levels
Respiratory Diseases
asthma
acute respiratory illnesses
Symptoms
Some of these chapters are currently in development or undergoing
revisions (e.g., skin cancer).
I.1.C.2 Chapter Format
Each chapter covering a specific illness follows the same general format;
the level of detail provided depends on the availability of information and
the goals of the analysis. First, the chapter provides the reader with
background information on the disease (Section A), including a definition
and description of the disease, adverse effects related to the disease,
associations with environmental pollutants, common medical approaches,
and likely disease outcomes (prognosis).
The second portion of each chapter (Section B) provides specific cost
information, including the methodology used to estimate costs, sources of
the data, and cost estimates. Costs are provided that were current in the
year in which the chapter was written or revised (1996 and forward). The
5Bone and liver cancer costs are also briefly discussed in the introductory cancer chapter (Chapter
II.l).
Link to Chapter II. 1
Chapter 1.1 1.1 -10 Introduction to the Cost of Illness Handbook
-------
Link to Appendix A
costs can be updated to the current year using the Consumer Price Index
(CPI) Medical Services inflation data provided in "Appendix A: Inflation
and Discounting Factors" on the sidebar at left.
Section B of each chapter also discusses results and limitations of the
methods used. In some cases an uncertainty and/or sensitivity analyses is
also provided. Studies that provide alternative cost estimates are presented
and discussed, when available. When more than one set of results is
discussed, recommendations for data use are given in a section titled
"Conclusions."
The headings that appear in the format of each disease chapter are listed
below and described in the text that follows:
A Background
A. 1 Description
A.2 Concurrent Effects
A.3 Causality & Special Susceptibilities
A.4 Treatments and Services
A. 5 Prognosis
B Costs of Treatment and Services
B.I Methodology
B.2 Results
B.3 Limitations
B.4 Other Studies
B.5 Conclusions
The chapters introducing each part of the handbook, such as cancer and
developmental effects, do not follow this format because they do not deal
with a specific disease.
I.1.C.3 Section Contents
The contents of each section are as follows:
A. Background:
A.I Description: provides a clear definition of the disease and what
subcategories of an illness are omitted. This section may also include data
on occurrence for some diseases, depending on the needs of the sponsoring
office.
A.2 Concurrent Effects: often there are other diseases associated with a
given disease. These may be attributable to the same causes (e.g.,
Chapter 1.1 1.1 -11 Introduction to the Cost of Illness Handbook
-------
environmental pollution). If the concurrent effects have been reported in
the reviewed medical literature, then they are listed in this section.
Treatment often incurs additional risk; radiation treatment, anti-cancer
drugs, and other therapies can cause serious illness while curing the target
disease. These secondary illnesses usually constitute a separate disease,
however, so their costs are not provided in the same chapter. In some
cases, the costs of these diseases are discussed in other chapters within the
handbook. Concurrent effects are listed even when cost data are not
provided, so that the analyst using the data can report the underestimate
and uncertainty associated with additional anticipated illness.
A.3 Causality and Special Susceptibilities: information on the
associations between environmental agents and diseases is presented in this
category. Factors that may increase susceptibility are also discussed; these
include many pre-existing conditions. Data are limited, however, on
special susceptibilities. A comprehensive evaluation of illnesses that would
increase susceptibility or severity of a disease was beyond the scope of this
analysis. In general, a pre-existing disease in the target organ causes
additional medical complications and higher costs. For example, coronary
artery disease in someone with pre-existing heart disease is likely to be
much more serious and costly.
1.4 Treatment and Services: includes a brief description of common
treatments and related services. In most cases, this description does not
include support services, such as specialized occupational training required
for rehabilitation. This information was available for some the chapters
that cover childhood diseases and disabilities, and so is included as
supplemental cost data.
1.5 Prognosis: contains quantitative or qualitative information on likely
disease outcomes. This information is important because survival
probabilities, and the duration between diagnosis and death among those
who do not survive, are used in cost of illness evaluations and have an
impact on the cost of a disease.
B. Costs of Treatment and Services: contains medical cost data and
methods used to estimate costs.
B.I Methodology: describes methods used in calculating the costs of
medical treatments and services, and the basic information used to calculate
costs.
B.2 Results: summarizes cost estimates. Discounted costs at zero, three,
five, and seven percent are provided for most illnesses.
B.3 Limitations: describes shortcomings of the methodology and results.
These typically include a discussion of factors such as the age of the data,
Chapter 1.1 1.1 -12 Introduction to the Cost of Illness Handbook
-------
sources of information, assumptions regarding treatment, and other
factors that may affect the applicability or accuracy of cost estimates.
B.4 Other Studies: when other study results are available, they are
presented and the advantages and disadvantages of the various studies are
discussed. In all cases, the study results recommended for use are
presented first and described in detail in the "Methodology" and "Results"
sections. Those studies with results of limited use are presented later in
this section.
B.5 Conclusions: when more than one set of results is discussed in the
medical cost and/or time sections, the final recommendations regarding the
optimal results are given in this section. The section is not included in
chapters having only one set of study results.
I.1.C.4 Selection of Illnesses
EPA selected diseases for inclusion in this handbook based on 1) the
known need for disease cost estimates for regulatory or policy activities, or
2) the anticipated need based on a review of the environmental health
literature and Agency activities.
Regulations and policy evaluations often require cost and benefit
information for specific illnesses anticipated to be affected by a rule or
policy. Because reductions in exposure to pollutants will result in
improvements in human health, evaluations of rules or policy changes may
incorporate consideration of the impacts on human health, including the
benefits of illness avoidance. For example, the medical cost estimates for
stomach, bladder, and lung cancer provided in this handbook were
developed in 1998, in response to a need for economic data for Office of
Water rules covering radon and arsenic.
Illnesses were also selected based on their likely occurrence as a result of
exposure to environmental pollution and anticipated future needs of the
Agency for cost data. Many illnesses that are frequently associated with
environmental pollutants, or that are clearly linked to pollution episodes
(e.g., asthma) were evaluated and included in the handbook. For these
health-based selections, the environmental health literature (i.e., toxicology
and epidemiology) was consulted. The illnesses in this handbook have
been linked to exposure to environmental hazards including both chemical
(e.g., PCBs) and physical (e.g., radiation) hazards.
The health-based selection process included the following considerations:
• a link to environmental exposures in the toxicological and/or
epidemiological literature;6
6Toxicological and epidemiological studies may strongly suggest, but rarely provide, unequivocal
evidence for links between exposure and effects. Credible studies in the peer-reviewed literature that
demonstrated statistically significant associations between exposure and effects were considered adequate
Chapter 1.1 1.1 -13 Introduction to the Cost of Illness Handbook
-------
• illnesses that could reasonably be linked to exposure levels likely to
occur in the environment;
the availability of costing data for an illness (either in the form of
multiple mergeable databases, or in the economic literature);
• occurrence linked to multiple chemicals, ubiquitous chemicals, or
those of particular interest to EPA based on policy considerations
(e.g., lead, mercury);
• an illness that does not result in death shortly after onset.
Professional journals were consulted for information on health effects. In
addition, some federal sources, such as the Hazardous Substances Data
Base (HSDB), the Integrated Risk Information System (IRIS), the Health
Effects Assessment Summary Tables (HEAST), and the Agency for Toxic
Substances and Disease Registry's (ATSDR) Toxicological Profiles, were
used as the sources of data linking illnesses and exposure to environmental
pollutants.7 In selecting illnesses, a convergence was sought between the
likelihood of an illness based on environmental exposure levels and the
availability of good quality aggregate cost data from databases and
research papers. For high priority illnesses, it was sometimes necessary to
construct cost estimates using a theoretical approach when aggregate data
were not available.
I.1.C.5 Linking Diseases to Agents
Matrices with preliminary information on environmental agents associated
with cancers and with birth defects are provided in Chapters III and III.l,
respectively. These contain lists of hundreds of agents that have been
associated with the disease categories in either toxicological or
epidemiological data, or both.
Some of the agents were identified with a single disease during research on
a single disease or disease category (e.g., birth defects) and have not been
reviewed regarding the induction of other diseases in the Handbook. In
some cases, the diseases were studied as a result of their link to an agent.
For example, the costs associated with stomach cancer were evaluated for
evidence for inclusion in this handbook. The associations listed in this handbook are presented at the
screening level; risk assessments and health evaluations require an in-depth review of the literature on each
chemical.
7 HSDB is an on-line toxicological database maintained by EPA and available through the National
Library of Medicine's TOXNET. IRIS and HEAST, developed by EPA, contain data regarding
carcinogenic and non-carcinogenic effects of chemicals.
Chapter 1.1 1.1-14 Introduction to the Cost of Illness Handbook
-------
a proposed radon rule. Consequently, when the stomach cancer chapter
was written, an extensive search for other stomach carcinogens was not
conducted.
In addition to those chemicals listed because the data were obtained for a
specific chapter, data on chemicals associated with health risks were
collected specifically for the matrix from general toxicological sources.
Most linkages between chemicals listed and categories of effects (i.e.,
cancer and birth defects) did not involve an extensive search of the
literature, and are often based on a quick review of relevant databases. The
reader should assume that the matrices do NOT list all adverse effects of an
agent, or all agents that can cause a disease. This limitation is due to the
very sizable scope of the work that would be required to fully evaluate all
potential effects of the hundreds of chemicals listed in the matrices. In the
future additional data may be added to the matrices.
Links to Chapters II. 1 and III. 1
I.1.D Methods used to Estimate Direct Medical Costs
As noted above, total costs associated with an illness incorporate many
elements. This handbook focuses on direct medical costs. The direct
medical costs of illness are calculated for the life cycle of each illness, (i.e.,
from diagnosis to cure or patient death). The goal of this handbook is to
provide cost estimates that are generalizable to any area of the United
States; therefore, cost data representative of the nation as a whole were
sought. Standard disease treatment methods, using generally acceptable
practices, were also considered appropriate. Finally, the average patient
experience regarding prognosis and survival was used in the cost estimate.
This approach is expected to yield representative cost estimates that are
generally applicable. They may be modified by changes in technology or
cost structure.
I.1.D.1 Overview of Method
Six basic steps are necessary to calculate direct medical costs:
1. Identify a cohort who has received the standard treatment for the
disease. If costs are to be determined by treatment component, list
the standard treatment elements.
2. Determine the costs of each phase of treatment or for each
treatment component and the timing of these costs.
3. Combine the cost estimates with probability data regarding the
likelihood of receiving specific treatments and their timing.
Chapter 1.1 1.1-15 Introduction to the Cost of Illness Handbook
-------
Incorporate survival data in probability estimates based on the age
of onset of the disease and life expectancy.
4. If total medical costs are used (rather than disease-specific cost
elements), determine the background medical costs that would be
incurred in the absence of the disease.8 Modify the disease-related
costs as needed to obtain incremental costs.
5. Discount the stream of treatment costs over time to estimate
present value treatment costs. All costs in the handbook are
adjusted to 1996 dollars using the medical care cost component of
the Consumer Price Index.
6. Aggregate the discounted stream to obtain an estimate of the total
medical costs of the disease.
Ideally, this process is carried out for both survivors and nonsurvivors of
the disease (these two subgroups are discussed below). These basic cost
estimation activities are carried out for all illnesses. As discussed above,
results reported in the literature are used for many of the diseases. In a few
cases, these steps were carried out specifically for the development of
chapters in this handbook.
The data obtained in the various steps above are used in a single
aggregating equation described by Hartunian et al. (1981) to calculate the
expected direct present value costs (PVC) for any individual of a given sex,
impairment category, and age at onset of impairment:
p;s (n) DC
M v 7
(1 + ry-'
where:
n = the various ages of the individual,
/ = the age at impairment onset,
P'iiS(n) = the probability that a person of sex s who acquires
condition /' at age / will survive to age n,
DC'ls(n-l+l) = the dollar value of the average annual incremental direct
costs generated by such persons during year n-l+\
following impairment onset, and
r = the discount rate.
8 Background costs are the average costs incurred by a population matched (e.g., in age, sex, etc.)
for those with the disease, and include care for both healthy and diseased people in the population.
Chapter 1.1 1.1 -16 Introduction to the Cost of Illness Handbook
-------
In reality, data rarely exist regarding the probability of survival and direct
costs for a specific disease for each age of diagnosis and sex. If there were
such data, however, the estimated average direct costs would be calculated
by weighting the direct costs for each age and sex by the percentage of
incidence in each sex/age grouping.
Appendix I.I-A, "Equations Describing the Expected Present Discounted
Value of the Per Capita Lifetime Stream of Costs Associated with a Given
Illness," contains a listing of equations and their input parameters used to
estimate the expected present discounted value of the per capita lifetime
stream of medical costs. The appendix includes more detail than is
provided in this section.
Link to Appendix 1.1-A.
I.1.D.2 Survivors and Nonsurvivors
Medical costs associated with a disease differ for those who survive a
disease and those who die of it. For purposes of the cost analysis in this
handbook, survivors are defined as those people who are diagnosed with a
disease, but do not die of it (although they may die of another cause).
Nonsurvivors are those who die of the disease at any point after diagnosis.
Separate cost estimates for survivors and nonsurvivors are provided in the
handbook for a few types of cancer, due to a specific Agency program
need. These estimates are listed separately in cases where it is important to
distinguish between the differing costs for the two groups, and in cases
where the value of a statistical life (VSL) is used for nonsurvivors. When
the VSL is used for nonsurvivors, their medical costs have already been
incorporated into the cost estimate. Under these circumstances, it would
be appropriate to use the medical and time (and any other costs) for
survivors, but not for nonsurvivors. The use of cost data depends on the
composite of all cost calculations that are being carried out for a benefits
assessment.
A patient's probability of receiving specific treatments is modified by the
likelihood that he or she will survive to receive that treatment. Survivors
incur initial treatment costs but not charges for services, such as terminal
care associated with the disease. They may die of other causes during the
treatment period, and their probability of receiving treatment is modified by
the probability that they will die of another cause. This probability is
determined from the background mortality rates for the U.S. population,
as reported by the Department of Health and Human Services publication
series, Vital Statistics in the United States (DHHS, 1994). The probability
of death due to the disease is determined from the medical literature (e.g.,
from the National Cancer Institute for cancers) for nonsurvivors.
Determining survivorship for some illnesses, such as asthma, is very
difficult due to rapid advances in treatments and the confounding effects of
Chapter 1.1 1.1 -17 Introduction to the Cost of Illness Handbook
-------
limited access to care among some socioeconomic groups. When survival
data are provided in a chapter, the source and confounding effects are
discussed.
The methods used to calculate medical costs for survivors and
nonsurvivors are described briefly below. Numerous additional
calculations are often required to determine survival rates, life expectancy,
etc., and are discussed in chapters where required.
I.1.D.2.1 Survivors (those who do not die of the disease)
The average cost among survivors of a disease for each year post-diagnosis
may be expressed as:
Average nth Year Costs = medical costs for nth year treatment x
probability of survival through the «th year + medical costs for «th
year of treatment 12 x probability of mortality in the nth year
As this description indicates, the costs decline as the population decreases
due to mortality. Costs are discounted back to the year of diagnosis.
Costs for those who die are calculated for one half year because they are
assumed to die at the midpoint of the year. Their survival rate is equal to
the background mortality rate at each age in the U.S. population as
reported in vital statistics reports (e.g., Vital Statistics of the United States,
DHHS, 1994).
The yearly costs are aggregated over the remaining life of the "average"
patient. For example, if the average age of a patient is 40 and medical
visits and drug treatment for asthma are anticipated to be required over the
expected lifetime for a 40-year-old in the general population, the costs for
each year are calculated, discounted from the time of diagnosis, and
summed over a lifetime. This generates a lifetime stream of costs.
Average Lifetime cost = Average 1st year cost + the sum of the
(discounted) average subsequent year costs
If there is a point at which treatment ceases (e.g., ten years after diagnosis
and treatment for cancer), the costs will be aggregated over time up to that
point. When this approach is compared to the equation supplied by
Hartunian et al. (1981) shown in section I.I.D.I, it can be seen that the
overall cost estimation method is the same.
Link to Section I. l.D. 1
I.1.D.2.2 Nonsurvivors (those who die of the disease)
The method for calculating medical costs for nonsurvivors is similar to that
shown above for survivors, but the costs themselves are often different, and
the probabilities used to calculate costs differ. For nonsurvivors, the
Chapter 1.1 1.1 -18 Introduction to the Cost of Illness Handbook
-------
probability of treatment is contingent on surviving the study disease for
each year after disease onset. The general descriptions are the same as
those shown above.
/. 1.D.2.3 Average Costs for Survivors and Nonsurvivors
The average lifetime medical costs of a disease are estimated for survivors
and for nonsurvivors separately. The average cost for both together is
calculated as a weighted average of costs for the two groups, using the
proportion of patients in each group. This can be expressed as:
Average Lifetime medical costs = costs for survivors x proportion of
survivors + lifetime costs for nonsurvivors x proportion of
nonsurvivors
I.1.D.3 Data Sources
Link to Chapter II. 2
Most cost estimates provided in this Handbook rely on an evaluation of
databases of actual costs incurred, either for this Handbook or by previous
researchers. Each chapter describes how various data sources were used
to calculate the final results, and contains a discussion of uncertainties
associated with data sources.
Well-designed studies in the literature that supplied recent cost estimates
were located for most illnesses and were preferred over data collection and
evaluation for the sake of efficiency. For example, many of the childhood
disease chapters in this handbook are based on work done by Waitzman et
al. This research group used 12 databases to accurately construct their
cost estimates.
Using the results of these studies is more efficient than constructing costs
directly from multiple databases. Extensive resources are required to
evaluate national databases, and confidentiality is often an issue that
requires a lengthy timetable for clearances. Most chapters use a
combination of data obtained from the literature and directly from data
sources. Results reported in the literature are supplemented by survival or
other essential data to obtain cost estimates.9
9 For example, Baker et al. (1989 and 1991) were used as a source of basic cost information for
many of the cancer chapters. Additional data required for the cost analyses regarding survival and
mortality estimates and cancer rates were obtained directly from the National Cancer Institute's databases.
These combined sources were used to calculate cost estimates.
Chapter 1.1 1.1 -19 Introduction to the Cost of Illness Handbook
-------
Cost estimates were developed for some diseases by constructing a typical
treatment course through physician panels, and evaluating the cost of each
treatment component using sources such as the Medicare database. This
approach relies on expert judgment by physicians who determine
treatments that the average patient receives, the timing of treatments, and
the likelihood that a percentage of patients will receive a particular
treatment. Survival data are obtained from a secondary source.
Limitations to this approach include physician errors in recall, physician
experience with a non-representative population, and incomplete
knowledge of variations in treatment patterns geographically and among
physician specialties. In addition, the approach tends to estimate ideal
costs because physician panels describe the treatment and costs that a
person with a disease should receive, whereas the direct cost estimation
approach described above is based upon actual costs. For example, skin
cancer costs were evaluated using physician recommendations. The
protocol described is one that provides the most appropriate care; it is not
necessarily the most expensive, but assumes that everyone receives the
services that would address their medical needs adequately. In practice this
does not happen, due to a variety of factors (e.g., limited access or funds,
avoidance of diagnosis).
Consequently, the chapters based on physician recommendations for
treatment describe costs associated with sound medical practice, and these
values may be higher than the "average" for the U.S. Adequate care for all
patients probably reduces time loss, however, so that what is theoretically
spent on complete medical care may be balanced by related reductions in
unnecessary morbidity and mortality. When these types of analyses have
been done, they generally find that preventive and/or adequate medical care
costs are more than offset by reductions in future morbidity, time loss, and
special services required (e.g., special education).
Due to the limitations of the approach and the resources required to carry
out an analysis based on this type of cost evaluation, it was used for very
few chapters in this handbook.
1.1.E Susceptible Subpopulations
Many diseases affect subgroups of the population disproportionately. The
subgroups may be defined by age, gender, racial, ethnic, socioeconomic, or
other differences within the U.S. population. For example, asthma is most
often reported and treated in children and the elderly. Most cancers occur
with increasing frequency in older populations (some leukemias being
notable exceptions). Very few diseases affect all population groups (ages,
sexes, races) equally. For purposes of evaluating costs and potential
benefits to different segments of the population, it is useful to evaluate
Chapter 1.1 1.1-20 Introduction to the Cost of Illness Handbook
-------
whether there are susceptible subpopulations that require consideration.
Their benefits may be considerably higher than those of the average
member of the general population.
Each chapter contains a section titled "Causality and Special
Susceptibilities" that contains information gathered to address this issue.
An exhaustive search was not carried out for these data, however, and new
information is being generated rapidly in this field. Consequently, there
may be susceptible subgroups not identified in this section. In addition,
there are usually pre-existing medical conditions that will increase
susceptibility to most diseases (as noted previously), such as a pre-existing
disease in the same organ.
Special susceptibilities are often indicated by higher-than-average rates of
the disease of interest. Increases in the rates of reported diseases may be
due to a variety of factors. Some of these indicate an increased
susceptibility; others are matters of personal choice and may not be
considered relevant to cost calculations. One way to approach this issue is
to evaluate increased susceptibility when it is based on an increased risk of
disease due to factors reasonably beyond the control of the subpopulation.
Factors that are usually beyond the control of the individual that may cause
increased susceptibility include:
• constitutional limitations (e.g., illnesses, genetic abnormalities, birth
defects such as enzyme deficiencies);
concurrent synergistic exposures that cannot reasonably be
controlled (e.g., at home or in the workplace);
• normal constitutional differences (i.e., differences based on sex,
age, race, ethnicity, etc).
Other factors that are not usually considered beyond the individual's
control include personal choices, such as smoking, drinking, and drug use.
These factors may be included in an analysis depending on the goals of the
analysis. Which types of factors should be included in an analysis is a
policy decision. Personal choice typically does not include the place of
residence or work, since these are not reasonably changed by many people.
For example, asthmatic smokers who increase their risk of asthma are not
discussed at length as susceptible subpopulations; however, asthmatic
residents of an area with high levels of acid aerosols may require additional
analysis of their risk of asthma and benefits of asthma avoidance. It may be
useful to evaluate the medical costs of people in the latter group using
different underlying risk factors than would be appropriate for the overall
U.S. population.
Chapter 1.1 1.1-21 Introduction to the Cost of Illness Handbook
-------
Link to Section I. l.B
Link to Chapter II. 2
These types of considerations are not used directly in the cost calculations
presented in this handbook. Much of the information regarding special
susceptibilities is incorporated, however, into the medical information
provided in the background sections of each chapter. In addition,
sensitivity analyses in some chapters include an analysis of the impact on
subgroups at highest risk.10 The data may be used in a variety of ways,
depending on the nature of the benefits assessment.
The degree to which special susceptibilities should be considered in an
assessment depends on the extent of impact that is expected in the cost
analysis (e.g., whether the differences will be substantial or minimal), as
well as equity considerations. For example, the rate of stomach cancer in
African-Americans is much greater than in non-African-Americans. There
is no conclusive information regarding the cause of this increase, so it is
assumed that it may be due to a genetically determined increase in
susceptibility to stomach cancer. Because these differences are consistent
across the ages, there is no modification required in the per capita cost
estimates for stomach cancer. If an area that was predominantly African-
American was the subject of a benefits assessment, however, the increased
risk in that area would merit consideration in the benefits assessment. If
risk factors for the general population were used with the per capita costs,
then the impact on the area would be estimated incorrectly. Most chapters
contain some information regarding known increases in susceptibility.
Some have considerable detail, such as the chapter on stomach cancer
(Chapter II.2) where basic information on differences in stomach cancer
rates based on race and sex from the National Cancer Institute are
provided. These data could be used by risk assessors or epidemiologists to
evaluate the potential for increased risk.11 Their results, together with the
economic data, could then be used in a benefits assessment. This same
approach may be used for any diseases for which a susceptible subgroup of
interest has been identified.
10 This type of analysis was begun in 1998 in response to specific needs regarding high-risk
subgroups. It is not contained in earlier chapters, but may be added in the future.
11 This type of analysis is complex because it requires an evaluation of the proportion of cases
expected in the population subgroup as well as in the overall population. This requires calculation of the
conditional probabilities of stomach cancer based on race. Because the overall stomach cancer rates
incorporate the rates for both blacks and non-blacks, the likelihood of occurrence in each of these two
groups would need to be determined. This information would be used with the racial distribution in the
target population to determine the estimated potential risk.
Chapter 1.1 1.1-22 Introduction to the Cost of Illness Handbook
-------
I.1.F Limitations
The limitations of the data provided in this handbook are related to two
primary constraints:
1) the scope of the analysis, and
2) uncertainty regarding the data used in the analysis.
The scope of the analysis, described in the introduction to this chapter,
includes only direct medical costs. Many aspects of the willingness to pay
to avoid these illnesses are not covered in the handbook, including direct
non-medical costs and pain and suffering (see discussion in Section I.l.B
and Figure 1.1-1).
Link to Figure 1.1-1
There are numerous sources of uncertainty associated with the values
presented in this handbook. The background and cost sections of the
chapters provide definitions and information on the "average" medical
experience and costs for a disease. Outside this experience, however,
numerous factors impact the costs associated with specific diseases and
lead to uncertainty in the cost estimates. The sources of uncertainty
specific to each disease cost analysis (i.e., the databases used in an analysis,
the methods of the analysis) are discussed in the individual disease
chapters. Those common to all diseases are discussed in this section.
I.1.F.1 Uncertainties Regarding the Market Value of Medical Goods and Services
As noted above, direct medical costs are provided in this handbook as a
component of WTP and may be used as a lower estimate of WTP. For a
variety of reasons, however, the price of medical services and the mix of
services purchased may not accurately reflect the market demand and value
for these services. These reasons are related to the nature of what is
purchased when buying medical services, and the way in which medical
services are paid for in the U.S.
I.1.F.1.1 The Nature of Medical Service Purchases
The nature of what is being purchased is often unclear when obtaining
medical services. Although a specific service or good is usually being
obtained at a point in time (e.g., surgery, a pharmaceutical, an X-ray), the
ultimate goal of the purchase is invariably an improvement in health. The
latter, often purchased through multiple related services and goods, is
generally far more valuable to the individual than the individual service or
item. For example, when faced with a serious illness, individuals would
Chapter 1.1 1.1-23 Introduction to the Cost of Illness Handbook
-------
expend the maximum funds they have available to avoid death or
permanent disability.12
Just as individuals may be willing to bear almost any medical costs required
for serious illnesses, society is often willing to spend very large sums to
avoid illness or death among otherwise healthy individuals. Extensive and
expensive health programs for indigent populations illustrate the interest in
the overall health of the population. This may be both the provision of a
public good and a self-protective strategy. Likewise, societies are often
willing pay very large medical costs under special circumstances. For
example, communities have raised millions of dollars to provide cancer
treatments for a relatively small number of children. What a society or
individual would be willing to pay for medical services may be strongly
affected by a society's response to potentially drastic consequences. The
willingness to pay very large medical costs to avoid dire consequences is
expressed both by the individual with the illness and by those who are
aware of the individual's plight.13
There is an interactive relationship between what medical costs society is
willing to pay and standard medical practice (which dictates costs), which
often reflects societal values. This dynamic may further diminish individual
control over the purchase of medical services and their impact on the
market for medical services. When insurance companies, the government,
or medical practice standards determine access to care or determine what
constitutes appropriate treatment, individual choice may be limited; this
may also affect cost. For example, both the costs of treating a disease such
as breast cancer and access to care for that disease may be quite different
for patients of different ages or with different health status (i.e., AIDS vs
good health).
I.1.F.1.2 The Sources of Payment for Medical Services
A second major source of possible inequality between medical costs and
market values is the system of medical payments in the U.S. The system
creates institutional reasons for a disjunction between costs, payments, and
WTP for both individuals and society. As Figure 1.1-1 shows, there are
often a multitude of sources of payment for medical costs. In the case of
medical services, people have typically prepaid insurance premiums or
taxes (e.g., for Medicare and Medicaid) that are used as the source of
payment. This prepayment may affect the demand for and the costs of
12 This discussion generally addresses people who prefer a cure over death. Although the latter
may occur, it is not the norm and is not directly relevant to determining willingness to pay for services to
improve health.
13 Alternatively, there is often opposition to high medical costs associated with prolonging life for
the very elderly, terminally ill, or those with certain types of health problems.
Chapter 1.1 1.1-24 Introduction to the Cost of Illness Handbook
-------
Link to Figure 1.1-1
services at the time the service is rendered, since the funds have already
been expended.
In addition, prepaid premiums or taxes are not designated for specific
services, but are held to be distributed on an as-needed basis (with need
being controversial in some cases). Consequently, insurance payments are
not provided equally to all premium holders due to differences in medical
services required. Thus, the premium holders have purchased an assurance
that their medical bills (or some portion of them) will be paid rather than
paying for a specific service. This is quite different than payment for a
specific treatment at the time it is needed based on decisions regarding its
value to the individual. There is no direct connection between payment for
and receipt of a good or service.14
When costs are high and exceed the amount of the premium (or individual
taxes), the excess is borne by a group rather than by the patient. This
system of payment also reinforces the concept that when the general public
purchases insurance or approves funding for public health programs, it is
buying some assurance of good health rather than specific services.
Patients may be willing to pay more for an improvement in health status
than for a specific service.
For some individuals and groups, the payment system is quite complex, and
determination of the costs of medical services and WTP is even more
difficult. For example, elderly patients, who comprise the majority of those
who suffer from cancer, chronic obstructive pulmonary diseases, heart
disease, and many other illnesses covered in this handbook, often have
multiple sources of payment, including Medicare (for which they often pay
premiums and taxes), private insurance, free clinic care for some services
(paid for by local funds), and direct out-of-pocket expenditures.
I.1.F.1.3 Conclusions Regarding the Market Value
These factors distort the simple connection between the price paid and the
actual costs that exists for purchasing many other goods and services.
Under these circumstances, it is difficult to determine precisely what an
average patient would be willing to pay. Although such uncertainties may
lead to an over- or underestimate of WTP, the illnesses in this handbook
are very serious in nature (i.e., either life-threatening or capable of resulting
in very debilitating conditions), and people are likely to be willing to pay
much more than they currently pay out-of-pocket or indirectly (e.g.,
through insurance premiums). The value to the afflicted individual and/or
friends and family of avoiding the anxiety, pain, and suffering associated
14 As Figure 1.1-1 shows, there are some direct payments of medical services by the patient;
however, full payment for major illnesses by a patient is relatively uncommon.
Chapter 1.1 1.1-25 Introduction to the Cost of Illness Handbook
-------
Link to Figure 1.1-1
with the illness (see Figure 1.1-1) may be considerably greater than those
cost components included in the cost-of-illness approach.
Because the payment of medical costs comes from multiple sources, both
direct and indirect, the purchase of medical services is actually determined
by multiple and diverse groups. Those who have a role in determining
medical costs and standard services for illnesses include insurance premium
payers; taxpayers and their elected representatives (who are involved in
determining payments for publicly funded services); state, local, and
national agencies, whose staff are also publicly accountable; corporations
with stockholders; medical personnel; and others. The overall cost of
living, employment characteristics, and cost for durable goods also play a
role. Consequently, although there are disjunctions and complications in
the determination of medical costs, the diversity of decision-makers in the
cost-setting process provide some assurance that medical costs reflect the
preferences of society.
I.1.F.2 Differences in Critical Patient Characteristics.
Some factors are related to individual characteristics. They introduce
uncertainty into the use of cost only if the population of concern (e.g., the
study population for a benefits assessment) has characteristics that differ
from those of the "average" members of the population with the disease.
These factors include:
• the organ systems affected by the disease (e.g., which systems,
multiple versus single systems);
the severity of the disease (may be related to the above);
the general health of the patients aside from the disease in question;
other complicating medical conditions in the patients; and
the life expectancy of the patients (this affects the length of
treatment)
I.1.F.3 Geographic Differences in Medical Practices and Services
Other factors that impact costs are associated with medical practices and
services in a geographic area. These factors may differ substantially among
various areas of the U.S. For example, urban areas generally have higher
direct medical costs than rural areas.
Chapter 1.1 1.1-26 Introduction to the Cost of Illness Handbook
-------
Factors of concern regarding differences in medical practices and services
include:
• the specific treatment protocols chosen;
• the hospital and professional fee rate structure in a particular area;
and
the support systems that provide care at no cost (e.g., home care).
Access to care is a particularly difficult factor in evaluating medical costs
and has a complex role in their calculation. Consequently, is it usually
discussed only qualitatively as a source of uncertainty. Its actual cost
impacts are rarely known. Access to services varies on a geographic and
socioeconomic basis and often increases both risks and costs for
economically disadvantaged patients. For example, access to medical care
has been found to be a critical factor in survival among people with asthma,
with limited access to care being closely linked to poverty status in some
major cities.
Access to care has multiple components. It includes the physical
availability of services. A one-hour bus trip to a clinic may be a major
impediment to care, and be replaced by the use of ambulance transport to
an emergency room rather than a doctor's office because an "emergency"
is required to obtain transportation (given that the bus ride is not a
reasonable option for someone who is sick). Moderately ill people who
could be seen in a clinic may be seen in a much more expensive emergency
room due to lack of access to designated Medicare/Medicaid clinics.
A second, and often more costly impact of limited access to care, is the
condition of the patient when he or she receives services. Access issues
may lead to delays in obtaining care so that disease management is poorer.
This care differential impacts the outcome and often leads to higher
mortality (as mentioned above with respect to asthma). The overall costs
of this factor, in direct and indirect costs (including pain and suffering), is
substantial. The vast majority of deaths from asthma occur among children
and the elderly who are below the poverty line.
When evaluating benefits, some aspects of the impacts of access to care
and other problems related to socioeconomic status may be relevant,
especially for chronic low-level diseases (e.g., COPD, asthma,
cardiovascular diseases). They are also relevant to cancers in which the
late diagnosis, in part due to less frequent medical checkups, is often noted
as a contributing factor to the higher mortality rate among lower
socioeconomic groups. An example is the strong positive association
between the probability of mammograms and family income. The reduced
use of this diagnostic tool among poorer women is linked to their increased
Chapter 1.1 1.1-27 Introduction to the Cost of Illness Handbook
-------
risk of breast cancer being diagnosed at later stages of the disease. This
delay leads to a poorer overall survival, higher medical costs, and increases
in lost time.
As noted previously, differences in medical practices and services will be
relevant ONLY if a benefits analysis focuses on an area or population
subgroup within the U.S. where the practices and services differ from the
"average" in the U.S. Most cost estimates in the handbook are derived
from databases that cover a wide socioeconomic and geographic spectrum.
1.1 .F.4 Uncertainty Regarding the Application And/or Accuracy of Input Data
Varied data sources are used for most cost analyses. These sources range
from scientific papers reported in the medical economics literature to
census and National Cancer Institute (NCI) data regarding demographics
and background rates of mortality, cancer diagnosis, and patient survival.
The quality and applicability of the input data are relevant to all uses of the
cost data. Uncertainty regarding the inputs to the cost estimates fall into
the following categories:
accuracy of estimates from primary sources,
accuracy of "background" cost estimates used with the disease cost
estimates to calculate the incremental costs,
• accuracy of life expectancy estimates for patients who are either
survivors or nonsurvivors of the disease,
accuracy of estimates of the percent of survivors and nonsurvivors,
accuracy of the estimates of survival among people without the
disease (used for background calculations),
/. 1.F.4.1 Geographically Representative Data
As noted previously, national data were sought to obtain the best
"average" estimate for each input parameter. Estimates were usually
obtained through the use of data on a cross-section of people with a similar
mix of ages, sexes, races, etc., to that of the overall U.S. population. Often
a subset of a national database was used, or the entire database for specific
years was obtained. This national approach provides a reasonably good
estimate of costs. As always with sampling, there is a small chance that the
data selected will not be entirely representative, thereby introducing some
uncertainty. This is not likely, however, to be a major source of
uncertainty.
Some diseases were evaluated based on data from a specific geographic
area. For example, some chapters on birth defects (e.g., cleft lip and
Chapter 1.1 1.1-28 Introduction to the Cost of Illness Handbook
-------
palate) were obtained from a detailed analysis conducted by Waitzman et
al. (1996) that used 18 databases in California to obtain a complex and
comprehensive description of direct and some indirect costs of the
disorders. California is economically and demographically diverse, and so,
as noted in the chapter, it is believed that the cost estimates are reasonable
approximations of a national average. Still, the use of location- or
population-specific data introduces additional uncertainty. When data were
used that generate this type of uncertainty, relevant concerns are described
in the chapter.
I.1.F.4.2 Treatment Estimates
As discussed previously, a few chapters rely on treatment protocols
provided by physicians to estimate medical costs. These protocols cause
uncertainty in estimating medical costs, due to the assumption that all
patients receive timely and adequate medical care, which is not always the
case in practice. Some patients don't seek care, some delay treatment, and
others are treated for only a short time. These behaviors are due to a
variety of factors. Evaluating the impact of this approach on costs is
difficult because a delay in care, while reducing immediate costs, usually
leads to increased costs in treating a more severe form of the disease and a
longer disability period, or in early mortality and an associated increase in
lost time. Rather than determining the direction of impact on costs (higher
or lower), it is simply noted that the cost estimate based on a physician
protocol introduces additional uncertainty.
I.1.F.4.3 Conversions and Calculations
Other types of uncertainty are introduced by the use of data that are
relevant, but not expressed in the units required for this analysis. Agencies
and researchers typically provide data in a format that is most useful for
their goals; this often does not match the requirements for cost estimation.
Consequently, the raw data obtained often require additional modifications
using inputs from multiple sources. For example, survival among cancer
patients is expressed as a relative survival rate (RSR) in the NCI databases.
It was necessary to convert the relative survival rate to an absolute survival
rate to carry out essential cost calculations. RSRs incorporate a
background mortality rate from the general population, as well as other
inputs, in their derivation. Various sources were consulted for the inputs
(e.g., the U.S. Census Bureau, mortality probabilities from NCI) and the
absolute survival rate was calculated. This type of calculation, which is a
derivation of a critical value, introduces uncertainty regarding the result.
The outcome is not likely to have substantial error, but there is not the
certainty that would be obtained if the absolute rates were available from a
primary source.
Chapter 1.1 1.1-29 Introduction to the Cost of Illness Handbook
-------
I.1.F.4.4 Changes in Medical Services Provision and Medical
Costs
The most common source of uncertainty is introduced by the use of data
that are not "current." There is no single definition of "current" because
the relevant concern is whether the cost estimates reflect ongoing practices
and costs. Medical approaches for some diseases have changed
considerably in recent years, while others remain very similar for many
years. Services and costs reflect changes and improvements in the
technical aspects of how medical care is delivered.
In addition to technical changes in how medicine is practiced, there have
been numerous changes in the way medical services are provided and in
medical costs during the 1980s and 1990s. Medical cost containment is a
relatively recent focus of private sector payers (e.g., insurance companies,
managed care providers) and has been the subject of considerable effort at
the federal level since the outset of the Medicare and Medicaid programs.
The rapidly escalating medical costs of the 1970s and 1980s led to a
national recognition of the need for medical cost containment. New
systems of care management and payment have evolved in recent years and
continue to evolve. Consequently, it is difficult to compare current costs
directly with those of the past. This is relevant to the costs presented in the
Handbook because many of the cost estimates are based on data collected
in past years.
Changes in payment structures and cost containment efforts proceeded
hand-in-hand with changes in the way in which services are provided. In
many cases, cost containment is partially accomplished through limitations
on the types of services or the specialty of the physicians to which a patient
has access. Thus, cost control efforts have been directed both at slowing
the increase in cost for a specific service, and in limiting the access to
expensive services. The consumer price index (CPI) medical care
component is used in the Handbook to inflate costs obtained in past years
to the most recent year available. The CPI was designed to estimate the
increases in costs associated with a specific service or item. It does not
address the issue of access to services. Although there have been studies
of the impacts of access limitation and other cost containment strategies,
there is not a single agreed-upon value that can be applied to compare
either services for a specific disease, or costs for that disease over time,
independent of factors other than access limitation.
Wicker et al (1999) contains a discussion of cost containment programs'
impact on patterns of care and the readmission of patients with respect to
children. They discussed utilization management (UM), which provides for
review and authorization of both inpatient and outpatient care for more
than 90 percent of enrollees in group insurance plans (HMOs, preferred
provide organizations, etc). Their study pointed out an ongoing problem
related to this cost reduction strategy: by limiting care through the review
Chapter 1.1 1.1-30 Introduction to the Cost of Illness Handbook
-------
process, UM decreases initial inpatient costs but increases the rate of
hospital readmission. This was noted in particular for low birth weight
infants and those with depression or drug or alcohol dependence problems
(Wicker et al., 1999). These are very common chronic conditions among
children and adolescents, and are also among the more costly conditions
due to their chronic nature. Consequently, the dynamic observed for these
conditions is potentially relevant to most chronic conditions. A similar
dynamic has been observed in adult patients, for whom delays in referrals
to specialists and specialized treatment led to more serious illnesses
requiring more costly care than would have originally been required. In the
most extreme cases, denial of services had led to death (and subsequent
court settlements).
What Wicker et al. and other studies suggest is that cost reductions using
resource limitation strategies may not reduce costs overall, at least in some
cases. In addition, the application of UM and other similar strategies varies
greatly in implementation, and its application is not universal. Making
simple assessments the impacts of trends in medical services, management,
and costs on the overall lifetime costs of an illness is therefore difficult.
The Handbook attempts to clearly state both the source of cost data and
limitations in the methods of data collection, study design, and other
factors that may affect how the cost data can be applied currently. It is
often not known if the costs for an illness are over- or underestimates of
current costs. Where new treatments are known to be offered and their
likely cost impact is known, however, that information is provided (usually
qualitatively).
When evaluating changes over time in costs and services, it is important to
consider all inputs to cost. For example, when the data on breast cancer
costs were collected, bone marrow transplants were not yet being done.
Now they are offered for the most advanced cases of the disease and they
are very expensive. Because they are provided to a small percentage of
women with the disease, they will not have a substantial impact on the
overall costs; however, bone marrow transplants may slightly increase the
average cost of treatment. Balancing this is the fact that women receiving
this treatment are less likely to die. Consequently, any application of
mortality percentages and the value of statistical life (VSL) to breast cancer
patients may overestimate the mortality-associated costs. In addition, there
is some evidence that women are being diagnosed with breast cancer at
earlier stages, which decreases the risk of death but also increases the
direct medical costs because they are living longer and requiring additional
services. Given the difficulty in evaluating changes in services and costs
over time, and the extreme difficulty in obtaining a national average lifetime
medical cost for most diseases, the cost estimates provided in this
Handbook offer a reasonable approximation, with the limitations associated
with the estimates acknowledged.
Chapter 1.1 1.1-31 Introduction to the Cost of Illness Handbook
-------
Appendix 1.1-A Equations Describing the Expected Present Discounted
Value of the Per Capita Lifetime Stream of Costs Associated with a
Given Illness
Illnesses are costly in many ways and often over long periods of time.15
Many illnesses result in costs for years after onset; some illnesses result in a
lifetime of costs. Some of these costs, such as hospitalization charges and
physician fees, are obvious. Other costs, such as the value of lost time due
to the illness, are less obvious but just as real. A complete accounting of
the total cost of an illness includes all the costs incurred as a result of the
illness from the time of onset to the time of cure or the death of the
individual — that is, the lifetime stream of costs associated with the
illness.16
Properly estimating the total value of this lifetime stream of costs requires
understanding several key considerations, including:
• costs incurred at a later time should be discounted,
there are several different kinds of cost, and
• the costs of an illness are incremental costs.
The lifetime stream of costs associated with an illness will vary from one
individual to another for a variety of reasons, including, for example, the
age of onset of the illness. For each year post-diagnosis, moreover, direct
costs can be incurred during that year only if the individual survives to that
year.17 If the individual dies in that year, then indirect costs are incurred.
The number of years of survival post-diagnosis, however, will also vary
from one individual to another. A fourth consideration, then, is that:
• it is not "a lifetime stream of costs" that is of interest, but rather the
expected, or average, lifetime stream of costs.
To estimate this average value, it is necessary to know the probabilities of
survival for each year post-diagnosis.
15 Birth defects are included within "illnesses."
16 Even a complete accounting of all costs of an illness will yield an underestimate of the true social
value of avoiding the illness, because it does not take into account the value of avoiding the pain and
suffering associated with the illness.
17 Estimates of the costs of an illness are usually used as lower-bound estimates of morbidity costs
rather than as estimates of the value of avoiding premature mortality from the illness. Estimates of the
value of avoiding premature mortality are generally substantially higher than cost of illness estimates.
Appendix 1.1-A I.1-A-1 Equations
-------
The expected present discounted value of per capita lifetime incremental
costs of an illness can be constructed from its component parts. Each of
the expressions or parameters used in this construction is explained in the
table below. When an expression is derived from other expressions and/or
parameters, the derivation is given in the table. All costs are average per
capita costs and are incremental (i.e., the costs of the illness beyond those
expected to be incurred by the same individual in the absence of the
illness).
Appendix 1.1-A I.1-A-2 Equations
-------
Table .1.A-1: Estimation of the Expected Present Discounted Value of Per Capita Lifetime Incremental Costs of an Illness
Parameter
Derivation
The cost of heightened morbidity:
j
pip medical
pip nonmedical
:«vlphm
ICj
:« vllthm
ICj
COStjhm
number of years post-diagnosis (an index)
direct medical costs j years post-diagnosis
direct nonmedical costs j years post-diagnosis
indirect costs j years post-diagnosis: value of lost time due
to heightened morbidity, estimated as the number of units
of productive time (e.g., hours or days) lost in the jth year
post-diagnosis due to the illness times the value per unit
time.
indirect costs j years post-diagnosis: value of lost leisure
time due to heightened morbidity, estimated as the
number of units of leisure time (e.g., hours or days) lost in
the jth year post-diagnosis due to the illness times the
value per unit time.
total costs of heightened morbidity incurred j years post-
diagnosis. Costjhm is an average cost among all those with
the illness who survive] years post-diagnosis. Any of the
components of costjhm may vary from one individual to
another because of factors such as sex or age.
^ hm _ .]„ medical , ,]„ nonmedical , ,•„ vlphm , ,•„ vllthm
The cost of premature mortality:
r
X
d
m
discount rate, reflecting individuals' positive rate of time
preference.
age of onset of the illness
age of death from the illness
expected age of death, in the absence of the illness
If death from the illness occurs] years post-diagnosis, d=x+j.
Appendix 1.1-A
1.1-A-3
Equations
-------
Table I.1.A-1: Estimation of the Expected Present Discounted Value of Per Capita Lifetime Incremental Costs of an Illness
Parameter
Derivation
vpk
value of time at age k (in the absence of the illness),
estimated as the number of units of time (e.g., days or
hours) of time at age k times the value per unit time.
vIL
value of leisure time at age k (in the absence of the
illness), estimated as the number of units of leisure time
(e.g., days or hours) at age k times the value per unit time.
indirect costs: value of lost time due to premature
mortality. This is the sum of discounted values of time for
each year that the individual would be expected to live in
the absence of the illness but did not live — i.e., from the
age at death (d) to the expected age at death (m).
1C
vlppm _
= E
k=d (1 + r)
k-d
jpVlltpm
indirect costs: value of lost leisure time due to premature
mortality. This is the sum of discounted values of leisure
time for each year that the individual would be expected to
live in the absence of the illness but did not live — i.e.,
from the age at death (d) to the expected age at death
(m).
ic
vlltpm _ V^ k
k=d (1 + ff
COStjprl
total costs of premature mortality for an individual who
dies j years post-diagnosis. Costjpm is an average cost
among all those who die from the illness j years post-
diagnosis. Any of the components of costjpm may vary
from one individual to another because of factors such as
sex or age. As noted above, if death from the illness
occurs j years post-diagnosis, then age at death is d=x+j.
Medical costs included in costjpm are those medical costs
that would not have been incurred in the absence of death
from the illness (e.g., terminal care costs of cancer).
costpm = dcmedlcal + i
where
1C
vlppm _
= E
k=d (1 + r)
k-d
, d = x+j
and similarly for ic.
vlltpm
Appendix 1.1-A
1.1-A-4
Equations
-------
Table .1.A-1: Estimation of the Expected Present Discounted Value of Per Capita Lifetime Incremental Costs of an Illness
Parameter
Derivation
Expected costs:
PSJ
probability of surviving j years post-diagnosis
Pd,
probability of dying j years post-diagnosis
~ Ps
E(costj)
expected costs incurred j years post-diagnosis. This is the
average cost of the illness j years post-diagnosis — the
average cost of heightened morbidity times the probability
of surviving j years post-diagnosis plus the average cost of
premature mortality times the probability of dying j years
post-diagnosis.
E(cost) = ps *costhm + pd *costpm
Age-of-onset-dependent costs: The costs and probabilities of surviving or dying j years post-diagnosis may depend not only on j but on the age
of onset of the illness, or, equivalently, on the individual's age (x+j) j years post-diagnosis. The probability of someone who gets lung cancer at
age 45 surviving to age 46, for example, may be very different from the probability of someone who gets lung cancer at age 70 surviving to age
71. The above parameters are therefore further refined below.
costjx
total costs of heightened morbidity incurred j years post-
diagnosis, given that age of onset is x. This is a
refinement of costjhm which acknowledges that one or more
components of the costs of heightened morbidity may
depend not only on the number of years post-diagnosis but
also on the age of onset or, equivalently, on current age
(x+j).
medical
nonmedical
vlphm
vllthm
costjxpm
total costs of premature mortality for an individual who
dies j years post-diagnosis, given that age of onset is x.
This is a refinement of costjpm which acknowledges that the
components of the costs of premature mortality may
depend not only on the number of years post-diagnosis but
also on the age of onset, or, equivalently, on current age
(x+j).
cost
J,X
pm = dc medlcal + ic vlppm + ic vlltpm
J,X J,X J,X
Appendix 1.1-A
1.1-A-5
Equations
-------
Table .1.A-1: Estimation of the Expected Present Discounted Value of Per Capita Lifetime Incremental Costs of an Illness
Parameter
PSi,x
Pdi,x
E(cosy
probability of surviving j years post-diagnosis, given that
age of onset is x.
probability of dying j years post-diagnosis, given that age
of onset is x.
expected total costs incurred j years post-diagnosis, given
that age of onset is x.
Derivation
Pdj,x =Psn,x*(l ~ Psj,>)
E(cost.) = psix*costixhm + pd.*cost.pm
jy\, jyA, JyA- jy\, jyA,
Discounted expected costs: Expected costs incurred] years post-diagnosis are discounted back to the time of diagnosis (onset) of the illness.
PDVECjx
PDVECX
Px
present discounted value of expected costs incurred j
years post-diagnosis, given age of onset x.
present discounted value of the lifetime stream of
expected costs when age of onset is x. This is the sum of
discounted expected costs from onset of the illness at age
x until either cure or the death of the individual.
probability that the age of onset of the illness is x
E(cost )
PF>VFC - J'
J' (1 + ry
E(cost )
PDVFC Y^ PDVFC Y^ J'
rL->^x Z^^-'^S* ^ (, v
y=o 7=0 (1 + ry
Appendix 1.1-A
1.1-A-6
Equations
-------
Table .1.A-1: Estimation of the Expected Present Discounted Value of Per Capita Lifetime Incremental Costs of an Illness
Parameter
EPDVEC
expected present discounted value of lifetime costs, i.e.,
the average over all possible ages of onset.
Derivation
,-, ,-, ,-, E(cost )
FPnVFC ^ H PDVFC VH V ^
r^rut £/u Z-^Px * ^x / j Pv/ -/
x=o x=o y=o (1 + ry
Approximations or alternatives to EPDVEC:
av
PDVECav
n
PDVCavghm
PDVCavg
the average age of onset of the illness
present discounted value of the lifetime stream of
expected costs when age of onset is the average age of
onset. PDVECav is PDVECX , where age of onset is the
average age of onset; av. PDVECav is an approximation to
EPDVEC.
average number of years post-diagnosis at which cure or
death occurs
present discounted value of the lifetime stream of costs of
heightened morbidity associated with the illness for the
"average individual" diagnosed with the illness.
present discounted value of the lifetime stream of costs
(including both heightened morbidity and premature
mortality costs) associated with the illness for the "average
individual" whose age of onset is av and who dies from the
illness at n years post-diagnosis. PDVCavg is a
simplification of PDVECav. It sets psjav = 1 forj n.
E(cost )
PDVFC Y^ PDVFC ^ J'
rut^L,av Z^ru>^j,av Z^ _
7=0 y=o (1 + ry
»^1 cost hm
PHFT hm - Y^ }'av
PD1Cavs ^(i+ry
pDVC _"^COS^hm + C°Stn,Jm
avg j-o (1 + rj (1 + /•)"
Appendix 1.1-A
1.1-A-7
Equations
-------
The hypothetical example below considers an "average individual" who
becomes ill at age 60 and who survives to age 68. The hypothetical
incremental costs incurred by this individual at each year post-onset, the
discounted age-specific costs, and the sum of these discounted costs (the
present discounted value of this average individual's costs at the time of
onset) are shown in the table below. Because the individual dies of the
illness at age 68, the value of lost leisure time for each year that he or she
would otherwise have lived is discounted back to age 68 and the
discounted values summed ($95,009).
For those individuals who ultimately die of an illness, it could be argued
that the value of a statistical life lost is the appropriate cost. This value
would subsume the cost-of-illness estimate. Based on evidence from
numerous value-of-life studies, the value of a statistical life far exceeds any
cost-of-illness estimates that might be derived from estimates of the four
cost components discussed above. For those individuals who do not die of
the illness, there will be no terminal care costs, nor will there be any lost
time or lost leisure time attributable to the individual dying prematurely.
Costs incurred during a period of remission, however, could exceed those
for terminal cases, if the remission period is substantially longer.
Appendix 1.1-A I.1-A-8 Equations
-------
Table 1.1. A-2: The Present Discounted Value of the Lifetime Stream of Hypothetical Costs of an
Illness of a Hypothetical Average Individual
Age
60*
61
62
63
64
65
66
67
68
Costs
Direct
Medical
$30,000
$10,000
$2,000
$2,000
$2,000
$2,000
$2,000
$2,000
$40,000
Nonmedical
$5,000
$5,000
$1,000
$1,000
$1,000
$1,000
$1,000
$1000
$500
Indirect
Value of Lost
Time
$20,000
$20,000
$200
$200
$200
$200
$0
$0
$0
Value of
Lost Leisure
Time
$7,000
$7,000
$0
$0
$0
$0
$200
$200
$95,009
Age-
specific
Costs
$62,000
$42,000
$3,200
$3,200
$3,200
$3,200
$3,200
$3,200
$135,509
Present discounted value of costs:
Discounte
d Age-
specific
Costs**
$62,000
$40,000
$2,902
$2,764
$2,633
$2,507
$2,388
$2,274
$91,718
$209,187
*Average age of onset.
** Using a discount rate of 5 percent, discounted back to the age of onset. For example, to get the present
discounted value (at age of onset) of the costs at age 63, these costs are divided by (1+0. 05)3
Appendix 1.1-A
1.1-A-9
Equations
-------
CHAPTER 11.1: INTRODUCTION TO THE COSTS OF CANCERS
Clicking on the sections below will take you to the relevant text.
II. 1. A Description
II. 1 .B Concurrent Effects
II.l.C Causality
II. 1 .D Chemicals Associated with Cancer Induction
II. 1 .E Genotoxicity
II. 1 .F Selection of Diseases
II.l.G Prognosis
II. 1 .G. 1 General Issues
II.1.G.2 Survival Estimates
II. 1 .H Typical Cancer Costs
II. I.H.I Source
II. 1 .H.2 Modifications to the Data
II. 1 .H.3 Total Non-incremental Costs of Treatment Phases
II. 1 .H.4 Maintenance Phase Costs
II. 1 .H. 5 Incremental Costs of Treatment Phases
II. 1 .H.6 Application to Specific Cancers
II. 1 .H.7 Conclusions Regarding Typical Cancer Cost Estimates
II. 1.1 Issues and Uncertainty in Cancer Medical Cost Estimation
II. 1.1.1 New Treatments
II. 1.1.2 Concurrent Effects
Chapter 11.1 11.1-1 Carcinogenic Effects
-------
CHAPTER 11.1: INTRODUCTION TO THE COSTS OF CANCERS
This section of the handbook contains chapters that describe costs of
medical treatments for a variety of cancers that have been associated with
exposure to environmental agents. Cancer is one of the three leading
causes of death in the United States and throughout the world (Williams
and Weisburger, 1993). It is a serious illness that has been associated with
environmental exposures in both human and animal studies. Cancers often
have similar treatment options (e.g., radiation, chemotherapy, surgery,
reconstructive and physical therapy treatments), and generally require
long-term medical care. Most occur with much greater frequency in
individuals in the second half of life.
This section contains an overview of the environmental causes of cancer
and general issues related to economic valuation of the medical treatment
of cancer. It also contains an estimate of the cost of a "typical" cancer case
and examples of how the typical cost estimates can be modified to obtain
more information on specific cancers (using liver and bone cancers as
examples). This chapter is followed by chapters containing medical cost
information on specific types of cancer that may be associated with
exposure to environmental agents.
11.1.A. Description
Cancer is the common term for all malignant tumors and includes
carcinomas and sarcomas, depending on the tissue of origin. Cancer is
characterized by abnormal growth that preys on the host. It may
metasticize to other locations in the body and often leads to debilitation
and/or death if left untreated. A critical distinction among tumors is made
between benign and malignant tumors. Although benign tumors may be
medically important and sometimes become malignant (Robbins et al.,
1984), they are not considered cancerous, and so are not included in the
cost estimates presented in this section.
11.1.B. Concurrent Effects
Concurrent effects commonly occur with cancer. These effects usually
arise from two main causes: 1) as a result of a metastatic (spreading)
process, leading to cancers at more than one site in the body, 2) as a result
of the impaired health status of the cancer patient. Impaired health can
arise from either the cancer's interference with normal functioning, or the
adverse effects associated with chemotherapy, radiation treatment, surgery,
or other medical treatments. Many of the side effects of cancer treatment
are well known. For example, the antineoplastic drug adriamycin causes
damage to the heart muscle. Radiation therapy may cause toxicity (e.g.,
sterility) and additional cancers while limiting the spread of cancer at the
Chapter 11.1 11.1-2 Carcinogenic Effects
-------
original site. A variety of other effects are associated with cancer
therapies. One focus in oncology is on balancing the toxic properties of the
anticancer therapies in the health portions of the body against the need to
cause toxicity to the tumor cells.
Because there are no benign cancer therapies, and cancer is often a
metastatic process, there are invariably some costs associated with cancer
that are not considered when only the direct medical costs of treating the
primary cancer are considered. Some researchers (e.g. Baker et al., 1989
and 1991) have taken many of the costs of concurrent effects into account
by evaluating the costs associated with the treatment of cancer patients in
relation to medical costs of individuals without cancer (referred to as
background medical costs). When background costs are subtracted from
the total medical costs to cancer patients, the remaining incremental costs
to cancer patients include costs of treating side effects, as well as the
original cancer treatment costs. Baker et al.'s values are reported in
Chapters H.2., II.3, II.4, II.5, II.7, and II.8.
This incremental approach captures medical costs associated with side
effects that occur during treatment for the original cancer, but those that
occur at a later date may not be included. For example, medical costs
associated with a second cancer that occurs years later, induced by
radiation therapy, would not be included using this approach. There are
currently no very long-term follow-up data on these types of costs. This
omission is likely to lead to an underestimate of total medical costs.
II.1.C. Causality and Special Susceptibilities
Carcinogens may act directly in causing cancer (initiators), or with other
chemicals or individual characteristics to promote the development of
cancer (promoters). Both initiators and promoters increase cancer risk.
Cancer involves a change in cells that eliminates the normal controls on the
growth of cells (Williams and Weisburger, 1993). Most carcinogens
interact with DNA to alter the basic genetic directions of cells. Common
characteristics of these carcinogens are that their effects are persistent,
cumulative, and delayed (Ibid). The delay in effects, often for decades,
make their identification and the quantification of their risks to humans
difficult.
Although thousands of studies of the carcinogen!city of chemicals are
conducted, most are carried out in animals due to:
1) the long delay after exposure in the development of tumors in
humans, as noted above;
2) ethical issues;
Chapter 11.1 11.1-3 Carcinogenic Effects
-------
3) costs of conducting human studies; and
4) difficulties with the confounding effects of carcinogens not under
study.1
Animals are used because they provide a controlled set of subjects whose
exposure can be measured accurately. Due to their relatively short life
span, cancer can be observed and quantified in a reasonable amount of time
in animals, especially in rodents. They are typically given large doses
because this allows a relatively small number of animals to be used (e.g., 50
or 100) to obtain a statistically significant result. The results are then
scaled to a human dose and response.
The chemical induction of cancer in animals is generally assumed to be
evidence that the chemical may pose cancer risks to humans. EPA's
Proposed Guidelines for Cancer Risk Assessment state that:
"The default assumption is that positive effects in animal
cancer studies indicate that the agent under study can have
carcinogenic potential in humans. Thus if no adequate
human data are present, positive effects in animal cancer
studies are a basis for assessing the carcinogenic hazard to
humans... The assumption is supported by the fact that
nearly all of the agents known to cause cancer in humans
are carcinogenic in animals in tests with adequate
protocols... Further support is provided by research on the
molecular biology of cancer processes, which has shown
that the mechanisms of control of cell growth and
differentiation are remarkably homologous among
species..." (EPA 1996).
These proposed guidelines are very similar to those which have been in
force for the last decade (EPA 1986).2
Although the assumption is made that cancer induction in animals may
indicate cancer risk in humans, it is not assumed that cancer will occur in
the same organ(s) in humans as in animals (EPA 1996). In addition,
carcinogens often act non-specifically, being capable of acting on multiple
organ systems throughout the body. Consequently, most cancer studies
1 The absolute prohibition against exposure to non-beneficial chemicals (e.g., pharmaceuticals) that
exists for humans does not exist for animals, and most chemicals of toxicological interest continue to be
tested on animals. Ethical issues still persist, however, when animal studies are conducted, due to
procedures that raise serious ethical concerns.
2 Many specific criteria are used in determining whether a chemical is carcinogenic. See the
proposed cancer guidelines for more information (EPA 1996).
Chapter 11.1 11.1-4 Carcinogenic Effects
-------
cannot be used to determine the specific site where cancer is likely to occur
in humans, but can be used to strongly suggest that a cancer risk exists.3
Because positive cancer study results in animals may be relevant to many
types of cancer in humans, the results of cancer studies in animals are listed
in this introductory cancer chapter rather than in the individual cancer
chapters that deal with a specific organ (e.g., kidney cancer, lung cancer).
11.1 .D. Chemicals Associated with Cancer Induction
Table II. 1-1 lists chemicals that have been associated with carcinogenic
effects in either animal or human studies, based on EPA's review of the
carcinogenicity data. The chemicals listed in Table II. 1-1 have been
identified as potential human carcinogens in one or more of EPA's large
toxicity databases: HSDB, IRIS, or HEAST. The table, compiled in 1996,
is not a comprehensive list of all carcinogens.
The chemicals listed in Table II. 1-1 are a sample of the potential
environmental agents associated with this disease. Although the table
contains many chemicals, it is incomplete for two reasons:
1. It does not include toxicological data from sources other than
HSDB, IRIS, and HEAST. The toxicological literature currently
available is vast, and a thorough review was beyond the scope of
this analysis.
2. Many chemicals have not been tested, or the results of the tests are
inconclusive. Consequently, the human health effects of many
environmental hazards are unknown, especially at concentrations
found in the environment.
3 When portal-of-entry effects or other location-specific interactions occur, cancer sites can be
more specifically predicted.
Chapter 11.1 11.1-5 Carcinogenic Effects
-------
Table 11.1-1. SUSPECTED CARCINOGENS LISTED IN IRIS, HEAST, AND HSDB
Chemicals are listed alphabetically and TRI chemicals as of August 2000 are shown with an asterisk. When
"compounds" of a metal were listed on TRI, all compounds of that metal in this table are considered to be TRI
chemicals. See text for discussion of inclusion criteria. This is not a comprehensive list of all carcinogens.
CHEMICAL
ACENAPHTHENE
ACENAPHTHYLENE
ACEPHATE*
ACRYLAMIDE*
ACRYLONITRILE*
ALACHLOR*
ALDRIN*
ALKANOLAMINE SALTS, 2,4-D,
ALUMINUM*
ALUMINUM FLUORIDE
ALUMINUM OXIDE*
ALUMINUM SODIUM FLUORIDE
AMMONIUM CHROMATE
AMMONIUM DICHROMATE
AMOSITE
AMPICILLIN
ANILINE*
ARAMITE
ARSENIC*
ARSENIC ACID*
ARSENIC PENTOXIDE*
ARSENIC TRIBROMIDE*
ARSENIC TRICHLORIDE*
ARSENIC TRIIODIDE*
ARSENIC TRIOXIDE*
ARSENIC TRISULFIDE*
ARSINE
ASBESTOS*
ASPHALT
ATRAZINE*
ATTAPULGITE
AZOBENZENE
BENZENE*
BENZENE HEXACHLORIDE
BENZIDINE*
BENZO(A)PYRENE*
BENZO(B)FLUORANTHENE*
BENZOTRICHLORIDE
BENZOYL CHLORIDE*
BENZO[A]PYRENE
BENZYL CHLORIDE*
BERYLLIUM*
BERYLLIUM CHLORIDE*
BERYLLIUM FLUORIDE*
BERYLLIUM HYDROXIDE*
BERYLLIUM NITRATE*
BERYLLIUM OXIDE*
BERYLLIUM PHOSPHATE*
SOURCE(S)
HSDB
HSDB
IRIS
IRIS
HSDB, IRIS
HEAST
IRIS
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
IRIS
IRIS
HSDB, IRIS
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HEAST
HSDB
IRIS
HSDB, IRIS
HSDB
HSDB, IRIS
HSDB
HSDB
HSDB, IRIS
HSDB
IRIS
IRIS
HSDB, IRIS
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
Chapter 11.1
1.1-6
Carcinogenic Effects
-------
Table 11.1-1. SUSPECTED CARCINOGENS LISTED IN IRIS, HEAST, AND HSDB
Chemicals are listed alphabetically and TRI chemicals as of August 2000 are shown with an asterisk. When
"compounds" of a metal were listed on TRI, all compounds of that metal in this table are considered to be TRI
chemicals. See text for discussion of inclusion criteria. This is not a comprehensive list of all carcinogens.
CHEMICAL
BERYLLIUM SULFATE*
BIS(2-CHLOROETHYL) ETHER*
BIS(2-CHLOROETHYL)SULFIDE
BIS(CHLOROETHYL)ETHER
BIS(CHLOROMETHYL) ETHER*
BROMO-2-CHLORO-1,1,1-TRIFLUOROETHANE, 2-
BROMOCHLORODIFLUOROMETHANE*
BROMODICHLOROMETHANE
BROMOETHENE (VINYL BROMIDE)*
BROMOFORM*
BUTADIENE, 1,3-*
BUTYRIC ACID, 4-(2,4-DICHLOROPHENOXY)
CALCIUM ARSENATE
CALCIUM ARSENITE
CALCIUM CHROMATE
CAPTAFOL
CAPTAN*
CARBAZOLE
CARBON BLACK
CARBON TETRACHLORIDE*
CHLORAMBUCIL
CHLORAMPHENICOL
CHLORANIL
CHLORDANE
CHLORO-1 ,1 ,1 ,2-TETRAFLUOROETHANE, 2-*
CHLORO-1.3-BUTADIENE, 2-
CHLORO-2-FLUOROETHANE, 1-
CHLORO-2-METHYLANALINE, 4-
CHLORO-2-METHYLANILINE HYDROCHLORIDE
CHLORO-2-METHYLPHENOL, 4-
CHLOROBENZILATE*
CHLORODIFLUOROMETHANE*
CHLOROFORM*
CHLOROMETHANE*
CHLOROMETHYL METHYL ETHER*
CHLORONITROBENZENE, O-
CHLORONITROBENZENE, P-
CHLOROPENTAFLUOROETHANE
CHLOROTETRAFLUOROETHANE
CHLOROTHALONIL
CHROMIC ACID, CHROMIUM(3+) SALT
CHROMIC OXIDE
CHROMIC SULFATE
CHROMIC TRIOXIDE
CHROMITE
CHROMIUM*
CHROMIUM CHROMATE*
CHROMIUM DIOXIDE*
CHROMIUM TRIHYDROXIDE*
CHROMIUM(III) ACETATE*
CHROMOUS CHLORIDE
SOURCE(S)
HSDB
HEAST
HSDB
IRIS
HSDB, IRIS
HSDB
HSDB
IRIS
HEAST
IRIS
HSDB
HSDB
HSDB
HSDB
HSDB
HEAST
HEAST
HEAST
HSDB
HSDB, IRIS
HSDB
HSDB
HEAST
HSDB, IRIS
HSDB
HSDB
HSDB
HEAST
HEAST
HSDB
HEAST
HSDB
IRIS
HEAST
HSDB
HEAST
HEAST
HSDB
HSDB
HEAST
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
Chapter 11.1
1.1-7
Carcinogenic Effects
-------
Table 11.1-1. SUSPECTED CARCINOGENS LISTED IN IRIS, HEAST, AND HSDB
Chemicals are listed alphabetically and TRI chemicals as of August 2000 are shown with an asterisk. When
"compounds" of a metal were listed on TRI, all compounds of that metal in this table are considered to be TRI
chemicals. See text for discussion of inclusion criteria. This is not a comprehensive list of all carcinogens.
CHEMICAL
CHROMOUS OXALATE
CHROMYL CHLORIDE
CHRYSOTILE ASBESTOS
CIS-DIAMINEDICHLOROPLATINUM
CLOMIPHENE
CLOZARIL
COAL TAR USP
COAL TAR CREOSOTE*
COAL TAR
COPPER*
CREOSOTE, WOOD*
CYANAZINE*
CYCLOPHOSPHAMIDE
D, 2,4-*
D BUTOXYETHYL ESTER, 2,4-*
D, BUTOXYPROPYL ESTER, 2,4-*
D BUTYL ESTER, 2,4-*
D CHLOROCROTYL ESTER, 2,4-*
D, DIMETHYLAMINE, 2,4-*
D ISOOCTYL ESTERS, 2,4-*
D ISOPROPYL ESTER, 2,4-*
D, PROPYLENE GLYCOL BUTYL ETHER ESTER, 2,4-*
DAUNORUBICIN
DDT
DI(2-ETHYLHEXYL)ADIPATE
DI(2-ETHYLHEXYL)PHTHALATE*
DIALLATE*
DIBENZ(A,H)ACRIDINE*
DIBENZ(A,J)ACRIDINE*
DIBENZO(A,E)PYRENE*
DIBENZO(A,H)PYRENE*
DIBENZO(A,L)PYRENE*
DIBENZO(C,G)CARBAZOLE, 7H-*
DIBROMO-3-CHLOROPROPANE, 1,2*
DIBROMOCHLOROMETHANE
DIBROMOETHANE, 1,2-*
DIBROMOTETRAFLUOROETHANE, 1,2-
DICHLORFOP-METHYL
DICHLORO-1 ,1 ,2-TRIFLUOROETHANE*
DICHLORO-1,1,1-TRIFLUOROETHANE, 2,2-*
DICHLORO-1 ,1 ,2,2-TETRAFLUOROETHANE
1,2-
DICHLORO-1 ,1-DIFLUOROETHANE, 1 ,2-*
DICHLORO-1-FLUOROETHANE, 1,1-*
DICHLORO-2-BUTENE, 1,4-*
DICHLOROBENZENE*
DICHLOROBENZENE, 1,4-*
DICHLOROBENZENE, 1,2-*
DICHLOROBENZENE, 1,3-*
DICHLOROBENZENE, 1,4-*
DICHLOROBENZIDINE, 3,3'-*
DICHLORODIFLUOROMETHANE*
SOURCE(S)
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HEAST
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
IRIS
IRIS
HEAST
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HEAST
IRIS
IRIS
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HEAST
HSDB
HSDB
HSDB
HSDB
HEAST
HSDB, IRIS
HSDB
Chapter 11.1
1.1-8
Carcinogenic Effects
-------
Table 11.1-1. SUSPECTED CARCINOGENS LISTED IN IRIS, HEAST, AND HSDB
Chemicals are listed alphabetically and TRI chemicals as of August 2000 are shown with an asterisk. When
"compounds" of a metal were listed on TRI, all compounds of that metal in this table are considered to be TRI
chemicals. See text for discussion of inclusion criteria. This is not a comprehensive list of all carcinogens.
CHEMICAL
DICHLORODIPHENYL DICHLOROETHANE, P,P'-
DICHLORODIPHENYLDICHLOROETHYLENE, P,P'-
DICHLORODIPHENYLTRICHLOROETHANE, P,P'-
DICHLOROETHANE, 1,2-*
DICHLOROETHYLENE, 1,1-
DICHLOROMETHANE*
DICHLOROPHENOL, 2,4-*
DICHLOROPROPANE, 1,2-*
DICHLOROPROPENE, 1,3-
DICHLOROTRIFLUOROETHANE*
DICHLORVOS*
DIELDRIN
DIENESTROL
DIETHYLSTILBESTROL
DIMEHTYLBENZIDINE, 3,3-*
DIMETHOXYBENZIDINE, 3,3-*
DIMETHYL SULFATE*
DIMETHYLANALINE HYDROCHLORIDE, 2,4-
DIMETHYLANILINE, 2.4-
DIMETHYLCARBAMOYL CHLORIDE*
DINITROTOLUENE MIXTURE, 2,4-/2,6-*
DIOXANE, 1,4-*
DIPHENYLHYDRAZINE, 1,2-*
DIRECT BLACK 38*
DIRECT BLUE 6*
DIRECT BROWN *95
EPICHLOROHYDRIN*
ESTRONE
ETHANOL
ETHYL ACRYLATE*
ETHYL CARBAMATE
ETHYLENE GLYCOL DINITRATE
ETHYLENE OXIDE*
ETHYLENE THIOUREA*
FERRIC ARSENATE
FERRIC OXIDE
FERROUS ARSENATE
FLUOROURACIL*
FOLPET*
FOMESAFEN*
FORMALDEHYDE*
FURAZOLIDONE
FURIUM
FURMECYCLOX
GASOLINE
GILSONITE
HCFC-123A*
HCFC-123B*
HCFC-124A*
HEMATITE
HEPTACHLOR*
SOURCE(S)
IRIS
IRIS
IRIS
IRIS
IRIS
HSDB, IRIS
HSDB
HEAST
HSDB
HSDB
HSDB, IRIS
IRIS
HSDB
HEAST
HEAST
HEAST
HSDB
HEAST
HEAST
HSDB
IRIS
IRIS
IRIS
HEAST
HEAST
HEAST
HSDB, IRIS
HSDB
HSDB
HEAST
HSDB
HSDB
HEAST, HSDB
HEAST
HSDB
HSDB
HSDB
HSDB
IRIS
IRIS
HSDB
HEAST
HEAST
IRIS
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB, IRIS
Chapter 11.1
1.1-9
Carcinogenic Effects
-------
Table 11.1-1. SUSPECTED CARCINOGENS LISTED IN IRIS, HEAST, AND HSDB
Chemicals are listed alphabetically and TRI chemicals as of August 2000 are shown with an asterisk. When
"compounds" of a metal were listed on TRI, all compounds of that metal in this table are considered to be TRI
chemicals. See text for discussion of inclusion criteria. This is not a comprehensive list of all carcinogens.
CHEMICAL
HEPTACHLOR EPOXIDE
HEXACHLOROBENZENE*
HEXACHLOROBUTADIENE
HEXACHLOROCYCLOHEXANE, ALPHA-
HEXACHLOROCYCLOHEXANE, BETA-
HEXACHLOROCYCLOHEXANE, GAMMA-
HEXACHLOROCYCLOHEXANE, TECHNICAL
HEXACHLORODIBENZO-P-DIOXIN*
HEXACHLOROETHANE*
HEXAHYDRO-1 ,3,5-TRINITRO-1 ,3,5-TRIAZINE
HYDRAZINE*
HYDRAZINE/HYDRAZINE SULFATE*
ISOPHORONE
KEROSENE
LEAD ARSENATE*
LEAD CHROMATE*
LEAD PHOSPHATE*
LITHIUM CHROMATE
MAGNESIUM ARSENATE
MECHLORETHAMINE
MELPHALAN
MERCURY, ELEMENTAL*
METHALLENESTRIL
METHOXSALEN
METHOXY-5-NITROANILINE, 2-
METHOXYPSORALEN, 5-
METHYL ISOBUTYL KETONE*
METHYL-5-NITROANILINE, 2-
METHYLANILINE, 2-
METHYLANILINE HYDROCHLORIDE, 2-
METHYLENE BIS(N,N'-DIMETHYL)ANILINE, 4,4'-
MOLYBDATE ORANGE
NAPHTHYLAMINE, 2-
NICKEL*
NICKEL CARBONATE*
NICKEL CARBONYL*
NICKEL CHLORIDE*
NICKEL FORMATE*
NICKEL HYDROXIDE*
NICKEL OXIDE*
NICKEL SULFATE*
NITROFURAZONE
NITROPROPANE, 2-*
NITROQUINOLINE-N-OXIDE, 4-
NITROSO-DI-N-BUTYLAMINE, N-*
NITROSO-M-ETHYLUREA, N-
NITROSO-N-METHYLETHYLAMINE, N-
NITROSODI-N-PROPYLAMINE, N-*
NITROSODIETHANOLAMINE, N-
NITROSODIETHYLAMINE, N-**
NITROSODIMETHYLAMINE, N-*
SOURCE(S)
IRIS
IRIS
IRIS
IRIS
IRIS
HEAST
IRIS
IRIS
IRIS
IRIS
HSDB
IRIS
IRIS
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HEAST
HSDB
HSDB
HEAST
HSDB
HSDB
HEAST
HEAST
HEAST
IRIS
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
HEAST
HEAST
HSDB
IRIS
HEAST
IRIS
IRIS
IRIS
IRIS
IRIS
Chapter 11.1
11.1-10
Carcinogenic Effects
-------
Table 11.1-1. SUSPECTED CARCINOGENS LISTED IN IRIS, HEAST, AND HSDB
Chemicals are listed alphabetically and TRI chemicals as of August 2000 are shown with an asterisk. When
"compounds" of a metal were listed on TRI, all compounds of that metal in this table are considered to be TRI
chemicals. See text for discussion of inclusion criteria. This is not a comprehensive list of all carcinogens.
CHEMICAL
NITROSODIPHENYLAMINE, N-*
NITROSOPYRROLIDINE, N-
OCHRATOXIN A
ORYZALIN*
OXYPHENBUTAZONE
PENTABROMO-6-CHLOROCYCLOHEXANE,
1,2,3,4,5-
PENTACHLORONITROBENZENE
PENTACHLOROPHENOL*
PENTACHLOROPHENOL, SODIUM SALT*
PETROLEUM ETHER
PHENOBARBITAL
PHENYLBUTAZONE
PHENYLENEDIAMINE, O-
PHENYLPHENOL, 2-*
POLYBROMINATED BIPHENYLS*
POLYCHLORINATED BIPHENYLS*
POLYVINYL CHLORIDE
POTASSIUM ARSENATE
POTASSIUM ARSENITE
PROCHLORAZ
PROPIONIC ACID, 2-(3-CHLOROPHENOXY)
PROPYLTHIOURACIL
PROPYLENE OXIDE*
QUINOLINE*
RADIUM
RADON
SILICON DIOXIDE
SIMAZINE*
SODIUM ARSENATE
SODIUM ARSENITE
SODIUM CHROMATE
SODIUM DICHROMATE
SODIUM DIETHYLDITHIOCARBAMATE*
STREPTOZOTOCIN
STRONTIUM
STRONTIUM CHROMATE
T, 2,4,5-
TALC
TETRACHLORODIBENZO-P-DIOXIN, 2,3,7,8- *
TETRACHLOROETHANE, 1,1,1,2-*
TETRACHLOROETHANE, 1,1,2,2-*
TETRACHLOROPHENOL, 2,3,4,6-
TETRACHLOROTOLUENE, PARA, ALPHA, ALPHA, ALPHA-
TETRACHLOROVINPHOS/(STIROFOS)
TETRAETHYL LEAD
THIO-TEPA
THORIUM DIOXIDE*
TITANIUM DIOXIDE
TOLUENE-2.4-DIAMINE
TOLUIDINE, P-
TOXAPHENE*
SOURCE(S)
IRIS
IRIS
HSDB
HSDB
HSDB
HEAST
HEAST
HSDB, IRIS
HSDB
HSDB
HSDB
HSDB
HEAST
HEAST
HEAST
IRIS
HSDB
HSDB
HSDB
IRIS
HSDB
HSDB
IRIS
HEAST
HSDB
HSDB
HSDB
HEAST
HSDB
HSDB
HSDB
HSDB
HEAST
HSDB
HSDB
HSDB
HSDB
HSDB
HSDB, HEAST
IRIS
IRIS
HSDB
HEAST
HEAST
HSDB
HSDB
HSDB
HSDB
HEAST
HEAST
HSDB, IRIS
Chapter 11.1
11.1-11
Carcinogenic Effects
-------
Table 11.1-1. SUSPECTED CARCINOGENS LISTED IN IRIS, HEAST, AND HSDB
Chemicals are listed alphabetically and TRI chemicals as of August 2000 are shown with an asterisk. When
"compounds" of a metal were listed on TRI, all compounds of that metal in this table are considered to be TRI
chemicals. See text for discussion of inclusion criteria. This is not a comprehensive list of all carcinogens.
CHEMICAL
TP, 2,4,5-
TREMOLITE ASBESTOS
TRICHLOROANILINE, 2,4,6-
TRICHLOROANILINE HYDROCHLORIDE, 2,4,6-
TRICHLOROETHANE, 1,1,2-*
TRICHLOROFLUOROMETHANE*
TRICHLOROPHENOL, 2,4,5-*
TRICHLOROPHENOL, 2,4,6-*
TRICHLOROPROPANE, 1,2,3-
TRIFLURALIN*
TRIMETHYL PHOSPHATE
TRINICKELDISULFIDE
TRINITROTOLUENE, 2,4,6-
URANIUM
URANYL ACETATE
URANYL NITRATE
URANYL SULFATE
VINCRISTINE
VINYL CHLORIDE*
VITAMIN A
ZINCCHROMATE*
ZINC CHROMATE HYDROXIDE*
ZINC DICHROMATE*
ZINC POTASSIUM CHROMATE*
SOURCE(S)
HSDB
HSDB
HEAST
HEAST
IRIS
HSDB
HSDB
IRIS
HEAST
IRIS
HEAST
HSDB
IRIS
HSDB
HSDB
HSDB
HSDB
HSDB
HEAST, HSDB
HSDB
HSDB
HSDB
HSDB
HSDB
For these reasons, Table II. 1-1 should not be used as a definitive source of
information on the links between chemicals and cancer. A comprehensive
literature search is necessary to identify the dose-response relationships
between chemicals of concern and this or other health effects.
The chemicals with asterisks in Table II. 1-1 are TRI chemicals (subject to
reporting under the Toxics Release Inventory, Section 313 of the
Emergency Planning and Community Right-to-Know Act). When
"compounds" of a metal were listed on TRI, all compounds of that metal in
this table are considered to be TRI chemicals. Chemicals included on the
TRI due to their human health effects are known or reasonably anticipated
to cause either significant adverse acute health effects or chronic health
effects as a condition of their listing on TRI.
The route of exposure (e.g., oral, inhalation, dermal) is often considered
when evaluating whether a chemicals poses a carcinogenic risk. The routes
of exposure are not listed in Table II. 1-1 for two reasons:
Chapter 11.1
11.1-12
Carcinogenic Effects
-------
1. A chemical that is carcinogenic by one route of exposure will
usually be assumed to be carcinogenic by other routes of exposure.
EPA's proposed cancer guidelines state that it is assumed "that an
agent that causes internal tumors by one route of exposure will be
carcinogenic by another route if it is absorbed by the second route
to give an internal dose" (EPA 1996). In effect, most carcinogens
will fall under this assumption under most circumstances.
2. This table provides preliminary information on many chemicals
identified as potential human carcinogens. Risk or health
assessment, however, requires considerably more information than
that provided in the table. Consequently, additional information
must be collected and evaluated by researchers to fully evaluate
cancer risks; an analysis of route-specific data is a part of this
evaluation.
In considering the potential impacts of carcinogens, it is useful to note that
a number of them are known to cross the placental barrier, and some
cancers are likely to be the result of this type of exposure (Williams and
Wei sburger, 1993).
.1.E. Genotoxicity
Genotoxicity assays usually provide information regarding a chemical's
ability to interact with DNA. Genotoxicity may be associated with cancer
induction because, in most cases, the alteration in the cells' normal
replication methods allows uncontrolled growth that characterizes cancer.
Table II. 1-2 contains a listing of chemicals associated with genotoxic
effects listed in a variety of sources.4 These chemicals have yielded
positive results in genotoxicity assays, which are usually cell-level studies
of a chemical's interaction with the genetic material (DNA) within a cell
and/or its ability to cause mutations. Genotoxins are not all necessarily
carcinogenic to humans; however, genotoxicity indicates the potential for
actions that may cause cancer. Table II. 1-2 contains only a small
percentage of all the chemicals that have had positive genotoxicity assays.
As of 1990, the Environmental Mutagen Information Center in Oak Ridge,
Tennessee, maintained mutagenicity data on 21,000 chemicals
(Hoffmann, 1991). The size of the database indicates the magnitude of the
chemicals of potential interest regarding their carcinogenic capabilities.5
4 Table II. 1-1 contains both genotoxic and non-genotoxic carcinogens with the criteria for inclusion
being a positive carcinogenicity assay. A positive genotoxicity assay was the criterion for inclusion in
Table II. 1-2.
5 Not all carcinogens are genotoxic (e.g., hormonally-mediated carcinogens). See EPA (1996) for
a discussion of this distinction.
Chapter 11.1 11.1-13 Carcinogenic Effects
-------
Chapter III. 1 contains an additional discussion of genotoxicity relevant to
birth defects.
Link to III.I.C.4
Table 11.1-2. CHEMICALS ASSOCIATED WITH GENOTOXIC EFFECTS
DATA FROM HUMAN, ANIMAL, AND IN VITRO STUDIES ARE INCLUDED.
CHEMICAL
ACETONE*
ACROLEIN*
ACRYLIC ACID*
ACRYLONITRILE*
ALACHLOR*
ALDICARB*
AMINOPTERIN
AMITRAZ*
AMITROLE*
ANTU
AROCLOR1016(APCB)*
ARSENIC COMPOUNDS*
ARSENIC*
ASULAM
ATRAZINE*
AVERMECTIN B1
BENOMYL*
BENZENE*
BENZO(A)PYRENE*
BIORESMETHRIN
BISULFAN
BORIC ACID
BRADIFACOUM
BUSULFAN
BUTACHLOR
CADMIUM*
CAPROLACTAM
CAPTAFAL
CAPTAN*
CARBARYL*
CARBOFURAN*
CARBON TETRACHLORIDE*
CARBON DISULFIDE*
CARBOPHENOTHION
CHLORDANE*
CHLORDECONE
CHLORDIMEFORM
CHLORFENVINPHOS
CHLORMEQUAT
CHLOROBENZILATE*
CHLOROBIPHENYLS (INCLUDES PCBS)*
CHLOROFORM*
CHLOROPHACINONE
CHLOROPROPHAM
CHLOROTHALONIL*
CHLORPROPHONE
CHROMIUM*
COPPER SULFATE*
REFERENCES
5
4
7
5
1
1
3
6
5
5
7
6
6
7
9
7
1,7
5
14
1
14
5
1
3
4
14
7
6
7,6
6
6
1
5
1
12
1
6
6
5
7
3
1
1
7
1
1
16
5
Chapter 11.1
11.1-14
Carcinogenic Effects
-------
Table 11.1-2. CHEMICALS ASSOCIATED WITH GENOTOXIC EFFECTS
DATA FROM HUMAN, ANIMAL, AND IN VITRO STUDIES ARE INCLUDED.
CHEMICAL
COUMACHLOR
COUMAFURYL
COUMATETRALYL
CYANIDES*
CYCLOHEXANE*
CYCLOHEXANONE
CYCLOHEXIMIDE
CYCLOPENTAPYRENE
CYCLOPHOSPHAMIDE
CYHALOTHRIN*
2,4-D*
DALAPON
DECAMETHRIN
DEET (DIETHYLTOLUAMIDE)
DI(2-ETHYL HEXYL) ADIPATE
DIBROMOCHLOROPROPANE*
DICAMBA*
DICHLOBENIL
O-DICHLOROBENZENE*
P-DICHLOROBENZENE*
DICHLOROETHYL ETHER
1 ,3-DICHLOROROPENE (2,3 ON TRI)
DICHLORVOS*
DIETHYLSTILBESTROL (DES)
DIFENACOUM
DIMETHOATE*
DIMETHYL SULFOXIDE
DINOSEB
DIOXANE*
DIPHACINONE
DIPHENYLHYDANTOIN
DIQUAT
DISULFOTON
DIURON*
ENDRIN
EPICHLOROHYDRIN*
EPN
EPTC
ETHANOL
ETHYL BENZENE*
ETHYLENE DIBROMIDE
ETHYLENE DICHLORIDE*
ETHYLENE THIOUREA*
ETHYLENE OXIDE*
ETHYLNITROSUREA
EUGENAL
FENBUTATIN OXIDE*
FERBAM*
FLUOMETURON*
FLURPRIMIDOL
FLUTOLANIL
FOLPET*
FORMALDEHYDE*
GLYCEROL FORMAL
REFERENCES
1
1
1
1
5
5
5
2
14
7
6
5
1
5
7
5,6
7
1
1
5
5
5
13
3
1
6
5
14
5
1
3
5
1
5
6
5
1
7
14
8,6
5
14
5
3
5
1
1
6
7
7
8
5
1
Chapter 11.1
11.1-15
Carcinogenic Effects
-------
Table 11.1-2. CHEMICALS ASSOCIATED WITH GENOTOXIC EFFECTS
DATA FROM HUMAN, ANIMAL, AND IN VITRO STUDIES ARE INCLUDED.
CHEMICAL
GLYPHOSATE
HALOXYFOP METHYL
HEXACHLOROBENZENE*
HEXACHLOROPHENE*
LEAD*
LINDANE*
LINURON*
LITHIUM (TRI LISTED AS LITHIUM CARBONATE)*
MALATHION*
MALEICHYDRAZIDE*
MANEB*
MCPA*
MERCURY*
MERCURY COMPOUNDS*
METALDEHYDE
METHIDATHION
METHIMAZOLE
METHOMYL
METHOXYCHLOR*
METHYL ETHYL KETONE (MEK) *
METHYL BROMIDE
METHYL METHACRYLATE*
METHYLCHOLANTHRENE*
METHYLENE CHLORIDE
METOLACHLOR
MEXACARBATE
MIREX
MNNG
MOLINATE*
NABAM*
NAPHTHALENES*
NAPROPAMIDE
NICKEL*
NICOTINES*
NITRATE*
NITRITE
NITROFEN*
NITROGUANIDINE
OXYFLUORFEN*
PARAQUAT*
PARATHION*
PCBS*
PENTACHLORONITROBENZENE
PENTACHLOROPHENOL*
PERCHLOROETHYLENE*
PERMETHRIN*
PHENMEDIPHAM
PHENOL*
O-PHENYLPHENOL*
PHOSMET
PICLORAM*
PIDRIN
PINDONE
PIPERONYL BUTOXIDE*
REFERENCES
7
7
5
1
14
1
5
3
1
5
6
1
7
15
5
1
3
6
5,7
7
1
5
14
5
1,7
1
14
3
1
5
1
7
15
1
7
7
4
7
6
5
1
1
1
1
1
1
1,7
5
1
1
7
1
1
Chapter 11.1
11.1-16
Carcinogenic Effects
-------
Table 11.1-2. CHEMICALS ASSOCIATED WITH GENOTOXIC EFFECTS
DATA FROM HUMAN, ANIMAL, AND IN VITRO STUDIES ARE INCLUDED.
CHEMICAL
PIRIMICARB
PIRIMIPHOS-ETHYL
PIRIMIPHOS-METHYL*
2-PIVALYL-1 ,3 INDANDIONE
PROPACHLOR*
PROPARGITE*
PROPHAM
PROPOXUR*
PROPYLENE OXIDE*
PROPYLENE DICHLORIDE
PYRAZON
PYRIDINE*
RADIONUCLIDES (ALPHA, BETA, & GAMMA EMITTERS)
RESMETHRIN*
RONNEL
ROTENONE
SODIUM CHLORATE
STRYCHNINE*
SULFUR DIOXIDE
TCDD
2,4,5-T
2,4,5-TP
TETRACHLORVINPHOS*
TETRACYCLINES
THIABENDAZOLE*
THIOPHANATE-METHYL 6*
THIRAM*
TOLUENE*
TOXAPHENE*
TRICHLORFON*
1 ,2,4-TRICHLOROBENZENE*
1,1,1-TRICHLOROETHANE*
TRICHLOROETHYLENE*
TRIDIPHANE
TRIFLURALIN*
TRIFORINE*
TRIMETHADONE
URETHANE*
VALPROICACID
VERNAM
WARFARIN*
WHITE PHOSPHORUS*
XYLENE*
ZINEB*
ZIRAM
REFERENCES
1
1
6
1
4
7
1
1
5
5
1
5
3
7
1
1
5
5
5
14
1
1
1
3
5
6
1
5
6
13
7
5
5
7
6
1
3
14
3
7
1
5
5
1
Chapter 11.1
11.1-17
Carcinogenic Effects
-------
Table 11.1-2. CHEMICALS ASSOCIATED WITH GENOTOXIC EFFECTS
DATA FROM HUMAN, ANIMAL, AND IN VITRO STUDIES ARE INCLUDED.
* = Listed in TRI as of August 2000 . When "compounds" of a metal were listed on TRI, all compounds of that
metal in this table are considered to be TRI chemicals
References
1. Cunningham and Hallenbeck (1985).
2. Archer and Livingston (1983).
3. Doulletal.(1980).
4. U.S. EPA 1983).
5. U.S. Department of Health and Human Services, NIOSH (1983).
6. U.S. EPA (1983).
7. IRIS, U.S. EPA online database.
8. Clayton and Clayton (1982).
9. Hayes (1982).
10. Vettorazzi(1979).
11. Council on Environmental Quality (1981).
12. International Agency for Research on Cancer (1979).
13. Chambers and Yarbrough (1982)
14. Keyetal. (1977).
15. Ticeetal. (1996).
16. U.S. Department of Health and Human Services, ATSDR (1993).
I.1.F. Selection of Diseases
The selection of cancers included in this section was based on input from a
variety of sources (as discussed in Chapter I.I). It is anticipated that
additional cancers will be added in the future.
Link to Chapter 1.1
II.1.G. Prognosis
II.1.G.1 General Issues
Cancers vary widely in the course of the diseases. Some types of cancer
have a relatively good prognosis (e.g., non-Hodgkins lymphoma), with
most patients surviving the course of the disease. Others, such as lung
cancer, are much more often fatal. Although generalizations can be made
regarding the "average" prognosis for each cancer, the prognosis for
survival and the length of time over which treatment is required vary
among individuals, even for the same type of cancer. This variability is
observed for all types of cancer. Consequently, the cost estimates
presented in the chapters in this section utilize estimates of the average
survival rates to obtain representative estimates of the medical costs.
Chapter 11.1 11.1-18 Carcinogenic Effects
-------
In addition to the individual variability in survival, patterns in survival for
specific types of cancer are based on patient characteristics. For example,
elderly patients with breast cancer typically have slower tumor growth than
younger patients. There may also be differences related to gender and
race. Although this type of pattern evaluation was beyond the scope of this
handbook, it may have an impact on the cost of medical treatment. (When
information was available in texts reviewed for other purposes, survival
patterns are reported.) Consequently, if an analysis is being conducted on a
homogeneous population (similar age, ethnicity, etc.), then it is advisable to
survey the literature to determine if patterns in disease course or survival
exist that may be relevant to an economic evaluation.
II.1.G.2 Survival Estimates
It is often important to obtain estimates of survival and mortality that are as
accurate as possible because medical costs depend on the duration of
treatment and whether patients are survivors or nonsurvivors. For
example, the value of a statistical life may be used for nonsurvivors,
whereas the summed direct medical and other costs may be used for
survivors. Obtaining accurate estimates of mortality due to a disease is
difficult unless the disease has a very short duration prior to death.6 When
the illness is protracted, as it is for most cancers, it is necessary to evaluate
multiple years of vital statistics for patients to obtain reliable mortality
estimates. For many illnesses there are scant data of this type; however,
the National Cancer Institute (NCI) maintains this data for most cancers
through their Surveillance, Epidemiology, and End Results (SEER)
Program and database, which is available both on line and in documents
published through the Biometry Branch of NCI.
Survival and mortality data are reported in SEER as the Relative Survival
Rate (RSR) for each cancer for each year post-diagnosis. It is usually
reported for the years 1973 to 1993, but for rare cancers may only be
provided for five years post-diagnosis. The stomach cancer chapter in this
handbook contains a detailed discussion of RSRs, and their derivation.
They are statistics based on the survival of cancer patients in relation to the
general population of the same age (hence the term "relative"). For
purposes of determining the percent of patients who are survivors and
nonsurvivors, a complex process that is described in the stomach cancer
chapter can be used to obtain a precise estimate of these percents. For
most uses, however, the RSRs provide a sufficiently close approximation
6 Mortality here refers to the risk that death will occur due to the illness under study, and can be
expressed as a rate (e.g., the percentage of all patients who ultimately die of the disease).
Chapter 11.1 11.1-19 Carcinogenic Effects
-------
of survival and mortality percents to be used without modification. An
example of a simplified approach to survival estimates is provided in the
discussion of bone and liver cancer in this chapter (Section II.l.H.6.4).
Links to Chapter II. 2 for detailed discussion ofRSRs and Section II. l.H. 6.4
II.1.H Typical Cancer Costs
Although cancer costs among individuals vary widely, there are similarities
in the average costs reported for cancers. This section reports and analyzes
some of these costs. The data in this section may be useful when
evaluating a cancer for which cost data aren't available (e.g., a rare cancer
such as bone cancer), or when the specific type of cancer is not known but
a cancer risk is projected (i.e. from an animal study) and a typical value is
sought.
The cost estimate provided in this section is referred to as a "typical
cancer" cost rather than an average cost because it is not a statistical
average of all cancer costs. Rather, the estimate is based on the average
cost calculated from the only long-term study of cancer costs available
(Baker et al., 1989). The costs reported here cannot be represented as
average costs because they are not based on an average of either all cancers
or a random sample of cancers. Cost estimates are based on a group of
cancers that represent the vast majority of cancers that occur in the U.S.,
however, and so offer a reliable estimate of typical costs.
.1.H.1. Source
The most recent source located for lifetime direct medical costs of a
number of cancers is Baker et al. (1989). Baker et al. evaluated the
continuous Medicare history sample file (CMHSF) from the Health Care
Financing Administration. The file contains a random sample of five
percent of all Medicare beneficiaries enrolled in 1974 in the United States,
and includes all Medicare activity from 1974 to 1981. They chose CMHSF
because:
1) it is a nationally representative sample of the Medicare population
(five percent), covering over 1.6 million patients;
2) it is longitudinal, dating from 1974 to 1981; and
3) it captures the majority of medical expenses for each beneficiary.
Five Medicare files are included in the CMHSF, which cover:
1) inpatient hospital stays,
2) skilled nursing facility stays,
3) home health agency charges,
Chapter 11.1 11.1-20 Carcinogenic Effects
-------
4) physicians' services, and
5) outpatient and other medical services.7
Because CMHSF provides no indication of initial diagnosis, Baker et al.
assumed that disease onset occurred when a diagnosis of cancer was listed on
a hospitalization record following a minimum of one year without a cancer
diagnosis. This assumption is reasonable due to the high frequency of
hospitalization associated with these diseases (i.e., individuals diagnosed with
cancer would usually be hospitalized).8
Baker et al. assigned costs associated with each cancer to three post-
diagnostic time periods:
• initial treatment, during the first three months following diagnosis;
• maintenance care, between initial and terminal treatment; and
• terminal treatment during the final six months prior to death.
Initial treatment includes all diagnostic work, and any treatments provided
in the first three months after diagnosis. This treatment may include
radiation therapy, surgery, antineoplastic drugs, etc. Terminal care
includes care provided only in the last six months of life. The care may be
palliative or aggressive in nature and covers the spectrum of all potential
cancer treatments. Maintenance care is defined as that care provided
between the initial care phase and terminal treatment (for nonsurvivors) or
cessation of care (for survivors). Maintenance includes any care provided
after the first three months, excluding terminal care. It may include
surgery, continued aggressive treatment with radiation or chemotherapy,
diagnostics to determine the patient's progress, or be limited to ongoing
monitoring and preventive therapies in cases where the cancer has been
minimized or eliminated.
II.1.H.2. Modifications to the Data
There are a number of limitations to using Medicare data; these were
addressed by Baker et al. with a variety of strategies. As noted in Chapter
I.I, the amount paid for service may differ from the actual medical costs
because many insurers and federal programs either 1) pay only a portion of
7 See Baker et al. (1989 and 1991) for further details.
8 Although there are some exceptions to this generalization, such as non-melanoma type skin
cancers, very few cancers exist for which hospitalization is not required. (Medical costs for non-melanoma
skin cancers are provided in this handbook in Chapter II..6)
Link to Chapter II. 6
Chapter 11.1 11.1-21 Carcinogenic Effects
-------
total costs or 2) pay more than actual costs to underwrite the care
providers' losses due to underpayment from other sources. Baker et al.
used provider charges, rather than Medicare reimbursements (which
represent only a portion of most total charges), thus providing a more
accurate cost estimate.
Link to Chapter 1-1
To improve the accuracy of the cost estimates, Baker et al. included the
costs of coinsurance, deductibles, and other cost components. They made
four adjustments to the cost estimates calculated from the CMHSF:
First, charges were added for skilled nursing facilities (SNFs) not
covered by Medicare by multiplying the "length of stay" at an SNF
(computed from admission and discharge dates) by the average
daily SNF charge.
• Second, the annual Medicare Part B deductible of $60 was added
to the reimbursed charges in the database.
Third, since Medicare pays only 80 percent of physicians' charges,
Baker et al. scaled these reimbursements to 100 percent of
physicians' charges to better reflect costs.
Finally, they inflated all dollar values to 1984 dollars using the
Medical Care component of the Consumer Price Index.
Costs that were not included are outpatient prescription medications and
nursing home care below the skilled level. The Mor et al. (1990) analysis
of the CMHSF data notes that including costs incurred only after the initial
diagnosis omits costs associated with prediagnostic tests and treatment.
Although these costs could be significant, substantial medical treatment
(e.g., tests requiring hospitalization) would also likely result in a diagnosis
and thus be included in Baker et al.'s estimates. This omission may lead to
an underestimate of costs by Baker et al. It is not likely to be substantial
when viewed in the context of the overall costs of treatment.
II.1.H.3. Total Non-incremental Costs of Treatment Phases
Costs were evaluated for initial care and for each year post-diagnosis (i.e.,
the patient may have been in any year post-diagnosis to be included in the
analysis). Patients with an initial diagnosis of cancer prior to or during the
1974 to 1981 time period numbered 125,832. Thirteen types of cancer had
sufficient Medicare beneficiaries (more than 1,000) to be analyzed.
Table II. 1-3 lists the cancer types; the number of patients diagnosed; and
the costs of initial, maintenance, and terminal phases of care in 1984
Chapter 11.1 11.1-22 Carcinogenic Effects
-------
dollars. As Table II. 1-3 shows, there is a relatively small variation in costs
among the cancers. Initial care costs vary by approximately a factor of 2,
continuing care by a factor of approximately 1.8, and terminal care by a
factor of approximately 1.3.
Table 11.1-3 Cancer Types, Number of Study Subjects and Treatment Phase Costs3
Treatment Phase Costs in 1984 Dollars"
Maintenance
Cancer Type (ICD9 Code) # Patients Initial (per year)
Colorectal (153-1 54)
Lung (162)
Prostate (185)
Breast (174)
Bladder (188)
Leukemia (204-208)
Pancreas (157)
Stomach (151)
Uterine corpus (182)
Kidney (189)
Ovary (183)
Uterine cervix (180)
Melanoma (172)
Mean Costs
a. Based on Baker et al.,
b. See text for definitions
19,673
15,381
14,002
12,486
6,843
3,740
3,231
3,228
3,042
1,953
1,605
1,448
1,105
1989. These are non-incremental
of treatment phases.
$14,190
$12,916
$8,112
$7,606
$8,470
$9,068
$14,009
$14,443
$9,260
$12,608
$11,055
$8,979
$6,954
$10,590
and not discounted.
$572
$690
$560
$483
$766
$676
$677
$660
$424
$670
$647
$493
$488
$600
Terminal
$15,776
$15,565
$14,613
$15,136
$18,577
$19,777
$14,790
$16,132
$17,623
$19,302
$18,650
$16,414
$16,194
$16,811
II.1.H.4. Maintenance Phase Costs
One complicating factor in evaluating cancer costs using the Baker et al.
data is determining a value for maintenance care. As Table II. 1-3 shows,
this value is reported as a yearly cost. The duration of maintenance care
for each cancer is not provided by the authors. It should be noted that
maintenance care refers to a time period rather than to the nature of the
care, and may include diagnostic tests, surgery, care during relapses, etc.,
and any other care provided more than three months after diagnosis and
more than six months prior to death due to the cancer (but not due to other
causes). Consequently, costs of maintenance care can vary widely among
patients. It may occur for only a few months or for decades, due to
variations in human disease patterns, disabilities, etc.
Chapter 11.1
11.1-23
Carcinogenic Effects
-------
A variety of strategies can be used to estimate an average maintenance
period. The most precise would be to determine the average length of
maintenance care for survivors and non-survivors of different ages, either
from the literature or through a national survey of medical practitioners for
each type of cancer. Literature has not been located with statistics on
maintenance care durations, and a survey of practitioners would be time-
consuming, expensive, and have considerable uncertainty. It would be
necessary to ascertain both how long a patient would be expected to live
(for both survivors and non-survivors) and how long they would receive
care if they lived for an extended time period. Some patients would not
live as long as the "recommended" period of maintenance care, due to
either death from cancer or from some other cause (i.e., background
mortality). The typical cancer costs estimated in this section are to be used
primarily for rare cancers that lack data; consequently, the necessary
information on mortality and care is not generally available.
Given the unknowns, some simplifying assumptions were made to estimate
maintenance costs for purposes of this "typical cancer" analysis. Rather
than evaluate maintenance care for each cancer separately, an average
duration of care was selected and an average cost calculated. Two
simplifying assumptions were used:
1) It was assumed that the average patient (survivors and nonsurvivors
combined) receives five years of maintenance care post-diagnosis.
2) Terminal costs were assumed to be applicable to 50 percent of
patients (a 50 percent mortality rate); survival actually varies
widely by cancer type.
Many patients will survive beyond the five-year maintenance period
assumed in this analysis and continue to incur cost due to diagnostic tests,
drugs, etc.9 Five years, however, is a reasonable estimate for follow-up
when non-survivors are included. Most cancers have a relatively high
mortality rate, as reflected in the 50 percent mortality rate used as an
assumption in this analysis. As a result of the two assumptions listed
above, the average maintenance cost for five years was added to the initial
costs plus one half of the terminal costs to obtain an estimate of the total
cost.
9 For example, the average age of diagnosis for most cancers is about 70 years. At this age the
average member of the general population has a life expectancy of 14 years. Patients may incur additional
costs over the full course of their lifespan.
Chapter 11.1 11.1-24 Carcinogenic Effects
-------
II.1.H.5. Incremental Costs of Treatment Phases
The costs shown in Table II. 1.3 are all medical costs incurred by a patient
with a cancer diagnosis. Consequently, the costs must be adjusted for
background medical expenses to obtain the incremental costs of cancer
treatment. Baker et al. (1991) provides an estimated background cost per
year of $2,988 (in 1984 dollars). The costs of each treatment phase were
adjusted for background costs, based on the duration of the treatment
phase, and the assumptions regarding maintenance care and survival
discussed above in sections II.1.H.3 and II.1.H.4. For example, the initial
care, which covers a three-month period, has a background medical cost of
$747 ($2,988 per year x 3/12 months). This background cost was
subtracted from the total cost for initial care of $10,590, to obtain an
incremental cost of $9,843.
Link to Sections II. l.H. 3 and II. l.H. 4
The costs have also been updated to 1996 using the Medical Care
Component of the Consumer Price Index (1984:1996 = 2.14 ). The results
are shown in Table II.1-4. The final value in the table, $82,581, is the
undiscounted estimate of the lifetime incremental direct medical costs for a
cancer case.
Depending on how this value is to be used, it may be possible to adjust the
cost components to better reflect the cancer(s) of interest. For example, if
it is known that there is a substantially higher mortality rate (50 percent
was used here), then the terminal cost component could be adjusted
accordingly. Any application should clearly state that this value was based
on numerous assumptions and represents and average of many, but not all,
cancers that occur in the U.S.
Chapter 11.1 11.1-25 Carcinogenic Effects
-------
Table 11.1-4 Incremental Undiscounted Direct Medical Costs for a Typical Cancer
Treatment Phase
Total Medical
Costs (1984$)
Incremental
Medical Costs3
Incremental
Medical Costs
in 1996
Dollars'3
Lifetime
Incremental
Costs0
Initial (3 months)
Maintenance
$10,590.00
$600.46
(per month)
Terminal (6 months) $16,811.46
Total Lifetime Costs in 1996 Dollars'1
$9,843.00
$351.46
(per month)
$15,317.46
$21,064.02
$752.12
(per month)
$21,064.02
$45,127.46
(5 years)
$32,779.36 $16,389.68
$82,581.16
a Adjusted for background medical costs of $2,988 per year (1984$), or $249 per month.
b Adjusted from 1984 to 1996 dollars using the medical care component of the Consumer Price Index
(1984:1996=2.14).
c Five years of maintenance care were assumed and a mortality rate of 50 percent was assumed (i.e., the
terminal care costs were multiplied by .5). See text for discussion.
d These costs can be updated to the current year using inflation factors accessible by clicking below.
Link to inflation factors
II.1.H.6. Application to Specific Cancers
II.I.H.6.1 Method
A more precise approach can be taken for specific cancers, if necessary,
when sufficient statistics are available. The typical costs per treatment
phase discussed above are used, with the maintenance phase and terminal
care evaluated in more detail. The following components were used:
1) initial care — all patients receive initial care, so there are no
modifications made to this phase's costs.
2) maintenance care — the length of the maintenance phase was
estimated based on the survival probability of people of the average
age of diagnosis. This duration of care was used to estimate costs
for this phase.
Two specific cancers were evaluated, bone and liver cancer, in response to
specific requirements of the Agency for an upcoming rule requiring benefits
evaluations. The rule required only the direct medical costs for survivors
of bone and liver cancer because the value of a statistical life (VSL) was to
be used for nonsurvivors. The percentage of survivors and nonsurvivors
for these cancer are discussed in Section II.I.H.6.4 below. Cost
estimations were made using steps above. Two different approaches to
estimating maintenance costs were used to illustrate alternative methods.
Chapter 11.1
11.1-26
Carcinogenic Effects
-------
As noted above, the initial care costs shown in Table II. 1-4 were used
without modification for the costs of this phase. To determine the
estimated cost for the maintenance period care for survivors, the length of
the maintenance period was evaluated using two statistics:
1) the average ages at diagnosis for the two cancers were determined
using the National Cancer Institute's SEER database, as described
in Section II. 1 .G.2 above. The percent of all patients diagnosed in
each age group was used to calculate the mean age at diagnosis
(this is illustrated graphically in the stomach cancer chapter).
Link to Chapter II. 2
Link to Chapter II. 1. G. 2
The average age at diagnosis for bone and liver cancer were
determined to be 69 and 66 years, respectively.
2) The life expectancy of an average individual in the general
population was determined for the two ages of diagnosis listed
above from vital statistics data (accessed in 1998 from National
Center for Health Statistics web site). They were determined to be
14.8 years (rounded to 15 years) for a 69-year-old bone cancer
patient, and 16.7 years (rounded to 17 years) for a 66-year-old liver
cancer patient. It was assumed that the life expectancy of
survivors is the same as that of the general population. In reality,
the treatments for cancer, including radiation, antineoplastic drugs,
etc., have toxic effects that may shorten the lives of cancer patients.
There are not sufficient data on these effects to quantitatively
determine the impact.
II.I.H.6.2 Approach I.
The full term of care was assumed to be ten years. This duration is
reasonable because nonsurvivors were not included, and the life expectancy
at the average ages of diagnosis (66 and 69 years) is considerable (15 to
17 years). Additional care associated with cancer may not be required over
the full remaining life of the individual. The first-year costs consisted of
initial care costs ($21,064) and nine months of maintenance care ($753,12
x 9 = $6,769). (See Table II. 1-4 for incremental costs for each phase of
care.) The remaining nine years of maintenance care were added to initial
costs to obtain the total estimated lifetime direct medical costs.
II.I.H.6.3 Approach II.
The maintenance phase was assumed to be equal to the life expectancy of
the general population at the average age of diagnosis, minus three months
of initial care. As in Approach I, the first year costs consisted of initial care
costs and nine months of maintenance care. The remaining years of life
(i.e., life expectancy at the average age at diagnosis minus the first year of
Chapter 11.1 11.1-27 Carcinogenic Effects
-------
services) were multiplied by the annual maintenance care cost and added
to initial costs to obtain the total estimated lifetime direct medical cost (i.e.,
14 years for bone cancer and 16 years for liver cancer). This approach is
reasonable because patients may require maintenance care over their
remaining lifetime due to the drastic nature of most cancers, and the likely
concurrent effects induced by surgery, radiation, and chemotherapy.
In the absence of accurate long-term treatment information for survivors,
either approach may be used. They are both offered to provide a range of
options for economists and to illustrate the impact of altering assumptions
regarding care on medical cost estimates.
The results obtained using both approaches are shown in Table II. 1-5 using
discount rates of 0, 3, 5, and 7 percent. Bone cancer lifetime medical cost
estimates for survivors range from $109,052 to $154,189 (undiscounted).
Liver cancer lifetime medical cost for survivors range from $109,052 to
$172,240 (undiscounted).
As the results indicate, maintenance care costs are a major portion of total
medical costs for survivors. Differing assumptions regarding the duration
of time over which these costs will occur lead to differences in overall
lifetime cost estimates that are not trivial ($87,000 versus $151,000,
undiscounted). These differences are relatively small, however, when
contrasted with costs associated with the value of a statistical life
(approximately $5,000,000). Although it is important to obtain medical
cost estimates that are as precise as possible, in the case of fatal cancers
(where the VSL is used for some patients) the differences between
Approach I and Approach II do not substantially alter the final results of a
benefits assessment.
As noted above, these costs are for survivors of the diseases only. It is
relatively simple to calculate the costs for nonsurvivors if data are located
on the timing of death. Terminal care costs are listed in Table II. 1-4 and
can be used, with the appropriate maintenance care costs, to estimate direct
medical costs for nonsurvivors. Note that the maintenance costs estimated
for survivors should not be used because they are likely to have a much
longer duration of care than do nonsurvivors.
Chapter 11.1 11.1-28 Carcinogenic Effects
-------
Table 11.1-5 Estimated Incremental Direct Medical Costs for Bone and Liver Cancer Survivors3 (1996 dollars)'5
Type of
Cancer and
Approach a
Life
Age at Expect- Initial Care
Diagnosis ancy Costs
Bone
Approach I
Approach II
Liver
Approach I
Approach II
69 15 $21,064
66 17 21,064
Maintenance Care Costs
Discount
Rates: 0
3
5
7
87,988
133,125
77,042
108,721
70,920
96,108
65,572
85,700
87,988
151,176
77,042
120,138
70,920
104,584
65,572
92,029
Total Lifetime Costs
Discount
Rates: 0
3
5
7
$109,052
$154,189
98,106
129,785
91,988
117,172
86,636
106,764
$109,052
$172,240
98,106
141,202
91,988
125,648
86,636
113,093
a. See text for discussion of approaches.
b. These costs can be updated to the current year using inflation factors accessible by clicking below.
Link to inflation factors
Chapter 11.1
1.1-29
Carcinogenic Effects
-------
II.1.H.6.4 Liver and Bone Cancer Survival Estimates
Because the VSL is sometimes used for nonsurvivors of bone and liver
cancer, it is important to estimate the survival and mortality rates for these
cancer patients. This was done using the RSR data from NCI as discussed
in Section II.1.G.2 and presented in detail in Chapter 112. Because the
cost estimates for these two cancers are not precise (the costs for a
"typical" cancer case were used, as described above), it was determined
that the RSRs provided a reasonable approximation of the survival rate.
Link to Section II. 1.G.2 and Chapter II. 2
Ideally, one would determine the lifetime mortality impacts of these cancers
on patients, which would require a lifetime follow-up. These data are not
available. NCI provides a twenty-one year database (1973-1993) of the
survival experience of liver cancer patients. That database was used in this
analysis and provides a lower-bound estimate of mortality impacts. (It may
slightly underestimate mortality because increased deaths may occur
beyond the twenty-first year post-diagnosis. This increase, however, is not
likely to be substantial.)
Bone cancer, which is rarer and less well studied, is included in the NCI
grouping "bone and joint cancers." Consequently, the survival estimates
are less precise for bone cancer. In addition, the RSR was available only
for five years post-diagnosis for this group of cancers. The actual mortality
rate is very likely to be greater than that observed at five years because
mortality is typically elevated for more than five years post-diagnosis.
Mortality will therefore be underestimated. Unfortunately, the dynamics of
survival and relapse differ considerably among cancers, so it is not possible
to estimate the longer-term survival for bone cancer based on mortality
patterns for other cancers.
The NCI RSR data indicate that the survival rate for liver cancer is
approximately 2.6 percent, indicating a 97.4 percent mortality rate (after 21
years). The bone and joint cancer survival rate is estimated to be 64.3
percent, indicating a 35.7 percent mortality rate (after five years).
II.1.H.7 Conclusions Regarding Typical Cancer Cost Estimates
There is clearly uncertainty when a "typical" cancer approach is used. As
Table II. 1-3 shows, however, there are relatively small differences among
the medical costs of various cancers when contrasted with the uncertainty
in risk estimations, changes in medical care and survival, and uncertainty
associated with other parameters in a benefits assessment. The value in
Table II. 1-4 and approaches described above provide a means to obtain an
estimate of cancer medical costs that may be useful in a benefits evaluation
when limited data are available to support a full and detailed analysis of
medical costs.
Chapter 11.1 11.1-30 Carcinogenic Effects
-------
II.1.1 Issues and Uncertainty in Cancer Medical Cost Estimation
Chapter I.I contains a detailed discussion of numerous sources of
uncertainty in medical cost estimation. Most issues related to estimating
medical costs of specific types of cancer are discussed in the individual
chapters, which also contain a detailed presentation of the methodologies
used to estimate costs. Some issues are common to all cancers and are
briefly discussed in this section. This section also contains the estimated
lifetime medical costs for a "typical" cancer case. In addition, there are
some uncertainties and issues that are particularly problematic for cancer
cost estimation. These issues, discussed below, include new treatments
that are developed (with attendant changes in cost) and concurrent effects
associated with either the occurrence of the cancer or medical treatments
for cancer.
Link to I.l.F: Limitations
II.1.1.1 New Treatments
The costs of new treatments are of particular concern in cancer therapy,
because they may be very expensive, and because what is considered
experimental at one time may soon become the treatment norm. For
example, advances have been made very rapidly during the 1990s in the
treatment of advanced stages of breast cancer. In the past, the prognosis
was poor for advanced stages and the treatments limited. Consequently,
economic valuations might include a value of life estimate rather than a
medical cost estimate as the predominating cost factor. In recent years,
however, more expensive and effective treatments, such as bone marrow
transplants and new pharmaceuticals, have shifted the balance for this
disease somewhat toward improved survival with a corresponding increase
in medical costs. As a result of this dynamic, it is appropriate for an
economic evaluation to include a review of recent literature to determine
whether new treatment approaches with substantially different costs are
being employed for a specific disease.
II.1.1.2 Concurrent Effects
As noted in Section II.l.B. above, concurrent effects are of particular
concern for certain types of illnesses, including cancer and developmental
effects (discussed in the next section). Cancer has a unique ability to
metasticize, leading to multiple types of cancer in an individual. Cancer
may also interfere with the functioning of various organs in the body,
requiring medical attention above and beyond the cancer-limiting
treatments. Finally, the treatments themselves, which often include ionizing
radiation and highly toxic chemotherapeutics, may cause serious illnesses,
including cancers at other sites in the body, impairment of the immune
system, disabilities, and damage to the nervous system or other organs.
Chapter 11.1 11.1-31 Carcinogenic Effects
-------
When data are not available on concurrent effects, and the chapter indicates
that they are likely to occur (as is the case for all cancers), the medical cost
estimates provided will underestimate total medical costs. This discrepancy
should be noted when the costs are used.
All of these concurrent effects may occur during or significantly after the
cancer occurrence. It is beyond the scope of this handbook to include a
discussion of the multiple associated diseases that can arise from a specific
cancer. This information may be important, however, to a comprehensive
economic analysis. Where data are available regarding concurrent effects,
they are described briefly in the cancer chapters. In addition, readers are
urged to consult with the medical and toxicological sources providing the
basic health risk information, in order to obtain additional data on likely
concurrent effects arising from the diseases or their treatment.
Some cancer cost evaluations, such as those based on Baker et al. (i.e.,
lung, breast, liver, kidney, bladder, colorectal, and the "typical" cancer
costs estimated in this chapter in the previous section) include all estimates
of the incremental medical costs associated with a cancer diagnosis. These
values are calculated by summing all medical costs and subtracting the
background costs to obtain incremental costs. Using this approach, costs
are included that may be cancer-related, but not specifically designed to
address cancer. For example, immune-suppressed patients who are
receiving radiation therapy may have greater costs associated with
infectious diseases. The additional required services are due to cancer, but
are not specifically designed to mitigate the cancer. Including these costs
provides a more comprehensive and realistic estimate of the total medical
cost of the disease.
Chapter 11.1 11.1-32 Carcinogenic Effects
-------
CHAPTER 11.2. COST OF STOMACH CANCER
Clicking on the sections below will take you to the relevant text.
II.2.A. Background
II.2. A.I Description
II.2.A.2. Concurrent Effects
II.2.A.3. Causality and Special Susceptibilities
II.2. A.4 Treatment and Services
II.2. A. 5 Prognosis
II.2.B Costs of Medical Treatment and Services for Stomach Cancer Patients
II.2.B.1 Methodology
II.2.B.2 Results
II.2.C. Sensitivity Analyses
II.2.C. 1 The Effect of Age at Diagnosis on Medical Costs
II.2.C.2 The Effect of Race on Medical Costs: An Analysis of African-
American Males
II.2.D. Uncertainties and Limitations
II.2.D. 1. Uncertainties Surrounding Key Inputs to the Analysis
II.2.D.2. Scope of the Analysis
Appendix II.2-A Deriving the Probabilities of Dying of Stomach Cancer and Dying of
Other Causes
Chapter 11.2 11.2-1 Cost of Stomach Cancer
-------
CHAPTER 11.2. COST OF STOMACH CANCER
.2.A. Background
This chapter contains a discussion of the methods used to estimate the
direct medical costs incurred by stomach cancer patients, and the results of
the analysis for these two cost elements. It does not include information on
elements such as indirect medical costs, pain and suffering, lost time of
patients and unpaid caregivers, etc.1 The reader is referred to Chapter I.I
for a discussion of the general methods and cost elements that are relevant
to all benefits estimates and for a discussion of the limitations of estimating
medical costs. In addition, Chapter II. 1 contains information regarding the
special characteristics of cost estimates for cancer.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and II. 1
Link to inflation factors
II.2.A.1 Description
Stomach cancer, also called gastric cancer, refers in most cases to
adenocarcinoma, which comprises 90 to 95 percent of all gastric
malignancies. Other types of stomach cancer include lymphoma,
leiomyosarcoma, carcinoid, adenocanthoma, and squamous cell carcinoma.
Stomach cancer can occur in various anatomical sections of the stomach
and may be limited to the mucosa of the stomach or include large portions
of the stomach and other organs. In addition to these distinctions, there
are two common classification systems, Lauren and Borrmanns, which
provide additional definition to the description of stomach cancers
(Gunderson et al., 1995). This Chapter will consider stomach cancer to
cover all of the above types, in keeping with the recent oncology texts and
the majority of information reviewed for this analysis.
There are approximately 24,000 cases of stomach cancer diagnosed per
year in the United States. Stomach cancer is fatal in over 80 percent of all
cases. It occurs with declining frequency in the United States. The
1 Some of these cost elements, especially pain and suffering may comprise a very large portion of
the cost of cancer. However, it was not feasible to estimate this cost component for this chapter.
Chapter 11.2 11.2-2 Cost of Stomach Cancer
-------
incidence in 1994 was 7.2 cases per 100,000 total population, down from
28.8 in 1973 (incidence represents newly diagnosed cases in a year).
Stomach cancer occurs with much greater frequency among the elderly,
which is typical of most cancers. The average age at diagnosis is
approximately 70 years. Only one percent of stomach cancers are
diagnosed before the age of 35, and the 5th percentile of age at diagnosis is
approximately 45 years. The 95th percentile is over 84 years of age (NCI,
1998). The age corresponding to the 95th percentile cannot be determined
precisely because the National Cancer Institute (NCI) aggregates all
occurrences over the age of 85, and 12 percent of stomach cancers are
diagnosed among people age 85 or greater.
The distribution of the age at diagnosis (onset) of stomach cancer is shown
in Figure II.2-1. The steep incline in the probability of stomach cancer
diagnosis is clear in this diagram, with a peak around 70 years of age. The
data used to generate this figure, as well as the cumulative percents of
stomach cancer in each five-year increment of life are shown in Table II.2-
1. The age-specific incidence data were used in Section B medical cost
calculations. Data on incidence and age at diagnosis were obtained from
NCI's Surveillance, Epidemiology, and End Results (SEER) reports and
tables. These were obtained on line through the NCI web site at:
http://www-seer.ims.nci.nih.gov in January, 1998.
.2.A.2. Concurrent Effects
As with all cancers, stomach cancer may spread to other organs. In
approximately 30 percent of patients, stomach cancer has spread to the
liver at the initial diagnosis (Gunderson et al., 1995). No data were located
indicating that concurrent effects unrelated to stomach cancer or its
treatment were likely to occur with this disease. As noted in Chapter III,
secondary cancers and other adverse health effects may occur due to
treatment and therapy. These can induce added medical costs not
considered in this chapter.
LinktoII.l.B
Chapter 11.2 11.2-3 Cost of Stomach Cancer
-------
Figure II.2-1. Age-specific Incidence of Stomach Cancer
Based on NCI, 1998
0.18 -r
0.16
0.14
0.12
I a'
I"*
0.06
0.04
0.02
0
Jll
0-4 5-9 10- 15- 20- 25- 30-35-40-45- 50- 55- 60- 65- 70- 75- 80- 85+
14 19 24 29 34 39 44 49 54 59 64 69 74 79 84
Age of Onset
Table 11.2-1. Age-specific Incidence of Stomach Cancer
Age Group
0-14
15-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75-79
80-84
85+
Age-specific Rate of
Diagnosis Per
100,000
0
0.3
1.5
2.7
4.4
8.3
14.1
21.7
31.8
45.6
55.9
67.5
79.4
Percent of All
Stomach Cancer
Occurring in Age
Group
0
1
1
2
3
5
7
9
14
17
16
13
12
Cumulative Percent
of Stomach Cancer
0
1
2
4
7
12
19
28
42
59
75
88
100
BasedonNCI, 1998
Chapter II.2
I.2-4
Cost of Stomach Cancer
-------
II.2.A.3. Causality and Special Susceptibilities
Link to Table II. 1-1
Some environmental pollutants are associated with stomach cancer. Radon
is of particular concern because it has been associated with stomach cancer
in numerous studies of occupationally-exposed workers (summary
provided in BEIR VI, 1998). Occupational studies also indicate possible
relationships between stomach cancer and coal mining and rubber and
asbestos manufacturing (Gunderson, et al., 1995).
Table II. 1-1 in Chapter III contains a list of chemicals known or suspected
of causing cancer (as reported in the EPA databases IRIS, HEAST, and
HSDB). Most chemicals in the table were carcinogenic in animal studies.
These studies do not provide organ-specific data because it is not generally
assumed that cancer induction will always occur at the same site in humans
as in animals. Consequently, the chemicals listed in Table II-l may cause
stomach cancer and/or other types of cancer. Evaluation of the likelihood
of this occurrence would require additional research (risk assessment).
Stomach cancer has been associated with dietary factors in some
populations. It occurs at a much higher rate in Japan and some other
countries than in the U.S. This difference is thought to be due to dietary
differences. Studies in the U.S. have shown that people with pernicious
anemia and some types of ulcers and polyps are at greater risk (Gunderson
et al., 1995). Very recent reports (1998) indicate an association between
the bacteria responsible for ulcers and stomach cancer, probably with ulcers
as an intermediate condition. This association is plausible based on current
knowledge of cellular proliferation and its role in cancer induction.
NCI provides age-, sex-, and race-specific data regarding diagnosis of
stomach cancer for the diagnosis years 1990 to 1994. Some statistics from
this data compilation illustrate the higher rates of stomach cancer among
males and African-Americans. The rate among males has been consistently
higher than in females over the years (10.8 versus 4.4 per 100,000
respectively in 1994). It is also consistently higher among African-
Americans than among whites (12.0 versus 6.1 per 100,000 respectively in
1994) (NCI, 1998), and is particularly high among African-American males
(NCI, 1998).2 For example, in the oldest age group of 85 years and up the
general population rate was 79.4 per 100,000, whereas among African-
American males it was 225.2 per 100,000.
The disproportionate occurrence of stomach cancer in males and African-
Americans may be an important consideration when environmental equity
2 The National Cancer Institute (NCI) data presented in this chapter are based on invasive cancer,
which comprises greater than 99 percent of stomach cancers (NCI, 1998).
Chapter 11.2 11.2-5 Cost of Stomach Cancer
-------
Link to Chapter 1.1
issues are evaluated. As noted above, the cause of the increased rate of
stomach cancer among men and African-Americans is not known. It may
be due to dietary, environmental, genetic, or other factors. In the absence
of causal data, it is reasonable to assume that the cause could be genetic
and that the increase in risk would be reflected in higher than average
medical costs and lives lost among males and African-Americans exposed
to a pollutant that causes stomach cancer. Issues related to susceptible
subgroups in benefits assessments are discussed in the Chapter I.I section
titled "Susceptible Subgroups."
The number of stomach cancer patients in the U.S. is relatively small and
the NCI cohort was also relatively small. Consequently the rates in the
younger age ranges (where the numbers are exceedingly small) are
somewhat erratic due to the high variability of estimates based on small
sample sizes. However, the data show a clear progression of increased
incidence over the ages and a consistent pattern of higher risk among
African-Americans than whites and males than females.
Section II.2.D.2 presents the results of a sensitivity analysis for stomach
cancer among African-American males. The analysis uses the higher
incidence rates observed among African-American males to estimate the
medical costs for this subgroup. The analysis raises complicated issues in
evaluating high-risk subgroups. In addition, this analysis evaluates costs
per stomach cancer patient and so does not reflect the additional costs to
society of the likely higher risks that would be incurred in an African-
American versus non-African-American population. These issues must be
dealt with by risk assessors in calculating the number of cases. It may also
be important to evaluate the disproportionate impact of environmental
stomach cancer risk factors on this population subgroup and the benefits
that would result for them from reducing pollution exposures.
II.2.A.4 Treatment and Services
Initial diagnosis may include gastrointestinal (GI) imaging, blood tests,
endoscopy, biopsy, computed tomography (CT) scans and laparoscopy.
Treatment includes surgery, chemotherapy, irradiation, and other general
medical services. Terminal care is eventually provided to most stomach
cancer patients due to the high mortality rate for this cancer. This care may
include a variety of services, including palliative surgery, drug therapy,
home visits, psychological counseling, and other medical care (Gunderson
etal., 1995).
Chapter 11.2 11.2-6 Cost of Stomach Cancer
-------
II.2.A.5 Prognosis
II.2.A.5.1 Background
The overall prognosis for stomach cancer patients is poor, with
approximately 80 to 85 percent of patients dying of the disease within ten
years. Most deaths (over 90 percent) occur in the first four years, and
approximately half of all patients die during the first year (NCI, 1998).
Factors such as tumor size and location, histology, involvement of nodes,
and the spread of cancer to other tissues affect outcome. Numerous new
biochemical and immunological tests are used to provide additional
information on the likely outcome. The importance of early detection and
a confined tumor is evidenced by the greater than 90 percent survival rate
among patients with a tumor confined to the mucosa or submucosa
(Gunderson, 1995). However, few patients in the U.S. are diagnosed with
stomach cancer at this early stage.
II.2.A.5.2 Relative Survival Rates (RSRs)
The NCI SEER data reports were accessed online to obtain information
regarding mortality and survival probabilities and the duration after
diagnosis until death (NCI, 1998).3 These data are presented in this section
because they relate to prognosis. Methods used to convert the NCI
statistics to survival probabilities for stomach cancer patients, used in
Section B to calculate medical costs, are discussed below.
The RSR is the number of observed survivors among patients, divided by
the number of "expected" survivors among persons with the same age and
gender in the general population (observed/expected). The RSR takes into
account that there are competing causes of death that increase with age.
The RSR for stomach cancer patients during the first year post-diagnosis is
46 percent. This value indicates that a person with stomach cancer would
have, on average, a one-year survival probability that is 46 percent of
someone of the same age and gender in the general population.
The RSRs provided by NCI for each year post-diagnosis are averages over
all ages at diagnosis. RSRs are also provided for different ages at
diagnosis, with five-year RSRs ranging from 19.0 to 21.8 percent (based
on 1986-1993 data, NCI, 1998). (Five-year RSRs are the only form in
which NCI provides RSRs that are specific to age at diagnosis.) Because
RSRs are very similar across the ages at diagnosis for stomach cancer,
relative survival rates are discussed without reference to the age at
diagnosis.
An evaluation of the RSRs over the past 20 years indicates that they have
increased overall through 1988, when they stabilized at approximately the
same level through 1993 (the most recent year for which there are data).
The Website is http://www-seer.ims.mci.hin.gov
Chapter 11.2 11.2-7 Cost of Stomach Cancer
-------
Based on this observation, the rates for 1988 through 1993 for the first
through fifth years post-diagnosis were used in this analysis. It was
necessary to use pre-1988 data to estimate survival beyond the fifth year.
These longer-term survival data are from a period when RSRs were slightly
lower. A very low rate of loss from stomach cancer exists after the first
four years (generally less than two percent for the fifth and sixth years
post-diagnosis, and less than one percent thereafter), so differences in
survival in recent years will not have a substantial impact on costs. Ten
years of data were used to estimate survival for the sixth through tenth
years post-diagnosis to increase reliability because the number of stomach
cancer deaths is very small during that period post-diagnosis.
II.2.A.5.3 Derivation of Survival and Mortality: Probabilities
for Stomach Cancer Patients
The RSRs reported by NCI for each year post-diagnosis are based on a
cohort followed over time. They are therefore estimates of the underlying
population RSRs (i.e., the RSRs for the entire population of stomach
cancer patients in the United States). A plot of the average RSRs for n
years post-diagnosis (n = 1, 2, ..., 10) described above (Figure II.2-2)
shows that the estimated RSRs follow a general exponential decay trend.
Chapter 11.2 11.2-8 Cost of Stomach Cancer
-------
Figure II.2-2. Average Relative Survival Rates (RSR) for n Years
Post-Diagnosis
0 £>
0.4
03
0.2
0.1 -
n -
I
<
>
*
RSR
>
I
0 2 4 6 f
I <
«RSR
»
3 10
Rather than use these average estimated RSRs from NCI, which display
some of the "bumpiness" that data often contain, the trend described by
these RSRs was estimated by regression. The model that fit the data best
was of the form:
\n(RSK) = a + b x \n (Years Post-Diagnosis) .
The intercept (a) was estimated to be -0.78613 and the slope (b) was
estimated to be -0.49508. The fit was excellent, with an R2 of 0.985.
Exponentiating the predicted natural logarithms of RSR yielded the
predicted RSRs shown in Table 112-2. A plot of the adjusted RSRs
against years post-diagnosis (Figure II.2-3) shows the generally smooth
trend.
Chapter 11.2
1.2-9
Cost of Stomach Cancer
-------
Table 11.2-2. Average RSRs* and Predicted (Adjusted) RSRs
Years Post-Diagnosis (n)
1
2
3
4
5
6
7
8
9
10
Average RSR for n Years
Post-Diagnosis
0.46
0.32
0.26
0.23
0.22
0.21
0.16
0.16
0.15
0.15
Adjusted (Predicted) RSR for
n Years Post-Diagnosis
0.456
0.323
0.264
0.229
0.205
0.188
0.174
0.163
0.154
0.146
The average RSR for n years post-diagnosis is the average of a set of RSRs reported by NCI (1998)
for n years post-diagnosis as described in the text above.
Figure II.2-3. Adjusted Relative Survival Rates (RSR) for n Years
Post-Diagnosis
Predicted RSR
0 £>
0.4 -
0 3
n 9
0.1 -
n -
I
4
>
* ,
* I <
0246!
I Predicted RSR
>
3 10
The adjusted RSRs shown in Table 11.2-2 were used to derive survival
probabilities for stomach cancer patients for each of the first ten years post-
diagnosis. The RSR, which expresses the survival of patients in relation to
the survival of the general population, can be converted to a survival
probability for stomach cancer patients by using the population survival
rate in the RSR equation:
Chapter 11.2
11.2-10
Cost of Stomach Cancer
-------
„„„ observed survival rate among stomach cancer patients
survival rate among age- and sex-matched cohort in the general population
The RSR is designed to enable the analyst to derive the probability of dying
specifically of the cancer of interest; this probability, however, is
conditional on having not already died of something else. Using the
definition of the RSR, it can be shown that 1 - RSR is the number in the
cohort who were expected to survive but died (and are therefore presumed
to have died of stomach cancer), divided by the number who were expected
to survive. This value effectively takes the original cohort of stomach
cancer patients and first subtracts those who die of other causes, then
calculates the proportion of the remaining subset who die of stomach
cancer specifically. This result is slightly different than the probability of
dying of stomach cancer, given that one is diagnosed with it. This latter
probability has the same numerator, but has as its denominator the entire
original cohort of stomach cancer patients.
To obtain an estimate of the survival rate for stomach cancer patients to
one-year post-diagnosis, the RSR for one-year post-diagnosis and the
background survival rate for one year were used in the above equation.
The survival rate for the general population at the average age at diagnosis
(70 years) during their 70th year (from age 70 to 71) is 0.97326. The
survival rate for stomach cancer patients to one-year post-diagnosis can be
calculated using this value and the RSR of 46.15 for the first year post-
diagnosis reported by NCI (1998) as follows:
46.15 = X
0.97326
(RSR) (background rate)
X = 44.91 (45 percent)
This equation converts the RSR to a survival probability for stomach
cancer patients. It tells us that among all persons diagnosed with stomach
cancer, approximately 45 percent will survive the first year and 55 percent
will die.
Although most stomach cancer patients die of the disease, some die of
other causes. The probability of a stomach cancer patient dying of other
causes is not the same as the probability of someone in the general
population dying of other causes, particularly in the first few years post-
diagnosis, when a stomach cancer patient's probability of dying of stomach
cancer is quite high.4 The probability of a stomach cancer patient dying of
4 This becomes clear in the extreme case in which the probability of dying of an illness is extremely
high. Suppose, for example, that the probability of dying of all causes except for illness X is 0.025 in the
general population. Suppose that in a cohort of patients diagnosed with illness X the probability of dying
Chapter 11.2 11.2-11 Cost of Stomach Cancer
-------
stomach cancer and the probability of a stomach cancer patient dying of
some cause other than stomach cancer in the nth year post-diagnosis, given
survival to the nth year, were each derived from two known probabilities:
1) the probability of a stomach cancer patient surviving through the
nth year post-diagnosis, given survival to the nth year, and
(2) the probability of dying of causes other than stomach cancer in a
matched cohort in the general population.
The derivation is explained in detail in Appendix II.2-A at the end of this
chapter.
Link to Appendix 11.2-A
Each of the known probabilities depends on the number of years post-
diagnosis and (minimally) on age at diagnosis. Consequently, separate
probabilities were calculated for each year post-diagnosis and for each age
at diagnosis considered in the analysis. They are shown for years one
through ten post-diagnosis for someone diagnosed with stomach cancer at
age 70 (the average age at diagnosis for stomach cancer) in Table II.2-3.
Using the probabilities in Table II.2-3, a hypothetical cohort of 100,000
stomach cancer patients diagnosed at age 70 was followed for ten years
after diagnosis. For each year post-diagnosis, the analysis calculated the
number of stomach cancer patients who:
1) survive through the year,
2) die of stomach cancer during the year, and
3) die of some other cause during the year.
from illness X in the first year post-diagnosis is 0.99. If the probability of dying of other causes in this
cohort were the same as in the general population (0.025), then their probability of dying would be greater
than 1.0.
Chapter 11.2 11.2-12 Cost of Stomach Cancer
-------
Table 11.2-3. Probabilities of Survival and Mortality for a Stomach Cancer Patient Diagnosed at Age 70
Years post-
diagnosis (n)
(1)
0
1
2
3
4
5
6
7
8
9
10
A Matched Cohort in the General
Population
Probability of
surviving n years
(2)
1.000
0.973
0.945
0.915
0.884
0.852
0.817
0.782
0.745
0.707
0.667
Probability of dying
in nth year of
causes other than
stomach cancer,
given survival to
the nth yeara
(3)
—
0.026
0.029
0.031
0.034
0.037
0.040
0.043
0.047
0.051
0.056
A Cohort of 100,000 Stomach Cancer Patients
(Adjusted)
Relative
Survival Rateb
(4)
—
0.456
0.323
0.264
0.229
0.205
0.188
0.174
0.163
0.154
0.146
Probability of
surviving n
years post-
diagnosis
((2)*(4))
(5)
—
0.4434
0.3055
0.2421
0.2028
0.1749
0.1534
0.1359
0.1212
0.1085
0.0972
Number
surviving
through the
nth year
(100,000*(5))
(6)
100,000
44,342
30,551
24,210
20,282
17,490
15,340
13,594
12,122
10,846
9,717
Probability of
surviving
through the
nth year,
given survival
to the nth year
( Wn/Wn.! )
(7)
—
0.4434
0.6890
0.7925
0.8378
0.8624
0.8771
0.8862
0.8917
0.8948
0.8959
Probability of
dying of
stomach
cancer in the
nth year,
given survival
to the nth
year"
(8)
—
0.5373
0.2865
0.1793
0.1307
0.1029
0.0849
0.0722
0.0628
0.0555
0.0497
Probability of
dying of other
causes in the
nth year,
given survival
to the nth
year0
(9)
—
0.0193
0.0245
0.0283
0.0315
0.0347
0.0380
0.0416
0.0455
0.0497
0.0544
aThe probabilities in the general population of dying from stomach cancer are 0.000256 in the 70-74 year age group, and 0.000348 in the 75-79 year age
group. The probabilities in column (3) were derived by subtracting these probabilities from the corresponding probabilities of dying from any cause in the
nth year given survival to the nth year.
"From Table II. 2-2.
c See Appendix to this chapter for derivation of these probabilities.
Chapter 11.2
1.2-13
Cost of Stomach Cancer
-------
From these numbers, the probabilities of:
1) surviving through the nth year,
2) dying of stomach cancer during the nth year, and
3) dying of other causes during the wth year
were derived. The probability of dying of stomach cancer in the wth year
post-diagnosis, for example, is calculated as the number of stomach cancer
patients who die of stomach cancer in the wth year post-diagnosis divided
by the original number in the cohort (100,000). The analysis is shown in
Table II.2-4. The probabilities of a stomach cancer patient (age 70 at
diagnosis) surviving through each year post-diagnosis, dying of stomach
cancer during each year, and dying of other causes during each year are
given in columns (7), (8) , and (9), respectively, of Table II.2-4. The
probabilities in columns (7), (8), and (9) of Table II.2-4 are used in Section
II.2.B to calculate the expected medical costs of stomach cancer patients
diagnosed at age 70.
Chapter 11.2 11.2-14 Cost of Stomach Cancer
-------
Table 11.2-4. Following a Cohort of 100,000 Stomach Cancer Patients (Diagnosed at Age 70) Over Ten Years: Derivation of
Probabilities of Survival and Mortality
Years Post-
Diagnosis (n)
(1)
0
1
2
3
4
5
6
7
8
9
10
Probability of
Surviving n years
post -diagnosis
(column (5) of
Table 11.2-3)
(2)
—
0.4434
0.3055
0.2421
0.2028
0.1749
0.1534
0.1359
0.1212
0.1085
0.0972
Number
Surviving
through nth year
(100,000*(2))
(3)
100,000
44,342
30,551
24,210
20,282
17,490
15,340
13,594
12,122
10,846
9,717
Number Dying in
nth year
«3)M-(3)n)
(4)
—
55,658
13,792
6,341
3,928
2,792
2,150
1,746
1,472
1,276
1,129
Number dying of
stomach cancer
in the nth year
((3)n.1*(8)n of Table
II.2-3 )
(5)
—
53,730
12,704
5,476
3,165
2,087
1,485
1,108
854
673
540
Number dying of
other causes in
the nth year
((3)n.1*(9)n of Table
II.2-3)
(6)
—
1,928
1,088
864
763
705
665
638
618
602
590
Probabilities
Probability of
surviving through
the nth year
((2)n)
(7)
—
0.4434
0.3055
0.2421
0.2028
0.1749
0.1534
0.1359
0.1212
0.1085
0.0972
Probability of
dying of stomach
cancer in the nth
year
((5)/1 00,000)
(8)
0.5373
0.1270
0.0548
0.0317
0.0209
0.0149
0.0111
0.0085
0.0067
0.0054
Probability of
dying of other
causes in the nth
year
((6)/1 00,000)
(9)
0.0193
0.0109
0.0086
0.0076
0.0071
0.0067
0.0064
0.0062
0.0060
0.0059
Chapter 11.2
1.2-15
Cost of Stomach Cancer
-------
11.2.B Costs of Medical Treatment and Services for Stomach
Cancer Patients
II.2.B.1 Methodology
II.2.B.1.1 Overview
Treatment of stomach cancer may occur over a brief or extended period of
time, and costs may be limited or substantial. There is no typical case
because of individual differences in the stage of cancer at diagnosis,
multiple treatment options, patient health and age, and other factors;
however, average costs can be calculated. Stomach cancer has a relatively
high mortality rate, as discussed in Section A. Approximately 82 percent
of people with stomach cancer eventually die of the disease. As discussed
in Chapter I.I of this handbook, the medical costs of those who die of the
disease are usually very different than for those who survive. Therefore,
although the focus of this chapter is on the costs incurred by the average
stomach cancer patient, survivors and nonsurvivors of stomach cancer are
considered as separate groups for purposes of this analysis.
Link to I. LD. 2
A data search was conducted for information regarding medical costs
associated with stomach cancer. In addition to a literature search, most
federal agencies dealing with cancer and medical costs were contacted for
information and the various federal databases were discussed with senior
staff at these agencies. Very recent cost data were not located. However,
current (1994) cancer data were obtained regarding incidence and survival,
which were used in the cost calculations described below. The cost
estimates presented in this are based primarily on the work of Baker et al.
(1989) and on two sources of statistical data: the National Cancer Institute
(1998) and Vital Statistics of the United States, 1993 (NCHS, 1997).
These data were evaluated and used to calculate appropriate estimates of
the direct medical costs due to stomach cancer.
\\.2.B. 1.2 Medical Cost Data
II.2.B.1.2.1 Sources
Medical cost data would ideally be obtained on current medical
expenditures for a specific illness. Although data files are maintained by
public and private sector sources, they are not readily available. In
addition, to obtain reliable cost estimates it is necessary to evaluate very
large databases of charges from a variety of sources, a method impractical
for the development of this chapter. A review of the medical economics
literature in 1997 did not identify very recent sources of cost estimates for
stomach cancer. Baker et al. (1989) was previously used for other
chapters of this handbook and has been used as the basis for the cost
estimates in this chapter. Based on the 1997 review of the literature
Chapter 11.2 11.2-16 Cost of Stomach Cancer
-------
carried out for the development of this chapter, there do not appear to be
new treatment methods for stomach cancer that substantially alter either
the medical costs or the survival rates. Consequently, the cost estimates
presented in this chapter may be considered appropriate under most
circumstances (e.g., regional costs may vary).
II.2.B.1.2.2 Baker et al.'s Cost Estimation Method
Baker et al. (1989) used the Continuous Medicare History Sample File
(CMHSF) to estimate the per-patient average lifetime medical cost of
treating stomach cancer based on data files from 1974 to 1981. They
chose CMHSF because:
1) it is a nationally representative sample of the Medicare population
(five percent), covering over 1.6 million patients;
2) it is longitudinal, dating from 1974 to 1981; and
3) it captures the majority of medical expenses for each beneficiary.
Five Medicare files are included in the CMHSF, which cover:
1) inpatient hospital stays,
2) skilled nursing facility stays,
3) home health agency charges,
4) physicians' services, and
5) outpatient and other medical services.5
Costs that were not included are outpatient prescription medications and
nursing home care below the skilled level.
Because CMHSF provides no indication of initial diagnosis, Baker et al.
assumed that disease onset occurred when a diagnosis of stomach cancer
was listed on a hospitalization record following a minimum of one year
without a stomach cancer diagnosis. This assumption is reasonable due to
the high frequency of hospitalization associated with the disease (i.e.
individuals diagnosed with stomach cancer would be hospitalized). Only
patients with an initial diagnosis during the years covered by the database
(1974-1981) were included.
Costs associated with stomach cancer were assigned to three post-
diagnostic time periods:
initial treatment, during the first three months following diagnosis;
• maintenance care, between initial and terminal treatment; and
• terminal treatment during the final six months prior to death.
5 See Baker et al. (1989 and 1991) for further details. Baker et al. (1991) contains additional
descriptive data regarding the database and methods used for the cost analysis; however, it does not contain
cost data for stomach cancer.
Chapter 11.2 11.2-17 Cost of Stomach Cancer
-------
Link to Chapter 1.1
As noted in Chapter 1.1, the amount paid for service may differ from the
actual medical costs because many insurers and federal programs either 1)
pay only a portion of total costs or 2) pay more than actual costs to
underwrite the care providers' losses due to underpayment from other
sources. Baker et al. used provider charges, rather than Medicare
reimbursements (which represent only a portion of most total charges),
thus providing a more accurate cost estimate.
To improve the accuracy of the cost estimates, Baker et al. included cost
data on coinsurance, deductibles, and other cost components. They made
four adjustments to the cost estimates calculated from the CMHSF. First,
charges were added for skilled nursing facilities (SNFs) not covered by
Medicare by multiplying the "length of stay" at an SNF (computed from
admission and discharge dates) by the average daily SNF charge. Second,
the annual Medicare Part B deductible of $60 was added to the reimbursed
charges in the database. Third, since Medicare pays only 80 percent of
physicians' charges, Baker et al. scaled these reimbursements to 100
percent of physicians' charges to better reflect social costs. Finally, they
inflated all dollar values to 1984 dollars using the Medical Care component
of the Consumer Price Index.
II.2.B.1.2.3 Cost Estimates by Treatment Period
Medical costs associated with the initial, maintenance, and terminal cancer
care treatment periods were itemized in Baker et al., 1989 and are shown in
Table II.2-5. To estimate the incremental costs, a co-morbidity cost of
$2,988 per year from Baker et al. (1991) was used in this analysis. To
account for costs of other medical services anticipated to occur while the
patient was receiving cancer treatment (i.e., co-morbidh^ackground
costs), the co-morbidity cost was pro-rated for this analysis using the
specified durations for the initial (three-month) and terminal (six-month)
treatment periods. These costs are listed in Table II.2-5 with the
incremental costs calculated for the three treatment periods. Total costs are
reported for the initial and terminal care periods. Annual costs for the
maintenance period are shown and are further discussed in the "Lifetime
Costs" section below. Using the Medical Care component of the
Consumer Price Index (CPI-U), all costs are inflated to 1996 dollars
(1984:1996 = 2.14).
Chapter 11.2 11.2-18 Cost of Stomach Cancer
-------
Table 11.2-5. Average Per Patient Costs for the Three Periods of Treatment for Stomach Cancer
in 1996 dollars
Treatment Period
Initial
(3 months)
Maintenance (per year)
Terminal
(6 months)
Cost3
$30,908
$16,949
$34,522
Co-morbidity Charge13
$1,599
$6,394
$3,197
Incremental Cancer
Treatment Cost
$29,309
$10,554
$31,325
a. From Baker et al. (1989 and 1991) adjusted using the Medical Care component of the Consumer Price Index
(CPI-U) 1984:1996 = 2.14.
b. Annual co-morbidity charges are $6,394 and were pro-rated for the duration of the treatment period.
II.2.B. 1.3 Calculation of Lifetime Cost Estimates for
Stomach Cancer Patients
Although Baker et al. provide useful cost estimates for the three treatment
periods, they do not provide information on two critical aspects of medical
costs:
1) costs for survivors versus nonsurvivors of stomach cancer. These
may differ substantially. For example, survivors would not have
terminal care costs and may receive maintenance services for an
extended time period.
2) estimates of the duration of the maintenance periods.
Data regarding age at diagnosis of stomach cancer were obtained from NCI
(1998). Survival and mortality probabilities for each year post-diagnosis
were derived from relative survival rates obtained from NCI (1998), as
discussed in Section II.2.A.5.3. This information was used to address
many time-related medical cost issues. For some aspects of the analysis,
however, detailed information was not available, and average values have
been used as a reasonable approximation (e.g., a ten-year maintenance
period was assumed for survivors of stomach cancer). When average
values or other assumptions are used in this analysis, they are so noted.
Link to Section II. 2.A. 5.3
As previously noted, there are not substantial differences in survival related
to age at diagnosis, and NCI does not provide age-specific relative survival
rates for each year post-diagnosis. Consequently, it was assumed for this
analysis that the relative survival rates for stomach cancer were the same
for all ages. The survival and mortality probabilities for stomach cancer
patients, which are incorporated into calculations of expected medical
costs, are based on this assumption.
Chapter 11.2
11.2-19
Cost of Stomach Cancer
-------
There is also a lack of information on age-specific medical costs incurred
by stomach cancer patients during the three treatment periods defined by
Baker et al. Because of this, any differences in expected medical costs for
stomach cancer patients diagnosed at different ages, based on current
information, would differ only because of differences in survival and
mortality probabilities. The discussion here focuses on the costs incurred
by stomach cancer patients diagnosed at age 70 (the average age at
diagnosis); a summary table of expected medical costs for patients
diagnosed at several different ages is presented in the "Results" section for
comparison.
The analysis assumes that death always occurs midyear. All stomach
cancer patients are therefore assumed to incur the costs of initial treatment
during the first three months of the illness. The costs incurred after that
during the first year depend on whether the patient (1) survives through the
year, (2) dies of stomach cancer during the year, or (3) dies of some other
cause during the year. Patients who survive through the year incur the
costs of initial treatment ($29,309) during the first three months, and then
incur nine months' worth of maintenance care costs (0.75 x $10,554 =
$7,916) during the remainder of the year. The total cost incurred during
the first year by those patients who survive the year is therefore $29,309 +
$7,916 = $37,225. Stomach cancer patients who die of stomach cancer
during the first year incur the initial treatment cost and then incur terminal
care costs for the remaining three months of their lives (because those who
die are assumed to die midyear). Total costs during the first year post-
diagnosis in this case are therefore $29,309 + 0.5 x $31,325 = $44,972.
Finally, the small percentage of stomach cancer patients who die of causes
other than stomach cancer during the first year post-diagnosis incur the
initial treatment costs and then incur three months' worth of maintenance
care costs. Total first-year costs for these patients are therefore $29,309 +
0.25 x $10,554 = $31,948.
The expected medical costs for stomach cancer patients during the first
year post-diagnosis, then, may be expressed as:
Expected First-Year Cost: initial treatment costs +
[maintenance care costs for nine months * probability of
survival through first year + terminal care costs for three
months * probability of dying of stomach cancer during first
year + maintenance care costs for three months x probability
of dying of other causes during the first year]
For each subsequent year, costs consist entirely of maintenance care costs
for those who survive the year. For those who do not survive the year,
costs depend on whether death was due to stomach cancer or other causes.
For those who die of stomach cancer during the wth year, costs incurred
Chapter 11.2 11.2-20 Cost of Stomach Cancer
-------
that year consist of six months of terminal care costs, or $31,325. For
those who die of other causes during the nth year, there are six months of
maintenance care costs, or 0.5 x $10,554 = $5,277.
The expected medical costs for stomach cancer patients during the wth year
post-diagnosis, for «>1, then, may be expressed as:
Expected nth Year (n>l) Cost: [maintenance care cost for one
year x probability of survival through «th year + terminal care
cost for six months x probability of dying of stomach cancer
during the «th year + maintenance care cost for six months x
probability of dying of other causes during the nth year]
Expected Lifetime cost =
Expected First-Year cost + the sum of the (discounted)
expected subsequent-year costs
The first year of treatment is calculated differently from other years
because the first three months of that year are spent in "initial" treatment
and the costs for that period of intensive medical care and surgery are
calculated separately. The mathematical equation for the expected lifetime
medical costs incurred by a stomach cancer patient over a ten-year period
is:
$29,309 + ($10,554 x 0.75 x pSi) + ($10,554 x 0.25 x pm^) + ($31,325 x 0.5
$10,554 ,
X '. ) +
(1
(pm; x $5'277 ) + (pm;
y /i \y—\ y
$31,325
where: y
ps
pms
pm
the year post-diagnosis
the probability of surviving through the year,
the probability of dying of stomach cancer during
the year
the probability of dying from other causes during the
year,
the discount rate
Example: Expected first-year medical costs of a stomach cancer patient
diagnosed at age 70
As noted above, all stomach cancer patients incur an initial treatment cost
of $29,309. Those who survive through the year (44.3 percent of those
Chapter 11.2
11.2-21
Cost of Stomach Cancer
-------
diagnosed at age 70) also incur maintenance care costs for the remaining
three quarters of the year. The total first-year costs of those who survive
the year are:
Initial treatment: $29,309
Maintenance treatment: $7,916 (.75 x $10,554)
Total First-Year Cost $37,225
More than half of stomach cancer patients (53.7 percent of those diagnosed
at age 70) will die of stomach cancer during the first year. Those who do
will incur the initial treatment costs plus half of the terminal care costs.
The total first year costs of those who die of stomach cancer during the
year are:
Initial treatment: $ 29,309
Terminal care: $15,663 (.50 x $31,325)
Total First-Year Cost $44,972
Finally, a few stomach cancer patients (1.9 percent of those diagnosed at
age 70) will die of competing illnesses during the first year. Because those
who die of causes other than stomach cancer are assumed to die at the
midpoint of the year, costs during the first half of the year are assumed to
consist of the initial treatment costs for three months, plus three months of
maintenance care costs as follows:
Initial treatment: $29,309
Maintenance treatment: $2,639 (.25 x $10,554)
Total First-Year Cost $31,948
The expected first year medical cost incurred by a stomach cancer patient
diagnosed at age 70 is just a weighted average of the costs of those who
survive the first year, those who die of stomach cancer during the first year,
and those who die of other causes during the first year, where the weights
are the probabilities of each of these occurrences (see Table II.2-4):
$37,225 x 0.443 + $44,972 x 0.537 + $31,948 x 0.019 = $41,286
The weighted average medical cost calculations were carried out for ten
years and expected costs were summed over all years from diagnosis to
year ten. This was assumed to be a reasonable period over which
additional medical costs associated with stomach cancer (i.e. maintenance
care costs) would be incurred by stomach cancer patients. In reality, there
may be follow-up care and continued testing over a longer period;
Chapter 11.2 11.2-22 Cost of Stomach Cancer
-------
II.2.B.2 Results
however, no data were available regarding those costs. They would
certainly be less than $10,554 per year.
II.2.B.2.1 Lifetime Cost Estimates for Survivors and
Nonsurvivors Combined
The cost estimates for each year post-diagnosis and the estimate of
expected total cost for a ten-year period are shown in Table 11.2-6 for
stomach cancer patients whose age of onset is 70 (the average age at
diagnosis for this cancer). The discounted results are shown in the
"Results" section which follows. The survival and mortality probabilities
necessary for the calculations are shown in columns (2), (3) and (4) (and
were taken from Table II.2-4). The cost components used in the
calculations are shown in columns (5), (6), and (7).
\\.2.B.2.2 Lifetime Cost Estimates for Stomach Cancer
Survivors and Nonsurvivors Separately
II.2B.2.2.1 Overview
There are differences in medical services provided to stomach cancer
patients who survive the disease (survivors) versus those who die of the
disease (nonsurvivors). Based on cost estimates by Baker et al. (1989),
terminal care is provided for approximately six months to terminally ill
cancer patients. The costs to nonsurvivors for this care ($31,325) is
considerably higher than costs for survivors who receive maintenance care
for the same period of time ($5,277).6 EPA may use the value of a
statistical life for nonsurvivors, and thus separate costs for survivors and
nonsurvivors were calculated. The method to calculate costs for all
patients described in Section II.2.B.1.3 uses the unconditional probabilities
of survival and mortality given in Table II.2-4. The method used to
calculate costs for survivors and nonsurvivors separately requires the
conditional probabilities of survival and mortality in each group — that is,
the probabilities conditional on being a stomach cancer survivor or being a
stomach cancer nonsurvivor.
6 Nonsurvivors include only those who die of stomach cancer and do NOT include those who die of
any other causes.
Chapter 11.2 11.2-23 Cost of Stomach Cancer
-------
Table 11.2-6. Expected Costs of Medical Services (in 1996$) for Stomach Cancer Patients (Age of Onset = 70)
Years Post-
Diagnosis (n)
(1)
1b
2
3
4
5
6
7
8
9
10
Probabilities9:
of surviving
through the nth
year
(2)
0.4434
0.3055
0.2421
0.2028
0.1749
0.1534
0.1359
0.1212
0.1085
0.0972
of dying of
stomach cancer
in the nth year
(3)
0.5373
0.1270
0.0548
0.0317
0.0209
0.0149
0.0111
0.0085
0.0067
0.0054
of dying of
other causes in
the nth year
(4)
0.0193
0.0109
0.0086
0.0076
0.0071
0.0067
0.0064
0.0062
0.0060
0.0059
Medical Costs in the nth Year (undiscounted)
if survive
through the nth
year
(5)
$37,225
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
if die of
stomach cancer
in the nth year
(6)
$44,972
$31 ,325
$31 ,325
$31 ,325
$31 ,325
$31 ,325
$31 ,325
$31 ,325
$31 ,325
$31 ,325
if die of other
causes in the
nth year
(7)
$31 ,948
$5,277
$5,277
$5,277
$5,277
$5,277
$5,277
$5,277
$5,277
$5,277
Expected Total Cost Through the 1 0th Year Post-Diagnosis for a Stomach Cancer Patient Diagnosed at Age 70:
Expected Medical Costs for
the nth Year Post-Diagnosis
( (2)x(5)+(3)x(6)+(4)x(7))
$41 ,286
$7,261
$4,316
$3,172
$2,537
$2,119
$1,815
$1 ,579
$1 ,387
$1 ,226
$66,700
a. The probabilities listed in this table are from Table II.2-4. The costs are listed in Table II. 2-5.
b. First year costs include the charge for "initial" therapy ($29,309). The duration of maintenance care is adjusted accordingly (see text for discussion).
Chapter 11.2
1.2-24
Cost of Stomach Cancer
-------
Link to Table II. 2-4
Probabilities of survival were calculated using data shown in Table II.2-4.
Summing the entries in column (5) of Table II.2-4, 81,822 of the 100,000
stomach cancer patients diagnosed at age 70 (or about 82 percent) die of
stomach cancer within ten years. The remainder (100,000 - 81,822 =
18,178) are survivors of stomach cancer. The probabilities of a stomach
cancer patient diagnosed at age 70 surviving stomach cancer and dying of
stomach cancer are therefore estimated to be 0.182 and 0.818 (about 0.18
and 0.82), respectively.
The conditional probability of a stomach cancer nonsurvivor dying in the
nth year is just the number of stomach cancer patients who die of stomach
cancer during the nth year (from column (5) of Table II.2-4) divided by the
total number of stomach cancer nonsurvivors. For example, the
conditional probability of a stomach cancer nonsurvivor dying during the
first year post-diagnosis is 53,730/81,822 = 0.657. Similarly, the
conditional probability of a stomach cancer survivor dying (of other causes)
in the nth year is just the number of stomach cancer patients who die of
other causes during the nth year (from column (6) of Table II.2-4) divided
by the total number of stomach cancer survivors. For example, the
conditional probability of a stomach cancer survivor dying of some other
cause during the first year post-diagnosis is 1,928/18,178 = 0.106. The
conditional probabilities of survival and mortality for survivors and
nonsurvivors of stomach cancer are given in Table II.2-7.
Table 11.2-7. Conditional Probabilities of Survival and Mortality for Survivors and Nonsurvivors
of Stomach Cancer (Age of Onset = 70)
Years Post-
Diagnosis
(n)
1
2
3
4
5
6
7
8
9
10
Stomach Cancer Survivors
Conditional probability of:
Surviving
through the nth
year
0.894
0.834
0.787
0.745
0.706
0.669
0.634
0.600
0.567
0.535
Dying of some
other cause
during the nth
year
0.106
0.060
0.048
0.042
0.039
0.037
0.035
0.034
0.033
0.032
Stomach Cancer Nonsurvivors
Conditional probability of:
Surviving
through the nth
year
0.343
0.188
0.121
0.082
0.057
0.039
0.025
0.015
0.007
0.000
Dying of
stomach cancer
during the nth
year
0.657
0.155
0.067
0.039
0.026
0.018
0.014
0.010
0.008
0.007
Chapter II.2
II.2-25
Cost of Stomach Cancer
-------
Link to Table II. 2-5
\\.2.B.2.2.2. Lifetime Cost Estimates for Stomach Cancer
Survivors
As shown in Table II.2-5, all stomach cancer patients will incur initial
treatment costs ($29,309) during the first three months of the first year
post-diagnosis. If they survive the first year, they will also incur nine
months worth of maintenance care costs (0.75 x $10,554 = $7,916) that
first year. The total cost incurred during the first year by those stomach
cancer survivors who survive the first year is therefore $29,309 + $7,916 =
$37,225.
Stomach cancer survivors who die of some other cause during the first year
incur the initial treatment costs during the first three months and then incur
three months (25 percent) of maintenance care costs (because they are
assumed to die midyear), for a total cost of $29,309 + 0.25 x $10,554 =
$31,948.
The expected medical costs for stomach cancer survivors during the first
year post-diagnosis may therefore be expressed as:
Expected First-Year Cost: initial treatment costs +
[maintenance care costs for nine months x probability of
survival through first year + maintenance care costs for three
months x probability of dying of other causes during the first
year]
For each subsequent year, costs consist entirely of maintenance care costs
for those who survive the year. For those who die of other causes during
the year, there are six months of maintenance care costs, or 0.5 x $10,554
= $5,277.
The expected medical costs for stomach cancer survivors during the nth
year post-diagnosis, for «>1, then, may be expressed as:
Expected nth Year (n>l) Cost: [maintenance care cost for one
year x probability of survival through nth year + maintenance
care cost for six months x probability of dying of other causes
during the nth year]
Expected Lifetime cost =
Expected first year cost + the sum of the (discounted) expected
subsequent-year costs
The calculations above use the conditional probabilities of patients who do
not die of stomach cancer, shown in Table II.2-7.
Chapter 11.2
11.2-26
Cost of Stomach Cancer
-------
Using the initial, maintenance, and terminal care costs from Table II.2-5,
the mathematical equation for the lifetime costs incurred by stomach cancer
survivors is:
$29,309 + pmf x 0.25 ($10,554) + ps^ x .75 x $10,554
s $10,554 s $5,277
10
+ E
y = 2
v-\
v-l
where: y
pm
the year post-diagnosis
the conditional probability of survival for that year,
conditional on being a survivor of stomach cancer
the conditional probability of mortality for that year,
conditional on being a survivor of stomach cancer
the discount rate.
The expected medical costs for stomach cancer survivors for each year
post-diagnosis, as well as the expected total medical costs over ten years
post-diagnosis are shown in Table II.2-8.
II.2.B.2.2.3 Lifetime Cost Estimates for Stomach Cancer
Nonsurvivors
Nonsurvivors of stomach cancer will incur initial, maintenance, and
terminal costs. Their lifetime medical costs associated with the disease can
be calculated from the costs per treatment period shown in Table II.2-5 and
the conditional probabilities for nonsurvivors of stomach cancer shown in
Table 11.2-7.
Link to Table II.2-5 andII.2-7
As Table 11.2-7 indicates, most stomach cancer patients who die of
stomach cancer die in the first few years post-diagnosis. About 80 percent
die in the first two years. Deaths from stomach cancer after the first four
years are minimal. As with stomach cancer survivors, medical costs for
nonsurvivors each year post-diagnosis were calculated as a weighted
average of the costs incurred by those who survive the year and those who
die (of stomach cancer) during the year.
Chapter 11.2
11.2-27
Cost of Stomach Cancer
-------
Table 11.2-8. Expected Costs of Medical Services (in 1996$) for Survivors of Stomach Cancer (Age of Onset = 70)
Years
Post-
Diag-
nosis
(n)
1d
2
3
4
5
6
7
8
9
10
Medical Costs Through the 10th Year Post-diagnosis9 (undiscounted)
Medical Cost if Survive
Through the nth Year
(1)
$37,225
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
Conditional Probability of
Survival Through the nth
Year"
(2)
0.894
0.834
0.787
0.745
0.700
0.669
0.634
0.600
0.567
0.535
Medical Cost if Die of
other Causes in the nth
Year
(3)
$31 ,948
$5,277
$5,277
$5,277
$5,277
$5,277
$5,277
$5,277
$5,277
$5,277
Conditional Probability of
Mortality in the nth Year"
(4)
0.106
0.060
0.048
0.042
0.039
0.037
0.035
0.034
0.033
0.032
Expected Total (Undiscounted) Cost Through the 1 0th Year Post-Diagnosis:
Total Cost Based on Weighted
Average0
= (1)x(2) + (3)x(4)
$36,666
$9,119
$8,553
$8,080
$7,654
$7,257
$6,878
$6,514
$6,159
$5,813
$102,693
a. Costs are based on data reported in Table II. 2-5, adapted from Baker et al., 1989.
b. Probabilities of survival and mortality, taken from Table II. 2-7, are conditional on surviving stomach cancer.
c. Weighted average of the costs incurred by survivors who survive the year and the costs incurred by survivors who die of other causes during the year. Weighting is based on the
conditional probabilities listed.
d. Costs during the first year include a charge for "initial" therapy ($29,309), and the duration of maintenance or terminal care is adjusted accordingly. See text for discussion.
Chapter II.2
I.2-28
Cost of Stomach Cancer
-------
It was assumed that those who die during a year receive six months of care
(as was done for the survivors above). It was also assumed that terminal
care lasting six months would be provided to all nonsurvivors. Therefore,
unless death occurred during the first year, when initial care was assumed
to occur, the care costs which were assigned to the last year of life were
terminal costs. If death occurred during the first year post-diagnosis, it
was assumed that initial care and three months (half of the total) of terminal
care were provided.
The general description of medical costs for nonsurvivors may be
expressed as:
Expected cost for each year post-diagnosis =
Expected First-Year Cost: [initial costs + half the terminal
costs] x probability of mortality during the first year + [initial
costs + maintenance care costs for nine months] x probability
of survival for first year
Expected nth Year (n>l) Cost: maintenance care cost for one
year x probability of survival through «th year + terminal
costs x probability of mortality in nth year
Expected Lifetime cost =
Expected First-Year cost + the sum of the (discounted)
expected subsequent-year costs
As with the cost calculations for stomach cancer survivors, the probabilities
used in these cost calculations are the conditional probabilities given in
Table II.2-7, which are conditional on dying of stomach cancer.
Using the initial, maintenance, and terminal care costs from Table II.2-5,
the mathematical equation for the expected lifetime costs incurred by
nonsurvivors is:
$29,309 + pm™ x 0.5 ($31,325) + ps™ x .75 x $10,554
10 m $10,554 m $31,325
ps + pm
Chapter 11.2 11.2-29 Cost of Stomach Cancer
-------
where: y = the year post-diagnosis
psns = the conditional probability of survival for that year,
conditional on being a nonsurvivor of stomach
cancer
pmns = the conditional probability of mortality for that year,
conditional on being a nonsurvivor of stomach
cancer
r = the discount rate.
The costs are summed over all years from diagnosis to death. Maintenance
care costs are not added in the last year of life because during the six
months that are assumed to constitute this period the patient is assumed to
receive terminal care. (The discounted results are shown in the "Results"
section that follows.)
Example: Nonsurvivors Year One
During the first year post-diagnosis, nonsurvivors of stomach cancer who
survive the year (34.3 percent) will, on average, incur the following costs:
Initial treatment: $29,309
Maintenance treatment: $7,916 (.75 x $10,554)
Total Cost $37,225
However, 65.7 percent of stomach cancer nonsurvivors will die during the
first year. Those individuals are assumed to die at the midpoint of the year
(to obtain an average survival for the time period and average costs). That
group will incur the initial costs for three months, plus three months of
terminal care as follows:
Initial treatment: $29,309
Terminal care: $15,663 (.5 x $31,325)
Total Costs: $44,972
A weighted average of the first-year costs incurred by stomach cancer
nonsurvivors who do and do not die during the first year was calculated as
follows:
$37,225 x 0.343 + $44,972 x 0.657 = $42,312
During the second and subsequent years up to but not including the year of
death, the medical costs of nonsurvivors will include the costs of
maintenance care. As noted above, the last year of life is composed of
Chapter 11.2 11.2-30 Cost of Stomach Cancer
-------
Link to Table II. 2-6
terminal care costs only, since all patients are assumed to receive six
months of terminal care. For example, if someone died during the third
year post-diagnosis, he would receive three months of initial care and nine
months of maintenance care during the first year; he would receive 12
months of maintenance care during the second year; and he would receive
terminal care for the six months that he is assumed to have survived during
the third year (as noted previously, all patients are assumed to die mid-year
to obtain average cost estimates for a year).
When the costs for each year are summed over a period often years post-
diagnosis, during which essentially all patients who will die of stomach
cancer have done so, the total cost per nonsurvivor is obtained. These
costs are shown in Table II.2-9.
II.2.B.2.3 A Comparison of the Expected Medical Costs of
Stomach Cancer Patients, Using Two Approaches
Section II.2B.1.3 discusses calculation of the average direct medical costs
for all patients (average costs) and Section II.2.B.1.4 provides separate
cost estimates for survivors and nonsurvivors of stomach cancer. The
average patient cost can be calculated, however, from the results in
II.2B.1.4 using the weighted average of the expected costs of stomach
cancer survivors and nonsurvivors, where the weights are the probabilities
of surviving stomach cancer and not surviving it, respectively.
As shown in Table II.2-6, the expected medical cost for ten years post-
diagnosis for a stomach cancer patient diagnosed at age 70 is $66,700.
The expected medical costs calculated separately for survivors and
nonsurvivors of stomach cancer (for those diagnosed at age 70) are
$102,693 and $58,704, respectively. The probability of being a stomach
cancer survivor when onset occurs at age 70 is 0.18178; the probability of
being a stomach cancer nonsurvivor is 0.81822. The expected cost incurred
by a stomach cancer patient, calculated as a weighted average of the costs
of those who survive stomach cancer and those who die from it, is
therefore
$102,693 x 0.18178 + $58,704 x 0.81822 = $66,700.
This is the same value that was calculated by following a cohort of stomach
cancer patients over the ten-year period, using their (unconditional)
probabilities of survival, death from stomach cancer, and death from other
causes for each year (shown in Table II.2-6).
Chapter 11.2
11.2-31
Cost of Stomach Cancer
-------
Table 11.2-9. Expected Costs of Medical Services (in 1996$) for Nonsurvivors of Stomach Cancer (Age of Onset = 70)
Years
Post-
Diag-
nosis
(n)
1d
2
3
4
5
6
7
8
9
10
Medical Costs Through the 10th Year Post-diagnosis3 (undiscounted)
Medical Cost if
Survive Through the
nth Year
(1)
$37,225
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
$10,554
Conditional Probability
of Survival Through
the nth Yearb
(2)
0.343
0.188
0.121
0.082
0.057
0.039
0.025
0.015
0.007
0.000
Medical Cost if
Die in the nth
Year
(3)
$44,972
$31,325
$31,325
$31,325
$31,325
$31,325
$31,325
$31,325
$31,325
$31,325
Conditional
Probability of
Mortality in the nth
Yearb
(4)
0.657
0.155
0.067
0.039
0.026
0.018
0.014
0.010
0.008
0.007
Expected Total (Undiscounted) Cost Through the 10th Year Post-Diagnosis:
Total Cost Based on
Weighted Average0
= (1)x(2) + (3)x(4)
$42,312
$6,849
$3,375
$2,082
$1,400
$978
$691
$483
$327
$207
$58,704
a. Costs are based on data reported in Table II. 2-6, adapted from Baker et al., 1989.
b. Probabilities of survival and mortality, taken from Table II.2-7, are conditional on dying of stomach cancer within ten years post-
diagnosis.
c. Weighted average of the costs incurred by nonsurvivors who survive the year and the costs incurred by nonsurvivors who die during
the year. Weighting is based on the conditional probabilities listed.
d. Costs during the first year include an additional charge for "Initial" therapy ($29,309), and the duration of maintenance or terminal care
is adjusted accordingly. See text for discussion.
Chapter II.2
I.2-32
Cost of Stomach Cancer
-------
II.2.B.2.4 Discounted Medical Costs for All Patients, Survivors,
and Nonsurvivors
The per patient lifetime direct medical costs were calculated for stomach
cancer patients (as shown in Table II.2-6), stomach cancer survivors (as
shown in Table II.2-8) and stomach cancer nonsurvivors (as shown in
Table II.2-9) diagnosed at age 70, undiscounted as well as using discount
rates of three, five, and seven percent. The discounted costs for each year
were discounted back to year one (time of diagnosis). This procedure was
carried out for ten years following diagnosis (which, for nonsurvivors,
comprises the full duration of treatment time because virtually all patients
that are going to die of stomach cancer do so within ten years) and
comprises the assumed full duration of maintenance care for survivors.
The results are shown in Table II.2-10.
Table 11.2-10. Costs of Medical Services (in 1996$) for Stomach Cancer Patients, Survivors,
and Nonsurvivors (Diagnosed at Age 70) Undiscounted and Discounted at 3, 5, and 7 Percent
Patient Group
Survivors
Nonsurvivors
Average Patient
Discount Rate
Undiscounted
$102,693
($102,700)
$58,704
($58,700)
$66,700
($66,700)
3
$94,229
($94,200)
$57,524
($57,500)
$64,229
($64,200)
5
$89,749
($89,700)
$56,826
($56,800)
$62,811
($62,800)
7
$85,654
($85,700)
$56,190
($56,200)
$61 ,546
($61 ,500)
* The costs in parenthesis have been rounded to the nearest hundred dollars.
II.2.C. Sensitivity Analyses
The calculation of expected medical costs incurred by a stomach cancer
patient depends on several input factors, including treatment methods,
survival percentages and durations, and the age of diagnosis. To illustrate
the sensitivity of medical costs to key demographic characteristics,
sensitivity analyses were carried out for age at diagnosis and for a sex and
race combination. Age at diagnosis was selected because environmental
pollutants may cause cancer to occur at earlier ages than the ages at which
these cancers typically occur. Many chemicals studied in controlled animal
cancer evaluations cause cancer at earlier ages than the ages at which
cancer "spontaneously" occurs in the animals. This same dynamic has been
observed among occupationally-exposed workers whose cancer results
from exposures to chemicals and radiation. For example, many studies of
radon and lung cancer indicate that radon-associated lung cancer occurs at
Chapter II.2
II.2-33
Cost of Stomach Cancer
-------
younger ages than would be expected in the general population. A
sensitivity analysis of African-American males was conducted because this
is a large high-risk group in the United States.
II.2.C.1 The Effect of Age at Diagnosis on Medical Costs
Expected medical costs incurred by stomach cancer patients, stomach
cancer survivors, and stomach cancer nonsurvivors were calculated for
ages at diagnosis of 22, 42, 52, 62, 70, and 82. For each age at diagnosis,
the methods used to calculate expected medical costs were the same as
those used when age at diagnosis is 70. These methods are discussed in
Section II.2.B and are illustrated using:
• age at diagnosis equal to 70 in Table 11.2-6 for expected medical
costs incurred by stomach cancer patients;
Tables II.2-8 and II.2-9 for expected medical costs incurred by
survivors and nonsurvivors, respectively;
Link To Tables II.2-6,8,9
There is no information on age-specific medical costs comprising initial
treatment, maintenance care, or terminal care (the three treatment phases
delineated by Baker et al. (1989)); however, costs would not be expected
to vary substantially by age. In addition, the relative survival rates obtained
from NCI (1998) are not age-specific. Consequently, differences in
expected medical costs across different ages at diagnosis are based solely
on differences in age-specific survival and mortality probabilities in the
general population. The results of the age-specific analysis are shown in
Table II.2-11.
Table 11.2-11. Summary Table of Expected Medical Costs Incurred by Stomach Cancer
Patients, Survivors, and Nonsurvivors, by Age at Diagnosis
Age at
Diagnosis
22
42
52
62
70*
82
(Undiscounted) Expected Medical Costs for 10 Years Incurred by a Stomach
Cancer:
Patient
$70,482
$70,227
$69,721
$68,479
$66,700
$61,581
Survivor
$130,789
$128,522
$124,298
$114,569
$102,693
$77,086
Non-survivor
$60,077
$59,987
$59,804
$59,360
$58,704
$56,798
* This is the average age used in the main analysis and is included as a point of reference for this
sensitivity analysis.
Chapter II.2
II.2-34
Cost of Stomach Cancer
-------
.2.C.2
As can be seen in Table II.2-11, differences in expected medical costs
across ages at diagnosis are greater among survivors than nonsurvivors.
This may be an artifact of several characteristics of the analysis, in which
medical costs and relative survival rates were not age-specific and a
survivor's cost of surviving through the year is twice what it is if death
from some other cause occurs during the year. Younger stomach cancer
survivors have a greater chance of not dying of other causes during each
year than older stomach cancer survivors, based simply on age-specific
general population survival rates. Stomach cancer survivors who survive
through a year incur twice the cost of those who die of other causes during
the year (a full year's worth of maintenance care cost versus half a year's
worth of maintenance care cost). Because of this, younger survivors,
whose survival probabilities are greater than older survivors, incur
substantially more costs.
The Effect of Race on Medical Costs: An Analysis of African-
American Males
II.2.C.2.1 Incidence of Stomach Cancer
Of the four gender-race categories for which NCI (1998) provides
information related to stomach cancer (white males, white females,
African-American males, African-American females), the group with the
highest rates of stomach cancer is African-American males. A comparison
of incidence rates at the different ages at diagnosis for the general U.S.
population versus African-American males is given in Table II.2-12.
Table 11.2-12. Stomach Cancer Incidence Rates for the General U.S. Population Versus African-
American Males
Age at Diagnosis
0-4
5-9
10-14
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
Incidence per 100,000
General U.S.
Population
0.0
0.0
0.0
0.1
0.1
0.4
0.7
1.5
2.7
4.4
8.3
14.1
21.7
31.8
African-American
Males
0.0
0.0
0.0
0.0
0.7
0.2
1.7
3.5
8.1
10.0
22.5
38.8
55.9
83.9
Ratio of Incidence
Rates (general
population/African-
American Males)
— -
— -
0.00
7.00
0.50
2.43
2.33
3.00
2.27
2.71
2.75
2.58
2.64
Chapter II.2
II.2-35
Cost of Stomach Cancer
-------
Table 11.2-12. Stomach Cancer Incidence Rates for the General U.S. Population Versus African-
American Males
Age at Diagnosis
70-74
75-79
80-84
85+
Incidence per 100,000
General U.S.
Population
45.6
55.9
67.5
79.4
African-American
Males
127.8
111.6
135.8
225.2
Ratio of Incidence
Rates (general
population/African-
American Males)
2.80
2.00
2.01
2.84
Link to Chapter 1.1
Incidence rates for the earliest ages at diagnosis are probably unreliable
because they are based on very small samples. Starting from about age 30,
however, African-American males have an incidence rate of stomach
cancer that is two to three times that of the general U.S. population. The
cause of the increased rate of stomach cancer among African-Americans
males is not known. It may be due to dietary, environmental, genetic, or
other factors. Issues related to susceptible subgroups in benefits
assessments are discussed in the Chapter I.I section titled "Susceptible
Subgroups."
II.2.C.2.2 Risk Versus Per Capita Costs
Incidence rates that are two to three times higher than those of the general
population suggest that exposure to pollutants associated with stomach
cancer may result in expected costs-of-illness for African-American males
that are two to three times higher per exposed African-American male as
compared with the expected costs incurred by the average exposed
individual in the general population. Suppose, for example, that stomach
cancer costs $50,000 per stomach cancer patient on average. Suppose also
that the incidence rate among African-American males is 84 per 100,000
while the incidence rate in the general population is 32 per 100,000. If the
cost per patient is the same among African-American males as among
individuals in the general population, then the expected cost of stomach
cancer per exposed African-American male is
$50,000 x 0.00084 = $42.
The expected cost of stomach cancer per exposed individual in the general
U.S. population is
$50,000 x 0.00032 = $16.
Chapter 11.2
11.2-36
Cost of Stomach Cancer
-------
Link to Table II. 2-2
II.2.C.2.3. Comparison of Per Capita Costs: African-
American Males versus the General U.S.
Population
The costs of illness analyzed in this handbook are not costs per exposed
individual, but rather costs per patient (per capita), unlike the analysis
above. This chapter estimates the expected costs incurred by an individual
who has been diagnosed with stomach cancer. Even though
disproportionately more African-American males are diagnosed with
stomach cancer than the general population, their medical costs per patient
may or may not be higher than those of stomach cancer patients in the
general population. To contrast the costs of these two groups, the analyses
that were carried out for the general U.S. population were also carried out
for African-American males.
Although the methodology used for this sensitivity analysis was the same as
that used in the main analysis, the values of the following inputs to the
analysis were altered to be specific to the population of African-American
males in the U.S.:
• age-specific survival rates,
age-specific life expectancies,
• age-specific probabilities of dying specifically of stomach cancer,
and
• RSRs for each year post-diagnosis.
RSRs were not available for African-American males for each year post-
diagnosis. These were therefore derived by multiplying the adjusted RSRs
used in the main analysis (see Table II.2-2) by the ratio of the five-year
RSR for African-American males, 1986-1993 (16.9), to the corresponding
five-year RSR for the general population (20.6).
Given values for these input parameters specific to African-American males
in the U.S., the expected per capita medical costs incurred by African-
American male stomach cancer patients were calculated. The resulting
medical costs is shown, and compared with the corresponding medical
costs of the average patient in the general population, in Tables II.2-13.
Chapter 11.2 11.2-37 Cost of Stomach Cancer
-------
Table 11.2-13. Expected Medical Costs Over 10 Years for a Stomach Cancer Patient, Survivor,
and Nonsurvivor for Selected Ages at Diagnosis: A Comparison Between the General U.S.
Population and African-American Males
Age at Diagnosis
22
42
52
62
82
Average age at
diagnosis for African-
American males: 66
Average age at
diagnosis in the
general U.S.
population: 70
General U.S. Population
Expected Medical Costs for 10 Years
for a stomach cancer:
Patient
$70,482
$70,227
$69,721
$68,479
$61,581
$66,700
Survivor
$130,789
$128,522
$124,298
$114,569
$77,086
$102,693
Non-
survivor
$60,077
$59,987
$59,804
$59,360
$56,798
$58,704
African -American Males
Expected Medical Costs for 10 Years
for a stomach cancer:
Patient
$65,629
$65,014
$64,205
$62,669
$56,026
$61,779
Survivor
$127,517
$121,002
$113,390
$100,569
$65,150
$94,397
Non-
survivor
$56,898
$56,678
$56,383
$55,829
$53,296
$55,489
Expected per capita medical costs of African-American male stomach
cancer patients are uniformly less than the corresponding per capita costs
for stomach cancer patients in the general U.S. population at all ages at
diagnosis considered. The patterns are similar for stomach cancer
survivors and nonsurvivors. As noted in Section II.2.C. 1, Baker et al.
(1989) do not report costs for any of the three treatment periods separately
by age at diagnosis or by any other demographic characteristic.
Differences in expected medical costs between African-American male
stomach cancer patients (or survivors or nonsurvivors of stomach cancer)
and their counterparts in the general population are therefore, in this
analysis, due solely to differences in survival and mortality probabilities.
II.2.C.2.3.1 Survivors' Medical Costs
Among survivors of stomach cancer, costs beyond the first year consist
only of maintenance care costs. The greater the probability of survival
(i.e., of not dying of some other cause), the longer the expected period of
maintenance care, and the greater the expected costs incurred by survivors.
African-American males have notably lower survival rates (death rates are
higher from non-stomach cancer causes) than the general population at
each age. Consequently, the expected costs incurred by an African-
American male survivor of stomach cancer are lower than the
corresponding costs of a stomach cancer survivor from the general
population. That is, at each age there is a smaller proportion of African-
American males surviving to incur the costs of maintenance care as
compared with stomach cancer survivors in the general U.S. population.
Chapter II.2
II.2-38
Cost of Stomach Cancer
-------
II.2.C.2.3.2 Nonsurvivors' Medical Costs
Among nonsurvivors of stomach cancer, costs beyond the first year are
three times as high for those who die of stomach cancer during the year
($31,325) as for those who survive the year ($10,554). In the first year
post-diagnosis, the cost differential between those who survive the year
and those who die of stomach cancer during the year is much less than in
subsequent years. This is because those who die are assumed to die
midyear, and during the first three months of the first year stomach cancer
patients are assumed to undergo initial treatment, leaving only three
months (as opposed to the full six months) of expensive terminal care in the
first year. Because a substantially larger percentage of African-American
male nonsurvivors die during the first year post-diagnosis as compared with
general population nonsurvivors, the overall costs incurred by African-
American male nonsurvivors are estimated to be less than those incurred
by general population nonsurvivors. This is to some extent due to the
assumption that, if the patient dies of stomach cancer during the first year
post-diagnosis, there are only three months of terminal care costs incurred.
11.2.D. Uncertainties and Limitations
As noted periodically in the above discussion, there is substantial
uncertainty surrounding various aspects of cost in the analyses. Information
concerning important inputs to the analysis was often limited to some
degree and, in some cases, was highly limited. Although a complete
uncertainty analysis is beyond the scope of this work, the significant
sources of uncertainty are discussed below in Section II.2.D.1. The scope
of the analysis had some limitations as well. These are discussed in Section
II.2.D.2.
II.2.D.1. Uncertainties Surrounding Key Inputs to the Analysis
The cost estimates based on Baker et al. (1989, 1991) have a number of
limitations, many of them noted by Baker et al. (1991) and Mor et al.
(1990) and Mor (1993). Most of these limitations arise from the use of
CMHSF. Medicare data has five limitations that decrease its value for
calculating the average lifetime direct medical costs of treating stomach
cancer. First, Medicare covers only medical services for most persons age
65 and over, disabled persons entitled to Social Security cash benefits for
at least 24 months, and most persons with end-stage renal disease. All
patients not covered by Medicare are excluded from the database, including
all non-disabled women under 65, and women over 65 using private health
insurance (Baker et al., 1991).
Given that diagnosis of stomach cancer occurs before age 65 in 28.7
percent of patients (NCI, 1998), the CMHSF excludes a significant number
of younger patients. According to Mor et al., treatment for younger
Chapter 11.2 11.2-39 Cost of Stomach Cancer
-------
women tends to be more intensive (and therefore more costly per unit time)
than treatment for older women, though older women tend to have longer
hospital stays. Because these differences counteract each other, the
omission of younger women from the analysis is not expected to affect the
results substantially. In addition, the majority of senior citizens are enrolled
in Medicare (Ibid); differences in medical costs incurred by senior citizens
not using Medicare should have little effect on overall cost estimates.7
Medicare also does not cover self-administered drugs, intermediate nursing
care, long-term nursing care, and expensive, extraordinary treatments (such
as bone marrow transplants). For some patients these may represent
significant percentages of total treatment costs. Most direct medical costs,
however, appear to be covered by the CMHSF database and are included in
Baker et al.'s analysis.
Another drawback is that Baker et al. were not able to identify stomach
cancer patients in CMHSF whose diagnosis and first course of therapy did
not involve hospitalization. In an analysis of Rhode Island stomach cancer
patients covered by Medicare, Mor et al. determined that approximately 13
percent of stomach cancer patients were initially diagnosed without
hospitalization, and had substantially lower initial and subsequent treatment
costs (Mor et al., 1990). This omission likely causes average treatment
costs to be overestimated, though by relatively little, since 87 percent of
stomach cancer patients on Medicare are initially diagnosed during
hospitalization and therefore would be recorded in CMHSF.
A fourth drawback is that Baker et al. (1989) provides no information
concerning the duration of the maintenance period for stomach cancer.
The analysis in this chapter assumed that stomach cancer survivors incur
maintenance care costs for ten years. If the average duration of
maintenance care among survivors of stomach cancer is shorter (longer)
than ten years, the estimates of the costs incurred by survivors would be
biased upward (downward). This is less of an issue for nonsurvivors' costs
because the great majority of stomach cancer nonsurvivors die within the
first few years. Because most stomach cancer patients (about 82 percent)
are ultimately nonsurvivors, the duration of the maintenance period is of
somewhat less importance for stomach cancer patients than for the 18
percent who ultimately survive the illness.
A fifth drawback is that the data used by Baker are from the period 1974 to
1981. Increased early detection and treatment modifications for stomach
cancer have increased the life expectancy of those diagnosed with the
disease.
7 This figure represents those enrolled in Medicare Part A; 95 percent of those enrolled in
Medicare Part A choose also to enroll in Medicare Part B.
Chapter 11.2 11.2-40 Cost of Stomach Cancer
-------
Finally, the reliability of the data contained in the database used by Baker et
al. varies. An independent analysis of CMHSF performed in 1977 by the
Institute of Medicine of the National Academy of Sciences found that the
frequency of discrepancies in principal diagnoses varied among diseases
(Baker et al., 1991). It is unclear whether the presence of misnamed
diagnoses contained in CMHSF potentially increases or decreases the
resultant cost estimates.
Overall, despite the limitations described above, Baker's analysis of the
CMHSF data represents the most nationally-representative, per-patient
lifetime estimate of the direct medical costs of treating stomach cancer to
date. Their cost estimates are based on sound criteria. Some of the data
limitations underestimate costs and others overestimate costs; the sum of
the data limitations therefore decreases the magnitude of error. More of the
uncertainties in their analysis appear to underestimate costs, however; the
net result is a likely underestimation of actual direct medical costs.
Although some uncertainties are associated with the estimation of the
survival and mortality probabilities used in the calculation of expected
medical costs, these uncertainties are likely to be relatively small. As noted
in the text, NCI RSRs used to estimate survival and mortality for this
analysis are based on the survival experience of a large group of stomach
cancer patients considered in relation to the survival experience of the
general population. Although age-specific RSRs for each year post-
diagnosis are not available, the age-specific five-year RSRs provided by
NCI (1998) suggest that there is relatively little variation in RSRs across
ages at diagnosis for stomach cancer patients.
An additional limitation of this analysis is that medical costs incurred as a
result of stomach cancer, but not considered by Baker et al., may arise as a
result of treatment for stomach cancer. Secondary cancers and other
adverse health effects may occur due to radiation, chemotherapy treatment,
and other therapies. These may occur substantially after stomach cancer
treatment has been completed and can incur added medical costs not
considered in this chapter.
Data have not yet been located regarding the average duration of
maintenance care. For purposes of this analysis, ten years of follow-up
care was assumed to be reasonable due to the severity of the disease and
the consequences of stomach surgery. This assumption may be revised in
the future if data are located.
Chapter 11.2 11.2-41 Cost of Stomach Cancer
-------
.2.D.2. Scope of the Analysis
The analysis in this chapter was confined to direct medical costs by the
patient. As noted in Chapter I.I, willingness-to-pay has many other cost
elements. The analysis does not include time lost by the patient; their
family and friends who provide care; pain and suffering on the part of the
patient, family, and friends; changes in job status among previously
employed patients, training for new job skills due to physical limitations; or
medical costs incurred after the ten-year maintenance period. These cost
elements may comprise a substantial portion of the total cost of stomach
cancer.
Link to Chapter 1.1
Chapter 11.2 11.2-42 Cost of Stomach Cancer
-------
Appendix II.2-A Deriving the Probabilities of Dying of Stomach Cancer
and Dying of Other Causes
This appendix contains a method to derive the probabilities of dying of
stomach cancer and dying of other causes in the nth year in a cohort of
stomach cancer patients who have survived to the nth year.
w
q1
The diagram above represents the stomach cancer cohort remaining at the
beginning of the nth year. The area of the entire box = the probability of
having survived to the beginning of the nth year post-diagnosis = 1. That
is, all probabilities described below are conditional on having survived to
the beginning of the nth year post-diagnosis.
p = the proportion of the cohort who survive through the nth year (=
the probability of surviving the nth year.)
q = the probability in the general population of dying of causes other
than stomach cancer. We assume that the proportion of the
stomach cancer cohort who would die of other causes if they were
not a cohort of stomach cancer patients is also q.
q = ql + q2 in the diagram.
z = the proportion of the cohort who would die of stomach cancer if
there were no other causes of death.
z = w + q2 in the diagram.
Chapter 11.2
11.2-43
Cost of Stomach Cancer
-------
q2 = the portion of the cohort who would die of other causes if they
were not stomach cancer patients but who could die instead of
stomach cancer (because they are in fact stomach cancer
patients).
= the portion of those who would die of stomach cancer if there were
no other causes of death who could die instead of other causes
(because there are in fact other causes of death).
= the intersection of z and q.
We assume that, of those who could die of either stomach cancer or other
causes (q2), half will die of stomach cancer and half will die of other
causes. Because this is a very small portion of the cohort, even if the split
is not exactly 50-50, this should affect the results only minimally.
Let:
a = the proportion of the cohort who actually die of stomach cancer (=
the probability of dying of stomach cancer in the stomach cancer
cohort), and
b = the proportion of the cohort who actually die of other causes (= the
probability of dying of other causes in the stomach cancer cohort).
Then:
a = w + 0.5xq2, and
b = ql+0.5xq2 = (q-q2) + 0.5xq2 = q-0.5xq2
Because:
w = 1 - p - q (see diagram), and
a = l-p-q + 0.5xq2
To solve for q2 in terms of known quantities (p and q), we use the
following (see the diagram):
q2 w
q (w + p)
(Recall that q = ql + q2). This implies that
w
q2 = q
(w + p)
Chapter 11.2 11.2-44 Cost of Stomach Cancer
-------
We also know that w = 1 - p - q (see diagram). Substituting this into the
above equation, we get q2 as a function of p and q, both of which are
known:
q2 = (-3-) x (1 - p - q ) .
\-q
We can now derive a and b, given q2.
In summary,
q2 = (--) x (1 - p - q )
\-q
where, recall, a is the probability of a stomach cancer patient dying of
stomach cancer during the wth year and b is the probability of a stomach
cancer patient dying of other causes during the nth year.
a = 1 - p - q + 0.5 x q2
b = q - 0.5 x q2
Chapter 11.2 11.2-45 Cost of Stomach Cancer
-------
CHAPTER 11.4. COST OF KIDNEY CANCER
Clicking on the sections below will take you to the relevant text.
II.4.A Background
II.4. A.I Description
II.4. A.2 Concurrent Effects
II.4.A.3 Causality
II.4.A.4 Treatments and Services
II.4. A. 5 Prognosis
II.4.B. Costs of Treatment and Services
II.4.B.1 Methodology
II.4.B.2 Results
II.4.B.3 Limitations
II.4.C. Conclusions
Chapter 11.4 11.4-1 Cost of Kidney Cancer
-------
CHAPTER 11.4. COST OF KIDNEY CANCER
II.4.A Background
This chapter contains a discussion of the methods used and results of
estimating the direct medical costs incurred by kidney cancer patients. It
does not include information on elements such as indirect medical costs,
pain and suffering, lost time of unpaid caregivers, etc. The reader is
referred to Chapter I.I for a discussion of the cost estimation methods and
cost elements that are relevant to all benefits estimates. In addition,
Chapter II. 1 contains information regarding cancer causality, a list of
known and suspected carcinogens, and information on cancer cost
estimation.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and II. 1
Link to inflation factors
II.4.A.1 Description
Kidney cancer is a malignancy within the kidneys and may be localized or
have spread to multiple sites (Bennet and Plum 1996). It represents one to
three percent of all adult cancers in the United States (Javadpour 1984,
Klein et al., 1993). Kidney cancer occurs most frequently in individuals in
their fifties through seventies, with two to three times as many males as
females developing the disease (NCI 1994, Javadpour 1984, Klein et al.,
1993, Montie et al., 1990). Approximately 25,300 new cases of kidney
cancer were diagnosed in 1991 in the U.S. (NCI 1994).
According to the National Cancer Institute (NCI 1994), the incidence of
kidney cancer in the U.S. has increased by over 30 percent over the past 20
years, from a rate of 6.7 per 100,000 in 1973 to 8.8 per 100,00 in 1991.
Although this increase may be attributed in part to improved detection
techniques, a concurrent rise in mortality due to kidney cancer indicates
that there has been a real increase in disease incidence in the U.S. over time
(McCredie 1994). Reported rates of kidney cancer incidence and mortality
have increased worldwide in the past decades, particularly in more
industrialized countries. According to McCredie (1994), kidney cancer is
likely to become the most common urinary cancer in the coming decades,
and one of the major cancers of affluent societies, unless its etiology can be
better identified and addressed.
Chapter 11.4 11.4-2 Cost of Kidney Cancer
-------
II.4.A.2 Concurrent Effects
II.4.A.3 Causality
No data were located indicating that concurrent effects unrelated to kidney
cancer or its treatment were likely to occur with this disease. Secondary
cancers and other adverse health effects may occur due to treatment and
therapy. These can incur added medical costs not considered in this
chapter.
Carcinogens suspected of causing renal cancer in human study groups
include hormones, radiation, certain viruses, tobacco smoking, phenacetin-
containing analgesics, paracetamol, obesity and hypertension (or drugs
used for their treatment), asbestos, and renal injury (Javadpour 1984,
McCredie 1994). Kidney cancer has been induced in experimental animals
via exposure to chemical, physical, viral, and hormonal agents; radiation;
and dietary deficiencies (Konishi et al., 1994, Javadpour 1984).
Epidemiological data show significantly greater frequencies of kidney
cancer in cigar smokers, smokers with occupational exposures to cadmium,
petroleum industry workers, and possibly workers with occupational
exposures to lead. Kidney toxins, such as unleaded gasoline, have
demonstrated carcinogenic potential in chronic bioassays in animals
(Konishi et al., 1994). Other risk factors appear to be linked to a more
affluent lifestyle, including diet; and perhaps to increasing industrial
exposure in some localities (McCredie 1994). As is often the case with
cancers, however, it is difficult to prove causality. McCredie (1994), for
example, hypothesizes that the general effect of changes in standard of
living is a significant cause of the increased incidence of kidney cancer.
Table II. 1-1 in Chapter III contains a list of chemicals known to cause or
suspected of causing cancer (as reported in the EPA databases IRIS,
HEAST, and HSDB). Most chemicals in the table were carcinogenic in
animal studies. These studies do not provide organ-specific data because it
is not generally assumed that cancer induction will always occur at the
same site in humans as in animals. Consequently, the chemicals listed in
Table II-1 may cause kidney cancer and/or other types of cancer.
Evaluation of the likelihood of this occurrence would require additional
research (risk assessment).
Link to Table II.I-l
Chapter 11.4
1.4-3
Cost of Kidney Cancer
-------
II.4.A.4 Treatments and Services
II.4.A.5 Prognosis
Kidney cancer is slow to develop, and may reach relatively advanced stages
before detection. Symptoms may include fever, weight loss, generalized
weakness, abdominal pain, abnormal increases in red blood cells or blood
calcium levels, anemia, bloody urine, cardiac enlargement, or liver
dysfunction without evidence of liver cancer. Advances in diagnostic
techniques, such as bone scanning, chest tomography, computed
tomography (CT), ultrasonography, and magnetic resonance imaging
(MRI) have greatly improved the diagnosis of kidney cancer, though have
not yet increased survival rates (Javadpour, 1984).
The only effective therapy for kidney cancer is surgical excision (often
involving removal of the kidney, adrenal gland, associated fat, and regional
lymph nodes) before it has a chance to spread to other organs or
metastasize to distant sites in the body (Javadpour 1984, Klein et al.,
1993). Chemotherapy, radiotherapy, and hormone therapy have all proven
ineffective as a systemic treatment for kidney cancer. Immunotherapy (also
known as biologic response modifier (BRM) therapy), though still
experimental, may prove valuable in treating advanced kidney cancer (Klein
et al., 1993, Montie et al., 1990).
Overall, the five-year survival rate in the U.S. for all kidney cancer patients
is only about 54 percent (McCredie 1994) and the six-year survival is
approximately 50 percent (NCI, 1998, based on patients diagnosed in
1987). Kidney cancer can and does metastasize to most areas of the body,
most commonly the lymph nodes, lungs, liver, and bones (Montie et al.,
1990). More than half of all new patients are diagnosed with regionally
advanced or metastatic kidney cancer (Klein et al., 1993). Although kidney
cancer can include spontaneous regressions and long survival in the
presence of metastatic disease, there is currently no effective treatment of
metastatic kidney cancer, and prognosis for these patients is very poor
(Javadpour 1984, Montie et al., 1990). Patients with advanced kidney
cancer face a median survival often months and a one to two percent
chance of surviving five years or more (Klein et al., 1993).
II.4.B. Costs of Treatment and Services
II.4.B.1 Methodology
This chapter estimates the per-patient lifetime direct medical costs of
treating kidney cancer based on the work of Baker et al. (1989 and 1991).
Baker et al. used the Continuous Medicare Flistory Sample File (CMHSF)
to estimate average per-patient medical costs of treating kidney cancer.
They chose CMHSF because: (1) it is a nationally representative sample of
Chapter 11.4 11.4-4 Cost of Kidney Cancer
-------
the Medicare population (5 percent), covering over 1.6 million patients; (2)
it is longitudinal, dating from 1974 to 1981; and (3) it captures the majority
of medical expenses for each beneficiary.
Five Medicare files are included in the CMHSF, which cover:
1) inpatient hospital stays,
2) skilled nursing facility stays,
3) home health agency charges,
4) physician services, and
5) outpatient and other medical services.1
Baker et al. calculated the average medical costs of Medicare patients with
kidney cancer, as well as the average medical costs of a randomly-selected
sample of Medicare patients without cancer (i.e., baseline costs). To
estimate costs attributable to kidney cancer, this report subtracts baseline
costs from the costs of patients with kidney cancer. An alternative
approach would have been to examine the medical services used by patients
with kidney cancer and make judgments, based on the nature of each
service, about whether its use was attributable to kidney cancer. This
second method is more complicated and requires that the motivation for
medical services be inferred.
Because CMHSF provides no indication of initial diagnosis, Baker et al.
assumed that a kidney cancer diagnosis appearing on a hospitalization
record after a minimum of one year without a kidney cancer diagnosis
indicated disease onset. This assumption would seem to hold true for a
majority of cases because of the high frequency of surgery and
hospitalization associated with initial treatment of kidney cancer (Klein et
al., 1993). Only patients with an initial diagnosis during the years covered
by the database were included.
The number of Medicare beneficiaries included in the kidney cancer subset
of the CMHSF, as defined by Baker et al., was 1,953. The random subset
of non-cancer patients, called the "co-morbidity subset" by the authors,
consisted of every 16th beneficiary contained in the CMHSF who had not
been hospitalized for cancer. Given that the CMHSF database contains
approximately 1.6 million accounts, the co-morbidity subset represents
about 100,000 individuals.
Baker et al. estimated total costs associated with three phases of treatment:
1 See Baker et al., 1991 for further details.
Chapter 11.4 11.4-5 Cost of Kidney Cancer
-------
Initial: all costs appearing on a beneficiary's record for up to three months
following diagnosis;
Terminal: all costs appearing on a beneficiary's record within six months
of death; and
Maintenance (intermediate phase): all costs incurred between these two
periods (calculated as an average monthly cost).
These periods differ significantly in intensity and cost of related medical
care. Initial therapy generally includes intensive diagnostic testing and
surgical removal of the tumor, incurring very high medical costs over the
approximately three months following initial diagnosis. If this treatment is
successful, a cancer patient will undergo a period of remission, during
which little medical treatment is given apart from monitoring for potential
cancer recurrence.
For the approximately fifty percent (NCI 1994) of patients who eventually
die of kidney cancer, a third phase of intensive terminal care (involving
further surgery, radiation, and/or other measures to alleviate symptoms)
takes place over approximately the last six months of their lives, which
again incurs substantial medical costs. According to Baker et al., the
pattern of initial, maintenance, and terminal care treatment phases is
apparently little affected by differences in total survival time of the patient;
the major difference is in the length of the maintenance phase (Baker et al.,
1991). Because Baker et al. (1989) calculated monthly maintenance phase
costs but not the duration of the maintenance phase, this report calculates
average lifetime maintenance phase costs based on survival rate data.
Baker et al. made four adjustments to the cost estimates calculated from
the CMHSF. First, they added charges for skilled nursing facilities (SNFs)
that were not covered by Medicare by multiplying the "length of stay" at an
SNF (computed from admission and discharge dates) by the average daily
SNF charge.2 Second, they added the annual Medicare Part B deductible
of $60 to the reimbursed charges in the database. Third, since Medicare
only pays 80 percent of physicians' charges, they scaled these
reimbursements up to 100 percent of physician charges to better reflect
social costs. Finally, they inflated all dollar values to 1984 dollars using the
Medical Care component of the Consumer Price Index.
2 Where no discharge date was given, Dec 31, 1981 (the end of the file) was used as the discharge
date. This likely underestimates SNF stays and therefore overall costs for some patients.
Chapter 11.4 11.4-6 Cost of Kidney Cancer
-------
II.4.B.2 Results
II.4.B.2.1 Treatment Phase Costs
Costs were estimated for each treatment phase, using the methodology
described above for Baker et al. Their cost estimates are given in 1984
dollars. Table II.4-1 shows average undiscounted total (non-incremental)
charges incurred over the three phases of kidney cancer treatment. The
comorbidity charges corresponding to the duration of each treatment
period are also listed.
Table 11.4-1
Average Patient Charges Per
Kidney
Initial
(Total Over 3
Months)
Maintenance
(Average per
Minth)
Terminal
(Total Over 6
Months)
Treatment Phase of Kidney
cancer treatment charges
$12,608
$670
$19,302
Cancer, in 1984 dollars
Co-morbidity charges3
$747
$249
$1,494
Source: Baker, et al., 1989 and 1991 (for background costs).
a From Baker et al. (1991) This is the average cost of medical care for patients not
undergoing treatment for cancer.
Table II.4-1 also includes co-morbidity charges (i.e., baseline charges) from
Baker et al. (1991). Co-morbidity charges for initial, maintenance, and
terminal phase treatment are $747 total, $249/month, and $1,494 total,
respectively.3 These charges are subtracted from the charges associated
with kidney cancer to yield net charges for the treatment of kidney cancer.
Baker et al. (1989) provides only summary charges for each of the
treatment phases; cost associated with each of the treatment components
(i.e., inpatient hospital stays, skilled nursing facility stays, home health
agency charges, physician services, outpatient services, and other medical
services) were not listed in the report.
3 Co-morbidity charges of $249 per month were reported in Baker et al.'s analysis of breast and
lung cancer (1991). The charges represent medical costs for a random sample of the Medicare population
without cancer with age distribution matching that of the breast and lung cancer subsets. Co-morbidity
charges for a sample matching the age distribution of the population with kidney cancer may vary slightly.
Chapter 11.4 11.4-7 Cost of Kidney Cancer
-------
II.4.B.2.2 Lifetime Cost Estimates
Lifetime costs of treating kidney cancer were calculated using the following
information:
1. the charges per phase (presented above),
2. an estimate of the average length of the maintenance phase of
treatment, and
3. the percentage of patients who survive the disease.
Survival data from the National Cancer Institute (NCI) indicate that
approximately 50 percent of the patients who die of kidney cancer do so
within one year of diagnosis. The NCI data cover a diagnostic period from
1974 to 1981. Although survival rates have improved, the distribution of
the occurrence of death is not expected to have changed substantially. This
is supported by Klein et al (1993), who found that more than half of all
new cases are diagnosed with regionally advanced or metastatic disease.
The median survival for patients with metastatic disease is approximately
ten months (Klein et al. 1993). The terminal charges are therefore assumed
to occur within one year of diagnosis, on average.
It is more difficult to determine the average period of maintenance. NCI
data indicate that the average relative survival rate is 49.8 percent at six
years after diagnosis.4 For the purposes of this analysis the average life
span for a person diagnosed with kidney cancer was assumed to be six
years post-diagnosis. Using this value, the average maintenance period
was assumed to be 5.25 years, after adjusting for initial care and terminal
care. Many patients will receive a longer or shorter period of maintenance
care; however, this is a reasonable estimate based on available information.
This approach is an approximation, and may be refined in the future. It is
likely to underestimate medical costs due to the improved prognosis and
survival duration of kidney cancer patients in recent years leading to longer
periods of maintenance care.
To determine the average lifetime incremental medical costs associated
with kidney cancer, the following calculations were used:
Initial phase costs attributable to kidney cancer were calculated using
$12,608 in gross charges from Table II.4-1 minus $747 in co-morbidity
charges to obtain a cost of $11,861.
4 The relative survival rate (RSR) is an approximation of the survival rate, adjusted for
background mortality. For a detailed discussion of RSR see Chapter II.2.
Link to Chapter II. 2
Chapter 11.4 11.4-8 Cost of Kidney Cancer
-------
Maintenance period charges were first calculated at $670/month from
Table II.4-1 multiplied by 63 months or 5.25 years. Sixty-three months
were estimated as the maintenance period based on a six-year average post-
diagnosis survival, minus three months initial phase and six months terminal
phase care. The average co-morbidity charges of $15,687 ($249/month
multiplied by 63 months) for the maintenance phase were subtracted from
the gross maintenance phase costs to obtain undiscounted annual
maintenance phase charges attributable to kidney cancer of $26,523.
Terminal costs attributable to kidney cancer were estimated by subtracting
co-morbidity charges of $1,494 from $19,302 (the gross charges for
terminal care for kidney cancer patients) to obtain an incremental cost of
$17,808. As noted previously, approximately 50 percent of patients with
kidney cancer ultimately die of the disease. Therefore the terminal costs
were multiplied by .5 to obtain an estimated average cost for this phase of
$8,904.
Total Costs. Table II.4-2 shows Baker et al.'s values for the incremental
medical costs for each phase. The total costs were modified by the
comorbidity costs (as shown in Table II.4-1).
Table 11.4-2
Average Incremental Per Patient Charges in 1984 dollars (undiscounted)
Incremental Costs Total cost
Initial (3 months care) $11,861
Maintenance ( 5.25 years care a) $26,523
Terminal (6 months careb) $8,904
$11,861 $47,288
Source: Table II.4-2, modified per text to obtain incremental charges.
3 Six-year assumed life span minus 3 months of initial treatment and 6 months of
terminal treatment.
b Terminal costs are incurred by only 50 percent of patients and were adjusted
accordingly.
The costs shown in Table II.4-2 were used to calculate the discounted
incremental per capital direct medical costs in 1996 dollars (based on the
Consumer Price Index ratio (CPI-U) for medical care, 1996:1984=2.14).
Both initial phase and terminal phase charges are present values because
they occur in the first year post-diagnosis. Those patients who will die of
kidney cancer do so, on average, within ten months of diagnosis. As
noted previously, terminal charges were adjusted to reflect the fact that
approximately 50 percent of patients incur terminal charges. Maintenance
care is assumed to be provided over a period of 5.25 years and is
discounted accordingly. (This calculation may be modified in the future as
Chapter 11.4 11.4-9 Cost of Kidney Cancer
-------
additional information is obtained on the duration of maintenance care.)
Present value maintenance phase charges are:
$26,523 using a zero percent discount rate
$24,747 using a three percent discount rate
$23,682 using a five percent discount rate
$22,702 using a seven percent discount rate
The undiscounted initial and terminal costs were summed with the various
discounted maintenance care charges to obtain the total incremental per
capita medical costs, as shown in Table II.4-3.
Table 11.4-3
Incremental Direct Medical Costs of Lifetime per Capita Treatment of Kidney
Cancer in 1984 and 1996 Dollars with Various Discount Rates a (1996:1984 =
2.14)
Present Value of Attributable Charges (Discount Rate %)
Lifetime Charges
1984 dollars
Lifetime Charges
1996dollarsd
47,288 45,512 44,417 43,467
96,916 97,396 95,117 93,019
a Treatment components are discounted based on when they occur in the course of
treatment. See text for discussion.
The costs presented in this chapter were current in the year the chapter was written. They
can be updated using inflation factors accessible by clicking below.
Link to inflation factors
II.4.B.3 Limitations
Link to Chapter II. 2
There are many limitations in cancer cost estimation. Those common to
most cancers are discussed in the introductory cancer chapter: 11.2.
There are a number of limitations to the use of the Baker et al. data, many
of which the authors have noted. These limitations are primarily related to
the use of CMHSF (Medicare) data. First, Medicare does not cover all
potential kidney cancer patients; it covers only most persons 65 and over,
disabled persons entitled to Social Security cash benefits for at least 24
Chapter 11.4
11.4-10
Cost of Kidney Cancer
-------
months, and most persons with end-stage kidney disease (Baker et al.
1991). All patients not covered by Medicare are excluded from the
database, including all non-disabled people under 65 and those over 65
using private health insurance. Given that approximately half of all kidney
cancer patients are under 65 (NCI 1994), the CMHSF excludes a
significant number of younger patients. Approximately 95 percent of
Americans 65 and over, however, are enrolled in Medicare (Baker et al.
1991).5 Differences in medical costs incurred by senior citizens not using
Medicare, or using it for only a portion of their costs, could lead to an
underestimate of medical costs.
Medicare also does not cover self-administered drugs, intermediate nursing
care, long-term nursing care, and expensive, extraordinary treatments (such
as bone marrow transplants). For some patients these may represent
significant percentages of their total treatment costs. To the extent that 1)
younger patients receive more aggressive (and therefore more expensive)
medical treatment for kidney cancer and 2) Medicare beneficiaries use
medical care not covered by Medicare, the values in the CMHSF database
will underestimate medical costs.
A minor drawback of the Baker et al. data is due to the researchers'
inability to identify cancer patients in CMHSF whose diagnosis and first
course of therapy did not involve hospitalization. For most kidney cancer
patients the first course of therapy involves major surgery. Consequently,
the omission of non-hospitalized patients is likely to result in a negligible
underestimate of medical costs.
The reliability of the data contained in the CMHSF database varies. An
independent analysis of CMHSF performed in 1977 by the Institute of
Medicine of the National Academy of Sciences found that the frequency of
discrepancies in principal diagnoses varied among diseases (Baker et al.,
1991). It is unclear whether the presence of misnamed diagnoses contained
in CMHSF potentially increase or decrease the resultant cost estimates.
Since Baker et al. include only those costs after the initial diagnosis, they
omit costs associated with prediagnostic tests and treatment. Although
these costs could be significant, substantial medical treatment (e.g., tests
requiring hospitalization) would also likely result in a diagnosis and thus be
included in Baker et al.'s analysis. This omission likely causes the values
generated by Baker et al. to somewhat underestimate direct medical costs
from the treatment of kidney cancer.
Another limitation to using the Baker data is its age. The most recent data
used by Baker is more than a decade old. Although many aspects of
5 This figure represents those enrolled in Medicare Part A; 95 percent of those enrolled in Medicare
Part A choose also to enroll in Medicare Part B.
Chapter 11.4 11.4-11 Cost of Kidney Cancer
-------
treatment have remained constant, new diagnostic methods and treatments
such as immunotherapy have potentially increased lifetime treatment costs.
Improved diagnostic procedures such as CT scans, ultrasonography, and
MRI (McCredie 1994) have led to earlier diagnoses and treatment. The
lack of data on new treatment methods may cause costs to be
underestimated.
Age of the data is also a problem with regard to survival statistics. NCI
maintains current survival information, but the survival prospects for a
patient diagnosed now are better than in previous decades. Since long-
term survival can only be evaluated retrospectively, there is uncertainty in
estimations of survival for current patients based on past survival patterns.
An additional limitation of this analysis is that medical costs incurred as a
result of kidney cancer, but not considered by Baker et al., may arise as a
result of treatment for kidney cancer. Secondary cancers and other adverse
health effects may occur due to radiation, chemotherapy treatment, and
other therapies. These effects may occur substantially after the cancer
treatment has been completed and can incur added medical costs not
considered in this chapter.
One limitation of the cost estimates that is directly related to the analytical
method used in this chapter is the lack of a yearly sequential analysis of
costs. A more precise method for estimating direct medical costs would
include year-to-year information on treatment and survival, with costs
determined using an estimate of the proportion of patients surviving each
year and those who died due to natural causes or kidney cancer. When this
chapter was developed, that method was not yet in use for this handbook.
(See Chapter III.2 for a more complete discussion of the new method.)
Average durations of treatment and survival were used in this analysis and
are likely to provide a good approximation of the costs that would be
obtained through a more detailed and complex analysis.
Link to Chapter II. 2
MAC. Conclusions
Overall, despite the limitations described above, Baker's analysis of the
CMHSF data represents the most nationally-representative per-patient
lifetime estimate of the direct medical costs of treating kidney cancer
available. Their cost estimates are based on sound criteria and reasonably
current data, since the treatment of kidney cancer has not changed
substantially since Baker et al.'s analysis. The authors have made
adjustments for many of the factors that can be quantitatively modified
(e.g., Medicare's underpayment for services).
Chapter 11.4 11.4-12 Cost of Kidney Cancer
-------
Because more of the uncertainties in their analysis appear to underestimate
costs (e.g., population covered, changes in treatment, and omission of pre-
diagnostic costs), the net result is a likely underestimation of actual direct
medical costs. This tendency toward underestimation can be noted when
using the cost estimates in this chapter in a benefits assessment.
Chapter 11.4 11.4-13 Cost of Kidney Cancer
-------
CHAPTER 11.5: COST OF LUNG CANCER
Clicking on the sections below will take you to the relevant text.
II. 5. A Background
U.S.A. 1 Description
II.5.A.2 Concurrent Effects
II.5.A.3 Causality & Special Susceptibilities
II.5.A.4 Treatments and Services
II.5.A.5 Prognosis
II.5.B Costs of Treatment and Services
II. 5.B.I Methodology
II.5.B.2 Results of Medical Cost Analysis
II.5.B.3 Other Studies
II.5.C Uncertainties and Limitations
II.5.C. 1 Uncertainties Surrounding Key Inputs to the Analysis
II.5.C.2 Scope of the Analysis
Chapter 11.5 11.5-1 Cost of Lung Cancer
-------
CHAPTER 11.5: COST OF LUNG CANCER
This chapter contains a discussion of the methods used and results of
estimating the direct medical costs incurred by lung cancer patients. It
does not include information on elements such as indirect medical costs,
pain and suffering, lost time of unpaid caregivers, etc. The reader is
referred to Chapter I.I for a discussion of the cost estimation methods and
cost elements that are relevant to all benefits estimates. In addition,
Chapter II. 1 contains information regarding cancer causality, a list of
known and suspected carcinogens, and information on cancer cost
estimation.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and II. 1
Link to inflation factors
II.5.A Background
II.5.A.1 Description
Lung cancer is the most frequent cause of cancer death in men and women
in the United States (Feld et al., 1995). There were approximately 173,000
cases of lung cancer diagnosed in 1994 in the United States and
approximately 150,000 lung cancer deaths occurred in that year (NCI,
1998). The increase in overall deaths per year from 18,000 in 1950 and
125,000 in 1988 to the current level is due, in part, to a more than five-fold
increase in lung cancer death rates (per 100,000) among women. This
increase is likely to be due to increased smoking rates (Bennett and Plum,
1996). Lung cancer is fatal in over 88 percent of all cases (NCI, 1998).
Lung cancer is a malignancy within the lungs and may be localized or have
spread to multiple sites (Bennet and Plum 1996). All types of lung cancer
likely originate from a common pluripotent stem cell. There are four types
of lung cancer: squamous (epidermoid), adenocarcinoma, large cell, and
small cell (oat cell). The first three types are often grouped together as
non-small cell lung cancer (NSCLC). These three types, which comprise
75 to 80 percent of all lung cancers, have different natural histories and
respond differently than small cell lung cancer to therapies. There are also
very rare types of lung cancer (with an approximate incidence of two
percent) that include adenosquamous mixed tumor or mixed small cell and
non-small cell histologies (Feld et al., 1995). Survival data from the
National Cancer Institute (1998) and cost data from Baker et al. (1989)
that are used in this chapter do not provide quantitative information for
different types of lung cancer. Consequently, this chapter contains an
Chapter 11.5 11.5-2 Cost of Lung Cancer
-------
evaluation of all types of lung cancer in aggregate. In addition, most risk
assessments that would be used in evaluating benefits do not specify the
type of lung cancer. If a specific type of lung cancer is of concern, Feld et
al. (1995) may be consulted for additional information regarding prognostic
information and treatment; however, the quantitative data are limited.
New discoveries may, have an impact on the diagnosis and understanding
of the causality of lung cancer in the future. Cytogenic abnormalities have
been demonstrated in lung cancer cells including lesions in chromosome
region 3p, which occurs as an early event in the biology of the tumor.
Mutation of the p53 gene is the most frequently identified genetic change in
human lung cancer. Activated proto-oncogenes are seen in lung cancer
and may arise from point mutations in the level of expression. Some of the
above changes may be used in the future for early detection of lung cancer.
Screening programs in the past, using more traditional diagnostic measures,
have not been successful in reducing lung cancer mortality among study
participants (Feld et al., 1995). l
Lung cancer occurs with much greater frequency among the elderly, which
is typical of most cancers. The average age at diagnosis is approximately
68 years. Less than two percent of lung cancers are diagnosed before the
age of 40 and five percent are diagnosed over the age of 85 (NCI, 1998).
The distribution of the age at diagnosis of lung cancer is shown in Figure
II.5-1. The steep incline in the probability of lung cancer diagnosis is clear
in this diagram, with a peak around 68 years of age. Approximately 50
percent of all lung cancer cases are diagnosed in the relatively small age
interval of 60 to 75 years. The data used to generate Figure II.5-1 are
shown in Table II.5-1. The cumulative percents of lung cancer at various
ages were calculated using the population-weighted distribution of
occurrence.
lrThe above information is not currently used in a manner that alters the survival or costs associated
with lung cancer.
Chapter 11.5 11.5-3 Cost of Lung Cancer
-------
Figure 11.5-1
-------
II.5.A.2 Concurrent Effects
As with all cancers, lung cancer may spread to other organs. In addition,
treatment of cancer, which usually includes chemotherapy, radiation, and
surgery, has numerous adverse side effects and may, in itself, lead to death.
For example, vinca alkaloids, used in lung chemotherapy, cause peripheral
neuropathy in some patients. Radiation treatments of cancer have led to
increased risks of other types of cancer, sterility, etc. Surgery, especially
the removal of a lung, may cause long-term changes in health status,
including reduced capacity or increased susceptibility to respiratory disease
that may lead to death. These effects are associated with additional
medical costs not considered in this chapter.
There is a strong link between lung cancer and smoking. Lung cancer
patients are much more likely to have smoked than people who have not
been diagnosed with lung cancer. Smoking is also associated with
increased risks of many other diseases, including other cancers. There is
no indication, however, that lung cancer causes these other diseases. The
simultaneous or sequential occurrence of the diseases are likely due to their
common causal link to smoking.
No data were located indicating that concurrent effects unrelated to lung
cancer or its treatment were likely to occur as a result of this disease,
although the same pollutants that cause lung cancer, especially respiratory
irritants (e.g., silica, nickel), may cause other adverse effects. These effects
can incur added medical costs not considered in this chapter. The risk
assessment that serves as the basis for a benefits evaluation should include
all adverse effects that are anticipated to result from exposure to the agent
of interest.
II.5.A.3 Causality & Special Susceptibilities
As noted above, lung cancer is caused by exposure to tobacco smoke. It is
also associated with certain air pollutants, such as radon and silica, and
chemical pollutants. Table II-l in Chapter II contains a list of chemicals
known to cause or suspected of causing cancer (as reported in the EPA
databases IRIS, HEAST, and HSDB). Most chemicals in the table were
carcinogenic in animal studies. These studies do not provide organ-specific
data because it is not generally assumed that cancer induction will
necessarily occur at the same site in humans as in animals. Consequently,
the chemicals listed in Table II-l may cause lung cancer and/or other types
of cancer. Evaluation of the likelihood of this occurrence would require
additional research (risk assessment).
Link to Table II. 1-1
Chapter 11.5 11.5-5 Cost of Lung Cancer
-------
Link to Chapter 1.1
NCI provides age-, sex-, and race-specific data regarding diagnosis of lung
cancer from 1990 to 1994, which may be used to evaluate susceptibilities
among population subgroups. The data must be used with care because
diagnostic rates indicate occurrence only, and may, or may not indicate
differences in susceptibility. See Chapter I.I for a more detailed discussion
of susceptibilities.
NCI lung cancer diagnosis and mortality data show higher diagnosis and
death rates among men than women. Lung cancer rates have been
declining over the past 22 years among males under 65 years of age,
however, while, lung cancer has been increasing in women. These
dynamics have been attributed to a tapering off of smoking rates in males
and a rapidly increasing rate of smoking among women in recent decades
(NCI, 1998).
The rate of diagnosis among black males in 1994 was approximately 50
percent higher than among white males, which cannot be fully explained by
smoking differences. The increased rate of lung cancer among black males
may be due to exposure to pollutants in the workplace or ambient
environment, exposure to carcinogens through other sources, or an
inherently greater susceptibility to lung cancer among blacks (NCI, 1998).
Some genetic factors have been identified that may increase the risk of lung
cancer. Individuals who metabolize debrisoquine readily, as well as those
lacking the MU phenotype of glutathione transferase, have an increased
lung cancer risk. There is also evidence for Mendelian inheritance of lung
cancer, indicating the importance of family history (Feld et al., 1995).
II.5.A.4 Treatments and Services
As noted above, lung cancer is usually treated with surgery,
chemotherapy, and/or radiation, depending on the type of lung cancer, the
stage of cancer at diagnosis, patient health, and other factors. The
treatment of lung cancer can be defined more precisely by histologic type
and specific location of the cancer in the lung. In this analysis, which is
concerned with the average cost for all lung cancers, all histologic types
and sub-sites are considered together.
Treatment is carried out in phases including initial diagnosis, initial
treatment, follow-up and maintenance treatment, and, for those who do not
survive, terminal treatment and palliative care. Some components of each
treatment are unique to each phase, but most medical activities and services
may occur more than once over the course of the disease from diagnosis to
death or cure. For example, X-rays may be used in diagnosis, to provide
ongoing status updates, to assist in determining initial and subsequent
surgical and other treatment interventions, etc.
Chapter 11.5 11.5-6 Cost of Lung Cancer
-------
Initial diagnostic activities may include an evaluation of signs and
symptoms, chest X-rays, computed tomography (CT) of the chest,
magnetic resonance imaging (MRI) of the chest, sputum cytological
analysis, percutaneous aspiration of pulmonary nodules, bronchoscopy,
mediastinoscopy, thorascopy, thoracotomy, and other procedures. Staging
of the disease occurs during this phase and is critical to determination of
subsequent medical actions (Feld et al., 1995). Surgery is usually
performed, and is associated with a relatively low mortality rate. Radiation
and/or chemotherapy may be done with or without surgery. In many
patients cancer has spread to the central nervous system, abdomen, bone
marrow, and other areas, requiring additional treatment strategies (Feld et
al., 1995).
Due to its poor prognosis, most patients receive terminal care. This care
may include a variety of medical services, long-term care in a nursing
facility, palliative care, family counseling, etc.
II.5.A.5 Prognosis
Link to Chapter II. 2
II.5.A.5.1 Background
As noted above, the overall prognosis for lung cancer patients is poor, with
an average of 88 percent of patients dying of the disease within 10 years.
Most deaths from lung cancer occur in the first four years, and
approximately 60 percent of all patients die during the first year (NCI,
1998). Patients with early stages (I and II) of the disease have a 40 to 85
percent five-year survival rate (Bennet and Plum, 1996). Unfortunately,
most diagnoses occur at later stages of the disease. Factors such as tumor
size and location, histology, involvement of nodes, and the spread of cancer
to other tissues affect outcome. Numerous new biochemical and
immunological tests are used to provide additional information on the likely
outcome (Feld et al., 1995).
II.5.A.5.2 Relative Survival Rates (RSRs)
The NCI SEER data reports were accessed online to obtain information
regarding mortality and survival probabilities and the duration after
diagnosis until death (NCI, 1998). Basic survival statistics on lung cancer
are provided in this section because they relate to prognosis. Methods
used to convert the NCI statistics to survival probabilities are discussed
briefly in this section and in detail in Chapter II.2 on stomach cancer.
NCI provides the RSR for each year post-diagnosis. The RSR is the
number of observed survivors among these patients, divided by the number
Chapter 11.5 11.5-7 Cost of Lung Cancer
-------
of "expected" survivors among persons with the same age and gender in
the general population (observed/expected). The equation for this is:
RSR =
observed survival rate among cancer patients
survival rate among age- and sex-matched cohort in the general population
The RSR takes into account that there are competing causes of death that
increase with age. The RSR for lung cancer patients during the first year post-
diagnosis is 41 percent (NCI, 1998). A person with lung cancer would
therefore have, on average, a one-year survival probability that is 41 percent
of someone of the same age and gender in the general population. The RSRs
provided by NCI for each year post-diagnosis are averages over all ages at
diagnosis. An evaluation of the RSRs over the past 20 years indicates that
they have increased by about 33 percent, with most of the progress occurring
in the early 1970s (NCI, 1998). The most current information, the rates for
1988 through 1993, was used for the first through fifth years post-diagnosis
Ten years of data were used to estimate survival for the sixth through tenth
years post-diagnosis due to the need for a longer time span. Table II.5-2 lists
the average RSRs for lung cancer for the first ten years post-diagnosis. The
RSRs shown in Table II.5-2 were used to derive survival probabilities for lung
cancer patients for each of the first ten years post-diagnosis.2
Table 11.5-2. Average RSRs* for Lung Cancer for the First 10
Years Post Diagnosis
Years Post-Diagnosis (n)
1
2
3
4
5
6
7
8
9
10
Average RSR for n Years
Post-Diagnosis
0.41
0.24
0.18
0.15
0.14
0.12
0.12
0.11
0.10
0.10
The average RSR for each year post-diagnosis is the average of a set of
RSRs reported by NCI (1998) as described in the text above.
2 All vital statistic data in this document applicable to the general population were obtained from
the National Center for Health Statistics (NCHS) Vital Statistics in the United States (NCHS, 1993).
Chapter 11.5
1.5-8
Cost of Lung Cancer
-------
Although most lung cancer patients will die of lung cancer, some may die
of other causes. The probability of a lung cancer patient dying of causes
other than lung cancer cannot be assumed to be the same as the probability
of someone in the general population dying of other causes, particularly in
the first few years post-diagnosis, when a lung cancer patient's probability
of dying of lung cancer is quite high.3
The probability of a lung cancer patient dying of lung cancer and the
probability of a lung cancer patient dying of some cause other than lung
cancer in the nth year post-diagnosis, given survival to the nth year, were
each derived from two known probabilities:
1) the probability of a lung cancer patient surviving through the nth year
post-diagnosis, given survival to the nth year, and
2) the probability of a lung cancer patient dying of causes other than lung
cancer in a matched cohort in the general population.
The derivation is explained in detail in the Appendix to Chapter 11.2.
Link to Chapter II. 2, Appendix II. 2-A
Because each of the known probabilities depends on the number of years post-
diagnosis and (minimally) on age at diagnosis, the derived probabilities were
calculated for each of the ten years post-diagnosis and for the average age at
diagnosis (68 years).4 The following probabilities are shown in Table II.5-3:
1) survival through the nth year,
2) dying of lung cancer during the nth year, and
3) dying of some other cause during the nth year.
Probabilities of survival and dying of all causes among all members of the
general population aged 68 were obtained from the National Center for
Health Statistics (NCHS) Vital Statistics in the United States (NCHS,
3 This difference becomes clear in the extreme case in which the probability of dying of an illness is
extremely high. Suppose, for example, that the probability of dying of all causes except for illness X is
0.025 in the general population. Suppose that in a cohort of patients diagnosed with illness X, the
probability of dying from illness X in the first year post-diagnosis is 0.99. If the probability of dying of
other causes in this cohort equaled that in the general population (0.025), then the probability of someone in
the cohort dying would be greater than 1.0.
4Ten years is a generous follow-up period during which most individuals who will die of lung
cancer have done so. It is also used as a reasonable maximum duration of maintenance care and treatment
for those who do not die of lung cancer.
Chapter 11.5 11.5-9 Cost of Lung Cancer
-------
1993). These probabilities are also shown in Table II.5-3. The values in
this table are used in Section II.5.B to calculate the expected medical costs
of lung cancer patients.
Table 11.5-3. Probabilities of Survival and Mortality for Lung Cancer Patient Diagnosed at Age
68a
Years post-
diagnosis
(n)
A Cohort in the General
Population (Matched)
Probability
of
surviving n
years
Probability
of dying in
nth year of
causes
other than
lung
cancer,
given
survival to
the nth
yearb
A Cohort of Lung Cancer Patients
Relative
Survival
Rate0
Probability
of
surviving
through
the nth
year post-
diagnosis'1
Probability
of dying of
lung
cancer in
the nth
year post-
diagnosis6
Probability
of dying of
other
causes in
the nth
year post-
diagnosis'
0
1.000
1.0
.977
.023
.41
.401
.586
.014
.953
.025
.24
.229
.165
.007
.928
.027
.18
.167
.057
.005
.901
.029
.15
.135
.028
.004
.872
.031
.14
.122
.009
.004
.843
.034
.12
.101
.018
.003
.812
.037
.12
.097
.000
.003
.779
.040
.11
.086
.008
.003
.745
.044
.10
.075
.008
.003
10
.710
.047
.10
.071
.000
.003
a. The survival and mortality probabilities for lung cancer patients presented here are derived from the RSRs
obtained from NCI and the survival probabilities for a matched cohort in the general population. They are
therefore only estimates of the underlying population survival and mortality probabilities for lung cancer patients.
Whereas the underlying population probabilities are likely to follow a smooth trend (over years post-diagnosis),
the estimates exhibit some of the "bumpiness" around that trend that typically results from normal sampling
variability. This variability will also be true of any other probabilities (such as the conditional probabilities
discussed below) that are derived from the estimated probabilities shown here.
b. The probabilities in the general population of dying from lung cancer are 0.000256 in the 70-74 year age
group, and 0.000348 in the 75-79 year age group. The probabilities in column (3) were derived by subtracting
these probabilities from the corresponding probabilities of dying from any cause in the nth year, given survival to
the nth year.
c. From Table II.5-2.
d. See Chapter II.2 for an explanation of the derivation of these probabilities.
Chapter II.5
11.5-10
Cost of Lung Cancer
-------
II.5.B Costs of Treatment and Services
II.5.B.1 Methodology
Link to Chapter 1.1
II.5.B.1.1 Overview
There is no single typical case or treatment pattern for lung cancer because
of individual differences in the stage of cancer at diagnosis, multiple
treatment options, patient health and age, and other factors; however,
average costs can be calculated. Treatment of lung cancer may occur over
a briefer extended period of time, and costs may be limited or substantial.
Lung cancer has a relatively high mortality rate of 88 percent, as discussed
in Section A. The medical costs of those who die of the disease are usually
very different than for those who survive (this is discussed in more detail in
Chapter 1.1). This chapter therefore provides costs for the "average" lung
cancer patient, as well as for survivors and nonsurvivors as separate patient
groups.
II.5.B. 1.2 Medical Cost Data
II.5.B.1.2.1 Sources
Medical cost data would ideally be obtained on current medical
expenditures. Although data files are maintained by public and private
sector sources, they are not readily available. In addition, to obtain reliable
cost estimates it is necessary to evaluate very large databases of charges
from a variety of sources. This activity was not practical for the
development of this chapter. A data search was conducted to locate
information in the medical economics literature regarding medical costs
associated with lung cancer. In addition to a literature search, most federal
agencies dealing with cancer, disabilities, medical costs and their
management, and related issues were contacted for information and the
various federal databases were discussed with senior staff at these agencies.
Very recent cost data were not located.5 Current (1994) cancer data were
obtained regarding incidence and survival (as reported in Section II.5.A,
above), and were used with cost data from the 1980s described below. The
cost estimates presented in this section are based primarily on the work of
Baker et al. (1989) and Hartunian et al. (1981) and on two sources of
statistical data: the National Cancer Institute (1998) and Vital Statistics of
the United States, 1993 (NCHS, 1997). These data were evaluated and
cost elements were used to calculate lifetime estimates of the direct medical
costs due to lung cancer.
5 Studies were located that used more recent cost data than the data used in this analysis. Serious
limitations existed, however (data were incomplete), and so the studies were not used. They are reported in
the "Other Studies" section at the end of Section B.
Chapter 11.5 11.5-11 Cost of Lung Cancer
-------
Based on the 1997 review of the medical literature carried out for the
development of this chapter, there do not appear to be widely-adopted new
treatment methods for lung cancer that substantially alter either the medical
costs or the survival rates for most patients. Consequently, the cost
estimates presented in this chapter may be considered appropriate under
most circumstances (e.g., regional costs may vary).
II.5.B.1.2.2 Baker et al.'s Cost Estimation Method
Baker et al. (1989) used the Continuous Medicare History Sample File
(CMHSF) to estimate the per-patient average lifetime medical cost of
treating lung cancer based on data files from 1974 to 1981. They chose
CMHSF because:
1) it is a nationally representative sample of the Medicare population
(five percent), covering over 1.6 million patients;
2) it is longitudinal, dating from 1974 to 1981; and
3) it captures the majority of medical expenses for each beneficiary.
Five Medicare files are included in the CMHSF, which cover:
1) inpatient hospital stays,
2) skilled nursing facility stays,
3) home health agency charges,
4) physicians' services, and
5) outpatient and other medical services.6
Costs that were not included are outpatient prescription medications and
nursing home care below the skilled level.
Because CMHSF provides no indication of initial diagnosis, Baker et al.
assumed that disease onset occurred when a diagnosis of lung cancer was
listed on a hospitalization record following a minimum of one year without
a lung cancer diagnosis. This assumption is reasonable due to the high
frequency of hospitalization associated with the disease (i.e., individuals
diagnosed with lung cancer would be hospitalized). Only patients with an
initial diagnosis during the years covered by the database (1974-1981) were
included.
6 See Baker et al. (1989 and 1991) for further details. Baker et al. (1991) contains additional
descriptive data regarding the database and methods used for the cost analysis; however, it does not contain
cost data for lung cancer.
Chapter 11.5 11.5-12 Cost of Lung Cancer
-------
Link to Chapter 1.1
Costs associated with lung cancer were assigned to three post-diagnostic
time periods:
• initial treatment, during the first three months following diagnosis;
maintenance care, between initial and terminal treatment; and
• terminal treatment during the final six months prior to death.
As noted in Chapter 1.1, the amount paid for service may differ from the
actual medical costs because many insurers and federal programs either 1)
pay only a portion of total costs or 2) pay more than actual costs to
underwrite the care providers' losses due to underpayment from other
sources.
Baker et al. used provider charges, rather than Medicare reimbursements
(which represent only a portion of most total charges), thus providing a
more accurate cost estimate. To improve the accuracy of the cost
estimates, Baker et al. included cost data on coinsurance, deductibles, and
other cost components. They made four adjustments to the cost estimates
calculated from the CMHSF. First, charges were added for skilled nursing
facilities (SNFs) not covered by Medicare by multiplying the "length of
stay" at an SNF (computed from admission and discharge dates) by the
average daily SNF charge. Second, the annual Medicare Part B deductible
of $60 was added to the reimbursed charges in the database. Third, since
Medicare pays only 80 percent of physicians' charges, Baker et al. scaled
these reimbursements to 100 percent of physicians' charges to better reflect
social costs. Finally, they inflated all dollar values to 1984 dollars using the
Medical Care component of the Consumer Price Index.
II.5.B.1.2.3 Cost Estimates by Treatment Period
Medical costs associated with the initial, maintenance, and terminal cancer
care treatment periods were itemized in Baker et al. (1989) and are shown
in Table II.5-4. The 1989 paper did not report incremental costs or the
costs of other medical services that would be anticipated to occur while the
patient was receiving cancer treatment (i.e., co-morbidity/background
costs). In order to estimate the incremental costs, a co-morbidity cost of
$2,988 per year (1984 dollars) from Baker et al. (1991) was used in this
analysis. (This is equivalent to $6,394 in 1996 dollars using the CPI
multiplier of 2.14 for 1984 to 1996.) The co-morbidity cost was pro-rated
for this analysis using the specified durations for the initial (three-month)
and terminal (six-month) treatment periods.
Table II.5-4 lists the incremental costs calculated for the three treatment
periods. Total costs are reported for the initial and terminal care periods.
Annual costs for the maintenance period are shown and are further
discussed in the "Lifetime Costs" section below. Using the Medical Care
Chapter 11.5 11.5-13 Cost of Lung Cancer
-------
component of the Consumer Price Index (CPI-U), all costs are inflated to
1996 dollars for purposes of this handbook. (The adjustment factor for
1984 to 1996 is 2.14; Bureau of Labor Statistics.)
Table 11.5-4. Average Per Patient Costs for the Three Periods of
Treatment for Lung Cancer in 1996 dollars
Costs adjusted for inflation using the Medical Care component of the
Consumer Price Index (CPI-U) 1 996:1 984 = 2.14 (Bureau of Labor)
Treatment Period
Initial
(3 months)
Maintenance (per year)
Terminal
(6 months)
Incremental Cancer Treatment
Cost
$26,042
$11,325
$30,112
(Based on Baker et al., 1989, with comorbidity charges from Baker et al., 1991.
II.5.B. 1.3 Calculation of Lifetime Cost Estimates for the
"Average" Lung Cancer Patient
This section contains a discussion of the calculation of lifetime medical
costs for the "average" lung cancer patient. Sections that follow discuss
methods and results of calculations for estimating costs for survivors and
nonsurvivors of lung cancer separately. These separate approaches were
used to address specific requirements of different activities that EPA
carries out using direct medical cost data. Although Baker et al. (1989)
provide useful cost estimates for the three treatment periods, they do not
provide information on two critical aspects of medical costs:
1) costs for survivors versus nonsurvivors of lung cancer. These values
may differ substantially. For example, survivors would not have
terminal care costs and may receive maintenance services for an
extended time period; and
2) estimates of the duration of the maintenance periods.
Data regarding age at diagnosis of lung cancer were obtained from NCI
(1998). Survival and mortality probabilities for each year post-diagnosis
were derived from relative survival rates obtained from NCI (1998), as
discussed in Section II.5.A.5.2.
Link to Section II. 5.A. 5.2
This information was used to address many time-related medical cost
issues. For some aspects of the analysis, however, detailed information
Chapter 11.5
11.5-14
Cost of Lung Cancer
-------
was not available and average values have been used as a reasonable
approximation (e.g., a ten-year maintenance period was assumed for
survivors of lung cancer). When average values or other assumptions are
used in this analysis, they are so noted.
As previously noted, there are not substantial differences in survival related
to age at diagnosis, and NCI does not provide age-specific RSRs for each
year post-diagnosis. Consequently, it was assumed for this analysis that the
relative survival rates for lung cancer were the same for all ages. The
survival and mortality probabilities for lung cancer patients, which are
incorporated into calculations of expected medical costs as discussed
below, are based on this assumption.
The analysis assumes that death always occurs midyear. All lung cancer
patients are therefore assumed to incur the costs of initial treatment during the
first three months of the illness. The costs incurred after that during the first
year depend on whether the patient:
(1) survives through the year,
(2) dies of lung cancer during the year, or
(3) dies of some other cause during the year.
Patients who survive through the year incur the costs of initial treatment
($26,042) during the first three months, and then incur nine months' worth of
maintenance care costs (0.75 x $11,325 = $8,494) during the remainder of
the year. The total cost incurred during the first year by those patients who
survive the year is therefore $26,042 + $8,493 = $34,535.
Lung cancer patients who die of lung cancer during the first year incur the
initial treatment cost and then incur terminal care costs for the remaining three
months of their lives (because those who die are assumed to die midyear).
Total costs during the first year post-diagnosis in this case are therefore
$26,042 + (0.5 x $30,112) = $41,098.
Finally, the small percentage of lung cancer patients who die of causes
other than lung cancer during the first year post-diagnosis incur the initial
treatment costs and then incur three months' worth of maintenance care
costs. Total first-year costs for these patients are therefore $26,042 + 0.25
x $11,325 = $34,536.
The expected medical costs for lung cancer patients during the first year
post-diagnosis, then, may be expressed as:
Chapter 11.5 11.5-15 Cost of Lung Cancer
-------
Expected First-Year Cost: initial treatment costs +
[maintenance care costs for nine months x probability of
survival through first year + terminal care costs for three
months x probability of dying of lung cancer during first
year + maintenance care costs for three months x
probability of dying of other causes during the first year]
Example: Expected first-year medical costs of a lung cancer patient
diagnosed at age 68
As noted above, all lung cancer patients incur an initial treatment cost of
$26,042. Those who survive through the year also incur maintenance care
costs for the remaining three quarters of the year. The total first-year costs
of those who survive the year are:
Initial treatment: $26,042
Maintenance treatment: $8,493 (.75 x $11,325)
Total First-Year Cost $34,535
More than half of lung cancer patients die of lung cancer during the first
year. Those who do will incur the initial treatment costs plus half of the
terminal care costs. The total first-year costs of those who die of lung
cancer during the year are:
Initial treatment: $ 26,042
Terminal care: $15,056 (.50 x $30,112)
Total First-Year Cost $41,098
Finally, a small percentage of patients will die of competing illnesses during
the first year. Because those who die of causes other than lung cancer are
assumed to die at the midpoint of the year, costs during the first half of the
year are assumed to consist of the initial treatment costs for three months,
plus three months of maintenance care costs as follows:
Initial treatment: $26,042
Maintenance treatment: $2,831 (.25 x $11,325)
Total First-Year Cost $28,873
For each subsequent year, costs consist entirely of maintenance care costs
for those who survive the year. For those who do not survive the year,
costs depend on whether death was due to lung cancer or other causes.
For those who die of lung cancer during the wth year, costs incurred that
year consist of six months of terminal care costs, or $30,112. For those
Chapter 11.5 11.5-16 Cost of Lung Cancer
-------
who die of other causes during the nth year, there are six months of
maintenance care costs, or 0.5 x $11,325 = $5663.
The expected first-year medical cost incurred by the "average" lung cancer
patient diagnosed at age 68 is a weighted average of the costs of those who
survive the first year, those who die of lung cancer during the first year,
and those who die of other causes during the first year, where the weights
are the probabilities of each of these occurrences. The weighted average
medical cost were calculated for ten years post diagnosis, and expected
costs were summed over the ten years. This was assumed to be a
reasonable period over which additional medical costs associated with lung
cancer (i.e., maintenance care costs) would be incurred by lung cancer
patients. In reality, there may be follow-up care and continued testing over
a longer period; however, no data were available regarding those costs.
They would certainly be less than $11,325 per year.
The expected medical costs for lung cancer patients during the wth year
post-diagnosis, for «>1, then, may be expressed as:
Expected nth Year (n>l) Cost: [maintenance care cost for
one year x probability of survival through «th year +
terminal care cost for six months x probability of dying of
lung cancer during the «th year + maintenance care cost for
six months x probability of dying of other causes during the
ttth year]
Expected Lifetime Cost = Expected first-year cost + the sum
of the (discounted) expected subsequent-year costs
The first year of treatment is calculated differently from other years
because the first three months of that year are spent in "initial" treatment,
and the costs for that period of intensive medical care and surgery are
calculated separately.
The mathematical equation for the expected lifetime medical costs incurred
by the "average" lung cancer patient over a ten-year period is:
$26,042 + ($11,325 x 0.75 x pSi) + ($11,325 x 0.25
10
+E
v=2
$11,325
(1 -
y-l'
_$5,663
(1
\y-i'
+ ($30,112 x 0.5
$30,112
(1
:)
Where:
y
ps
pm
sc =
the year post-diagnosis,
the probability of surviving through the year,
the probability of dying of lung cancer during the year
Chapter 11.5
11.5-17
Cost of Lung Cancer
-------
Link to Table II. 5-3
pm° = the probability of dying from other causes during the
year,
r = the discount rate
The cost estimates for each year post-diagnosis and the estimate of
undiscounted expected total cost for a ten year period are shown in Table
II.5-5 for the "average" lung cancer patients diagnosed at age 68. The
survival and mortality probabilities necessary for the calculations of costs
are shown in Table II.5-3.
II.5.B.1.4 Calculation of Lifetime Cost Estimates Separately
for Lung Cancer Survivors and Nonsurvivors
II.5.B.1.4.1 Survivors and Nonsurvivors
As noted above, there are differences in medical services provided to lung
cancer patients who survive the disease (survivors) versus those who die of
the disease (nonsurvivors). Based on cost estimates by Baker et al. (1989),
terminal care is provided for approximately six months to terminally ill
cancer patients. The costs to nonsurvivors for this care ($30,112) is
considerably higher than costs for survivors who receive maintenance care
for the same period of time ($5,662).7
EPA may use the value of a statistical life (VSL) for nonsurvivors, and thus
separate costs for survivors and nonsurvivors were calculated. The method
shown above to calculate costs for the "average" patient uses the
unconditional probabilities of survival and mortality listed in Table II.5-3.
The method used to calculate costs for survivors and nonsurvivors
separately requires the probabilities that are conditional on being either a
survivor or nonsurvivor of lung cancer.
7 Nonsurvivors include only those who die of lung cancer and do NOT include those who die of
any other causes.
Chapter 11.5 11.5-18 Cost of Lung Cancer
-------
Table 11.5-5. Expected Costs of Medical Services (in 1996$) for Lung Cancer Patients
(Age of Onset = 68)
Years Post-
Diagnosis (n)
1b
2
3
4
5
6
7
8
9
10
Medical Costs in the nth Year (undiscounted)
if survive
through the
nth year
34,535
11,325
11,325
11,325
11,325
11,325
11,325
11,325
11,325
11,325
if die of
lung cancer
in the nth
year
41,098
30,112
30,112
30,112
30,112
30,112
30,112
30,112
30,112
30,112
if die of other
causes in the nth
year
28,873
5.662
5,662
5,662
5,662
5,662
5,662
5,662
5,662
5.662
Expected Total Cost Through the 10th Year Post-Diagnosis for a Lung
Cancer Patient Diagnosed at Age 68
Expected Medical
Costs for the nth Year
Post-Diagnosis
(Discounted)
38,300
7,601
3,636
2,393
1,684
1,694
1,133
1,239
1,101
831
59,612
a. The probabilities listed in this table are from Table II. 5-3. The costs are listed in Table II. 5-4.
b. First-year costs include the charge for "initial" therapy ($26,042). The duration of maintenance care is adjusted
accordingly (see text for discussion).
c. Calculated using the probabilities in Table II. 5-3 and the costs in Columns (5), (6), and (7) of this table.
The conditional probability of a lung cancer nonsurvivor dying in the nth
year is the number of nonsurviving lung cancer patients who die of lung
cancer during the nth year divided by the total number of lung cancer
nonsurvivors. Likewise, the conditional probability of a lung cancer
survivor dying in the nth year is the number of surviving lung cancer
patients who die of lung cancer during the nth year divided by the total
number of lung cancer survivors. A detailed explanation of the derivation
of these values is provided in Chapter 112. The conditional probabilities of
survival and mortality for survivors and nonsurvivors of lung cancer are
given in Table 11.5-6.
Link to Chapter II. 2
Chapter 11.5
11.5-19
Cost of Lung Cancer
-------
Table 11.5-6. Conditional Probabilities of Survival and Mortality for Survivors and Nonsurvivors
of Lung Cancer (Age of Onset = 68) a
Years Post-
Diagnosis
(n)
1
2
3
4
5
6
7
8
9
10
Lung Cancer Survivors
Conditional probability of:
Surviving
through the nth
year
.886
.829
.791
.758
.727
.699
.671
.643
.616
.589
Dying of some
other cause
during the nth
year
.114
.056
.039
.033
.030
.029
.028
.028
.027
.027
Lung Cancer Nonsurvivors
Conditional probability of:
Surviving
through the nth
year
.334
.146
.081
.050
.039
.019
.019
.009
.00
.00
Dying of lung
cancer during
the nth year
.666
.188
.065
.032
.011
.020
.000
.009
.009
.000
a. As noted for Table II. 5-3, the survival and mortality probabilities for lung cancer patients presented here are
derived from the relative survival rates obtained from NCI and the survival probabilities for a matched cohort in
the general population. They are therefore only estimates of the underlying population survival and mortality
probabilities for lung cancer patients. Whereas the underlying population probabilities are likely to follow a
smooth trend (over years post-diagnosis), the estimates exhibit some of the "bumpiness" around that trend that
typically results from normal sampling variability. This variability will also be true of any other probabilities (such
as the conditional probabilities discussed below) that are derived from the estimated probabilities shown here.
II.5.B.1.4.2 Calculation of Lifetime Cost Estimates for Lung
Cancer Survivors
As shown in the example portion of Section II.5.B.1.3, cost estimates are
calculated by summing the costs of the different treatment phases over the
lifetime of the lung cancer patient. The expected medical costs for lung
cancer survivors during the first year post-diagnosis may therefore be
expressed as:
Expected First-Year Cost: initial treatment costs +
[maintenance care costs for nine months x probability of
survival through first year + maintenance care costs for
three months x probability of dying of other causes during
the first year]
Chapter II.5
II.5-20
Cost of Lung Cancer
-------
The expected medical costs for lung cancer survivors during the wth year
post-diagnosis, for ri>l, then, may be expressed as:
Expected nth Year («>1) Cost: [maintenance care cost for 1
year x probability of survival through «th year +
maintenance care cost for six months x probability of dying
of other causes during the «th year]
Expected Lifetime Cost = Expected first-year cost + the sum
of the (discounted) expected subsequent-year costs
Note that the probabilities used in these calculations are the conditional
probabilities given in Table II.5-6. They are conditional on the lung cancer
patient not dying of lung cancer.
Using the initial, maintenance, and terminal care costs from Table II.5-6,
the mathematical equation for the lifetime costs incurred by lung cancer
survivors is:
$26,042 + pm^ x 0.25 ($11,325) + ps^ x .75 x $11,325
10
PSy
$11,325
(1
v-i
+ pm
$5,662
1 + ry
-i
where: y
pss
pm
the year post-diagnosis
the conditional probability of survival for that year,
conditional on being a survivor of lung cancer
the conditional probability of mortality for that year,
conditional on being a survivor of lung cancer
the discount rate.
The expected medical costs for lung cancer survivors for each year post-
diagnosis, as well as the expected total medical costs over ten years post-
diagnosis, are shown in Table II. 5-7.
Chapter 11.5
11.5-21
Cost of Lung Cancer
-------
Table 11.5-7. Expected Undiscounted Costs of Medical Services (in 1996$) for
Survivors of Lung Cancer (Age of Onset = 68)
Years
Post-
Diag-
nosis
(n)
1d
2
3
4
5
6
7
8
9
10
Medical Costs Through the 10th Year Post-diagnosis3 (undiscounted)
Medical Cost if
Survive Through
the nth Year
34,535
11,325
11,325
11,325
11,325
11,325
11,325
11,325
11,325
11,325
Medical Cost if Die
of other Causes in
the nth Year
28,873
5,662
5,662
5,662
5,662
5,662
5,662
5,662
5,662
5,662
Expected Total (Undiscounted) Cost Through the
10th Year Post-Diagnosis:
Total Cost Based on
Weighted Average0
33,889
9,713
9,173
8,768
8,411
8,077
7,756
7,438
7,124
6,818
107,167
a. Costs are based on data reported in Table II. 5-4, adapted from Baker et al., 1989.
b. Probabilities of survival and mortality, taken from Table II. 5-6, are conditional on surviving
lung cancer.
c. Weighted average of the costs incurred by survivors who survive the year and the costs
incurred by survivors who die of other causes during the year. Weighting is based on the
conditional probabilities provided in Table II. 5-6.
d. Costs during the first year include a charge for "initial" therapy ($26,042), and the duration
of maintenance or terminal care is adjusted accordingly. See text for discussion.
II.5.B. 1.4.3 Calculation of Lifetime Cost Estimates for Lung
Cancer Nonsurvivors
Nonsurvivors of lung cancer will incur initial, maintenance, and terminal
costs. Their lifetime medical costs associated with the disease can be
calculated from the costs per treatment period shown in Table 11.5-4 and
the conditional probabilities for nonsurvivors of lung cancer shown in Table
II.5-6.
As Table II.5-6 indicates, most lung cancer patients who will ultimately die
of lung cancer do so in the first few years post-diagnosis. About 85
percent die in the first two years. Deaths from lung cancer after the first
four years are minimal. As with lung cancer survivors, medical costs for
nonsurvivors each year post-diagnosis were calculated as a weighted
average of the costs incurred by those who survive the year and those who
die (of lung cancer) during the year.
Chapter 11.5
11.5-22
Cost of Lung Cancer
-------
It was assumed that those who die during a year receive six months of care
(as was done for the survivors above). It was also assumed that terminal
care lasting six months would be provided to all nonsurvivors. Therefore,
unless death occurred during the first year, when initial care was assumed
to occur, the care costs which were assigned to the last year of life were
terminal costs. If death occurred during the first year post-diagnosis, it
was assumed that initial care and three months (one-half of the total) of
terminal care were provided.
The general description of medical costs for nonsurvivors may be
expressed as:
Expected First-Year Cost: [initial costs + one-half terminal
costs] x probability of mortality during the first year +
[initial costs + maintenance care costs for nine months] x
probability of survival for first year
Expected nth Year (n>l) Cost: maintenance care cost for 1
year x probability of survival through nth year + terminal
costs x probability of mortality in «th year
Expected Lifetime Cost = Expected first-year cost + the sum
of the (discounted) expected subsequent year costs
As with the cost calculations for lung cancer survivors, the probabilities
used in these cost calculations are the conditional probabilities given in
Table II.5-6: in this case, conditional on dying of lung cancer.
Using the initial, maintenance, and terminal care costs from Table II.5-6,
the mathematical equation for the expected lifetime costs incurred by
nonsurvivors is:
$26,042 + pm™ x 0.5 ($30,112) + ps™ x .75 x $11,325
10
+ E
y = 2
$11,325
_! + /»V
$30,112
v-l
where: y
ps'
,ns
pm
ns
the year post-diagnosis
the conditional probability of survival for that year,
conditional on being a nonsurvivor of lung cancer
the conditional probability of mortality for that year,
conditional on being a nonsurvivor of lung cancer
the discount rate.
Chapter 11.5
11.5-23
Cost of Lung Cancer
-------
The costs are summed over all years from diagnosis to death. Maintenance
care costs are not added in the last year of life because during the six
months that are assumed to constitute this period the patient is assumed to
receive terminal care. (The discounted results are shown in the "Results"
section that follows.) The approach is the same as that shown in the
example in Section II.5.B. 1.3 When the costs for each year are summed
over a period often years post-diagnosis, during which essentially all
patients who will die of lung cancer have done so, the total cost per
nonsurvivor is obtained. These costs are shown in Table II.5-8.
Link to Section II. 5.B. 1.3
The results shown above can be used to calculated costs for an "average" lung
cancer patient, from the costs calculated for survivors and nonsurvivors. The
expected medical costs of a lung cancer patient can be calculated as a
weighted average of the expected costs of survivors and nonsurvivors of lung
cancer. This approach, which was not used to calculated costs for the
"average" patient in this chapter, yields the same results as the approach
shown in Section II.5.B.1.3. A discussion of why these to approaches yield
the same results is provided in Chapter II.2 (Section II.2.B.2.3). In brief, the
approach used in this chapter for the average patient uses cost data for all
patients, weighted by their average utilization of services. If the survivor and
nonsurvivor data were used, which incorporates utilization of services, cost
results obtained through separate calculations for the two subgroups are
simply re-aggregated based on each group's proportional contribution to the
cost.
Link to Chapter II.2.B.2.3
Chapter 11.5 11.5-24 Cost of Lung Cancer
-------
Table 11.5-8. Expected Undiscounted Costs of Medical Services (in 1996$) for
Nonsurvivors of Lung Cancer (Age of Onset = 68)
Years Post-
Diagnosis
(n)
1d
2
3
4
5
6
7
8
9
10
Medical Costs Through the 10th Year Post-diagnosis3
(undiscounted)
Medical Cost if
Survive Through the
nth Year
34,535
11,325
11,325
11,325
11,325
11,325
11,325
11,325
11,325
11,325
Medical Cost if Die
in the nth Year
41,098
30,112
30,112
30,112
30,112
30,112
30,112
30,112
30,112
30,112
Expected Total (Undiscounted) Cost Through the 10th Year Post-
Diagnosis:
Total Cost Based
on Weighted
Average0
38,905
7,311
2,877
1,519
761
818
224
389
274
9
53,088
a. Costs are based on data reported in Table II. 5-5, adapted from Baker et al., 1989.
b. Probabilities of survival and mortality, taken from Table II. 5-6, are conditional on dying of lung
cancer within 10 years post-diagnosis.
c. Weighted average of costs incurred by nonsurvivors who survive the year and those who die
during the year. Weighting is based on the conditional probabilities shown in Table II. 5-6.
d. Costs during the first year include "Initial" therapy ($26,042) , and pro-rated maintenance or
terminal care. See text for discussion.
II.5.B.2 Results of Medical Cost Analysis
The per patient lifetime direct medical costs calculated for the "average"
lung cancer patient (as shown in Table II.5-5), lung cancer survivors (as
shown in Table II.5-7) and lung cancer nonsurvivors (as shown in Table
II.5-8) diagnosed at age 68 are listed in Table II.5-9. Undiscounted costs
and costs discounted at three, five, and seven percent back to year one
(time of diagnosis) are shown. Discounting was carried out for ten years
following diagnosis (which, for nonsurvivors, comprises the full duration of
treatment time because virtually all patients that are going to die of lung
cancer do so within ten years) and comprises the assumed full duration of
maintenance care for survivors.
Chapter 11.5
11.5-25
Cost of Lung Cancer
-------
Table 11.5-9. Incremental Per-capita Medical Costs for the Average Lung Cancer Patient,
Survivors, and Nonsurvivors (Diagnosed at Age 68) Undiscounted and Discounted at 3, 5, and
7 Percent ($1996)
Patient Group
Survivors
Nonsurvivors
Average Patient
Discount Rate
Undiscounted
$107,167
$53,088
$59,612
3%
$97,822
$52,215
$57,716
5%
$92,572
$51,692
$56,624
7%
$87,966
$51,211
$55,645
See text for a definitions of patient groups.
The results show much higher costs for survivors than nonsurvivors, due
primarily to their ongoing maintenance care. It is noted that although a
ten-year maintenance period for lung cancer survivors is assumed (with
adjustment for background mortality that reduces utilization), the actual
average period of maintenance is not known and is likely to vary
considerably among individuals, depending on age, health status, access to
care, and other factors. Most lung cancer patients (88 percent) die of the
disease, and their costs are the major cost element in determining the
"average" patient costs. The uncertainty surrounding the period of
maintenance care for survivors therefore does not have a substantial impact
on the cost estimates for the "average" patient.
II.5.B.3 Other Studies
The results of these studies are examined: Mor et al. (Draft 1990),
Hartunian et al. (1981), Oster et al. (1984), Riley et al., (1995). The Baker
et al. study has a combination of characteristics of study design and data
quality that makes it preferable to the other studies, as discussed below.
II.5.B.3.1 Moretal.
The Mor et al. (1990) study tracked prospectively eligible Medicare
beneficiaries in Rhode Island from 1984-1986 by "examining pathology
reports in nine Rhode Island hospitals." The medical records were linked
to Medicare inpatient and outpatient claims data. Medicare claims files
were reviewed, and the charges associated with lung cancer were
aggregated for the first year post-diagnosis. The study has three limitations
that make it less appropriate for use than the Baker et al. study:
1) the study was limited to Medicare beneficiaries in Rhode Island;
2) the study has not yet been submitted to a refereed journal;
3) the study covers only one year of treatment;
4) costs are only those reimbursed by Medicare, rather than all costs.
Chapter 11.5
11.5-26
Cost of Lung Cancer
-------
Link to Chapter 1.1
The study does, however, use more recent data than those used by Baker
et al. Mor et al.'s estimates for the cost of lung cancer in the first year
after diagnosis (i.e., lifetime cost, since the majority of lung cancer patients
survive less than one year) is $40,051 (inflated from 1986 to 1996 dollars
using the CPI-U for Medical Care). This value agrees closely with the
results of Baker et al.'s work.
II.5.B.3.2 Hartunian et al.
Hartunian et al.'s (1981) method of estimating the costs of illness has been
discussed in Chapter 1.1. The authors defined expected treatment on a
yearly basis, developed annual costs of the treatment, and combined the
cost data with survival data. Using this method, they estimated the costs
of cancer at eight sites, including cancer of the respiratory system.
Hartunian et al. estimated the costs of inpatient stays for respiratory cancer
using a 1976 study by Scotto and Chazze, in which 6,332 newly diagnosed
cancer patients were followed over a two-year period to establish
hospitalization and payment patterns. For other costs, the authors relied on
a questionnaire, called the Patient Interview Book (PIB), delivered as part
of the Third National Cancer Survey. Using the PIB, about 8,500 cancer
patients or surviving relatives were interviewed regarding the costs of non-
hospital medical services. The PIB presents the proportional distribution of
total medical expenditures between hospital and non-hospital charges.
Applying these proportions to the hospital costs from Scotto and Chiazze,
Hartunian et al. estimated non-hospital costs. Hartunian et al. presented
their estimates of the present value of total direct costs of respiratory
cancers in 1975 dollars, discounted by six percent, by the age of onset for
males and females. Inflated to 1996 dollars using the CPI-U for Medical
Care (Bureau of Labor Statistics), their cost estimates, which approximate
Baker et al.'s estimates are:
Males
Age 65-74 $37,201
Age 75+ $34,821
Females
Age 65-74 $36,838
Age 75+ $34,508
The Hartunian data are quite old (over 20 years) and both survival and
treatment methods have changed since that time.
Chapter 11.5 11.5-27 Cost of Lung Cancer
-------
II.5.B.3.3 Oster et al.
Oster et al. (1984) estimated the cost of lung cancer as part of an analysis
of the cost of smoking. Their methodology followed that laid out by
Hartunian et al. Annual cost estimates were multiplied by survival rates
and discount factors and summed to arrive at a present value of the direct
costs of treating lung cancer. Oster et al.'s cost estimates were presented
in 1980 dollars for all ages. Using the CPI-U for Medical Care (Bureau of
Labor Statistics), these prices are inflated to 1996 dollars. The resulting
cost estimates (using a discount rate of three percent) are:
Males $57,861
Females $60,047
Although these estimates are substantially higher than the Baker et al.,
estimate, Oster et al.'s estimate of the cost of a lung cancer patient who
survives between zero and one year past disease onset is $45,013 in 1996
dollars. This cost estimate is similar to that developed based on Baker et
al. A potential explanation for the remaining difference in the cost
estimates is that Oster et al. assume that all patients incur the same costs in
their first year of disease regardless of their survival period. This cost
scheme may be realistic, but it is also possible that patients with relatively
long survival periods were diagnosed with less advanced cancer cases.
These cancer cases may require less extensive treatment and may therefore
be associated with lower first-year costs.
II.5.B.3.4 Riley et al
A study of medicare payments from diagnosis to death in elderly cancer
patients was carried out by Riley et al. (1995). The cost estimates are
based on Medicare payments only, which do not include: most nursing
home care, home health care, pharmaceuticals unless supplied for
inpatients, out-of-pocket expenses, deductibles, charges in excess of
Medicare paid by other sources (e.g., coinsurance), and other related
medical services that are not covered by Medicare.
Medicare patients younger than 65 were not included, and the average age
at diagnosis of the lung cancer cohort was 73.6 years, in contrast with the
68-year national average. Riley et al. note that patients diagnosed at
younger ages have higher costs. In addition, those diagnosed at earlier
stages have a better prognosis, but may have higher medical costs (due to
longer continuing care).
Medical costs are reported for all patients who were diagnosed with lung
cancer, regardless of other diseases or their ultimate causes of death. Due
to the link between lung cancer, smoking, and numerous other diseases,
this method is especially problematic because costs associated with other
illnesses may be commingled with the lung cancer costs.
Chapter 11.5 11.5-28 Cost of Lung Cancer
-------
The background cost per year for medical services was estimated by Riley
et al. to be $2,250 ($3,154 in 1996 dollars), based on the experience of all
people over the age of 65 who received Medicare-compensated care. The
study excluded those costs that occur during the last year of a person's life.
Consequently, the estimated background value may underestimate
background costs, especially as age and associated mortality risks increase
over the age of 65.
Riley et al. estimated that the total average Medicare payment from
diagnosis to death for persons diagnosed with lung cancer was $29,184 in
1990 dollars ($40,908 in 1996 dollars). This value is considerably lower
than the estimates obtained from Baker et al. The difference is most likely
due to the exclusion of many costs that are not covered by Medicare and
the various other factors described above. Due to these limitations, the
Riley et al. study is not recommended for a benefits evaluation.
II.5.C Uncertainties and Limitations
As noted periodically in the above discussion, there is uncertainty
surrounding various aspects of the analysis. Information concerning some
inputs to the analysis was often limited. Although a complete uncertainty
analysis is beyond the scope of this work, the significant sources of
uncertainty are discussed. Limitations of the scope of the analysis are also
discussed.
II.5.C.1 Uncertainties Surrounding Key Inputs to the Analysis
11.5. C.1.1. Analysis of Medical Costs
The cost estimates based on Baker et al. (1989, 1991) have a number of
limitations, many of them noted by Baker et al. (1991) and Mor et al.
(1990) and Mor (1993). Most of these limitations arise from the use of
CMHSF. Medicare data have five limitations that decrease its value for
calculating the average lifetime direct medical costs of treating lung cancer.
First, Medicare covers medical services for only most persons age 65 and
over, disabled persons entitled to Social Security cash benefits for at least
24 months, and most persons with end-stage renal disease. All patients not
covered by Medicare are excluded from the database, including all non-
disabled women under 65, and women over 65 using private health
insurance (Baker et al. 1991).
Given that diagnosis of lung cancer occurs before age 65 in 34 percent of
patients (NCI, 1998), the CMHSF excludes a significant number of
younger patients. According to Mor et al., treatment for younger women
tends to be more intensive (and therefore more costly per unit time) than
treatment for older women, though older women tend to have longer
hospital stays. Because these differences counteract each other, the
Chapter 11.5 11.5-29 Cost of Lung Cancer
-------
omission of younger women from the analysis is not expected to affect the
results substantially. In addition, the majority of senior citizens are enrolled
in Medicare (Ibid); differences in medical costs incurred by senior citizens
not using Medicare should have little effect on overall cost estimates.8
Medicare also does not cover self-administered drugs, intermediate nursing
care, long-term nursing care, and some expensive new treatments (such as
bone marrow transplants). For some patients these may represent
significant percentages of total treatment costs. Most direct medical costs,
however, appear to be covered by the CMHSF database and are included in
Baker et al.'s analysis. In addition, Baker et al. made adjustments for some
cost elements not covered by Medicare (see Section B).
Another drawback is that Baker et al. were not able to identify lung cancer
patients in CMHSF whose diagnosis and first course of therapy did not
involve hospitalization. In an analysis of Rhode Island lung cancer patients
covered by Medicare, Mor et al. determined that a small percentage of
lung cancer patients were initially diagnosed without hospitalization, and
had substantially lower initial and subsequent treatment costs (Mor et al.
1990). This omission likely causes average treatment costs to be
overestimated, though by relatively little.
A fourth drawback is that Baker et al. (1989) provides no information
concerning the duration of the maintenance period for lung cancer. The
analysis in this chapter assumed that lung cancer survivors incur
maintenance care costs for ten years. If the average duration of
maintenance care among survivors of lung cancer is shorter (longer) than
ten years, then the estimates of the costs incurred by survivors would be
biased upward (downward). This bias is less of an issue for nonsurvivors'
costs because the great majority of lung cancer nonsurvivors die within the
first few years. Because most lung cancer patients (about 88 percent) are
ultimately nonsurvivors, the duration of the maintenance period is of
somewhat less importance for lung cancer patients than for the 12 percent
who ultimately survive the illness.
A fifth drawback is that the data used by Baker are from the period 1974 to
1981. This limitation causes uncertainty regarding changes in treatment
methods and costs.
Finally, the reliability of the data contained in the database used by Baker et
al. varies. An independent analysis of CMHSF performed in 1977 by the
Institute of Medicine of the National Academy of Sciences found that the
frequency of discrepancies in principal diagnoses varied among diseases
(Baker et al., 1991). It is unclear whether the presence of misnamed
This figure represents those enrolled in Medicare Part A; 95 percent of those enrolled in Medicare
Part A choose also to enroll in Medicare Part B.
Chapter 11.5 11.5-30 Cost of Lung Cancer
-------
diagnoses contained in CMHSF potentially increases or decreases the
resultant cost estimates.
Overall, despite the limitations described above, Baker's analysis of the
CMHSF data represents the most nationally-representative, per-patient
lifetime estimate of the direct medical costs of treating lung cancer to date.
Their cost estimates are based on sound criteria. Some data limitations
underestimate costs and others overestimate costs; the sum of the data
limitations therefore decrease the magnitude of error. More of the
uncertainties in their analysis appear to underestimate costs, however; the
net result is a likely underestimation of actual direct medical costs.
Although there are some uncertainties associated with the estimation of the
survival and mortality probabilities used in the calculation of expected
medical costs and lost time (discussed below), these uncertainties are likely
to be relatively small. As noted in the text, NCI RSRs used to estimate
survival and mortality for this analysis are based on the survival experience
of a large group of lung cancer patients considered in relation to the
survival experience of the general population. Although age-specific RSRs
for each year post-diagnosis are not available, the age-specific five-year
RSRs provided by NCI (1998) suggest that there is relatively little variation
in RSRs across ages at diagnosis for lung cancer patients.
An additional limitation of this analysis is that medical costs incurred as a
result of lung cancer, but not considered by Baker et al., may arise as a
result of treatment for lung cancer. Secondary cancers and other adverse
health effects may occur due to radiation, chemotherapy treatment, and
other therapies. These effects may occur substantially after lung cancer
treatment has been completed, and can incur added medical costs not
considered in this chapter.
Data have not yet been located regarding the average duration of
maintenance care. For purposes of this analysis, ten years of follow-up
care was assumed to be reasonable due to the severity of the disease and
the consequences of lung surgery. This assumption may be revised in the
future if data are located.
II.5.C.2 Scope of the Analysis
The analysis in this chapter was confined to direct medical costs by the
patient. As noted in Chapter I.I, willingness-to-pay has many other cost
elements.
Link to Chapter 1.1
Chapter 11.5 11.5-31 Cost of Lung Cancer
-------
The analysis does not include time lost by the patient and his or her family
and friends who provide care, pain and suffering on the part of the patient
and his or her family and friends, changes in job status among previously
employed patients, training for new job skills due to physical limitations, or
medical costs incurred after the ten-year maintenance period. These cost
elements may comprise a substantial portion of the total cost of lung
cancer.
Chapter 11.5 11.5-32 Cost of Lung Cancer
-------
CHAPTER 11.7: COST OF COLORECTAL CANCER
Clicking on the sections below will take you to the relevant text.
II.7.A. Background
II.7. A.I. Description
II.7.A.2. Concurrent Effects
II.7.A.3. Causality & Special Susceptibilities
II.7.A.4. Treatments and Services
II.7.A.5. Prognosis
II.7.B. Costs of Treatment and Services
II.7.B.1. Methodology
II.7.B.2. Results of Medical Cost Analysis
II.7.B.3. Other Studies
II.7.C. Uncertainties and Limitations
II.7.C. 1. Uncertainties Surrounding Key Inputs to the Analysis
II.7.C.2. Scope of the Analysis
Chapter 11.7 11.7-1 Cost of Colorectal Cancer
-------
CHAPTER 11.7: COST OF COLORECTAL CANCER
This chapter contains a discussion of the methods used to estimate and the
results of estimating the direct medical costs incurred by surviving
colorectal cancer patients.1 It does not include information on elements
such as indirect medical costs, pain and suffering, lost time of unpaid
caregivers, etc. The reader is referred to Chapter I.I for a discussion of the
cost estimation methods and cost elements that are relevant to all benefits
estimates. In addition, Chapter II. 1 contains information regarding cancer
causality, a list of known and suspected carcinogens, and information on
cancer cost estimation.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapter 1.1 and II. 1
Link to inflation factors
Survival data from the National Cancer Institute (NCI, 1998) and cost data
from Baker et al (1989) that are used in this chapter do not provide
quantitative information for different types of colorectal cancer.
Consequently, this chapter contains an evaluation of all types in aggregate.
In addition, most risk assessments used in evaluating benefits do not
specify the type of colorectal cancer.
II.7.A. Background
II.7.A.1. Description
Colorectal cancers are malignancies of the colon or rectum. They are most
often adenocarcinomas that are thought to develop through genetic
alterations in the cells. Colorectal cancers can be differentiated, based on
the site of the tumor(s). As noted above, however, they are considered as
a single cancer type for this cost analysis.
Most cases of colorectal cancer occur among the elderly, which is typical
of cancer. The average age at diagnosis is 70.4 years. Less than two
percent of colorectal cancers are diagnosed before the age of 40, and 42
percent are diagnosed over the age of 75 (NCI, 1998). The age
distribution at diagnosis of colorectal cancer is shown in Figure II.7-1. The
1 Survivors are those who do not die of this specific disease. This chapter was prepared in
response to EPA's specific requirement for a proposed rule. Because they were using the value of a
statistical life (VSL) for those who die of the disease, they required only cost data for disease survivors.
Sources provided in the chapter can be used to calculate medical costs for nonsurvivors, if required.
Chapter 11.7 11.7-2 Cost of Colorectal Cancer
-------
steep incline in the probability of diagnosis is clear in this diagram, with a
peak around 70 years of age. The data used to generate Figure II.7-1 are
shown in Table II. 7-1. The cumulative percents of colorectal cancer at
various ages were calculated using the population-weighted distribution of
occurrence. These percents are also shown in Table II.7-1. Approximately
60 percent of all colorectal cancer cases are diagnosed in the relatively
small age interval of 70 to 85 years.
Figure 11.7-1
0.18 -r
0.16
0.14
0.12 -
5 0.1 J
-------
Table 11.7-1. Age-specific Incidence of Colorectal Cancer
Age Group
0- 14
15-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75-79
80-84
85+
Age-specific Rate of
Diagnosis Per
100,000
0.1
5.5
6.4
12.7
24.7
49.9
88.4
140.3
206.4
281.5
367.8
457.5
473.1
Percent of All
Colorectal Cancer
Occurring in Age
Group
0.0
0.8
1.0
1.7
2.8
4.5
6.6
9.7
14.1
16.9
16.6
13.7
11.4
Cumulative Percent
of Colorectal Cancer
0.0
0.8
1.8
3.5
6.3
10.8
17.5
27.2
41.3
58.3
74.9
88.6
100.0
Basedon NCI, 1998
II.7.A.2. Concurrent Effects
As with all cancers, colorectal cancer may spread to other organs.
Colorectal cancer often spreads to distant sites, involving regional lymph
nodes, or to the liver or lung (Abeloff et al., 1995). In addition, treatment
of cancer, which usually includes chemotherapy, radiation, and surgery, has
numerous adverse side effects and may in itself lead to death. Radiation
treatments of cancer have led to increased risks of other types of cancer,
sterility, etc. Surgery may cause long-term changes in health status that
may also lead to death. These effects are associated with additional
medical costs not considered in this chapter.
The material herein was specifically developed for the U.S. EPA for use in
the analysis of a particular Rule, and does not include information
regarding additional medical costs incurred from concurrent effects of
colorectal cancer or its treatment.
II.7.A.3. Causality & Special Susceptibilities
Dietary, environmental, and heriditary factors are important in colorectal
cancer causality (Abeloff et al., 1995). Table II. 1-1 in Chapter III
contains a list of chemicals known to cause or are suspected of causing
cancer (as reported in the EPA databases IRIS, HEAST, and HSDB).
Most chemicals in the table were carcinogenic in animal studies. These
studies do not provide organ-specific data because it is not generally
assumed that cancer induction will necessarily occur at the same site in
humans as in animals. Consequently, the chemicals listed in Table II. 1-1
Chapter 11.7
1.7-4
Cost of Colorectal Cancer
-------
Link to Table II. 1-1
may cause colorectal cancer and/or other types of cancer. Evaluation of
the likelihood of this occurrence would require additional risk assessment
research.
Epidemiologic studies worldwide have documented a direct correlation
between colorectal cancer mortality and per capita consumption of calories,
meat protein, and dietary fat and oil (Mayer, 1998). Other risk factors
include hereditary syndromes (as many as 25 percent of colorectal cancer
patients have a family history of the disease), and inflammatory bowel
disease (colorectal cancer risk is relatively small among this population
during the first ten years of the disease, but it increases at a rate of
approximately 0.5-1 percent per year) (Mayer, 1998). Various syndromes
affecting the intestine are linked to a much higher risk of colorectal cancer,
and both primary and secondary mutations are found in about 15 percent of
cases. In addition to genetic alterations, abnormal DNA methylation is an
important factor in colorectal cancer. There has been, and continues to be,
intensive study of the genetic aspects of this disease (Abeloff et al., 1995).
Colorectal cancer generally affects the over-50-year-old population. Most
colorectal cancers, regardless of etiology, are believed to arise from
adenomatous polyps, which have been found in the colons of
approximately 30 percent of middle-aged or elderly people. Less than one
percent of these polyps, however, ever become malignant (Mayer, 1998)
Other risk factors include prior history of breast, ovarian, endometrial,
genital, colon, or bladder cancer.
NCI provides age-, sex-, and race-specific data regarding diagnosis of
colorectal cancer from 1990 to 1994, which may be used to evaluate
susceptibilities among population subgroups. The data must be used with
care, however, because diagnostic rates indicate occurrence only, and may
or may not indicate differences in susceptibility. See Chapter 1.1 for a
more detailed discussion of susceptibilities.
NCI colorectal cancer diagnosis and mortality data show higher diagnosis
and death rates among men than women. From the 1973 to 1994 time
period, males generally had a 50 percent higher mortality rate than females.
Over the time period of 1978-1994, black women consistently showed
higher incidence rates than white women. The percent change from 1973
to 1994 were +14 percent and -11 percent, respectively.
The rate of diagnosis among black males from 1973-1994 was higher than
that in the white male. Over the period 1973 to 1994, incidence rates
among black men increased 34 percent, while it dropped five percent in
Chapter 11.7 11.7-5 Cost of Colorectal Cancer
Link to Chapter 1.1
-------
white men. A similar trend was also observed of mortality rates over that
same time period.
II.7.A.4. Treatments and Services
Colorectal cancer is usually treated with surgery, chemotherapy, and/or
radiation, depending on the type of colorectal cancer, the stage of cancer at
diagnosis, patient health, and other factors. The treatment is defined more
precisely by histologic type and specific location of the cancer. In this
analysis, all histologic types and sub-sites are considered together. Most
surgery involved en bloc resection, which entails removing large sections of
the intestinal tract. This procedure may be modified if cancer has not
spread to regional lymph nodes (Abeloff et al., 1995). Treatment for this
type of cancer often requires permanent lifestyle changes due to the nature
of the surgical intervention required.
Treatment is carried out in phases including initial diagnosis, initial
treatment, follow-up and maintenance treatment, and, for those who do not
survive, terminal treatment and palliative care. Although there are some
components of each treatment that are unique to each phase, most medical
activities and services may occur more than once over the course of the
disease from diagnosis to death or cure. For example, X-rays may be used
in diagnosis, to provide ongoing status updates, to assist in determining
initial and subsequent surgical and other treatment interventions, etc.
Initial diagnostic activities may include an evaluation of signs and
symptoms, X-rays and other types of imaging, laboratory tests, and other
procedures. Staging of the disease occurs during this phase and is critical
to determination of subsequent medical actions (Feld et al., 1995). Surgery
is usually performed, as well as radiation and/or chemotherapy. In some
patients, cancer has spread to other organs requiring additional treatment
strategies. Colorectal cancer is fatal in approximatley 50 percent of
patients. Consequently, most patients receive terminal care that may
include a variety of medical services, long-term care in a nursing facility,
palliative care, family counseling, etc.
II.7.A.5. Prognosis
Cancer of the large bowel is the second highest cause of cancer death in the
U.S. Although approximately 70-80 percent of all patients survive
colorectal cancer for three years following diagnosis (NCI, 1998),
Chapter 11.7 11.7-6 Cost of Colorectal Cancer
-------
approximately 47 percent of patients die of colorectal cancer within ten
years of diagnosis (53 percent survive).2 This value is used as a reasonable
approximation of the lifetime mortality rate in later sections of this chapter.
Factors such as tumor size and location, histology, involvement of nodes,
and the spread of cancer to other tissues affect outcome. As with other
cancers, the prognosis of colorectal cancer is determined partially by the
occurrence of cancer at other sites. When it has metasticized, the survival
rate is poorer. As noted above, colorectal cancer generally spreads to
distant sites, involving regional lymph nodes or the liver. The latter is
reportedly the initial site of distant cancer spread in 33 percent of recurring
colorectal cancer patients, and involved in over 66 percent of such patients
at their time of death. In fact, colorectal cancer rarely disseminates to the
lungs, bone, or brain without hepatic involvement first. The median
survival duration for colorectal cancer-associated liver metastases ranges
from 6-9 months to 24-30 months (Mayer, 1998).
The prognosis for this cancer has improved in recent decades. An
estimated six to eight percent increase in survival over the past two
decades has been reported (Abeloff et al., 1995). The change has been
attributed to earlier detection and a decrease in treatment-related mortality.
The mortality rates among African-American patients are typically higher
than within other racial groups.
1.7.B. Costs of Treatment and Services
.7.B.1. Methodology
II.7.B.1.1 Overview
As noted above, this chapter examines the direct medical cost for the
"average" colorectal cancer survivor and does not include nonsurvivors of
colorectal cancer.
II.7.B.1.2 Medical Cost Data
II.7.B.1.2.1 Sources
Medical cost data would ideally be obtained on current medical
expenditures. Although data files are maintained by public and private
sector sources, they are not readily available. In addition, it is necessary to
evaluate very large databases of charges from a variety of sources to obtain
2 The SEER data reports were accessed online to obtain information regarding mortality and
survival probabilities (RSRs) (NCI, 1998). The RSR is the number of observed survivors among these
patients, divided by the number of "expected" survivors among persons with the same age and gender in the
general population (observed/expected). The RSR takes into account that there are competing causes of
death that increase with age. Methods used to convert the NCI statistics to survival probabilities are
described in detail in Chapter II.2.
Link to Chapter II. 2
Chapter 11.7 11.7-7 Cost of Colorectal Cancer
-------
reliable cost estimates. That method was not practical for the development
of this chapter. A data search was conducted to locate information in the
medical economics literature regarding medical costs associated with
colorectal cancer. In addition to a literature search, most federal agencies
dealing with cancer, disabilities, medical costs and their management, and
related issues were contacted for information; the various federal databases
were discussed with senior staff at these agencies. Very recent cost data
were not located.3 Current (1994) cancer data were obtained regarding
incidence and survival (as reported in Section II.7.A, above), and were
used with cost data from the 1980s, described below.
The cost estimates presented in this section are based primarily on the work
of Baker et al. (1989) and Hartunian et al. (1981) and on two sources of
statistical data: the National Cancer Institute (1998) and Vital Statistics of
the United States, 1995, Preprint of Volume II, Mortality, Part A Section 6
Life Tables (NCHS, 1998). These data were evaluated and cost and time
elements were used to calculate lifetime estimates of the direct medical
costs due to colorectal cancer. Based on a 1998 review of the literature,
carried out for the development of this chapter, there do not appear to be
new treatment methods for lung cancer that alter either the medical costs
or the survival rates for most patients substantially. Consequently, the cost
estimates presented in this chapter may be considered appropriate under
most circumstances (e.g., regional costs may vary).
II.7.B.1.2.2 Baker et al.'s Cost Estimation Method
Baker et al. (1989) used the Continuous Medicare History Sample File
(CMHSF) to estimate the per-patient average lifetime medical cost of
treating lung cancer based on data files from 1974 to 1981. They chose
CMHSF because:
1) it is a nationally representative sample of the Medicare population
(five percent), covering more than 1.6 million patients;
2) it is longitudinal, dating from 1974 to 1981; and
3) it captures the majority of medical expenses for each beneficiary.
Five Medicare files are included in the CMHSF, which cover:
1) inpatient hospital stays,
2) skilled nursing facility stays,
3) home health agency charges,
3Studies were located that used more recent cost data than were used in this analysis. The studies
were not used due to serious limitations (e.g., data were incomplete). The studies are reported in the "Other
Studies" section at the end of Section II.7.B.
Chapter 11.7 11.7-8 Cost of Colorectal Cancer
-------
4) physicians' services, and
5) outpatient and other medical services.4
Costs that were not included are outpatient prescription medications and
nursing home care below the skilled level.
Because CMHSF provides no indication of initial diagnosis, Baker et al.
assumed that disease onset occurred when a diagnosis of colorectal cancer
was listed on a hospitalization record following a minimum of one year
without a colorectal cancer diagnosis. This assumption is reasonable due
to the high frequency of hospitalization associated with the disease (i.e.,
individuals diagnosed with colorectal cancer would be hospitalized). Only
patients with an initial diagnosis during the years covered by the database
(1974-1981) were included.
Costs associated with colorectal cancer were assigned to three post-
diagnostic time periods:
initial treatment, during the first three months following diagnosis;
• maintenance care, between initial and terminal treatment; and
terminal treatment, during the final six months prior to death.
As noted in Chapter 1.1, the amount paid for service may differ from the
actual medical costs because many insurers and federal programs either 1)
pay only a portion of total costs or 2) pay more than actual costs to
underwrite the care providers' losses due to underpayment from other
sources. Baker et al. used provider charges, rather than Medicare
reimbursements (which represent only a portion of most total charges),
thus providing a more accurate cost estimate.
To improve the accuracy of the cost estimates, Baker et al. included cost
data on coinsurance, deductibles, and other cost components. They made
four adjustments to the cost estimates calculated from the CMHSF. First,
charges were added for skilled nursing facilities (SNFs) not covered by
Medicare by multiplying the "length of stay" at an SNF (computed from
admission and discharge dates) by the average daily SNF charge. Second,
the annual Medicare Part B deductible of $60 was added to the reimbursed
charges in the database. Third, since Medicare pays only 80 percent of
physicians' charges, Baker et al. scaled these reimbursements to 100
percent of physicians' charges to better reflect social costs. Finally, they
inflated all dollar values to 1984 dollars using the Medical Care component
of the Consumer Price Index.
4 See Baker et al. (1989 and 1991) for further details. Baker et al. (1991) contains additional
descriptive data regarding the database and methods used for the cost analysis; however, it does not contain
cost data for lung cancer.
Chapter 11.7 11.7-9 Cost of Colorectal Cancer
-------
II.7.B.1.2.3 Cost Estimates by Treatment Period
Medical costs associated with the initial, maintenance, and terminal cancer
care treatment periods were itemized in Baker et al., 1989. These figures
are reported as incremental costs in Table II.7-2, because the 1989 paper
did not specifically report incremental costs or the costs of other medical
services anticipated to occur while the patient was receiving cancer
treatment (i.e., co-morbidit>Vbackground costs). To estimate the
incremental costs, a co-morbidity cost of $2,988 per year (1984 dollars)
from Baker et al. (1991) was used in this analysis.
The total cost for the initial three-month treatment period is reported in
Table II.7-2, which includes the pro-rated co-morbidity cost for that three-
month time period. Annual costs for the maintenance period are also
shown and are further discussed in the "Lifetime Cost Estimates" section
below (see II.7.B.1.3). Note that only the initial and maintenance costs are
relevant to this analysis on colorectal cancer survivor population.
Nonsurvivors, and hence terminal treatment costs, are not further
addressed herein.
Using the Medical Care component of the Consumer Price Index (CPI-U),
all costs were inflated to 1996 dollars for purposes of this handbook. For
example, the 1984 annual co-morbidity cost of $2,988 would be equivalent
to $6,394 in 1996 dollars, using the CPI adjustment multiplier factor of
2.14 for the period 1984 to 1996.
Table 11.7-2. Average Per Patient Costs for the Treatment Periods for
Colorectal Cancer (in 1996$)
Costs adjusted for inflation using the Medical Care component of the
Consumer Price Index (CPI-U) 1996:1984 = 2.14 (Bureau of Labor Statistics)
Treatment Period
Initial
(3 months)
Maintenance (per year)
Terminal
(6 months)
Incremental Cancer Treatment
Cost
$28,768
$8,295
$30,563
(Based on Baker et al., 1989, with co-morbidity charges from Baker et al., 1991.
II.7.B. 1.3 Calculation of Lifetime Cost Estimates for the
"Average" Colorectal Cancer Survivor
This section contains a discussion of the calculation of lifetime medical
costs for the "average" colorectal cancer patient, identified in this analysis
as an individual diagnosed at age 70.4 (the average age of colorectal cancer
diagnosis from SEER, NCI 1998), with a life expectancy period of 13.8
years beyond that age (as determined by linear interpolation of NCHS 1998
Chapter 11.7
11.7-10
Cost of Colorectal Cancer
-------
data for the years 70 and 71). The approach described below was used to
address specific EPA rulemaking requirements of the direct medical cost
data. It therefore focuses specifically on the lifetime costs of colorectal
cancer survivors over the life expectancy period from the average age of
diagnosis. As in the previous chapters of this handbook, lifetime costs for
nonsurvivors or for other compounding illnesses will not be presented.
The analysis assumes that death may occur only after the full life expectancy
period from the average age of diagnosis has elapsed. All patients are
therefore assumed to incur initial treatment costs during the first three-month
period of the illness as defined by Baker et al. (1989). The costs incurred
during the remaining months of the first year of illness are calculated by pro-
rating the annual maintenance costs. For example, in the first year, the
average colorectal cancer survivor incurs the costs of initial treatment
($28,768) over the first three months, and then incurs nine months' worth of
maintenance care costs ($8,295 x 0.75 = $6,221) (see Table H7-2). The total
cost of colorectal cancer incurred during the first year to survivors is therefore
$28,768 + $6,221 = $34,989, representing the intensive medical care
treatment a patient would initially receive.
The expected medical costs for colorectal cancer patients during the first year
post-diagnosis, then, is defined as:
Expected First-Year Cost: initial treatment costs over a three-
month period + maintenance care costs for nine months
Example: Expected first-year medical costs of a colorectal cancer patient
diagnosed at age 70.4
As noted above, all colorectal cancer patients incur an initial treatment cost
of $28,768. Those who survive through the year also incur maintenance
care costs for the remaining three quarters of the year. Recall from above
that the total first-year costs of those who survive the year were:
Initial treatment: $28,768
Maintenance treatment: $6,221 ($8,295 x 0.75)
Total First-Year Cost $34,989
For each subsequent year post-diagnosis, medical costs consist entirely of
annual maintenance care costs.
In this analysis, each patient would incur 13.8 years of maintenance costs,
assumed to be a reasonable average period over which additional medical
costs associated with colorectal cancer would also be incurred.
Chapter 11.7 11.7-11 Cost of Colorectal Cancer
-------
The expected medical costs for lung cancer patients during the nth year
post-diagnosis, for ri>l, then, is defined as:
Expected nth Year (n>l) Cost: maintenance care cost for the year,
pro-rated as necessary.
Maintenance care costs are not assumed at full value in the last year (i.e., in
the fourteenth year) of life expectancy. The treatment costs must be pro-
rated based on how long a colorectal cancer survivor is anticipated to live
in the final year of expected life. As previously discussed, linear
interpolation of life expectancy data between the ages of 70 and 71 resulted
in a forecast of 13.8 years, which means that only 80 percent of the annual
maintenance costs ($8,295 x 0.80) will be incurred by the patient, or
$6,636.
The expected lifetime or total cost to a colorectal cancer survivor is
subsequently derived by summing all the expected medical costs over the
entire period from diagnosis to death.
Expected Lifetime Cost = Expected first-year cost + the sum of the
(discounted) expected subsequent-year costs
Using the initial treatment and maintenance costs listed in Table II.7-2, the
mathematical equation for the expected lifetime medical costs incurred by
the "average" colorectal cancer survivor over a 13.8-year period may be
expressed as:
13
$28,768 + ($8,295 x 0.75)
$8,295
r
y-l
($8,295 x 0.8)
Where:
y
r
the year post-diagnosis, and
the discount rate.
The cost estimates for each year post-diagnosis and the estimate of
expected undiscounted and discounted (at three, five, and seven percent)
total costs for a fourteen- year period are shown in Table II.7-3 for the
"average" colorectal cancer survivor diagnosed at age 70.4.
Chapter 11.7
11.7-12
Cost of Colorectal Cancer
-------
Table 11.7-3. Expected Costs of Medical Services (in 1996$) for Surviving Colorectal
Cancer Patients (Age of Onset = 70.4) Over a Life Expectancy Period of 13.8 Years
Years
Post-
Diagnosis
(n)
1b
2
3
4
5
6
7
8
9
10
11
12
13
14C
Expected
Total
Costd
Expected
Medical Costs in
the nth Year
Post-Diagnosis
(Undiscounted)3
$34,989
8,295
8,295
8,295
8,295
8,295
8,295
8,295
8,295
8,295
8,295
8,295
8,295
6,636
$141,160
Expected Medical
Costs in the nth
Year Post-
Diagnosis
(Discounted 3%)
$34,989
8,053
7,818
7,591
7,370
7,155
6,947
6,744
6,548
6,357
6,172
5,992
5,818
4,519
$122,072
Expected Medical
Costs in the nth
Year Post-
Diagnosis
(Discounted 5%)
$34,989
7,900
7,523
7,165
6,824
6,499
6,190
5,895
5,614
5,347
5,092
4,850
4,619
3,519
$112,026
Expected Medical
Costs in the nth
Year Post-
Diagnosis
(Discounted 7%)
$34,989
7,752
7,245
6,771
6,328
5,914
5,527
5,165
4,828
4,512
4,217
3,941
3,683
2,754
$103,624
a. The undiscounted initial and maintenance costs used in this table are from Table II. 7-2, as adapted
from Baker et al. , 1 989. Link to Table II. 7-2
b. First-year costs include the charge for "initial" therapy ($28,768) and an adjusted maintenance cost
pro-rated for the initial year (see text for discussion).
c. Final-year costs are pro-rated according to the average life expectancy (13.8 years) at the average
age of diagnosis (70.4) (see text for discussion).
d. Sums may not equal reported totals due to rounding.
Chapter II.7
11.7-13
Cost of Colorectal Cancer
-------
.7.B.2.
Results of Medical Cost Analysis
The per-patient lifetime direct medical costs calculated for the "average"
colorectal cancer patient (as shown in Table II.7-3), diagnosed at age 70.4
are listed in Table II.7-4.
Undiscounted costs and costs discounted at three, five, and seven percent,
are shown. Discounting was carried out for 14 years following diagnosis,
which represents the full duration of treatment and maintenance care
duration and the life expectancy at that age.
Table 11.7-4. Summary of Total Costs of Medical Services (in 1996$)
for Surviving Colorectal Cancer Patients
Undiscounted
$141,160
Discount Rate
3
$122,072
5
$112,026
7
$103,624
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
below.
Link to inflation factors
The actual average period of maintenance is not known and is likely to vary
considerably among individuals, depending on age, health status, access to
care, and other factors. For the EPA requirements under which this
analysis was developed, however, this method is assumed to represent the
average colorectal survivor.
.7.B.3.
Other Studies
Riley et al. (1995) studied cancer costs, but their results are not
recommended due to characteristics of their study design and data quality.
They studied Medicare payments from diagnosis to death in elderly cancer
patients. Cost estimates presented in the paper were based only on
Medicare payments, data that do not include most nursing home care,
home health care, pharmaceuticals unless supplied for inpatients, out-of-
pocket expenses, deductibles, charges in excess of Medicare paid by other
sources (e.g., coinsurance), and other related medical services not covered
by Medicare.
Medicare patients younger than 65 years old were not included, and the
average age at diagnosis of the colorectal cancer cohort was 76.2 years, in
contrast with the national average of 70.4. Riley et al. noted that patients
diagnosed at younger ages often incurred higher costs. In addition, those
Chapter 11.7
11.7-14
Cost of Colorectal Cancer
-------
diagnosed at earlier stages had a better prognosis, but may have had higher
medical costs (due to longer continuing care).
Medical costs were reported for all patients diagnosed with colorectal
cancer, and did not differentiate between colorectal cancer costs and those
of other diseases. Determination of colorectal medical costs were
calculated by subtracting background costs. The background cost per year
for medical services was estimated by Riley et al., based on the experience
of all people over the age of 65 who received Medicare-compensated care.
This value was estimated to be $2,250 in 1990 dollars ($3,154 in 1996
dollars, using the CPI 1990-1996 multiplier of 1.4). This estimate excluded
costs that occurred during the last year of a person's life. Consequently,
the estimated background value reported may underestimate background
costs and this omission would lead to a slight overestimate of incremental
costs.
Riley et al. estimated that the total average incremental Medicare payment
from diagnosis to death for persons diagnosed with colorectal cancer was
$51,865 in 1990 dollars ($72,611 in 1996 dollars). This estimate is
considerably lower than the estimates obtained from Baker et al. The
difference is most likely due to the exclusion of many costs that are not
covered by Medicare, in addition to the various other factors described
above. As a result of such limitations, the Riley et al. study is not
recommended for a benefits evaluation.
1.7.C. Uncertainties and Limitations
There are many limitations in cancer cost estimation. Those common to
most cancers are discussed in the introductory cancer chapter: III
Link to Chapter II. 1
Several aspects of this analysis contain underlying uncertainties based
mainly on the limited information concerning some analytical inputs. A
discussion of the uncertainty and limitations regarding the data sources of
the analysis (Section II.7.C.1) and the scope of the analysis (Section
II.7.C.2) follows below.
I.7.C.1. Uncertainties Surrounding Key Inputs to the Analysis
//. 7.C.1.1. Analysis of Medical Costs
The cost estimates based on Baker et al. (1989, 1991) have a number of
limitations, many of them noted in Baker et al., 1991. Most of these
limitations are related to the use of CMHSF. Medicare data have five
limitations that decrease their value for calculating the average lifetime
direct medical costs of treating lung cancer. First, Medicare covers
Chapter 11.7 11.7-15 Cost of Colorectal Cancer
-------
medical services for most persons age 65 and over, disabled persons
entitled to Social Security cash benefits for at least 24 months, and most
persons with end-stage renal disease. All patients not covered by Medicare
are excluded from the database, including all non-disabled women under
65, and women over 65 using private health insurance (Baker et al., 1991).
Medicare also does not cover self-administered drugs, intermediate nursing
care, long-term nursing care, and some expensive new treatments (such as
bone marrow transplants). For some patients these may represent
significant percentages of total treatment costs. Most direct medical costs,
however, appear to be covered by the CMHSF database and are included in
Baker et al.'s 1989 analysis. In addition, Baker et al. made adjustments for
some cost elements not covered by Medicare (see Section II.7.B). Another
drawback is that Baker et al. were not able to identify colorectal cancer
patients in CMHSF whose diagnosis and first course of therapy did not
involve hospitalization.
A fourth drawback is that Baker et al. (1989) provide no information
concerning the duration of the maintenance period for colorectal cancer.
The analysis in this chapter assumed that colorectal cancer survivors incur
maintenance care costs for 13.8 years. If the average duration of
maintenance care among survivors of lung cancer is shorter (or longer)
than 13.8 years, the estimates of the costs incurred by survivors would be
biased upward (or downward).
A fifth limitation with using Medicare data is that the data used by Baker
are from the period 1974 to 1981. The age of the data causes uncertainty
regarding changes in treatment methods and costs.
The reliability of the data contained in the database used by Baker et al.
also varies. An independent analysis of CMHSF performed in 1977 by the
Institute of Medicine of the National Academy of Sciences found that the
frequency of discrepancies in principal diagnoses varied among diseases
(Baker et al. 1991). It is unclear, however, whether the presence of
misnamed diagnoses contained in CMHSF potentially increases or
decreases the resultant cost estimates.
Overall, despite the limitations described above, Baker et al.'s analysis of
the CMHSF data represents the most nationally-representative, per-patient
lifetime estimate of the direct medical costs of treating colorectal cancer to
date. Their cost estimates are based on sound criteria. Because some of
the data limitations underestimate costs and others overestimate costs, the
sum of the data limitations decreases the magnitude of error. More of the
uncertainties in their analysis appear to underestimate costs, however, and
poses the problem that the net result may likely be an underestimation of
actual direct medical costs.
Chapter 11.7 11.7-16 Cost of Colorectal Cancer
-------
Although there are some uncertainties associated with the estimation of the
survival and mortality probabilities used in the calculation of expected
medical costs (discussed below), these uncertainties are likely to be
relatively small. As noted in the text, NCI RSRs used to estimate survival
and mortality for this analysis are based on the survival experience of a
large group of colorectal cancer patients considered in relation to the
survival experience of the general population. Although age-specific RSRs
for each year post-diagnosis are not available, the age-specific five-year
RSRs provided by NCI (1998) suggest that there is relatively little variation
in RSRs across ages at diagnosis.
An additional limitation of this analysis is that medical costs incurred as a
result of colorectal cancer, but not considered by Baker et al., may arise as
a result of treatment. Secondary cancers and other adverse health effects
may occur due to radiation, chemotherapy treatment, and other therapies.
These effects may occur substantially after colorectal cancer treatment has
been completed and can incur added medical costs not considered in this
chapter. Data have not yet been located regarding the average duration of
maintenance care. For purposes of this analysis, an approximately 14-year
period of follow-up care was assumed to be reasonable, due to the severity
of the disease and the consequences of colorectal surgery. This assumption
may be revised in the future if data are located.
.7.C.2. Scope of the Analysis
The analysis in this chapter was confined to direct medical costs by the
patient. As noted in Chapter I.I, willingness-to-pay has many other cost
elements. The analysis does not include time lost by the patient or their
family and friends who provide care. Also omitted from cost of illness
estimates are pain and suffering on the part of the patient or their family
and friends, changes in job status among previously employed patients,
training for new job skills due to physical limitations, or medical costs
incurred after the ten-year maintenance period. These cost elements may
also comprise a substantial portion of the total cost of colorectal cancer.
Link to Chapter 1.1
Chapter 11.7 11.7-17 Cost of Colorectal Cancer
-------
CHAPTER 11.8. COST OF BLADDER CANCER
Clicking on the sections below will take you to the relevant text.
II.8.A Background
II.8. A.I Description
II.8.A.2 Concurrent Effects
II.8.A.3 Causality & Special Susceptibilities
II.8.A.4 Treatments and Services
II.8.A.5 Prognosis
II.8.B Costs of Treatments and Services
II. 8.B.I Methodology
II.8.B.2 Results of Medical Cost Analysis
II.8.B.3 Other Studies
II.8.C Uncertainties and Limitations
II.8.C. 1 Uncertainties Surrounding Key Inputs to the Analysis
II.8.C.2 Scope of the Analysis
Chapter 11.8 11.8-1 Cost of Bladder Cancer
-------
CHAPTER 11.8. COST OF BLADDER CANCER
II.8.A Background
This chapter contains a discussion of the methods used and results of
estimating the direct medical costs incurred by bladder cancer patients. It
does not include information on elements such as indirect medical costs,
pain and suffering, lost time of unpaid caregivers, etc. The reader is
referred to Chapter I.I for a discussion of the cost estimation methods and
cost elements that are relevant to all benefits estimates. In addition,
Chapter II. 1 contains information regarding cancer causality, a list of
known and suspected carcinogens, and information on cancer cost
estimation.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and II. 1
Link to inflation factors
II.8.A.1. Description
Bladder cancers are tumors that arise from the transitional cell lining of the
urinary tract. These are a part of a larger group of tumors that are all
related and are referred to as urothelial cell cancers. Urothelial cell cancers
may occur in the kidneys, ureter, bladder, urethra, and the ducts of the
prostate. The most common of these, bladder cancer, is the only cancer
discussed in this chapter. Ninety percent of urothelial cell tumors are
transitional cell carcinomas, with the remainder composed of squamous cell
carcinomas and adenocarcinomas (Bennett and Plum, 1996).
Although a small percentage of bladder cancers differ somewhat from the
majority in their cell origin or composition, this chapter contains an
evaluation of all types of bladder cancer in aggregate. In addition, most
risk assessments that would be used in evaluating benefits do not specify
the type of bladder cancer. If a specific type of bladder cancer is of
concern, Bennett and Plum (1996) may be consulted for additional
information regarding prognostic information and treatment; however, the
quantitative data are limited.
Bladder cancer is a common cause of cancer death in men and women in
the U.S. (Feld et al., 1995), accounting for two percent of all cancer cases
in the U.S.(Abeloff et al., 1995). Approximately 51,200 cases of bladder
cancer were diagnosed in 1994 in the United States; approximately 10,600
bladder cancer deaths occurred in that year (Bennett and Plum, 1996). In
Chapter 11.8 11.8-2 Cost of Bladder Cancer
-------
1996 the incidence rate was 27.7 per 100,000 in men and 7.4 in women
(NCI, 1999).1 The highest risk group are white men over the age of 64,
who have an incidence rate of 217.8 per 100,000 (NCI, 1999).
The incidence of bladder cancer has increased overall by 7.7 percent
between 1973 and 1996, due primarily to a 14.5 percent increase among
those over the age of 64. The most dramatic increase has occurred among
women, with a 22.6 percent increase in white women and a 24.8 percent
increase among black women.2 As discussed under "causality" below, this
increase may be due to the increased rate of smoking among women.
Incidence rates among black and white men over the age of 65 during the
1973 to 1996 interval have also increased substantially: 24.0 and 22.6
percent, respectively. Fortunately, this trend toward increased incidence
has turned around slightly in very recent years (1992-1996) (NCI, 1999).
During the period 1973 to 1996, there has been a small drop in the
incidence of bladder cancer among those under the age of 65 years (3.4
percent), driven solely by a decline in bladder cancer among white men and
black women under the age of 65 years. Mortality rates have also
decreased (discussed under prognosis, below).
Bladder cancer is observed in three times as many men as women. As with
most cancers, it also occurs with much greater frequency among the
elderly. The average age at diagnosis is between 70 and 75 years. Less
than 1.6 percent of bladder cancers are diagnosed before the age of 40, and
21.7 percent are diagnosed over the age of 85 (NCI, 1999). The age
distribution at diagnosis of bladder cancer is shown in Figure II.8-1. The
steep incline in the probability of bladder cancer diagnosis with age is clear
in this diagram. The data used to generate Figure II.8-1 are shown in
Table II. 8-1. The cumulative percents of bladder cancer at various ages
were calculated using the population-weighted distribution of occurrence;
these are also shown in Table II.8-1. The age-specific incidence data were
used in the Section II.8.B medical cost calculations.
1 Data on incidence and age at diagnosis were obtained from the National Cancer Institute's (NCI)
Surveillance, Epidemiology, and End Results (SEER) reports and tables. These data were obtained online
through the NCI web site at: http://www-seer.ims.nci.nih.gov in 1999.
2 Racial designations are listed as specified by NCI.
Chapter 11.8 11.8-3 Cost of Bladder Cancer
-------
Table 11.8-1. Age-specific Incidence of Bladder Cancer
Age Group
0- 14
15-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75-79
80-84
85+
Age-specific Rate of
Diagnosis Per
100,000
0.0
0.5
2.0
4.0
9.0
18.1
32.2
51.9
84.4
110.7
130.2
148.8
138.2
Percent of All
Bladder Cancer
Occurring in Age
Group
0.0
0.8
0.8
1.5
2.9
4.6
6.7
10.0
16.1
18.5
16.4
12.4
9.3
Cumulative Percent
of Bladder Cancer
0.0
0.8
1.6
3.1
6.0
10.6
17.3
27.3
43.4
61.9
78.3
90.7
100.0
Basedon NCI, 1999
II.8.A.2. Concurrent Effects
As with all cancers, bladder cancer may spread to other organs. In
addition, treatment of cancer, which usually includes chemotherapy,
radiation, and surgery, has numerous adverse side effects and may, in itself,
lead to death. This is particularly true of the agents used to treat bladder
cancer (Abeloff et al., 1995). Radiation treatments of cancer have led to
increased risks of other types of cancer, sterility, etc. Surgery, especially
the removal of a bladder, may cause long-term changes in health status,
including reduced capacity or increased susceptibility to respiratory disease
that may lead to death. These effects are associated with additional
medical costs are considered in this chapter if they occur during the
treatment period.3
3 The source of direct medical costs for this chapter (Baker et al., 1979 and 1981) include all
medical costs for cancer patients, minus usual background medical costs. This incremental approach allows
for the inclusion of medical costs that are associated with treatment and side effects, and discussed in more
detail in Section B.
Chapter 11.8
1.8-4
Cost of Bladder Cancer
-------
Figure II.8.1. Age Distribution of Bladder Cancer
(0 0.10
£ °-08
79 80-84 85+
Age
Bladder cancer has multiple concurrent symptoms that require treatment in
addition to the treatments directed at the primary medical goal of cancer
eradication. Many of these additional symptoms are related to
chemotherapy. Effects observed include: hematuria (blood in urine) and
irritative bladder symptoms, bladder obstruction leading to hydronephrosis,
tumor infiltration of regional nerves or bone causing pain, and lymphedema
as a result of lymphatic obstruction due to lymph node metastasis (Bennett
and Plum, 1996; Abeloff et al., 1995), increased risk of epididymitis,
orchitis (male reproductive disorders), pneumonitis, hepatitis, and sepsis
(Abeloff etal., 1995).
There is a strong link between bladder cancer and smoking. Bladder
cancer patients are much more likely to have smoked than people who have
not been diagnosed with bladder cancer (causality is discussed below).
Smoking is also associated with increased risks of many other diseases,
including other cancers. There is no indication, however, that bladder
cancer causes these other diseases. The simultaneous or sequential
occurrence of the diseases are likely due to their common causal link to
smoking.
The same pollutants that cause bladder cancer may cause other adverse
effects, especially of the urogenital system. These effects can incur added
medical costs not considered in this chapter. The risk assessment that
serves as the basis for a benefits evaluation should include all adverse
effects anticipated to result from exposure to the agent of interest.
Chapter 11.8
1.8-5
Cost of Bladder Cancer
-------
U.S.A.3. Causality & Special Susceptibilities
The causality and progression of this disease are not fully understood. It
has been hypothesized that bladder cancer may develop from a
preneoplastic and preinvasive localized condition to hyperplasia and then to
atypical hyperplasia and dysplasia. In some cases, the pathology
progresses further to neoplasms (Abeloff et al., 1995). As discussed
below, chemical irritants and other irritants are associated with bladder
cancer and may cause the observed cell proliferation and hyperplasia, as
well as the sometimes observed sequelae — bladder cancer. Experimental
evidence suggests that the DNA-damaging effects of carcinogens on the
urothelium causes cell proliferation. When the body is unable to repair
DNA adequately, progression to bladder cancer may occur. For more
detailed information on this topic see Abeloff et al. (1995).
Bladder cancer has been associated with environmental exposures for more
than 100 years. Exposure to aromatic amines and working in the dye
industry (especially with 2-naphthylamine) were known to cause high rates
of bladder cancer among workers. More recently, workers in the rubber,
electric, cable, paint, and textile industries had substantially higher
incidences of bladder cancer, with exposures to benzidine, auramine, and 4-
nitrophenol (Bennett and Plum, 1996), and arylamines from cigarettes and
other sources (Jones and Ross, 1999) were particularly noted. Arsenic is
also known to increase bladder cancer occurrence (ATSDR, 1998). It is
difficult to identify exposures that result in bladder cancer because the
latency period is between 15 and 50 years (Abeloff et al., 1995).
As noted above, cigarette smoking is also associated with a higher bladder
cancer risk and may account for one-half of all cases (Bennett and Plum,
1996; Abeloff et al., 1995). The prevalence of smoking and occupational
exposure among men may account for some or all of the increased
incidence seen in men versus women. The dramatic increases in bladder
cancer among women in recent years may be due to the relatively recent
entrance of women into the workforce, and their recent increase in tobacco
use. These factors would result in bladder cancer rates that are only
recently observed to increase, due to the long latency of most solid tumors
and the typically elderly age of diagnosis for bladder cancer.
In other parts of the world, infection with Schistosoma haemotobium is
responsible for a large proportion of bladder cancer cases in less-developed
countries (e.g., as studied in Egypt (Abeloff et al., 1995; Bennett and Plum,
1996)). Pharmaceuticals, including cyclophosphamide used in treating
malignancies, and the analgesic phenacetin, are also associated with
increased bladder cancer risk (Bennett and Plum, 1996; Abeloff et al.,
1995).
Chapter 11.8 11.8-6 Cost of Bladder Cancer
-------
Link to Table II. 1-1
Link to Chapter 1.1
Although genetic abnormalities are associated with bladder cancer, it is not
clear wither these occurred as a result of the disease (or biomarker) or
preceded the disease. Chromosome 9 abnormalities, particularly
monosomy, occurs early in bladder cancer. 1 Ip and 18p abnormalities are
found in more advanced tumors (Bennett and Plum, 1996). For a more
complete discussion of cellular-level changes and genetic markers for
bladder cancer, see Jones and Ross (1999) and Abeloff et al. (1995).
Individuals with a family history of bladder cancer before the age of 45
have a risk that is approximately 50 percent greater than the general
population risk (Abeloff et al., 1995). Other genetic risks are suggested by
data regarding cigarette smokers. Some individuals who detoxify cigarette
toxins more slowly are theorized to have higher risks associated with
smoking (Abeloff et al., 1995).
Other factors that may increase the risk of bladder cancer are chronic
bladder irritation, bladder infections, and urinary nitrites (Abeloff et al.,
1995).
Table II. 1-1 in Chapter II contains a list of many of the chemicals known to
cause or suspected of causing cancer (as reported in the EPA databases
IRIS, HEAST, and HSDB). Most chemicals in the table were carcinogenic
in animal studies. These studies do not provide organ-specific data because
it is not generally assumed that cancer induction will necessarily occur at
the same site in humans as in animals. Consequently, the chemicals listed in
Table II-1 may cause bladder cancer and/or other types of cancer.
Evaluation of the likelihood of this occurrence would require additional
research (e.g., risk assessment).
Bladder cancer is much more prevalent in the United States and in Spain
than in many other countries. Rates here are 25 to 30 cases per 100,000, in
contrast with a baseline rate of 2 per 100,000 in a typical low-rate area
(Bennett and Plum, 1996). Bladder cancer also has a positive association
with socioeconomic status (Jones and Ross, 1999; Abeloff et al., 1995),
which has not yet been explained.
NCI provides age-, sex-, and race-specific data regarding diagnosis of
bladder cancer from 1990 to 1994, which may be used to evaluate
susceptibilities among population subgroups. The data must be used with
care because diagnostic rates indicate occurrence only, and may or may not
indicate differences in susceptibility. See Chapter 1.1 for a more detailed
discussion of susceptibilities.
Chapter 11.8 11.8-7 Cost of Bladder Cancer
-------
U.S.A.4. Treatments and Services
As noted above, bladder cancer is usually treated with surgery,
chemotherapy, and/or radiation, depending on the type of bladder cancer,
the stage of cancer at diagnosis, patient health, and other factors. The
treatment of bladder cancer can be defined more precisely by histologic
type and specific location of the cancer in the bladder. In this analysis,
which is concerned with the average cost for all bladder cancers, all
histologic types and sub-sites are considered together.
Treatment is carried out in phases including initial diagnosis, initial
treatment, follow-up and maintenance treatment, and, for those who do not
survive, terminal treatment and palliative care. Although there are some
components of each treatment that are unique to each phase, most medical
activities and services may occur more than once over the course of the
disease from diagnosis to death or cure. For example, X-rays may be used
in diagnosis, to provide ongoing status updates, to assist in determining
initial and subsequent surgical and other treatment interventions, etc.
Initial diagnostic activities may include an evaluation of signs and
symptoms, abdominal, pelvic, bone, and chest scans, intravenous
pyelogram, cystoscopy, urinary cytology, computed tomography (CT)
scans, magnetic resonance imagery (MRI), biopsies, and other procedures
(Bennett and Plum, 1996; Abeloffet al., 1995). Staging of the disease
occurs during this phase and is critical to determination of subsequent
medical actions. Most tumors are confined to the transitional cell layer and
these are generally treated only with surgery. The tumors often recur,
which requires frequent cystoscopy with subsequent removal of recurrent
tumors as necessary. Higher-grade tumors require chemotherapy or
immunotherapy. Invasive cancers with a higher metastatic potential often
require total removal of the bladder, and subsequent reconstruction of an
alternative urinary reservoir (Bennett and Plum, 1996)
Although there is an 80 to 90 percent survival rate among bladder cancer
patients during initial diagnosis and treatment, tumors recur in 30 to 80
percent of patients, and 30 percent progress to a higher stage or grade.
Often chemotherapy is used to control or prevent this progression;
however, this entails the use of chemicals that are toxic to other organ
systems (Abeloffet al., 1995). Photodynamic therapy has also been used
recently with success, without the chemically-induced side effects (Abeloff
etal., 1995).
The small percentage of patients who cannot be cured receive terminal
care, which may include a variety of medical services, long-term care in a
nursing facility, palliative care, family counseling, etc.
Chapter 11.8 11.8-8 Cost of Bladder Cancer
-------
II.8.A.5. Prognosis
II.8.A.5.1 Background
The prognosis for bladder cancer is relatively good, compared to many
cancers. Approximately 26 percent of patients with bladder cancer die of
the disease.4 Mortality rates for bladder cancer patients have decreased
overall by 24 percent from 1973 to 1996. The most dramatic
improvements have been made among blacks under the age of 65 years,
with a 46.5 percent reduction in mortality among those diagnosed with the
disease (NCI, 1999). Improvements in survival are due to both better
diagnostic methods and improved treatments.
The overall prognosis for bladder cancer patients is good, with an average
of only 19 percent of patients dying of the disease within five years.
Among younger patients the survival rates are better, with those diagnosed
under the age of 45 having a 7.9 percent mortality rate over the first five
years post-diagnosis. Those with more limited tumors have a much better
prognosis than those with metastatic tumors (discussed below), with a
range of 6.9 percent mortality with localized tumors to 93.6 percent
mortality among those with distant tumors (based on 1989 to 1995 data)
(NCI, 1999).
Tumors that are restricted to a single site, or that occur in multiple sites
within the bladder, provide the best prognosis. Most bladder cancer cases
are superficial tumors of the transitional cell layer with a low potential for
metastatic spread. Those that invade multiple layers of the organ wall
and/or metastasize to other organs yield a poorer prognosis. Death occurs
primarily as a result of uncontrolled growth of metastatic tumors.
A dynamic observed with bladder cancer, but not common to cancers, is a
long-term increase in mortality over 20+ years. Most cancers cause high
rates of mortality during the first few years, but are often considered
"cured" if the patient survives without recurrence of the cancer for five
years. Based on National Cancer Institute (NCI) statistics, bladder cancer
continued to cause increased mortality for at least 20 years (the maximum
length tracked) post-diagnosis. This trend is discussed in more detail
below.
II.8.A.5.2 Relative Survival Rates (RSRs)
The NCI Surveillance, Epidemiology, and End Results (SEER) data
reports were accessed online to obtain information regarding mortality and
survival probabilities and the duration after diagnosis until death (NCI,
1999). Basic survival statistics on bladder cancer are provided in this
4 This value is relevant for patients diagnosed at the age of 70 years (the average age of diagnosis
for bladder cancer) and is based on a follow-up period of 20 years. The method of calculating this value is
discussed below in Section U.S.A.5.2, and source values are listed in column (6) of Table II.8-3.
Chapter 11.8 11.8-9 Cost of Bladder Cancer
-------
section because they relate to prognosis. Methods used to convert the NCI
statistics to survival probabilities are discussed briefly in this section and in
detail in Chapter II.2 on stomach cancer.
Link to Chapter II. 2
NCI provides the relative survival rate (RSR) for each year post-diagnosis.
The RSR is the number of observed survivors among these patients,
divided by the number of "expected" survivors among persons with the
same age and gender in the general population (observed/expected). The
equation for this is:
„„„ observed survival rate among bladder cancer patients
survival rate among age- and sex-matched cohort in the general population
The RSR takes into account that there are competing causes of death that
increase with age. The RSR for bladder cancer patients during the first year
post-diagnosis is 0.86 (NCI, 1999). This value indicates that a person with
bladder cancer would have, on average, a one-year survival probability that
is 86 percent of someone of the same age and gender in the general
population. The RSRs provided by NCI for each year post-diagnosis are
averages obtained from all ages at diagnosis.
An evaluation of the RSRs for bladder cancer over the past 20 years
indicates that (1) survival has increased notably (up to ten percent) over the
20 years, and (2) mortality from bladder cancer, while at much lower rates
than for some other cancers, continues at non-negligible rates for at least
20 years post-diagnosis. This trend differs from other cancer evaluations
previously carried out for this handbook. Due to the long-term dynamic of
increasing mortality for bladder cancer, the medical and opportunity costs
incurred by bladder cancer patients were estimated for 20 years post-
diagnosis (previous chapters considered ten years).
Because the RSRs for many years (e.g., 20 years) post-diagnosis
incorporate the survival probabilities of bladder cancer patients who were
diagnosed many years ago (e.g., in 1975), direct reliance on the RSRs
provided by NCI will result in downward-biased estimates of what RSRs
would be for patients who are currently being diagnosed with bladder
cancer. This bias occurs because the RSRs at each year post-diagnosis are
currently significantly higher than they were many years ago (i.e., the
survival for years one through five post-diagnosis in the late 1980s is
higher than for years one though five in the 1970s).
To provide a more accurate estimate of what the RSRs (and the
corresponding survival and mortality probabilities) for bladder cancer
patients are likely to be for the next 20 years, we estimated RSRs for each
Chapter 11.8 11.8-10 Cost of Bladder Cancer
-------
year post-diagnosis using a two-step procedure. This procedure focuses
on using the most current data available for each year post-diagnosis. In
the first step, we assumed that the ratio of the RSR at n years post-
diagnosis to the RSR at (n-1) years post-diagnosis in the most recent year
for which we have data is what that ratio will be in future years. For
example, the most recent year for which we have an RSR for one year
post-diagnosis for bladder cancer is 1995. The RSR in 1995 is 0.91, which
we assume will be the RSR for one year post-diagnosis in future years.
The most recent year for which there is an RSR for two years post-
diagnosis is 1994. We assume that:
RSRfuture
so that
RSR?94
RSRt994 •
Using the most recent RSR for one year post-diagnosis (0.91), and the
RSRs we have from 1994 for one year and two years post-diagnosis (0.911
and 0.865, respectively), the RSR for two years post-diagnosis is estimated
tobeO.91 x (0.865/0.911) = 0.8641. This estimate of the RSR for two
years post-diagnosis is then used to estimate the RSR for three years post-
diagnosis, using the above formula. We continue this process until we have
generated RSRs for each of twenty years post-diagnosis.
The RSRs, derived as described above, are only estimates of the underlying
population RSRs (i.e., the RSRs for the entire population of bladder cancer
patients in the United States). As such, they display some of the
"bumpiness" that data often contain. In step two, a plot of these estimated
RSRs against years post-diagnosis was generated. It shows that they
follow a general exponential decay trend. Rather than use these estimated
RSRs, regression was used to estimate the smooth trend described by the
estimates derived in step one. In particular, we estimated the model:
In(RSR) = a + b x ln(years post-diagnosis)
using the RSRs estimated in step one. The intercept (a) was estimated to
be -0.131866 and the slope (b) was estimated to be -0.0134114. The fit
was excellent, with an R2 of 0.952. Exponentiating the predicted natural
logarithms of RSR yielded the predicted RSRs shown in Table II.8-2.5
5 All vital statistics data in this document applicable to the general population were obtained from
the National Center for Health Statistics (NCHS) Vital Statistics in the United States (NCHS, 1993).
Chapter 11.8 11.8-11 Cost of Bladder Cancer
-------
Some bladder cancer patients will die of bladder cancer, but most die of
other causes. The probability of a bladder cancer patient dying of causes
other than bladder cancer cannot be assumed to be the same as the
probability of someone in the general population dying of other causes,
particularly in the first few years post-diagnosis, when a bladder cancer
patient's probability of dying of bladder cancer is not trivial.6 This
becomes clear in the extreme case in which the probability of dying of an
illness is extremely high. Suppose, for example, that the probability of
dying of all causes except for illness X is 0.025 in the general population.
Suppose that in a cohort of patients diagnosed with illness X the
probability of dying from illness X in the first year post-diagnosis is 0.99.
If dying of other causes in this cohort were the same as in the general
population (0.025), then their probability of dying would be greater than
1.0.
6 This difference becomes clear in the extreme case in which the probability of dying of an illness is
extremely high. Suppose, for example, that the probability of dying of all causes except for illness X is
0.025 in the general population. Suppose that in a cohort of patients diagnosed with illness X, the
probability of dying from illness X in the first year post-diagnosis is 0.99. If the probability of dying of
other causes in this cohort were the same as in the general population (0.025), then the probability of
someone in the cohort dying would be greater than 1.0.
Chapter 11.8 11.8-12 Cost of Bladder Cancer
-------
Table 11.8-2. Estimated RSRs* for Bladder Cancer for the First
20 Years Post Diagnosis
Years Post-Diagnosis (n)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Estimated RSR for n Years
Post-Diagnosis
0.86
0.85
0.84
0.83
0.82
0.81
0.80
0.79
0.78
0.77
0.76
0.75
0.74
0.73
0.72
0.71
0.70
0.69
0.68
0.67
The estimated RSR for each year post-diagnosis is the result of a two
step procedure, using the set of RSRs reported by NCI (1999), as
described in the text above.
The probability of a bladder cancer patient dying of bladder cancer and the
probability of a bladder cancer patient dying of some cause other than
bladder cancer in the nth year post-diagnosis, given survival to the nth
year, were each derived from two known probabilities:
1) the probability of a bladder cancer patient surviving through the
nth year post-diagnosis, given survival to the nth year; and
(2) the probability of dying of causes other than bladder cancer in a
matched cohort in the general population.
The derivation is explained in detail in the Appendix to Chapter 11.2.
Link to Chapter II. 2, Appendix II. 2-A
Because each of the known probabilities depends on the number of years
post-diagnosis and (minimally) on age at diagnosis, the derived
Chapter 11.8
11.8-13
Cost of Bladder Cancer
-------
probabilities were calculated for each of the 20 years post-diagnosis and for
the average age at diagnosis (70 years).7 The following probabilities are
shown in Table II. 8-3:
1) survival through the wth year,
2) dying of bladder cancer during the nth year, and
3) dying of some other cause during the wth year.
Probabilities of survival and dying of all causes among all members of the
general population aged 70 were obtained from the National Center for
Health Statistics (NCHS) Vital Statistics in the United States (NCHS,
1993). They are also shown in Table II.8-3. The values in this table are
used in Section II. 8.B to calculate the expected medical costs of bladder
cancer patients. The probabilities in the general population of dying from
bladder cancer are 0.00020 in the 70-74 year age group, 0.00031 in the 75-
79 year age group, 0.00047 in the 80-84 year age group, and 0.00069 in
the 85+ age group. The probabilities in column (3) were derived by
subtracting these probabilities from the corresponding probabilities of dying
from any cause in the wth year, given survival to the nth year. The Chapter
II.2 Appendix contains a detailed explaination of the derivation of survival
and mortality probabilities.
Link to Chapter II. 2, Appendix II. 2-A
The mortality rate of 26 percent, cited in the introduction to this section,
was calculated for patients who are diagnosed at age 70 as the sum of the
probabilities of their dying of the disease in each year post-diagnosis. This
rate was calculated for 20 years post-diagnosis, using the data shown in
Table II.8-3. The probabilities of dying of bladder cancer during each year
post-diagnosis, shown in the column titled "probability of dying of bladder
cancer in the nth year post-diagnosis" were summed to obtain a value of 26
percent.
7 Twenty years is period that captures most of the deaths due to bladder cancer among those
diagnosed with the disease. This period is a reasonable maximum duration of maintenance care and
treatment for those who do not die of bladder cancer.
Chapter 11.8 11.8-14 Cost of Bladder Cancer
-------
Table 11.8-3. Survival and Mortality Probabilities for the Average Bladder Cancer Patient3
Years
post-
diagnosis
(n)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
A Cohort in the General
Population (Matched)
Probability
of
surviving n
years
1.000
.973
.945
.915
.884
.852
.817
.782
.745
.707
.667
.626
.584
.541
.497
.451
.410
.373
.339
.308
.280
Probability of
dying in nth year of
causes other than
bladder cancer,
given survival to
the nth year
—
.027
.029
.031
.034
.037
.040
.043
.047
.051
.056
.061
.067
.073
.081
.090
.090
.090
.090
.090
.090
A Cohort of Bladder Cancer Patients
Relative
Survival
Rate
(RSR)
—
0.86
0.85
0.84
0.83
0.82
0.81
0.80
0.79
0.78
0.77
0.76
0.75
0.74
0.73
0.72
0.71
0.70
0.69
0.68
0.67
Probability
of surviving
through the
nth year
post-
diagnosis
1.0
.842
.806
.771
.735
.698
.661
.624
.586
.549
.511
.473
.436
.398
.361
.324
.290
.260
.234
.210
.188
Probability of
dying of
bladder
cancer in the
nth year post-
diagnosis"
—
.134
.011
.011
.010
.010
.009
.009
.008
.008
.007
.007
.006
.006
.005
.005
.004
.004
.004
.003
.003
Probability of
dying of other
causes in the
nth year post-
diagnosis
—
.025
.024
.025
.026
.027
.028
.028
.029
.030
.030
.031
.031
.032
.032
.032
.029
.026
.023
.021
.019
a. The survival and mortality probabilities for bladder cancer patients presented here are derived from the RSRs
estimated from RSRs obtained from NCI and the survival probabilities for a matched cohort in the general
population. The average age of diagnosis of 70 years was used. See text for an explanation calculation
methods.
b. When the probalities in this column are summed, they yield the probability of dying of bladder cancer over 20
years post-diagnosis, which is equal to 26 percent.
Chapter II.8
11.8-15
Cost of Bladder Cancer
-------
There is likely to be additional loss, although at a very low rate, beyond 20
years; however, data were not available on those values. Impacts beyond
20 years are unlikely to have a substantial impact on cost analyses because
there are many competing causes of death when a person reaches 90 years
of age.
The RSR can be used to approximate the probability of mortality at young
ages, when the background death rate is minimal. The bladder cancer
mortality rate is approximated by 1 - RSR when background age-related
mortality from other causes is not considered. In the case of bladder
cancer, the RSR is 0.67 and 1 - RSR = 0.33. In reality, there is always a
background death rate in a population, and this rate increases with age. As
noted previously, most bladder cancer patients do not die of bladder
cancer. Only 18.8 percent of the population diagnosed with bladder cancer
survive to 20 years post-diagnosis, but most of these losses are due to
other causes of death than bladder cancer. The RSR, used with the
background mortality rates of the population at the average age at
diagnosis, provides clear information on the survival and mortality dynamic
of that specific population. If the cancer occurs at a younger age than
usual due to its genesis (e.g., chemical induction), however, then the
mortality statistics obtained through the method described above will
underestimate bladder cancer-related mortality. In these cases, the RSR
itself is a better approximation of survival (and derived mortality), allowing
estimation of mortality at young ages when background mortality is
negligible.
Some environmentally-induced cancers, such as arsenic-induced skin
cancer, occur at much younger ages than those at which the cancers are
typically observed in the general population. Consequently, the use of the
inverse of survival, 1 - RSR (e.g., 0.33 in the case of bladder cancer), as
an estimate of mortality may be very relevant for calculating benefits
associated with the avoidance of some environmentally-induced cancers.
Due to the higher mortality estimates that this approach will always
generate, the benefits of avoiding the disease will be larger when the RSR
is used directly to estimate cancer mortality. When there is no evidence
that the disease will occur at an age that is younger than that of the general
population, the calculations that precede this — that link morbidity and
mortality to the average age at diagnosis — are used to estimate direct
medical costs.
Chapter 11.8 11.8-16 Cost of Bladder Cancer
-------
11.8.B Costs of Treatments and Services
II.8.B.1. Methodology
Link to Chapter 1.1
II.8.B.1.1 Overview
There is no single typical case or treatment pattern for bladder cancer due
to individual differences in the stage of cancer at diagnosis, multiple
treatment options, patient health and age, and other factors; however,
average costs can be calculated. Treatment of bladder cancer may occur
over a briefer extended period of time, and costs may be limited or
substantial. As discussed in Section II.8.A, bladder cancer has a relatively
low mortality rate, with a relative survival rate of 0.67. The medical costs
of those who die of the disease are usually very different than for those
who survive (this is discussed in more detail in Chapter I.I). This chapter
therefore provides costs for the "average" bladder cancer patient, as well as
for survivors and nonsurvivors as separate patient groups.
II.8.B. 1.2 Medical Cost Data
II.8.B.1.2.1 Sources
Medical cost data would ideally be obtained on current medical
expenditures. Although data files are maintained by public and private
sector sources, they are not generally available for public use. In addition,
to obtain reliable cost estimates it is necessary to evaluate very large
databases of charges from a variety of sources. This method was neither
practical nor cost-effective for the development of this chapter, given the
availability of summary data from other sources. A data search was
conducted to locate information in the medical economics literature
regarding medical costs associated with bladder cancer. In addition to a
literature search, most federal agencies dealing with cancer, disabilities,
medical costs and their management, and related issues were contacted for
information, and the various federal databases were discussed with senior
staff at these agencies.
Very recent cost data were not located.8 However, current (1994) cancer
data were obtained regarding incidence and survival (as reported in Section
II.8.A, above), and were used with cost data from the 1980s described
below. The cost estimates presented in this section are based primarily on
the work of Baker et al. (1989) and Hartunian et al. (1981), respectively,
and on two sources of statistical data: the National Cancer Institute (1999)
and Vital Statistics of the United States, 1993 (NCHS, 1997).
8 Studies were located that used more recent cost data than those used in this analysis. Due to
serious limitations (i.e., data were incomplete), the studies were not used. They are reported in the "Other
Studies" section at the end of Section II.8.B.
Chapter 11.8 11.8-17 Cost of Bladder Cancer
-------
Based on the 1997 review of the medical literature carried out for the
development of this chapter, there do not appear to be widely-adopted new
treatment methods for bladder cancer that alter either the medical costs or
the survival rates for most patients substantially. Consequently, the cost
estimates presented in this chapter may be considered appropriate under
most circumstances (e.g., regional costs may vary).
II.8.B.1.2.2 Baker et al.'s Cost Estimation Method
Baker et al. (1989) used the Continuous Medicare History Sample File
(CMHSF) to estimate the per-patient average lifetime medical cost of
treating bladder cancer based on data files from 1974 to 1981. They chose
CMHSF because:
1) it is a nationally representative sample of the Medicare
population (five percent), covering over 1.6 million patients;
2) it is longitudinal, dating from 1974 to 1981; and
3) it captures the majority of medical expenses for each beneficiary.
Five Medicare files are included in the CMHSF, which cover:
1) inpatient hospital stays,
2) skilled nursing facility stays,
3) home health agency charges,
4) physicians' services, and
5) outpatient and other medical services.9
Costs not included are outpatient prescription medications and nursing
home care below the skilled level.
Because CMHSF provides no indication of initial diagnosis, Baker et al.
assumed that disease onset occurred when a diagnosis of bladder cancer
was listed on a hospitalization record following a minimum of one year
without a bladder cancer diagnosis. This assumption is reasonable due to
the high frequency of hospitalization associated with the disease (i.e.,
individuals diagnosed with bladder cancer would be hospitalized). Only
patients with an initial diagnosis during the years covered by the database
(1974-1981) were included.
9 See Baker et al. (1989 and 1991) for further details. Baker et al. (1991) contains additional
descriptive data regarding the database and methods used for the cost analysis; however, it does not contain
cost data for bladder cancer.
Chapter 11.8 11.8-18 Cost of Bladder Cancer
-------
Costs associated with bladder cancer were assigned to three post-
diagnostic time periods:
• initial treatment, during the first three months following
diagnosis;
• maintenance care, between initial and terminal treatment; and
• terminal treatment during the final six months prior to death.
As noted in Chapter 1.1, the amount paid for service may differ from the
actual medical costs because many insurers and federal programs either
1) pay only a portion of total costs, or 2) pay more than actual costs to
underwrite the care providers' losses due to underpayment from other
sources.
Link to Chapter 1.1
Baker et al. used provider charges, rather than Medicare reimbursements
(which represent only a portion of most total charges), thus providing a
more accurate cost estimate. To improve the accuracy of the cost
estimates, Baker et al. included cost data on coinsurance, deductibles, and
other cost components. They made four adjustments to the cost estimates
calculated from the CMHSF. First, charges were added for skilled nursing
facilities (SNFs) not covered by Medicare by multiplying the "length of
stay" at an SNF (computed from admission and discharge dates) by the
average daily SNF charge. Second, the annual Medicare Part B deductible
of $60 was added to the reimbursed charges in the database. Third, since
Medicare pays only 80 percent of physicians' charges, Baker et al. scaled
these reimbursements to 100 percent of physicians' charges to better reflect
social costs. Finally, they inflated all dollar values to 1984 dollars using the
Medical Care component of the Consumer Price Index.
II.8.B.1.2.3 Cost Estimates by Treatment Period
Medical costs associated with the initial, maintenance, and terminal cancer
care treatment periods were itemized in Baker et al. (1989) and are shown
in Table II.8-4. The 1989 paper did not report incremental costs or the
costs of other medical services, which would be anticipated to occur while
the patient was receiving cancer treatment (i.e., co-morbidity^ackground
costs). In order to estimate the incremental costs, a co-morbidity cost of
$2,988 per year (1984 dollars) from Baker et al. (1991) was used in this
analysis. (This is equivalent to $6,394 in 1996 dollars using the CPI
multiplier of 2.14 for 1984 to 1996.) The co-morbidity cost was pro-rated
for this analysis using the specified durations for the initial (three-month)
and terminal (six-month) treatment periods.
Table II.8-4 lists the incremental costs calculated for the three treatment
periods. Total costs are reported for the initial and terminal care periods.
Annual costs for the maintenance period are shown and are further
Chapter 11.8 11.8-19 Cost of Bladder Cancer
-------
discussed in the "Lifetime Costs" section below. Using the Medical Care
component of the Consumer Price Index (CPI-U), all costs are inflated to
1996 dollars for purposes of this handbook. (The adjustment factor for
1984 to 1996 is 2.14; Bureau of Labor Statistics.)
Table 11.8-4. Average Per Patient Costs for the Three Periods of
Treatment for Bladder Cancer in 1996 dollars
Costs adjusted for inflation using the Medical Care component of the
Consumer Price Index (CPI-U) 1996:1984 = 2.14 (Bureau of Labor)
Treatment Period
Initial
(3 months)
Maintenance (per year)
Terminal
(6 months)
Incremental Cancer Treatment
Cost
$16,527
$13,277
$36,558
Based on Baker et al., 1989, with comorbidity charges from Baker et al., 1991.
Link to U.S.A.5.2
II.8.B. 1.3 Calculation of Lifetime Cost Estimates for the
"Average" Bladder Cancer Patient
This section contains a discussion of the calculation of lifetime medical
costs for the "average" bladder cancer patient. The sections that follow
discuss methods and results of calculations for estimating costs for
survivors and nonsurvivors of bladder cancer separately. These separate
approaches were used to address specific requirements of different
activities that EPA carries out using direct medical cost data. Although
Baker et al. (1989) provide useful cost estimates for the three treatment
periods, they do not provide information on two critical aspects of medical
costs:
1) costs for survivors versus nonsurvivors of bladder cancer. These
may differ substantially. For example, survivors would not have
terminal care costs and may receive maintenance services for an
extended time period.; and
2) estimates of the duration of the maintenance periods.
Data regarding age at diagnosis of bladder cancer were obtained from NCI
(1999). Survival and mortality probabilities for each year post-diagnosis
were derived from relative survival rates obtained from NCI (1999), as
discussed in Section II.8.A.5.2.
This information was used to address many time-related medical cost
issues. For some aspects of the analysis, however, detailed information
Chapter 11.8
11.8-20
Cost of Bladder Cancer
-------
was not available and average values have been used as a reasonable
approximation (e.g., a 20-year maintenance period was assumed for
survivors of bladder cancer). When average values or other assumptions
are used in this analysis, they are so noted.
As previously noted, there are no substantial differences in survival related
to age at diagnosis, and NCI does not provide age-specific relative survival
rates for each year post-diagnosis. Consequently, it was assumed for this
analysis that the relative survival rates for bladder cancer were the same for
all ages. The survival and mortality probabilities for bladder cancer
patients, which are incorporated into calculations of expected medical costs
as discussed below, are based on this assumption.
The analysis assumes that death always occurs midyear. All bladder cancer
patients are therefore assumed to incur the costs of initial treatment during
the first three months of the illness. The costs incurred after that during the
first year depend on whether the patient:
1) survives through the year,
2) dies of bladder cancer during the year, or
3) dies of some other cause during the year.
Patients who survive through the year incur the costs of initial treatment
($16,527.2) during the first three months, and then incur nine months'
worth of maintenance care costs (0.75 x $13,276.6 = $9,957.4) during the
remainder of the year. The total cost incurred during the first year by those
patients who survive the year is therefore $16,527.2 + $9,957.4 = $26,485.
Bladder cancer patients who die of bladder cancer during the first year incur
the initial treatment cost and then incur terminal care costs for the remaining
three months of their lives (because those who die are assumed to die
midyear). Total costs during the first year post-diagnosis in this case are
therefore $16,527.2 + (0.5 x $36,557.6) = $34,806.
Finally, the small percentage of bladder cancer patients who die of causes
other than bladder cancer during the first year post-diagnosis incur the
initial treatment costs and then incur three months' worth of maintenance
care costs. Total first-year costs for these patients are therefore $16,527.2
+ 0.25 x $13,276.6 = $19,846.
Chapter 11.8 11.8-21 Cost of Bladder Cancer
-------
The expected medical costs for bladder cancer patients during the first year
post-diagnosis, then, may be expressed as:
Expected First-Year Cost: initial treatment costs +
[maintenance care costs for nine months x probability of
survival through first year + terminal care costs for three
months x probability of dying of bladder cancer during first
year + maintenance care costs for three months *
probability of dying of other causes during the first year]
Example: Expected first-year medical costs of a bladder cancer patient
diagnosed at age 70
As noted above, all bladder cancer patients incur an initial treatment cost of
$16,527. Those who survive through the year also incur maintenance care
costs for the remaining three quarters of the year. The total first-year costs
of those who survive the year are:
Initial treatment: $16,527.2
Maintenance treatment: $9,957.4 (.75 x $13,276.6)
Total First-Year Cost $26,485
About nine percent of bladder cancer patients die of bladder cancer during
the first year. Those who do will incur the initial treatment costs plus half
of the terminal care costs. The total first-year costs of those who die of
bladder cancer during the year are:
Initial treatment: $16,527.2
Terminal care: $18,278.8 (.50 x $36,557.6)
Total First-Year Cost $34,806
Finally, a small percentage of patients will die of competing illnesses during
the first year. Because those who die of causes other than bladder cancer
are assumed to die at the midpoint of the year, costs during the first half of
the year are assumed to consist of the initial treatment costs for three
months, plus three months of maintenance care costs as follows:
Initial treatment: $16,527.2
Maintenance treatment: $3,319.1 (.25 x $13,276.6)
Total First-Year Cost $19,846
For each subsequent year, costs consist entirely of maintenance care costs
for those who survive the year. For those who do not survive the year,
costs depend on whether death was due to bladder cancer or other causes.
Chapter 11.8 11.8-22 Cost of Bladder Cancer
-------
For those who die of bladder cancer during the nth year, costs incurred that
year consist of six months of terminal care costs, or $36,558. For those
who die of other causes during the nth year, there are six months of
maintenance care costs, or 0.5 x $13,276.6 = $6,638.
The expected first-year medical cost incurred by the "average" bladder
cancer patient diagnosed at age 70 is a weighted average of the costs of
those who survive the first year, those who die of bladder cancer during the
first year, and those who die of other causes during the first year, where the
weights are the probabilities of each of these occurrences.10 The weighted
average medical costs were calculated for 20 years post-diagnosis and
expected costs were summed over the 20 years. This timeframe was
assumed to be a reasonable period over which additional medical costs
associated with bladder cancer (i.e., maintenance care costs) would be
incurred by bladder cancer patients.
Although the actual average period of maintenance care for bladder cancer
is not known, the resulting uncertainty about the expected maintenance
costs during a 20-year period is somewhat lessened by the fact that a large
percentage of bladder cancer patients diagnosed at age 70 die within the
20-year period (mostly of other causes), and would therefore not incur
maintenance costs for the full 20 years anyway.
The expected medical costs for bladder cancer patients during the nth year
post-diagnosis, for ri>\, then, may be expressed as:
Expected nth-Year (n>l) Cost: [maintenance care cost for
one year x probability of survival through nth year +
terminal care cost for six months x probability of dying of
bladder cancer during the nth year + maintenance care cost
for six months x probability of dying of other causes during
the nth year]
Expected Lifetime Cost = Expected first-year cost + the sum
of the (discounted) expected subsequent-year costs
10 Although this analysis focuses on costs incurred by a patient diagnosed with bladder cancer at
the average age of diagnosis (70 years), some environmentally-induced cancers are diagnosed at earlier
ages than those commonly reported for the cancers (e.g., arsenic-induced skin cancer). As noted in Section
II.8.A, this earlier diagnosis has an impact on the mortality dynamics and on the direct medical costs. The
uncertainty analysis contained in Section II.8.C includes an age-specific analysis of direct medical costs,
that demonstrates the differences that can result in earlier ages of onset than those used in the basic analysis
for this chapter.
Chapter 11.8 11.8-23 Cost of Bladder Cancer
-------
The first year of treatment is calculated differently from other years
because the first three months of that year are spent in "initial" treatment
and the costs for that period of intensive medical care and surgery are
calculated separately.
The mathematical equation for the expected lifetime medical costs incurred
by the "average" bladder cancer patient over a 20-year period is:
$16,527 + ($13,277 x 0.75 x psj + ($13,277 x 0.25
$13,277 , , ,__. 0 „ $6,638 , ,
20
+E
v=2
(pm °
y
(pm
bc
+ ($36,558 x 0.5
$36,558 .
Where:
y
ps
pm
pm°
r
be
Link to Table II.8-3
= the year post-diagnosis,
= the probability of surviving through the year,
= the probability of dying of bladder cancer during the year,
= the probability of dying from other causes during the year, and
= the discount rate.
The cost estimates for each year post-diagnosis and the estimate of
undiscounted expected total cost for a 20-year period are shown in Table
II.8-5 for the "average" bladder cancer patient diagnosed at age 70. The
survival and mortality probabilities necessary for the calculations of costs
are shown in Table II.8-3.
II.8.B.1.4 Calculation of Lifetime Cost Estimates Separately
for Bladder Cancer Survivors and Nonsurvivors
II.8.B.1.4.1 Survivors and Nonsurvivors
As noted above, there are differences in medical services provided to
bladder cancer patients who survive the disease (survivors) versus those
who die of the disease (nonsurvivors). Based on cost estimates by Baker et
al. (1989), terminal care is provided for approximately six months to
terminally ill cancer patients. The costs to nonsurvivors for this care
($36,558) is considerably higher than costs for survivors who receive
maintenance care for the same period of time ($6,638).n
EPA may use the value of a statistical life (VSL) for nonsurvivors and thus
calculate separate costs for survivors and nonsurvivors. The method
11 Nonsurvivors include only those who die of bladder cancer and do not include those who die of
any other causes.
Chapter 11.8
11.8-24
Cost of Bladder Cancer
-------
Link to Table II.8-3
shown above to calculate costs for the "average" patient uses the
unconditional probabilities of survival and mortality listed in Table II.8-3.
The method used to calculate costs for survivors and nonsurvivors
separately requires the probabilities that are conditional on being either a
survivor or nonsurvivor of bladder cancer.
The conditional probability of a bladder cancer nonsurvivor dying in the nth
year is the number of nonsurviving bladder cancer patients who die of
bladder cancer during the nth year divided by the total number of bladder
cancer nonsurvivors. Likewise, the conditional probability of a bladder
cancer survivor dying in the nth year is the number of bladder cancer
survivors who die (of causes other than bladder cancer) during the nth year
divided by the total number of bladder cancer survivors. A detailed
explanation of the derivation of these values is provided in Chapter 11.2.
The conditional probabilities of survival and mortality for survivors and
nonsurvivors of bladder cancer are given in Table II.8-6.
Link to Chapter II. 2
II.8.B.1.4.2 Calculation of Lifetime Cost Estimates for Bladder
Cancer Survivors
As shown in the example portion of Section II.8.B.1.3, cost estimates are
calculated by summing the costs of the different treatment phases over the
lifetime of the bladder cancer patient.
Link to Section II. 8.B. 1.3
Chapter 11.8 11.8-25 Cost of Bladder Cancer
-------
Table 11.8-5. Expected Costs of Medical Services (in 1996$) for Bladder Cancer Patients (Age of
Onset = 70)a
Years Post-
Diagnosis (n)
1b
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Medical Costs in the nth Year (undiscounted)
if survive
through the
nth year
26,485
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13.277
13.277
13.277
13.277
13.277
13.277
13.277
13.277
13.277
13.277
if die of
bladder
cancer in
the nth year
34,806
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
if die of other
causes in the nth
year
19,846
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
Expected Total Cost Through the 20th Year Post-Diagnosis for a Bladder
Cancer Patient Diagnosed at Age 70
Expected Medical
Costs for the nth Year
Post-Diagnosis0
(Undiscounted)
27,432
1 1 ,276
10,790
10,299
9,802
9,301
8,794
8,283
7,769
7,254
6,740
6,224
5,708
5,195
4,685
4,205
3,772
3,384
3,035
2,723
156,670
a. The probabilities used in this table are from Table II. 8-3. The costs are from Table II. 8-4.
Links to Tables 11.8-3 and 11.8-4
b. First-year costs include the charge for "initial" therapy ($16,527). The duration of maintenance care is adjusted
accordingly (see text for discussion).
c. Calculated using the probabilities in Table II. 5-3 and the costs in Columns (2), (3), and (4) of this table.
Link to Table II. 5-3
Chapter II.8
II.8-26
Cost of Bladder Cancer
-------
Table 11.8-6. Conditional Probabilities of Survival and Mortality for Survivors and Nonsurvivors
of Bladder Cancer (Age of Onset = 70)a
Years Post-
Diagnosis
(n)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Bladder Cancer Survivors
Conditional probability of:
Surviving
through the nth
year
.966
.934
.900
.865
.828
.791
.752
.713
.672
.631
.589
.546
.503
.459
.415
.376
.341
.309
.281
.255
Dying of some
other cause
during the nth
year
.034
.033
.034
.035
.036
.037
.039
.040
.041
.041
.042
.043
.043
.044
.044
.039
.035
.032
.028
.025
Bladder Cancer Nonsurvivors
Conditional probability of:
Surviving
through the nth
year
.494
.451
.410
.371
.334
.299
.266
.234
.205
.177
.151
.127
.105
.085
.067
.051
.036
.023
.011
.000
Dying of bladder
cancer during
the nth year
.506
.043
.041
.039
.037
.035
.033
.031
.030
.028
.026
.024
.022
.020
.018
.016
.015
.013
.012
.011
a. As noted for Table II. 8-3, the conditional survival and mortality probabilities for bladder cancer patients
presented here are derived from the RSRs estimated from RSRs obtained from NCI and the survival probabilities
for a matched cohort in the general population. See Section II.8.A.5.2 for an explanation of the estimation of
RSRs.
Link to Section II. 8. A. 5.2
The expected medical costs for bladder cancer survivors during the first
year post-diagnosis may therefore be expressed as:
Expected First-Year Cost: initial treatment costs + [maintenance care
costs for nine months x probability of survival through first year +
maintenance care costs for three months * probability of dying of
other causes during the first year]
Chapter II.8
II.8-27
Cost of Bladder Cancer
-------
The expected medical costs for bladder cancer survivors during the nth
year post-diagnosis, for ri>l, then, may be expressed as:
Expected nth-Year («>1) Cost: [maintenance care cost for
one year x probability of survival through «th year +
maintenance care cost for six months x probability of dying
of other causes during the «th year]
Expected Lifetime Cost = Expected first-year cost + the sum
of the (discounted) expected subsequent-year costs
Note that the probabilities used in these calculations are the conditional
probabilities given in Table II.8-6. They are conditional on the bladder
cancer patient not dying of bladder cancer.
Using the initial, maintenance, and terminal care costs from Table II.8-6,
the mathematical equation for the lifetime costs incurred by bladder cancer
survivors is:
$16,527 + pm^ x 0.25 ($13,277) + ps^ x .75 x $13,277
20
+ E
y = 2
s $13,277
_x + /»»,
$6,638
where:
y
pss
pms
the year post-diagnosis;
the conditional probability of survival for that year,
conditional on being a survivor of bladder cancer;
the conditional probability of mortality for that year,
conditional on being a survivor of bladder cancer; and
the discount rate.
The expected medical costs for bladder cancer survivors for each year post-
diagnosis, as well as the expected total medical costs over 20 years post-
diagnosis, are shown in Table II.8-7.
Chapter 11.8
11.8-28
Cost of Bladder Cancer
-------
Table 11.8-7. Expected Undiscounted Costs of Medical Services (in 1996$) for
Survivors of Bladder Cancer (Age of Onset = 70)
Years Post-
Diagnosis
(n)
1C
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Medical Costs Through the 20th Year Post-diagnosis3
(undiscounted)
Medical Cost if
Survive Through
the nth Year
26,485
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
Medical Cost if Die of
other Causes in the
nth Year
19,846
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
6,638
Expected Total (Undiscounted) Cost Through the 20th Year
Post-Diagnosis:
Total Cost Based on
Weighted Average13
26,262
12,613
12,172
11,713
11,238
10,748
10,243
9,724
9,193
8,649
8,096
7,535
6,965
6,389
5,808
5,255
4,759
4,315
3,916
3,559
179,153
a. Costs are based on data reported in Table II. 8-4, adapted from Baker et al., 1989.
Probabilities of survival and mortality, taken from Table II. 8-6, are conditional on surviving bladder
cancer.
b. Weighted average of the costs incurred by survivors who survive the year and the costs incurred
by survivors who die of other causes during the year. Weighting is based on the conditional
probabilities provided in Table II. 8-6.
c. Costs during the first year include a charge for "initial" therapy ($16,527), and the duration of
maintenance or terminal care is adjusted accordingly. See text for discussion.
II.8.B.1.4.3 Calculation of Lifetime Cost Estimates for Bladder
Cancer Nonsurvivors
Nonsurvivors of bladder cancer will incur initial, maintenance, and terminal
costs. Their lifetime medical costs associated with the disease can be
calculated from the costs per treatment period shown in Table II.8-4 and
the conditional probabilities for nonsurvivors of bladder cancer shown in
Table II.8-6.
Links to Tables II. 8-4 and II. 8-6
Chapter II.8
II.8-29
Cost of Bladder Cancer
-------
As Table II.8-6 indicates, about 55 percent of bladder cancer nonsurvivors
die within one or two years of diagnosis. Of the remaining 45 percent of
nonsurvivors, death from bladder cancer may occur anywhere within the
remaining 18 years of the 20-year period considered in this analysis, with
the conditional probabilities of death ranging from four percent in the third
year post-diagnosis to about one percent in the twentieth year. As with
bladder cancer survivors, medical costs for nonsurvivors each year post-
diagnosis were calculated as a weighted average of the costs incurred by
those who survive the year and those who die (of bladder cancer) during
the year.
It was assumed that those who die during a year receive six months of care
(as was done for the survivors above). It was also assumed that terminal
care lasting six months would be provided to all nonsurvivors. Therefore,
unless death occurred during the first year, when initial care was assumed
to occur, the care costs assigned to the last year of life were terminal costs.
If death occurred during the first year post-diagnosis, it was assumed that
initial care and three months (half of the total) of terminal care were
provided.
The general description of medical costs for nonsurvivors may be
expressed as:
Expected First-Year Cost: [initial costs + half the terminal
costs] x probability of mortality during the first year +
[initial costs + maintenance care costs for nine months] x
probability of survival for first year
Expected nth-Year (n>l) Cost: maintenance care cost for
one year x probability of survival through nth year +
terminal costs x probability of mortality in «th year
Expected Lifetime Cost = Expected first-year cost + the sum
of the (discounted) expected subsequent-year costs
As with the cost calculations for bladder cancer survivors, the probabilities
used in these cost calculations are the conditional probabilities given in
Table II. 8-6, in this case, conditional on dying of bladder cancer.
Link to Table II.8-6
Chapter 11.8
11.8-30
Cost of Bladder Cancer
-------
Using the initial, maintenance, and terminal care costs from Table II.8-6,
the mathematical equation for the expected lifetime costs incurred by
nonsurvivors is:
$16,527 + pm™ x 0.5 ($36,558) + ps™ x .75 x $13,277
20 „, $13,277 m $36,558
ns , + ns
y t-t . _.\v-i "
y = 2
(1 + rF1 ' (1
where:
y = the year post-diagnosis;
psns = the conditional probability of survival for that year,
conditional on being a nonsurvivor of bladder cancer;
pmns = the conditional probability of mortality for that year,
conditional on being a nonsurvivor of bladder cancer; and
r = the discount rate.
The costs are summed over all years from diagnosis to death. Maintenance
care costs are not added in the last year of life because the patient is
assumed to receive terminal care during the six months assumed to
constitute this period. (The discounted results are shown in the "Results"
section that follows.) The approach is the same as that shown in the
example in Section II.8.B. 1.3. When the costs for each year are summed
over a period of 20 years post-diagnosis, the total 20-year cost per
nonsurvivor is obtained. These costs are shown in Table II.8-8.
Link to Section II. 8.B. 1.3
The results shown above can be used to calculate costs for an "average"
bladder cancer patient, from the costs calculated for survivors and
nonsurvivors. The expected medical costs of a bladder cancer patient can
be calculated as a weighted average of the expected costs of survivors and
nonsurvivors of bladder cancer. This approach, which was not used to
calculate costs for the "average" patient in this chapter, yields the same
results as the approach that was shown in Section II.8.B. 1.3. In brief, the
approach used in this chapter for the average patient (in Section II.8.B.1.3)
uses cost data for all patients, weighted by their average utilization of
services. If the survivor and nonsurvivor data were used, which
incorporate utilization of services, the cost results obtained through
separate calculations for the two subgroups are simply re-aggregated based
on each group's proportional contribution to the cost. A discussion of why
these two approaches yield the same results is provided in Chapter II.2
(Section II.2.B.2.3)
Link to Chapter II.2.B.2.3
Chapter 11.8 11.8-31 Cost of Bladder Cancer
-------
Table 11.8-8. Expected Undiscounted Costs of Medical Services (in 1996$) for
Nonsurvivors of Bladder Cancer (Age of Onset = 70)
Years Post-
Diag-nosis
(n)
1C
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Medical Costs Through the 20th Year Post-diagnosis3
(undiscounted)
Medical Cost if
Survive Through the
nth Year
26,485
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
13,277
Medical Cost if Die
in the nth Year
34,806
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
36,558
Expected Total (Undiscounted) Cost Through the 20th Year Post-
Diagnosis:
Total Cost Based
on Weighted
Average13
30,699
7,543
6,936
6,352
5,793
5,264
4,750
4,261
3,797
3,359
2,955
2,564
2,200
1,862
1,551
1,274
1,016
784
577
390
93,927
a. Costs are based on data reported in Table II. 8-4, adapted from Baker et al., 1989.
Probabilities of survival and mortality, taken from Table II. 8-6, are conditional on dying of bladder
cancer within 10 years post-diagnosis.
Links to Tables 11.8-4 and 11.8-6
b. Weighted average of costs incurred by nonsurvivors who survive the year and those who die
during the year. Weighting is based on the conditional probabilities shown in Table 11.8-6.
c. Costs during the first year include "Initial" therapy ($16,527) , and pro-rated maintenance or
terminal care. See text for discussion.
II.8.B.2. Results of Medical Cost Analysis
The per-patient lifetime direct medical costs calculated for the "average"
bladder cancer patient (as shown in Table II.8-5), bladder cancer survivors
(as shown in Table II.8-7) and bladder cancer nonsurvivors (as shown in
Table II.8-8) diagnosed at age 70 are listed in Table II.8-9. Undiscounted
costs and costs discounted at three, five, and seven percent back to year
Chapter II.8
II.8-32
Cost of Bladder Cancer
-------
one (time of diagnosis) are shown.12 Discounting was carried out for 20
years following diagnosis and comprises the assumed full duration of
maintenance care for survivors.
Links to Tables II.8-5 andII.8-7
Table 11.8-9. Incremental Per-capita Medical Costs for the Average Bladder Cancer Patient,
Survivors, and Nonsurvivors (Diagnosed at Age 70) Undiscounted and Discounted at 3, 5, and
7 Percent ($1996)
Patient Group
Survivors
Nonsurvivors
Average Patient
Discount Rate
Undiscounted
$179,153
$93,927
$156,670
3
$148,149
$83,449
$131,081
5
$132,653
$77,983
$118,231
7
$120,132
$73,424
$107,811
See text for a definitions of patient groups.
The costs presented in this chapter were current in the year the chapter was written. They can be updated using
inflation factors accessible by clicking below.
Link to inflation factors
The results show substantially higher costs for survivors than nonsurvivors,
due primarily to their ongoing maintenance care. It is noted that although a
20-year maintenance period for bladder cancer survivors is assumed (with
adjustment for background mortality that reduces utilization), the actual
average period of maintenance is not known. Maintenance periods are
likely to vary considerably among individuals, depending on age, health
status, access to care, and other factors. Maintenance care costs comprise
the major cost element in determining the "average" patient costs, due to
the relatively low mortality attributable to bladder cancer. The uncertainty
surrounding the period of maintenance care for survivors could therefore
have an impact on the cost estimates for the "average" patient and for
survivors of bladder cancer. Because less than half of bladder cancer
patients diagnosed at age 70 survive beyond the first ten years post-
diagnosis (most die of causes other than bladder cancer), the impact of
maintenance care costs on the costs of the "average" 70-year-old patient is
less than it would be if there were a lower overall mortality rate (i.e., if
diagnosis occurred at younger ages).
12 As noted previously, costs will be higher if ages of diagnosis are earlier. The uncertainty
analysis in Section II.8.C contains additional cost estimates for earlier ages of diagnosis.
Chapter 11.8
11.8-33
Cost of Bladder Cancer
-------
II.8.B.3. Other Studies
Link to Chapter 1.1
A number of other studies were reviewed for this analysis. Most had
shortcomings with respect to the duration of the study (e.g., only one year
of medical cost data obtained) or the age of the data. A select group of
these are discussed below.
II.8.B.3.1 Hartunian et al.
Hartunian et al.'s (1981) method of estimating the costs of illness has been
discussed in Chapter 1.1. The authors defined expected treatment on a
yearly basis, developed annual costs of the treatment, and combined the
cost data with survival data. Using this method, they estimated the costs
of cancer at eight sites, including cancer of the urinary system, which
includes bladder, kidney, and related structures.
Hartunian et al. estimated the costs of inpatient stays using a 1976 study by
Scotto and Chazze of newly diagnosed cancer patients followed over a
two-year period to establish hospitalization and payment patterns. These
data were supplemented by questionnaire data. The Hartunian data are
quite old (over 20 years) and both survival and treatment methods have
changed since that time. Their estimated survival for bladder cancer was
0.15 percent, which is considerably lower than the current survival
statistics and impacts the percentage of patients who access care
components (reducing medical cost estimates through high death rates).
Medical treatment has also changed considerably since their data were
obtained. In addition, they also limit their analysis to two years post-
diagnosis, which is not sufficient due to both the prolonged period of
detailed follow up care that is required (note the 20 year post-diagnosis
time frame for increased mortality) and the frequency of relapse.
II.8.B.3.2 Riley et al.
A study of Medicare payments from diagnosis to death in elderly cancer
patients was carried out by Riley et al. (1995). The cost estimates are
based on Medicare payments only, which do not include: most nursing
home care, home health care, pharmaceuticals unless supplied for
inpatients, out-of-pocket expenses, deductibles, charges in excess of
Medicare paid by other sources (e.g., coinsurance), and other related
medical services not covered by Medicare.
Medicare patients younger than 65 were not included and the average age
at diagnosis of the bladder cancer cohort was 75.2 years, in contrast with
the 70 year national average. Riley et al. note that patients diagnosed at
younger ages have higher costs. In addition, those diagnosed at earlier
stages have a better prognosis but may have higher medical costs (due to
longer continuing care).
Chapter 11.8 11.8-34 Cost of Bladder Cancer
-------
Medical costs are reported for all patients who were diagnosed with
bladder cancer, regardless of other diseases or their ultimate causes of
death. Due to the links among bladder cancer, smoking, and numerous
other diseases, this method is especially problematic because costs
associated with other illnesses may be commingled with the bladder cancer
costs.
The background cost per year for medical services was estimated by Riley
et al. to be $2,250 ($3,154 in 1996 dollars), based on the experience of all
people over the age of 65 who received Medicare-compensated care. The
study excluded those costs that occur during the last year of a person's life.
Consequently, the estimated background value may underestimate
background costs, especially as age and associated mortality risks increase
over the age of 65.
Riley et al. estimated that the total average Medicare payment from
diagnosis to death for persons diagnosed with bladder cancer was $57,629
in 1990 dollars, adjusted to $80,796 in 1996 dollars (CPI 1990:1996 =
1.402). This estimate is considerably lower than those obtained from
Baker et al. The difference is most likely due to the exclusion of many
costs not covered by Medicare, and the various other factors described
above that tend to reduce the cost estimate. Due to these limitations, the
Riley et al. study is not recommended for a benefits evaluation.
11.8.C. Uncertainties and Limitations
As noted periodically in the above discussion, there is uncertainty
surrounding various aspects of the analysis. Information concerning some
inputs to the analysis was limited. Although a complete uncertainty
analysis is beyond the scope of this work, the significant sources of
uncertainty are discussed. Limitations of the scope of the analysis are also
discussed.
II.8.C.1. Uncertainties Surrounding Key Inputs to the Analysis
11.8. C.1.1. Analysis of Medical Costs
The cost estimates based on Baker et al. (1989, 1991) have a number of
limitations, many of them noted by Baker et al. (1991), Mor et al. (1990),
and Mor (1993). Most of these limitations arise from the use of CMHSF.
Medicare data have five limitations that decrease its value for calculating
the average lifetime direct medical costs of treating bladder cancer. First,
Medicare covers medical services only for most persons age 65 and over,
disabled persons entitled to Social Security cash benefits for at least 24
months, and most persons with end-stage renal disease. All patients not
covered by Medicare are excluded from the database, including all non-
Chapter 11.8 11.8-35 Cost of Bladder Cancer
-------
disabled women under 65, and women over 65 using private health
insurance (Baker et al., 1991).
Given that diagnosis of bladder cancer occurs in almost half of all patients
before age 65, the CMHSF excludes a significant number of younger
patients. According to Mor et al., treatment for younger women for other
cancers tends to be more intensive (and therefore more costly per unit
time) than treatment for older women, though older women tend to have
longer hospital stays. This is likely to be the case for bladder cancer as
well, and may be supported by the improved cancer-specific five-year
survival observed among younger versus older patients. Because these
differences counteract each other, the omission of younger patients from
the Baker et al. analysis is not expected to affect the results substantially.
In addition, the majority of senior citizens are enrolled in Medicare (Ibid);
differences in medical costs incurred by senior citizens not using Medicare
should have little effect on overall cost estimates.13
Medicare also does not cover self-administered drugs, intermediate nursing
care, long-term nursing care, and some expensive new treatments (such as
bone marrow transplants). For some patients, these costs may represent
significant percentages of total treatment costs. Most direct medical costs,
however, appear to be covered by the CMHSF database and are included in
Baker et al.'s analysis. In addition, Baker et al. made adjustments for some
cost elements not covered by Medicare (see Section II.8.B).
Another drawback is that Baker et al. were not able to identify bladder
cancer patients in CMHSF whose diagnosis and first course of therapy did
not involve hospitalization. In an analysis of Rhode Island non-bladder
cancer patients covered by Medicare, Mor et al. determined that a small
percentage of patients were initially diagnosed without hospitalization, and
had substantially lower initial and subsequent treatment costs (Mor et al.,
1990). This omission likely causes average treatment costs to be
overestimated, though by relatively little.
A fourth drawback is that Baker et al. (1989) provide no information
concerning the duration of the maintenance period for bladder cancer. The
analysis in this chapter considers a 20-year period. If the average duration
of maintenance care among patients of bladder cancer is shorter (longer)
than 20 years, the estimates of the costs incurred would be biased upward
(downward). This function is true for survivors as well as nonsurvivors of
bladder cancer.
13 This figure represents those enrolled in Medicare Part A; 95 percent of those enrolled in
Medicare Part A choose also to enroll in Medicare Part B.
Chapter 11.8 11.8-36 Cost of Bladder Cancer
-------
A fifth drawback is that the data used by Baker are from the period 1974 to
1981, leading to uncertainty regarding changes in treatment methods and
costs.
Finally, the reliability of the data contained in the database used by Baker et
al. varies. An independent analysis of CMHSF performed in 1977 by the
Institute of Medicine of the National Academy of Sciences found that the
frequency of discrepancies in principal diagnoses varied among diseases
(Baker et al., 1991). It is unclear whether the presence of misnamed
diagnoses contained in CMHSF potentially increases or decreases the
resultant cost estimates.
Overall, despite the limitations described above, Baker's analysis of the
CMHSF data represents the most nationally-representative, per-patient
lifetime estimate of the direct medical costs of treating bladder cancer to
date. Their cost estimates are based on sound criteria. Some of the data
limitations underestimate costs, and others overestimate costs; the sum of
the data limitations therefore decreases the magnitude of error. More of the
uncertainties in their analysis appear to underestimate costs, however; the
net result is a likely underestimation of actual direct medical costs.
Although there are some uncertainties associated with the estimation of the
survival and mortality probabilities used in the calculation of expected
medical costs (discussed below), these uncertainties are likely to be
relatively small. As noted in the text, NCI RSRs used to estimate survival
and mortality for this analysis are based on the survival experience of a
large group of bladder cancer patients considered in relation to the survival
experience of the general population.
An additional limitation of this analysis is that medical costs incurred as a
result of bladder cancer, but not considered by Baker et al., may arise as a
result of treatment for bladder cancer. Secondary cancers and other
adverse health effects may occur due to radiation, chemotherapy treatment,
and other therapies. These effects may occur substantially after bladder
cancer treatment has been completed and can incur added medical costs not
considered in this chapter. Many concurrent and related costs are included,
however, due to the methods used by Baker et al. The authors considered
all costs due to any medical services, minus the background costs of all
individuals who did not have cancer.
As with all chronic diseases, it is difficult to estimate the period following
diagnosis and initial treatment, during which additional medical monitoring
and follow-up care take place. Data have not yet been located regarding
the average duration of maintenance care for bladder cancer. For purposes
of this analysis, 20 years of follow-up care was assumed to be reasonable,
due to the severity of the disease and the consequences of bladder surgery.
Chapter 11.8 11.8-37 Cost of Bladder Cancer
-------
As noted above, there is a prolonged period during which mortality
continues to increase, indicating that morbidity must be monitored for
many years post-diagnosis. The estimated time required for maintenance
care may be revised in the future if data are located.
There is additional uncertainty associated with the age of onset of the
disease and the resulting medical costs. If bladder cancer is induced at an
earlier than average age (70 years), as may be the case with arsenic-induced
bladder cancer if it follows the skin cancer pattern, the direct medical costs
will be greater, particularly the maintenance costs. Younger patients will
survive to require medical treatment and follow-up for a longer period (on
average) than those who are diagnosed at age 70, because they have fewer
competing causes of death. In addition, bladder cancer frequently recurs
and patients require careful monitoring; some require treatment for
recurrent cancer. The latter is more likely with young patients who may
survive for many decades after initial diagnosis.
.8.C.2. Scope of the Analysis
Link to Chapter 1.1
The analysis in this chapter was confined to direct medical costs by the
patient. As noted in Chapter I.I, willingness-to-pay has many other cost
elements.
The analysis does not include time lost by the patient or their family and
friends who provide care. Also omitted from cost of illness estimates are
pain and suffering on the part of the patient or their family and friends,
changes in job status among previously employed patients, training for new
job skills due to physical limitations, or medical costs incurred after the 20-
year maintenance period. These cost elements may comprise a substantial
portion of the total cost of bladder cancer.
Chapter 11.8 11.8-38 Cost of Bladder Cancer
-------
CHAPTER 111.1. INTRODUCTION TO THE COSTS OF DEVELOPMENTAL
ILLNESSES AND DISABILITIES
Clicking on the sections below will take you to the relevant text.
III.l.A Overview
III. I.A.I Description
III.l.A.2 Overall Costs
III. 1 .B. Causes of Developmental Effects
III. 1 .B. 1 Somatic Cell Damage
III. 1 .B .2 Heritable Cell Damage
III.IB.3 Developmental Toxicity Studies
III. 1 .B .4 Genotoxicity
III. 1 .B. 5 Categories of Causes of Developmental Damage
III. l.C. Valuation Issues
III. I.C.I Short- Versus Long-term Effects
III.l.C.2 Occurrence of Illnesses
III.l.C.3 Multiple Effects
III. l.C.4 Prognosis
III. 1 .D. Environmental Agents Associated with Developmental Toxicity
III. 1 .D. 1 Developmental Toxicity Agents
III. 1. D. 2 Genotoxi c Agents
Chapter III.1 III.1-1 Adverse Developmental Effects
-------
CHAPTER 111.1. INTRODUCTION TO THE COSTS OF DEVELOPMENTAL
ILLNESSES AND DISABILITIES
1.1.A Overview
Link to Chapter 1.1.
This section of the handbook focuses on developmental illnesses and
disabilities that may be associated with exposure to environmental agents.
Its chapters (III.2 through III.9) provide data on the direct medical costs of
individual effects or groups of similar types of developmental effects. As in
previous chapters, information is not included on all elements of
willingness to pay (WTP) to avoid the illness. A summary of the cost
elements that comprise WTP is provided in Chapter 1.1. Due to the
availability of data on some additional cost elements in the source used to
obtain direct medical costs, data are provided on the additional cost
elements at the end of some chapters in this section.. Some of elements
for which cost estimates are not available, such as the WTP to avoid pain
and suffering in children, are likely to be very substantial in the case of
adverse developmental effects.
This chapter contains a description of some aspects of the environmental
causes of developmental toxicity, types of effects, and general issues
related to economic valuation of developmental effects, as well as a list of
agents associated with developmental toxicity and genotoxicity.
I.1.A.1 Description
Developmental effects cover a wide spectrum of short- and long-term
illnesses, disabilities, and conditions. Pre- and postnatal exposure of
children carries risks that differ in many respects from those of adults due
to the unique physiological characteristics of developing individuals, their
food consumption patterns, other exposure characteristics, and other
factors (NAS, 1994).1 For example, children (especially infants) have a
much higher food intake than adults in relation to their body weight. They
may have much greater contact with soil, and their skin absorption differs
from adults.
Developmental illnesses and disabilities may be referred to in some EPA
sources as "adverse developmental effects," and are defined in this
handbook as any adverse health effect resulting from exposure of the
mother or father pre-conception, of the mother during pregnancy, or of a
1 These differences are described in a recent NAS report, developed by pediatricians,
epidemiologists, toxicologists, and environmental scientists.
Chapter III.1 III.1-2 Adverse Developmental Effects
-------
child to toxic substances. Adverse developmental health effects range from
physical deformities such as cleft palate, cleft lip, and shortened limbs, to
cognitive impairments such as mental retardation, hyperactivity, and
delayed cognitive development. Effects also include reduced birth weight,
growth retardation, adverse hormonal changes, and abnormalities in
physiology (e.g., liver and kidney function) and effects resulting from
genetic abnormalities (e.g., Down syndrome).2
There are hundreds of developmental effects listed under the general
category of congenital anomalies in the International Classification of
Diseases (ICD-9-CM).3 ICD-listed effects include structural abnormalities
that result from errors in embryogenesis or the fetal period (Bennett and
Plum, 1996) and are usually identifiable at, or shortly after, birth.
Synonyms include congenital malformations, birth defects, or structural
anomalies (Bennett and Plum, 1996). Other developmental effects include
non-structural abnormalities such as physiological disturbances. Still other
effects may occur during childhood; for example, lead-induced brain
damage can occur up through late adolescence (discussed in Chapter III.9).
Those developmental effects chosen for inclusion in this handbook include
illnesses and disabilities that could reasonably be associated with
environmental exposures to chemicals and that had sufficient medical cost
information to be of use in economic valuations.
Major congenital anomalies, which comprise the majority of developmental
effects discussed in this handbook, are identified in two to three percent of
all newborn infants, independent of ethnic group or country of origin.
Many anomalies are not detected at birth. During the first year of
observing the child, as developmental milestones are anticipated, the rate
of diagnosis doubles. By school age (five years) approximately five to
seven percent of children have been diagnosed with a major congenital
anomaly or learning disability (Bennett and Plum, 1996). (Learning
disabilities are not included in this edition of the handbook but may be
included in the future.)
.1.A.2 Overall Costs
The medical and related costs of developmental effects are substantial.
Low birth weight (LEW), which is a relatively common occurrence in the
2 Many of these are referred to as "birth defects" in the popular press, especially physical effects.
They fall into the broad categories of structural abnormalities, functional deficiencies, and growth
alterations (Kimmel, 1993).
3 Adverse developmental effects, which include all of the types of effects discussed above, are
referred to as "developmental effects" subsequently in this chapter, for brevity. The term implies adverse
effects, rather than simply any effect that deviates from the norm.
Chapter III.1 III.1-3 Adverse Developmental Effects
-------
United States, has been estimated to cost approximately 35 percent ($4
billion) of the $11 billion spent on health care for infants. Medical care
resulting from LEW may cost in excess of one million dollars for a single
child (Lewitt et al., 1995). Congenital anomalies, which are a subset of all
developmental effects, were estimated to cost $6.3 billion in 1980 in the
United States and represented 1.4 percent of the total cost of illness (Rice
et al., 1985). This estimate included direct medical costs and lost
productivity, but not special educational costs and developmental services.
A similar type of evaluation in 1991 found the national cost of a single
effect, cerebral palsy, to be $1.2 billion. This estimate did not include long-
term institutional care, which is often very costly (National Foundation for
Brain Research, 1992).
Over one half of children evaluated in subspecialty medical clinics or
admitted to hospitals in North America are treated due to disorders arising
from congenital anomalies. Two thirds of deaths in pediatric hospitals are
also due to these anomalies. In addition, the rate of congenital anomalies
in early miscarriages is approximately ten times higher than that observed in
infants (Bennett and Plum, 1996). Serious developmental effects are likely
to be the cause of pregnancy loss in many cases.
.1.B Causes of Developmental Effects
The biology of developmental toxicity proceeds by two primary routes:
damage to heritable cell lines and damage to somatic cell lines. Heritable
cells carry the genetic materials from generation to generation (egg and
sperm) and damage typically involves alterations in chromosomes.
Genotoxic chemicals cause mutations or death of heritable cell lines and the
mutations (genetic changes) may be perpetuated through generations.4
When cell death occurs, there is no perpetuation of cell lines, whether
mutated or not. Somatic cells (non-reproductive cells) may also be
damaged by genotoxins or other types of toxins. Somatic cell damage or
death affects those cells directly exposed or proceeding from the exposed
cells in the same individual (damage is not conveyed from one generation
to the next). This section is provided to explain the rationale for including
heritable and non-heritable birth defects in the COI handbook. It also
provides information on the biological mechanisms behind some birth
defects.
4 Genotoxicity deals with the effects of chemicals on DNA and on mechanisms of inheritance in
cells and organisms, focusing on the process of mutagenesis (Hoffmann, 1991).
Chapter III.1 III.1-4 Adverse Developmental Effects
-------
.1.B.1 Somatic Cell Damage
Chemicals may cause cellular toxicity, including toxicity to genetic
material. Somatic cell damage affects the exposed individual primarily by
altering cell structure or function. Exposure to chemicals that are
genotoxic or toxic to the cells in other ways during the prenatal period can
cause damage to somatic cell lines during development by altering or killing
cells. Exposure thus results in damage to cells, or in the absence of cells
that would arise from the damaged cell. Although redundancy occurs in
many developing systems and damage does not always lead to birth
defects, somatic cell damage from cellular toxicity may be responsible for
many of the birth defects discussed in this section. For example,
genotoxicity, and other types of cellular toxicity that prevent cell
replication, limit the production in the embryo of neurons in the central
nervous system (CNS). They also limit the development of limb buds that
eventually become limbs and digits, resulting in shortened or absent limbs
(discussed in Chapter III.4). Cleft lip or palate (discussed in Chapter III.3),
some heart anomalies (discussed in Chapter III.5), and spina bifida
(discussed in Chapter III.6) may also result from failure of early cells to
form and migrate properly. (There may also be a heritable component to
some occurrences of these effects that affects cell replication and
movement, as discussed in Section III.1.B.2 below.)
Changes in somatic cells may also affect the offspring of an exposed
woman when the cellular damage alters her physiology or reproductive
system in such a way that it impacts her children (e.g., via maternal
toxicity). For example, a woman whose toxic exposure caused severe
kidney damage may not have the necessary capacity in her kidneys to
adequately process the substantial increase in fluid load that accompanies
pregnancy. Renal failure and the related toxicity to the mother will have a
serious and sometimes fatal impact on the child. Generally, when past
toxicity to the mother is the cause of an adverse developmental effect, it is
not reported in the toxicological literature. This omission occurs because
the studies are looking for toxicity in the offspring that is independent of
maternal toxicity, and is appropriate from a scientific perspective to
evaluate some aspects of developmental toxicity. Information is then
lacking, however, that may be appropriate for a benefits assessment on the
simultaneous occurrences of maternal and offspring toxicity.
.1.B.2 Heritable Cell Damage
Heritable cells carry the genetic materials from generation to generation
(via egg and sperm) and damage typically involves alterations in
chromosomes in these cells. Changes in heritable cells may be perpetuated
through generations. The mutations include alterations in specific
Chapter III.1 III.1-5 Adverse Developmental Effects
-------
chromosomes, as well as in the number of chromosomes (e.g., Down
syndrome occurs when three of chromosome number 21 are present rather
than the usual two chromosomes).
The mutations may be observable in all persons with the chromosomal
alteration (dominant disorders), or may be expressed only when two
chromosomes (one from each parent) with the same alteration are present
(recessive disorders). Consequently, many mutations are not continuously
observable because they are recessive, but may appear when both parents
have the same recessive mutation (Hoffmann, 1991). Many important
areas of the chromosomes have substantial redundancy so that mutations in
small portions of the chromosome are compensated for by other portions
that carry out the same function and remain intact.
Numerous repair mechanisms operate in cells so that exposure to
genotoxins does not necessarily result in permanent damage to cells
(Hoffmann, 1991). When a chemical is shown by studies to be genotoxic,
however, it has the potential to cause chromosome damage. The specific
site of chromosome damage cannot be predicted because chemicals alter
the basic structure or mechanisms related to DNA replication.
Consequently, damage may occur at any site and the nature of any resulting
defect can vary accordingly.
In 1984, the National Academy of Sciences estimated that about 0.3 to 0.4
percent of infants have syndromes associated with chromosomal
abnormalities (NAS, 1984). More recently these have been associated
with approximately seven to ten percent of stillbirths and infant deaths
(Bennett and Plum, 1996). As the human genome is mapped, understanding
and identification of genetic defects will increase. Some relatively common
effects that are based on genetic changes include Down syndrome
(discussed in Chapter III.8), Turner's syndrome, and Klinefelter syndrome.
Some genetic diseases that are recessive (e.g., cystic fibrosis, Tay-Sachs,
and phenylketonuria) are thought to occur primarily through inheritance
rather than new mutations (Hoffmann, 1991).
.I.B.3 Developmental Toxicity Studies
Most evidence of developmental toxicity is obtained from human
epidemiological studies or animal toxicity studies. The majority of data are
obtained from animal studies due to ethical issues and difficulties in
conducting large-scale controlled human studies. As discussed above,
many developmental effects are not observed until many years after birth;
consequently, birth defects are poorly tracked in the population with only
a small percentage reported formally (i.e., on birth certificates). Other
complicating factors include large variations in exposure, lifestyle, and
genetically-determined risks of birth defects within most populations.
Chapter III.1 III.1-6 Adverse Developmental Effects
-------
Animal studies are often preferred due to the difficulty in conducting
human studies and for a variety of other reasons. These reasons include
cost, ethical considerations, and the level of control over exposure
concentrations and durations. Confounding study factors are limited in an
animal study and an evaluation of all observable structural and functional
effects is possible. There are many limitations to animal studies, including
considerable interspecies differences in neurological capabilities and
potential damage, and some physiological differences between study
mammals and humans.
Often the implications of effects observed in human and animal studies are
not clear from a health or economic perspective. For example, a recent
study by Kanitz et al. (1996) found that some water disinfection chemicals
were associated with reduced infant head size and body length in some
communities (characteristics easily extracted from birth certificates).
These features usually suggest retarded development. Although giving
clear reason for concern, the immediate and long-term health effects of
these measures aren't clear. Such external measures don't provide specific
information on the more important internal effects on the brain or other
organ systems or on future development. Consequently, the data cannot be
easily used to establish the benefits of avoidance, even though most people
would agree that a considerable benefit is to be gained from avoiding the
observed effects. Some human studies (e.g., of mercury) provide more
straightforward indications of developmental toxicity.
Animal studies of developmental toxicity have been carried out for many
environmental agents, and are required by the federal government for some
groups of chemicals (e.g., as part of the registration process for pesticides
and some pharmaceuticals). Although effects seen in humans may be
different than those seen in animal studies, it is assumed that developmental
effects in animals indicate the potential for developmental toxicity in
humans (EPA, 1991). Agents associated with developmental toxicity are
listed in Section III. 1 .D below.
.1.B.4 Genotoxicity
Genotoxicity, as discussed above, may result in mutations or death of
somatic or heritable cell lines. Ethical considerations preclude the testing
of toxic agents on humans, but data on genotoxicity exist from studies of
occupationally-exposed workers. Incidental exposure to some chemicals
present in the workplace has been shown to cause chromosomal damage in
some workers. For example, exposure to vinyl chloride, styrene, benzene
and ethylene oxide have been positively associated with chromosomal
aberrations (Hoffmann, 1991). A high incidence of spontaneous abortions
was observed in nurses exposed to genotoxic cancer therapy drugs.
Even when these studies have been carried out, it is very difficult to
Chapter III.1 III.1-7 Adverse Developmental Effects
-------
establish unequivocally whether a chemical causes genotoxicity in humans
using epidemiological studies, due to a number of factors, including:
• the inability to fully characterize all exposures that may contribute
to genotoxicity;
the likelihood that severe damage will result in early miscarriage
and not be observable ;
• the largely random occurrence of many birth defects (e.g., Down
syndrome); and
the separation in time between germ-cell mutations and effects in
subsequent generations (Hoffmann, 1991).
In addition, damage to cells, whether heritable or somatic, may be followed
by cell repair or death, which may not result in an adverse outcome.
Multiple factors determine whether an adverse effect at the cellular level
will result in damage to a child.
Some studies in animals have been performed. The availability, however,
of relatively quick and inexpensive in vitro assays on human or other
mammalian cell cultures using somatic and reproductive cells, or on small
animals or microorganisms, make these the methods of choice for most
genotoxicity tests. In addition, researchers are beginning to study the
mutagenicity of environmental mixtures that occur in their natural setting.
Hannigan et al. (1996) recently evaluated the mutagenicity of urban
particulate air pollution samples. This type of research has the potential to
provide very useful information regarding the mutagenic potential of
environmental pollutants.
Thousands of chemicals have been evaluated using various in vitro tests of
genotoxicity to determine the potential for genotoxic effects.5 This
information has been commonly used as an indicator of a chemical's
carcinogenic potential, because there is a link between genotoxicity and
cancer (Hoffmann, 1991). This information also has implications for a
chemical's potential to cause developmental toxicity. Genotoxicity is not,
however, used at this time to establish a causal link between developmental
effects and exposure. Even when human data are available indicating that
an agent causes mutations, it is not possible to predict the type of
developmental effect that would arise from the mutation because damage
can occur at thousands of sites on the chromosomes.
Agents associated with genotoxicity are listed in Section III.l.D. below.
5 The Environmental Mutagen Information Center in Oak Ridge, Tennessee, had data on 21,000
chemicals in their database in 1990 (Hoffmann, 1991).
Chapter III.1 III.1-8 Adverse Developmental Effects
-------
.1.B.5 Categories of Causes of Developmental Damage
The contributions to congenital anomalies by various factors have been
estimated:
congenital infection 2-3%
maternal diabetes 1.5%
other maternal illness < 1.5%
chromosomal 6%
monogenic 7.5%
multifactorial 20%
maternal medication 1 - 2%
unknown >50%
(Bennett and Plum, 1996)
As this list indicates, most congenital anomalies have no clear origin (more
than 50 percent). They may arise from a single factor or a specific
combination of factors. Interference with the normal progress of
development during a pregnancy may result from genotoxic effects, cellular
toxicity, or other factors. Adding to the complexity of evaluating
developmental toxicity is the fact that the nature of the developmental
effects often depend on the period of development during which exposure
occurred.
III.1.C Valuation Issues
III.1.C.1 Short-Versus Long-term Effects
Developmental effects can be organized in many different ways, and are
typically approached by the medical community on an organ system basis
(e.g., neural damage, skeletal abnormalities) or by syndrome (e.g., Down).
Approaching these effects from an economic perspective may require a
different approach that takes into account issues of cost, such as long-
versus short-term care. Economic evaluations differ substantially
depending on whether effects are time-limited or not. For example,
treatment and monitoring for elevated blood-lead levels is usually time-
limited, while medical treatment and services for cerebral palsy require
long-term medical and other interventions. Most developmental effects
required long-term follow-up and care, often including a variety of medical
specialities, leading to relatively high lifetime care costs.
Federal and state agencies tend to group many of the effects that require
long-term care under the heading of "disabilities." This term includes
physical and mental disabilities, which share the common requirement that
long-term care is required. Many of the disabilities are classified by the
Chapter III.1 III.1-9 Adverse Developmental Effects
-------
Department of Health and Human Services and enable families to receive
special funding for medical care, maintenance, and other costs incurred by
the disabled child and adult. Funding is carried out through programs such
as Supplemental Security Income (SSI). The availability of federal
databases that track these types of disabilities and costs improves our
ability to estimate the treatment patterns and costs.
Fewer data are available for disabilities not covered under federal
programs. Care may be received from a variety of specialists and is often
difficult to estimate, due to variations in the types of services that are
required. There is considerable variation in services and costs among
children with the same disabilities, due to differences in the severity of their
disabilities. An example of this can be observed among children with
Down syndrome, who may range from mildly affected by the chromosomal
abnormality to severely incapacitated. With the advent of managed care,
there has been an effort to evaluate the overall costs of caring for children
with disabilities that do not qualify for government supplements, so that
capitation costs can be estimated. This process is still in its infancy,
although some data are available that have been developed for this purpose.
.1.C.2 Occurrence of Illnesses
Tracking of occurrence is poor because only a small percentage of
developmental effects is reported on birth certificates. Many effects are not
noted at birth and may be observable only after a child begins to develop
and is observed to miss critical developmental milestones. Severe birth
defects are more often reported because they are observed at birth and are
typically reported; however, many of these are very rare, (e.g., spinabifida,
Down Syndrome) (McManus et al., 1996 in Altman and Reinhardt, Eds.,
1996). Their rarity occurs in part because severe impairments often result
in miscarriage rather than a live birth. When the occurrence of a disease is
infrequent, it is particularly important to access data on large population
groups to obtain reasonable estimates of average costs.
Recent work using a large set of California databases has been carried out
by Waitzman et al. (1996), used in Chapters III.3 through III.8 in this
section and described in Chapter III.3. In addition, the Agency for Toxic
Substances and Disease Reduction is funding studies in numerous states
that provide training for medical staff and follow-up evaluations for
children to better ascertain the actual rates of developmental effects. These
efforts are linked to studies of environmental pollutants and developmental
effects. A better understanding of the occurrence and causation of
developmental effects is expected to result from these studies.
Chapter III.1 III.1-10 Adverse Developmental Effects
-------
.1.C.3 Multiple Effects
A single chemical often induces an array of effects on different organ
systems. The occurrence of multiple effects has been reported in both the
human epidemiological and animal toxicological literature. Most animal
studies report multiple effects when developmental effects are observed.
This observation is supported by the public health literature on children
with disabilities, which indicates that disabled children often have multiple
disabilities. For example, low birth weight is associated with physical and
cognitive disabilities, and with liver, kidney and brain damage (Hacket et
al., 1994; see also Chapter III.2). The multiple simultaneous occurrences
of effects are reported in the chapters within this section that discuss
specific developmental effects. When data were available, the chapters
provide information on the likelihood of co-occurrence of effects. The
information provided on the costs of developmental effects reflects these
complexities. The clustering of effects differs considerably among
individuals, however, and introduces uncertainty into the analysis.
The developmental effects of chemicals differ, depending on when
exposure occurs, because different components of organs are developing at
different stages during a pregnancy and postnatally. For example,
development within the CNS occurs over many years, with especially rapid
development during early life. Some milestones in the development of the
human nervous system are listed below:
Prenatal:
neural tube closes 22-26 days
first neurons born 22-26 days
cortical neurons migrate 6 weeks
mesencephelon expands considerably 9 weeks
cerebellum is visible 12 weeks
reflex actions are observable 3 months
most major nerve tracks have been formed 6-9 months
Postnatal:
cortical migration complete 5 months
neuron proliferation complete 12 months
myelin 50 percent complete 18 months
visual system connections complete 3-4 years
brain is mature in form 20 years
Cell proliferation and neural connections occur continuously during early
life, and interference in this process may lead to impaired or altered neural
functioning. The CNS is particularly susceptible because it produces many
cell types over an extended period and thus is subject to injury at more
stages than other organs. It also does not have the ability to replace
missing neurons when the developmental period for that neuron is past
Chapter III.1 III.1-11 Adverse Developmental Effects
-------
(Rodier, 1994). Although the CNS is particularly vulnerable, all organ
systems go through many stages during the prenatal period and have
varying susceptibilities and manifestation of toxicity during early life.
The effects discussed in this section often occur in multiples, and the costs
of concurrent effects are incorporated into the costs for some effects
(those discussed in Chapters III.2 through III.8). Generally, however,
medical costs are presented for each illness or disability separately (this is
discussed in more detail in Chapters III.2 and III.3). Based on the nature
of the economic analysis and the supporting scientific data for the
chemicals of concern, it may be appropriate to include an illness with its
related effects. Inclusion is a matter of judgement. For example, a large
percentage of children with Down syndrome have hearing loss and serious
visual problems. Inclusion of treatments would be logical, since these
conditions are so strongly associated with the syndrome. Alternatively,
children with many of the cardiac defects presented in Chapter III. 5 often
have related cardiovascular problems. These problems are often treated
with the predominant effect presented in the chapters, and are part of the
costs of both immediate surgical intervention and follow-up care.
Consequently, separate calculations of the benefits of avoiding these related
effects would not be necessary.
.1.C.4 Prognosis
Developmental diseases and disabilities vary widely in their progression.
Some types of anomalies have a relatively good prognosis (e.g., correction
of cleft lip and palate), with most patients achieving a normal quality of life.
Others, such as Down syndrome patients, have lifelong costs associated
with their multiple symptoms and have a shortened lifespan. Some effects,
such as very low birth weight are usually fatal. Although generalizations
can be made regarding the "average" prognosis for specific effects, the
prognosis for survival and the length of time over which treatment is
required varies widely among individuals. Consequently, the cost estimates
presented in the chapters in this section use estimates of the average
treatment and survival rates to obtain representative estimates of the
medical costs.
.1.D Environmental Agents Associated with Developmental Toxicity
This section contains lists of agents associated with developmental toxicity
and genotoxicity. As previous sections discussed, these data have many
limitations. Additional information on each chemical may be obtained from
the listed source and from a Medline or Toxnet search.
Chapter III.1 III.1-12 Adverse Developmental Effects
-------
.1.D.1 Developmental Toxicity Agents
Table III. 1-2 lists many of the chemicals that have evidence of
developmental toxicity in animals or humans. Due to the orientation of this
handbook toward environmentally-induced illnesses, the discussion focuses
on induction by environmental agents. The information in this table was
obtained from journal articles, toxicology texts, and study reports
submitted to the government by pesticide registrants and other chemical
industry sources. The table does NOT provide an exhaustive list of
chemicals that cause developmental toxicity, but instead gives an indication
of the diversity of chemicals that have been associated with these effects.
Information is provided below on genotoxins, which are potential causative
agents with a less established but no less important link to adverse
developmental effects. (Many chemicals have evidence of both
developmental toxicity and genotoxicity.)
TABLE 111.1-1.
CHEMICALS ASSOCIATED WITH DEVELOPMENTAL EFFECTS1
DATA FROM HUMAN AND ANIMAL STUDIES ARE INCLUDED.
CHEMICAL
•ACETONE
*ACROLEIN
•ACRYLIC ACID
*ACRYLONITRILE
*ALACHLOR
*ALDICARB
AMINOPTERIN
*AMITRAZ
*AMITROLE
ANTU
*AROCLOR1016(APCB)
•ARSENIC COMPOUNDS
•ARSENIC
ASULAM
*ATRAZINE
AVERMECTIN B1
*BENOMYL
•BENZENE
*BENZO(A)PYRENE
BIORESMETHRIN
BISULFAN
BORIC ACID
BRADIFACOUM
BUSULFAN
BUTACHLOR
•CADMIUM
CAPROLACTAM
CAPTAFAL
"CAPTAN
"CARBARYL
"CARBOFURAN
•CARBON TETRACHLORIDE
REFERENCE
1
3
7
1
4
1
3
1
3,5
1
7
3
3
7
9
7
7,1
5
14
10
14
3
3,4
3
1
14
7
6
7,1,2
1,2,5
1
5
Chapter 111.1
1.1-13
Adverse Developmental Effects
-------
TABLE 111.1-1.
CHEMICALS ASSOCIATED WITH DEVELOPMENTAL EFFECTS1
DATA FROM HUMAN AND ANIMAL STUDIES ARE INCLUDED.
CHEMICAL
•CARBON DISULFIDE
CARBOPHENOTHION
"CHLORDANE
CHLORDECONE
CHLORDIMEFORM
CHLORFENVINPHOS
CHLORMEQUAT
"CHLOROBENZILATE
"CHLOROBIPHENYLS (INCLUDES PCB'S)
•CHLOROFORM
CHLOROPHACINONE
CHLOROPHENOXY HERBICIDES
CHLOROPROPHAM
"CHLOROTHALONIL
CHLORPROPHAM
•CHROMIUM
•COPPER SULFATE
COUMACHLOR
COUMAFURYL
COUMATETRALYL
•CYANIDES
"CYCLOHEXANE
CYCLOHEXANONE
CYCLOHEXIMIDE
CYCLOPHOSPHAMIDE
"CYHALOTHRIN
*2,4-D
DALAPON
DECAMETHRIN
DEET (DIETHYLTOLUAMIDE)
DFP
DI(2-ETHYL HEXYL) ADIPATE
*DIBROMOCHLOROPROPANE
*DICAMBA
DICHLOBENIL
*0-DICHLOROBENZENE
*P-DICHLOROBENZENE
DICHLOROETHYL ETHER
1 ,3-DICHLOROROPENE (2,3 ON TRI)
*DICHLORVOS
DIETHYLSTILBESTROL (DES)
DIFENACOUM
*DIMETHOATE
DIMETHYL SULFOXIDE
DINOSEB
*DIOXANE
DIOXIN
DIPHACINONE
DIPHENYLHYDANTOIN
DIQUAT
DISULFOTON
*DIURON
ENDRIN
REFERENCE
5,7
6
12
11
6
6
5
7
3
5
1
15
7
4
1
17
5
1
3,4
3,4
3
1
8
5
14,3
7
4,6
1
9
9
2
7
5
7
4
1
1
1
1
1
3,14
3,4
5,6
3,5
14,7
1
2
3,4
3
4
6
5
6
Chapter 111.1
1.1-14
Adverse Developmental Effects
-------
TABLE 111.1-1.
CHEMICALS ASSOCIATED WITH DEVELOPMENTAL EFFECTS1
DATA FROM HUMAN AND ANIMAL STUDIES ARE INCLUDED.
CHEMICAL
*EPICHLOROHYDRIN
EPN
*EPTC
ETHANOL
ETHYL CHLORIDE
*ETHYL BENZENE
ETHYLENE DIBROMIDE
•ETHYLENE DICHLORIDE
*ETHYLENE THIOUREA
•ETHYLENE OXIDE
ETHYLNITROSUREA
EUGENAL
•FENBUTATIN OXIDE
•FERBAM
•FLUOMETURON
FLURPRIMIDOL
FLUTOLANIL
•FOLPET
•FORMALDEHYDE
GLYCEROL FORMAL
GLYPHOSATE
HALOXYFOP METHYL
•HEXACHLOROBENZENE
•HEXACHLOROPHENE
•LEAD
•LINDANE
•LINURON
•LITHIUM (TRI LISTED AS LITHIUM CARBONATE)
•MALATHION
•MALEIC HYDRAZIDE
•MANEB
•MANGANESE
*MCPA (METHOXONE)
•MERCURY
METALDEHYDE
METHIDATHION
METHIMAZOLE
METHOMYL
•METHOXYCHLOR
•METHYL ETHYL KETONE (MEK)
METHYL BROMIDE
•METHYL METHACRYLATE
METHYLCHOLANTHRENE
METHYLENE CHLORIDE
METOLACHLOR
MEXACARBATE
MIREX
MNNG
•MOLINATE
•NABAM
•NAPHTHALENES
NAPROPAMIDE
•NICOTINES
REFERENCE
5
6
7
14
7
7
5,6
1
14
5
3
1
4,6
5
1
7
7
9,3,6
5
5
7
7
5
5
14
6
4,6
3
5
10
1
16
4,6
7
5
5
3
1
5,7
7
6
3,5
14
5
4,7
1
14
3
9
5
9
7
8
Chapter 111.1
1.1-15
Adverse Developmental Effects
-------
TABLE 111.1-1.
CHEMICALS ASSOCIATED WITH DEVELOPMENTAL EFFECTS1
DATA FROM HUMAN AND ANIMAL STUDIES ARE INCLUDED.
CHEMICAL
•NITRATE
NITRITE
*NITROFEN
NITROGUANIDINE
"OXYFLUORFEN
•PARAQUAT
*PARATHION
*PCBS
PENTACHLORONITROBENZENE
*PENTACHLOROPHENOL
*PERCHLOROETHYLENE
*PERMETHRIN
PHENMEDIPHAM
•PHENOL
"O-PHENYLPHENOL
PHOSMET
*PICLORAM
PIDRIN
PINDONE
*PIPERONYL BUTOXIDE
PIRIMICARB
PIRIMIPHOS-ETHYL
*PIRIMIPHOS-METHYL
2-PIVALYL-1 ,3 INDANDIONE
*PROPACHLOR
*PROPARGITE
PROPHAM
*PROPOXUR
*PROPYLENE OXIDE
PROPYLENE DICHLORIDE
PYRAZON
*PYRIDINE
RADIONUCLIDES (ALPHA, BETA, & GAMMA EMITTERS)
*RESMETHRIN
RONNEL
ROTENONE
SODIUM CHLORATE
•STRYCHNINE
SULFUR DIOXIDE
2,4,5-T
TCDD
2,4,5-TP
*TETRACHLORVINPHOS
TETRACYCLINES
THERAM
THIABENDAZOLE
*THIOPHANATE-METHYL 6
THIRAM
•TOLUENE
TOXAPHENE
*TRICHLORFON
*1 ,2,4-TRICHLOROBENZENE
*1 ,1 ,1 -TRICHLOROETHANE
REFERENCE
7
7
4
7
1
8
5,2
2
10,9,6,5
5
5
6
1
7,1
5
6
5
7
3,4
5
4,6
6
1
3,4
1
7,6
1
1,9
1
1
5
8
3
7
5
9,7
1
1
5
4
14
6
6
3
2
5
1
5
5
6
6
7
5
Chapter 111.1
1.1-16
Adverse Developmental Effects
-------
TABLE 111.1-1.
CHEMICALS ASSOCIATED WITH DEVELOPMENTAL EFFECTS1
DATA FROM HUMAN AND ANIMAL STUDIES ARE INCLUDED.
CHEMICAL
REFERENCE
TRICHLOROETHYLENE
TRIDIPHANE
TRIFLURALIN
4,6
TRIFORINE
TRIMETHADONE
"URETHANE
14
VALPROICACID
VERNAM
•WARFARIN
3,4
•WHITE PHOSPHORUS
*XYLENE
*ZINEB
3,6
ZIRAM
9.10
* = Listed in TRI. When "compounds" of a metal were listed on TRI, all compounds of that
metal in this table are considered to be TRI chemicals.
Footnotes:
1. This includes a wide range of effects including skeletal deformities (increased or reduced
number of digits or ribs, limb shortening or malformation, spina bifida, cleft lip or palate, etc),
neurological impairment (altered cognitive functioning, learning disorders including
retardation, palsy, inappropriate responses to stimuli, etc), and structural or functional changes
in organs (cardiomyopathy, renal agenesis, sterility, etc).
References
1. Cunningham and Hallenbeck (1985).
2. Stellman, J.M. (1983).
S.Doulletal. (1980).
4. U.S. EPA(1983a).
5. U.S. Department of Health and Human Services, NIOSH (1983a).
6. U.S. EPA (1983b).
7. IRIS, EPA online database.
8. Clayton and Clayton (1982).
9. Hayes, W.H. (1982).
10. Vettorazzi, G. (1979).
11. Council on Environmental Quality (1981).
12. International Agency for Research on Cancer (1979).
13. Chambers and Yarbrough (1982).
14. Key et al. (1977).
15. Garry et al. (1996).
16. Webster and Valois (1987).
17. ATSDRQ993).
Chapter III.1 III.1-17 Adverse Developmental Effects
-------
.1.D.2 Genotoxic Agents
Table III. 1-2 lists agents for which there is evidence of genotoxicity. The
evidence varies for each agent and is based on a variety of in vitro and/or in
vivo analyses.6
TABLE 111.1-2. CHEMICALS ASSOCIATED WITH GENOTOXIC EFFECTS
DATA FROM HUMAN, ANIMAL, AND IN VITRO STUDIES ARE INCLUDED.
CHEMICAL
•ACETONE
*ACROLEIN
•ACRYLIC ACID
*ACRYLONITRILE
*ALACHLOR
*ALDICARB
AMINOPTERIN
*AMITRAZ
*AMITROLE
ANTU
*AROCLOR1016(APCB)
•ARSENIC COMPOUNDS
•ARSENIC
ASULAM
*ATRAZINE
AVERMECTIN B1
*BENOMYL
•BENZENE
*BENZO(A)PYRENE
BIORESMETHRIN
BISULFAN
BORIC ACID
BRADIFACOUM
BUSULFAN
BUTACHLOR
•CADMIUM
CAPROLACTAM
CAPTAFAL
"CAPTAN
"CARBARYL
"CARBOFURAN
•CARBON TETRACHLORIDE
•CARBON DISULFIDE
CARBOPHENOTHION
"CHLORDANE
CHLORDECONE
CHLORDIMEFORM
CHLORFENVINPHOS
CHLORMEQUAT
"CHLOROBENZILATE
"CHLOROBIPHENYLS (INCLUDES PCBS)
REFERENCES
5
4
7
5
1
1
3
6
5
5
7
6
6
7
9
7
1,7
5
14
1
14
5
1
3
4
14
7
6
7,6
6
6
1
5
1
12
1
6
6
5
7
3
6 Results of genotoxic tests are often mixed because they evaluate different aspects of a chemical's
ability to cause genetic damage and impair cell replication.
Chapter 111.1
1.1-18
Adverse Developmental Effects
-------
TABLE 111.1-2. CHEMICALS ASSOCIATED WITH GENOTOXIC EFFECTS
DATA FROM HUMAN, ANIMAL, AND IN VITRO STUDIES ARE INCLUDED.
CHEMICAL
•CHLOROFORM
CHLOROPHACINONE
CHLOROPROPHAM
"CHLOROTHALONIL
CHLORPROPHONE
•CHROMIUM
•COPPER SULFATE
COUMACHLOR
COUMAFURYL
COUMATETRALYL
•CYANIDES
"CYCLOHEXANE
CYCLOHEXANONE
CYCLOHEXIMIDE
CYCLOPENTAPYRENE
CYCLOPHOSPHAMIDE
"CYHALOTHRIN
*2,4-D
DALAPON
DECAMETHRIN
DEET (DIETHYLTOLUAMIDE)
DI(2-ETHYL HEXYL) ADIPATE
*DIBROMOCHLOROPROPANE
*DICAMBA
DICHLOBENIL
*0-DICHLOROBENZENE
*P-DICHLOROBENZENE
DICHLOROETHYL ETHER
1 ,3-DICHLOROROPENE (2,3 ON TRI)
*DICHLORVOS
DIETHYLSTILBESTROL (DES)
DIFENACOUM
*DIMETHOATE
DIMETHYL SULFOXIDE
DINOSEB
*DIOXANE
DIPHACINONE
DIPHENYLHYDANTOIN
DIQUAT
DISULFOTON
*DIURON
ENDRIN
*EPICHLOROHYDRIN
EPN
EPTC
ETHANOL
*ETHYL BENZENE
ETHYLENE DIBROMIDE
*ETHYLENE DICHLORIDE
•ETHYLENE THIOUREA
*ETHYLENE OXIDE
ETHYLNITROSUREA
EUGENAL
*FENBUTATIN OXIDE
REFERENCES
1
1
7
1
1
16
5
1
1
1
1
5
5
5
2
14
7
6
5
1
5
7
5,6
7
1
1
5
5
5
13
3
1
6
5
14
5
1
3
5
1
5
6
5
1
7
14
8,6
5
14
5
3
5
1
Chapter 111.1
1.1-19
Adverse Developmental Effects
-------
TABLE 111.1-2. CHEMICALS ASSOCIATED WITH GENOTOXIC EFFECTS
DATA FROM HUMAN, ANIMAL, AND IN VITRO STUDIES ARE INCLUDED.
CHEMICAL
*FERBAM
•FLUOMETURON
FLURPRIMIDOL
FLUTOLANIL
*FOLPET
•FORMALDEHYDE
GLYCEROL FORMAL
GLYPHOSATE
HALOXYFOP METHYL
•HEXACHLOROBENZENE
•HEXACHLOROPHENE
*LEAD
•LINDANE
•LINURON
•LITHIUM (TRI LISTED AS LITHIUM CARBONATE)
•MALATHION
*MALEIC HYDRAZIDE
*MANEB
*MCPA
•MERCURY
•MERCURY COMPOUNDS
METALDEHYDE
METHIDATHION
METHIMAZOLE
METHOMYL
•METHOXYCHLOR
•METHYL ETHYL KETONE (MEK)
METHYL BROMIDE
•METHYL METHACRYLATE
METHYLCHOLANTHRENE
METHYLENE CHLORIDE
METOLACHLOR
MEXACARBATE
MIREX
MNNG
•MOLINATE
•NABAM
•NAPHTHALENES
NAPROPAMIDE
•NICKEL
•NICOTINES
•NITRATE
NITRITE
•NITROFEN
NITROGUANIDINE
•OXYFLUORFEN
•PARAQUAT
•PARATHION
*PCBS
PENTACHLORONITROBENZENE
•PENTACHLOROPHENOL
•PERCHLOROETHYLENE
•PERMETHRIN
PHENMEDIPHAM
REFERENCES
1
6
7
7
8
5
1
7
7
5
1
14
1
5
3
1
5
6
1
7
15
5
1
3
6
5,7
7
1
5
14
5
1,7
1
14
3
1
5
1
7
15
1
7
7
4
7
6
5
1
1
1
1
1
1
Chapter 111.1
1.1-20
Adverse Developmental Effects
-------
TABLE 111.1-2. CHEMICALS ASSOCIATED WITH GENOTOXIC EFFECTS
DATA FROM HUMAN, ANIMAL, AND IN VITRO STUDIES ARE INCLUDED.
CHEMICAL
•PHENOL
"O-PHENYLPHENOL
PHOSMET
*PICLORAM
PIDRIN
PINDONE
*PIPERONYL BUTOXIDE
PIRIMICARB
PIRIMIPHOS-ETHYL
*PIRIMIPHOS-METHYL
2-PIVALYL-1 ,3 INDANDIONE
*PROPACHLOR
*PROPARGITE
PROPHAM
*PROPOXUR
*PROPYLENE OXIDE
PROPYLENE DICHLORIDE
PYRAZON
*PYRIDINE
RADIONUCLIDES (ALPHA, BETA, & GAMMA EMITTERS)
*RESMETHRIN
RONNEL
ROTENONE
SODIUM CHLORATE
•STRYCHNINE
SULFUR DIOXIDE
TCDD
2,4,5-T
2,4,5-TP
*TETRACHLORVINPHOS
TETRACYCLINES
THIABENDAZOLE
*THIOPHANATE-METHYL 6
*THIRAM
•TOLUENE
TOXAPHENE
TRICHLORFON
*1 ,2,4-TRICHLOROBENZENE
*1,1,1-TRICHLOROETHANE
TRICHLOROETHYLENE
TRIDIPHANE
*TRIFLURALIN
TRIFORINE
TRIMETHADONE
"URETHANE
VALPROICACID
VERNAM
•WARFARIN
•WHITE PHOSPHORUS
*XYLENE
*ZINEB
ZIRAM
REFERENCES
1,7
5
1
1
7
1
1
1
1
6
1
4
7
1
1
5
5
1
5
3
7
1
1
5
5
5
14
1
1
1
3
5
6
1
5
6
13
7
5
5
7
6
1
3
14
3
7
1
3
5
5
1
Chapter 111.1
1.1-21
Adverse Developmental Effects
-------
* = Listed in TRI. When "compounds" of a metal were listed on TRI, all compounds of that metal in this table are
considered to be TRI chemicals.
References
1. Cunningham and Hallenbeck (1985).
2. Archer and Livingston (1983).
3. Doulletal. (1980).
4. U.S. EPA1983A).
5. U.S. Department of Health and Human Services, NIOSH (1983).
6. U.S. EPA(1983B).
7. IRIS, EPA online database.
8. Clayton and Clayton (1982).
9. Hayes (1982).
10. Vettorazzi(1979).
11. Council on Environmental Quality (1981).
12. International Agency for Research on Cancer (1979).
13. Chambers and Yarbrough (1982)
14. Keyetal. (1977).
15. Ticeetal. (1996).
16. ATSDR(1993).
Chapter III.1 III.1-22 Adverse Developmental Effects
-------
CHAPTER III.2. COST OF Low BIRTH WEIGHT
Clicking on the sections below will take you to the relevant text.
III.2.A Background
III.2.A. 1 Description
III.2. A.2 Concurrent Effects
III.2.A.3 Causality and Special Susceptibilities
III.2. A.4 Treatment and Services
III. 2. A. 5 Prognosis
III.2.B Costs of Medical Treatment and Other Services
III.2.B.1 Methodology
III.2.B.2 Results
III.2.B.3 Other Studies
III.2.B.4 Conclusions
Chapter III.2 III.2-1 Cost of Low Birth Weight
-------
CHAPTER III.2. COST OF Low BIRTH WEIGHT
I.2.A Background
This chapter contains a discussion of the methods used and the results of
estimating the direct medical costs incurred by individuals with low birth
weight (LEW). It does not include information on elements such as
indirect medical costs, pain and suffering, lost time of unpaid caregivers,
etc. The reader is referred to Chapter I.I for a discussion of the cost
estimation methods and cost elements that are relevant to all benefits
estimates. Chapter III. 1 contains information regarding the causes and
special characteristics of developmental illnesses and disabilities, as well as
environmental agents linked to these disorders.
Link to Chapters 1.1 and III. 1
Low birth weight is a serious medical condition that occurs in
approximately seven percent of all infants born in the United States (Oski,
1993). It is associated with multiple adverse effects in numerous organ
systems and carries a much higher risk of death than normal birth weight.
Consequently, considerable medical resources are devoted to the treatment
of LEW infants and the medical expenditures on these infants is estimated
to be $5 billion per year (Lewit et al.,1995). This chapter contains a
detailed estimate of the medical and related costs associated with LEW for
children through age ten and limited additional information for older
individuals. Data regarding special education and aggregate costs are
included due to their availability, although they are not in the usual scope
of this handbook. The chapter also contains a brief discussion of the
medical science related to LEW.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to inflation factors
.2.A.1 Description
LEW babies are defined as having a weight less than 2,500 grams (5.5 Ibs),
and very low birth weight infants weight is defined as a birth weight under
1,500 grams (3.3 Ibs). "Low birth weight results when infants are born
prematurely or grew too slowly in utero, or a combination of the two"
(Paneth, 1995). These children may be 1) full-term infants with
intrauterine growth retardation (IUGR, see below); 2) born prematurely,
Chapter III.2 III.2-2 Cost of Low Birth Weight
-------
with IUGR; or 3) appropriate for their gestational age, but with LEW due
to their prematurity. Regarding the full-term infants, IUGR is considered a
"final common pathway by which genetic and environmental influences
result in low birth weight for gestational age" infants (Oski, 1994). These
are infants that are born after a normal duration pregnancy, but have
retarded growth, and are born smaller and in a more immature condition
than normal infants.
There are varying degrees of severity of LEW. Infants who are 1,501 to
2,500 grams are considered moderately low birth weight and comprise 82
percent of LEW infants. At 1,001 to 1,500 grams they are very low birth
weight infants; this occurs among 12 percent of LEW infants. Six percent
of LEW infants weigh 1,000 grams or less at birth are designated as
extremely low birth weight children (Oski, 1993).
Regardless of the child's status with respect to pregnancy duration, there
are serious consequences associated with low birth weight in many
children. More than three-quarters of all infant deaths have been attributed
to LEW (Paneth, 1995). When compared to normal-weight infants,
moderate, very low, and extremely low birth weight infants have a risk of
death that is 40 times, 200 times, and 600 times higher, respectively, than
the risk of death among normal-weight infants. Children who survive often
face multiple medical disorders arising from LEW (Oski, 1993).
In LEW infants, the basic mechanisms for survival outside the womb are
often lacking. Premature LEW infants typically have very limited body fat
(energy stores), inability to maintain an acceptable body temperature, and
less available energy resources that are required for vital cellular processes,
including protein synthesis and other basic physiological processes. Organ
systems may be functionally immature, including the respiratory system. An
inability to maintain fluid balance leads to patent ductus arteriosus in 15 to
35 percent of very LEW and extremely LEW infants. This condition is
associated with myocardial stress, pulmonary congestion, progressive heart
failure, hepatic congestion, and a constellation of progressively serious
disorders. Intracranial hemorrhage occurs in 40 to 50 percent of LEW
infants and is a major cause of morbidity and mortality (Oski, 1993).
LEW is associated with other developmental abnormalities including
delayed cognitive development and other central nervous system (CNS)
disabilities. For example, approximately 20 percent of infants born at
weights between 500 and 1500 grams have brain injury (Hack et al., 1994).
These acute problems often result in chronic medical conditions in the
LEW child. The disorders and diseases may occur in any organ system,
but are often seen in the gastrointestinal system (necrotizing enterocolitis),
respiratory system (bronchopulmonary dysplasia), the nervous system, and
the eyes (retinopathy of prematurity) (Oski, 1993).
Chapter III.2 III.2-3 Cost of Low Birth Weight
-------
Many babies with birth weights as low as 750 grams (1 pound, 10 ounces)
are surviving due to technology available in neonatal intensive care units
(Hack et al., 1994). This technology is expensive. The medical costs
associated with LEW have been estimated to be $5.4 billion per year
(incremental direct costs among children ages birth to 15 years in 1988)
(Lewit, et al., 1995). This value does not include the costs of long term
care, special services, and special education, which are estimated to add an
additional $500 million per year. The medical expenditures for LEW
children is approximately one-tenth of the total expenditure for health care
for all children (Lewit et al., 1995).
III.2.A.2 Concurrent Effects
Concurrent effects are those that often occur with, but are not necessarily
caused by, the disease or disorder under discussion. LEW is a condition
that causes many adverse effects, and also occurs concurrently with certain
types of conditions and diseases. It is not always possible to determine
whether LEW was a causative agent for a specific medical condition, or
whether the disorders occurred independently, but in tandem, perhaps
arising from the same cause. Often the factors or agents responsible for
LEW are linked to concurrent disorders.
LEW has been linked with most developmental defects. As noted above,
children with low birth weight are often afflicted with severe brain damage,
cerebral palsy, and disorders of multiple organs (e.g., lung and liver
disease, learning disabilities, asthma, and attention disorders). Children
may be born with low birth weight because of birth defects that cause
growth retardation. Alternatively, children may have a number of birth
defects because their birth weight and development are considerably
retarded. Finally, as noted above, the low weight at birth may be simply
concurrent with other birth defects, when the factors that lead to low birth
weight also cause other birth defects (Oski, 1993).
Although long-term disabilities and illnesses may be associated with LEW,
the medical costs associated with these effects are not discussed
individually in this chapter, but are included in the final cost estimates.
The estimated costs in this handbook for LEW incorporate all costs of
medical treatment that exceeded those for a non-LBW child.
Consequently, these incremental medical costs include the costs of treating
concurrent effects, are included in the final incremental cost estimate.
Many of the long-terms effects are discussed in subsequent chapters (e.g.,
cerebral palsy) because these effects also occur independently of LEW.
The occurrence and costs estimates presented in this chapter are based on
the incidence in the overall LEW population. If an agent is known to cause
LEW plus another disorder (e.g., cerebral palsy) at a higher rate than is
usually found in the LEW population as a result of environmental agents
Chapter III.2 III.2-4 Cost of Low Birth Weight
-------
than is typically observed with LEW in the general population, it would be
necessary to add costs associated with the excess probability of the
disorder. By providing cost estimates of each effect separately in
subsequent chapters, the reader can use the information as required by their
risk data.
I.2.A.3 Causality and Special Susceptibilities
As described above, LEW may result from prematurity (a pregnancy of less
than 38 weeks with or without growth retardation) or from delayed growth
of an infant that is full-term (38+ weeks).
III.2.A.3.1 Introduction
Prematurity may result from a variety of factors, including external factors,
maternal health, and the condition of the fetus. The causes in humans are
difficult to evaluate through the usual animal toxicology studies because
rodents are usually used in studies of reproductive outcomes. The rodent
gestation period is short, and there is a small window in time when
pregnancy could be observed to result in a live, but premature birth. More
often the studies yield results that include miscarriage, which is a delivery
prior to full-term that results in the death of the offspring. Numerous
chemicals are associated with miscarriage in toxicological studies of
pollutants. Many of these are listed in Table HI. 1-1 in Chapter III.l
Link to Table III. 1-1
III.1.A.3.2 Data on Causality
Intrauterine growth retardation (IUGR) results in infants who have
retarded growth, and are born smaller and in a more immature condition
than normal infants. They are often LEW infants. Among chemicals that
cause reproductive disorders, it is very common to observe retarded
growth as one of the effects. Retarded growth is often coupled with other
effects, including structural abnormalities and functional anomalies. For
example, chromium, which has been the subject of multiple developmental
toxicity studies, is associated with delayed bone formation, decreased fetal
size, skeletal anomalies, increased resorptions (fetal death) and
postimplantation loss (embryonic loss). This cluster suggests a variety of
developmental impacts, including growth retardation.
Many of the objective measures of developmental toxicity in animals are
highly specific (e.g., delayed bone ossification, decreased fetal weight);
however, when considered as an overall description of the condition of the
offspring, they are analogous to growth retardation in humans.
Developmental toxicity studies in animals usually evaluate the offspring
very near the time of, but prior to, birth, and in the process they sacrifice
the animals (this is why studies report "fetal" rather than "offspring"
abnormalities). Consequently, animal studies provide limited predictive
Chapter III.2 III.2-5 Cost of Low Birth Weight
-------
information regarding the postnatal disorders associated with low birth
weight in humans. Multigeneration studies and postnatal studies may be
carried out, but are more expensive and so are less frequently available.
There are a very limited number of studies evaluating low birth weight and
intrauterine growth retardation in humans in relation to environmental
exposure. It is both difficult and expensive to carry out well-controlled
human development studies. This is illustrated by a recent study by
Munger et al. (1997) that found an association between herbicide-
contaminated drinking water supplied in the Midwestern United States and
intrauterine growth retardation. The presence of very commonly
encountered pesticides (e.g., atrazine, metolachlor, and cyanazine) were
linked to growth retardation after controlling for multiple potential
confounders (e.g., smoking, nutritional status, alcohol consumption). Low
birth weight was also marginally increased among the exposed population.
As is the case for most environmental studies, exposures occurred
simultaneously to the mixture of chemicals, with varying concentrations
over time. Conclusions can be drawn regarding an association between the
cluster of chemical and growth retardation. However, the ability to
establish causality for specific chemicals in this study is questionable.
III.1.A.3.3 Environmental Agents of Concern
Human data on environmental risk factors associated with LEW are very
limited, but dozens of chemicals have been demonstrated to cause LEW in
experimental animals, including many common pollutants. Table III. 1-1 in
Chapter III.l lists some of the chemicals associated with developmental
toxicity. Many of these are linked to growth retardation and/or
prematurity. Many chemicals have been shown to cause LEW in
experimental animals. Delayed development and reduced body weight are
relatively common observations in developmental toxicity studies on
chemical pollutants.
Link to Chapter III. 1, Table III. 1-1
Some chemicals associated specifically with LEW are acrylic acid, aroclor
1016 (aPCB), chlorobenzilate, captan, benomyl, cylothrin, dimethyl
sulfoxide, dicamba, dinoseb, flutolanil, methyl ethyl ketone, metholchlor,
napropamide, nitroguanidine, phenol, propagite, resmethrin, rotenone,
vernam (IRIS), and chromium (ATSDR, 1993). LEW is also strongly
associated with pre-term delivery due to substance abuse, smoking,
diabetes, poor prenatal nutrition and care, and other causes.
II.2.A.3.4 Susceptible Subpopulations
African-American infants are twice as likely as white infants to be born
with LEW and their risk is even greater for very LEW births. The causes
of this increased rate are not know. Although social and economic factor,
access to medical care, and other societal issues are considered to be
Chapter III.2 III.2-6 Cost of Low Birth Weight
-------
responsible in part (Shiono and Behrman., 1995), very recent studies that
have controlled for these factors have found an increase in LEW among
African-American infants. It has been theorized that some groups may be
genetically at greater risk for LEW, suggesting a sensitive subpopulation
that may merit consideration in a benefits assessment. Additional support
for this argument can be obtained from the literature, which indicates that
the rates of LEW among other ethnic minorities in the United States (i.e.,
Hispanic, Native American, and Asian American) are similar to those of
the white population (Paneth, 1995). Consequently, biological and genetic
factors may be important considerations (Shiono and Behrman, 1995). In
an environmental risk benefit context, African Americans may be
considered to be at higher risk, and benefits of avoiding environmental
agents that pose risks of LEW should be considered accordingly.
Additional research is required in this area to fully explore the risk factors
involved.
.2.A.4 Treatment and Services
Medical services in the neonatal period address the numerous acute medical
problems described above (see "Description"). They are designed to
eliminate the short-term crisis and mitigate long-term health problems.
Extensive medical services are often required whether the LEW results
from a preterm delivery or a IUGR condition. Treatment often includes
monitoring in a neonatal intensive care unit; extensive testing to determine
the functional status of various organ systems; surfactant therapy to
preserve lung structure; specialized nutrition; fluid management; and
management of respiration, glucose levels, temperature, and other basic
physiological processes at a normal level to minimize damage (Oski, 1993).
Prenatal evaluations may provide information regarding intrauterine growth
retardation (IUGR), and medical intervention may begin considerably prior
to the birth of the child to address problems that have been diagnosed. The
costs of these types of services are not included in the estimates provided
below. They typically would be considered a part of prenatal costs and
would be attributed (in most accounting methods) to the mother, rather
than the child. Their lack of inclusion in this chapter generates a cost
estimate that underestimates true costs by the amount spent prenatally to
minimize exposure.
III.2.A.5 Prognosis
Although mortality among LEW infants declined rapidly into the 1970s, the
morbidity among the survivors has increased, due to the survival of much
smaller and more seriously ill infants. Table III.2-1 shows the morbidity
pattern for some major categories of illnesses arising from LEW in relation
to the size of the infant at birth. It demonstrates that there is an increasing
risk of morbidity and mortality associated with progressively smaller birth
Chapter III.2 III.2-7 Cost of Low Birth Weight
-------
weight. The table includes selected outcomes for the most seriously
underweight infants, and does not include those in the moderate LEW
category, defined above as having a birth weight of 1,501 to 2,500 grams.
Table III. 2-1. Morbidity and Mortality Among Very Low Birth weight Infants.
Based on data from the National Institute of child Health and Human Development Neonatal
Network. Adapted from Oski, 1993 and originally based on Hack et a/., 1991.
Effect
morbidity among survivors (%)
chronic lung disease (%) among survivors
intracranial hemorrhage (%) among
survivors
enterocolitis(%) among survivors
death (%)
Weight at Birth ( in grams)
501-750
56
26
26
3
66
751-1000
39
14
17
8
34
1001-1250
25
7
13
6
13
1251-1500
15
3
6
4
7
Although survival among LEW has improved considerably in recent years,
those that survive often have long-term health and developmental problems
(Oski, 1993). The long-term prognosis for LEW children depends greatly
on the severity of their initial condition, as well as the medical services
provided shortly after birth. Researchers have found that "Infants with
IUGR secondary to environmental insult or decreased growth potential
generally have outcomes that are poor and reflect the underlying
neuropathology of conditions caused by the environmental or genetic
insult." (Oski notes: "environmental" in this context refers to all factors
outside the mother and child). Infants with normal brain growth generally
have a more favorable prognosis. Most full-term IUGR infants have
normal intelligence, and even many preterm IUGR infants achieve normal
intelligence by school age (Oski, 1993). Conversely, children with LEW
are more likely than children with normal birth weight to have attention
disorders, developmental impairments, breathing problems (e.g., asthma),
and learning disabilities (Shiono and Behrman, 1995). All categories of
LEW children are more likely to be enrolled in special education classes
than normal birth weight children, and half of all children who were very
LEW are enrolled in special education classes (Hack et al., 1995).
.2.B Costs of Medical Treatment and Other Services
The overall costs associated with LEW infants are very high and comprise
a substantial portion of all pediatric costs. While costs vary considerably
depending on the individual and the severity of their condition, the costs for
a single infant with LEW may exceed $1 million. The average cost for
initial hospitalization (only) for surviving infants weighing 500 to 600
Chapter III.2
1.2-8
Cost of Low Birth Weight
-------
grams was $1 million per child (Pomerance et al., 1993). This weight
group is the most seriously affected, but their costs give some indication of
the potential magnitude of the medical costs. Most LEW infants require
some additional care, at an increase in cost over the usual pediatric care
expenditures. It has been estimated that of the $11 billion spent on health
care for infants, approximately 35 percent ($4 billion ) is spent on the
incremental costs of medical care for LEW infants (Lewit et al., 1995).
There are two studies of the costs of LEW discussed in this section. The
Lewit et al. (1995) study, discussed in detail below, is more comprehensive
and is recommended for use in cost estimates. A study by McCormick et
al. (1991) is also briefly discussed, but is not recommended for use in
evaluating LEW costs due to its limitations.
III.2.B.1 Methodology
III.2.B. 1.1 Data from Lewit et al.
A recent comprehensive analysis of LEW was published in 1995 by Lewit
et al. They analyzed the incremental direct costs of low birth weight using
a prevalence approach. These costs, estimated for children aged 0 to 15,
include expenditures for health care, child care, special education, and
grade repetition (considered to be related to special education). Each type
of cost is itemized, and although the handbook focuses on medical costs,
the costs of related professional services (e.g., special education) are
included in the material presented in this chapter. Child care, while a direct
cost, is not estimated in other chapters of this handbook and was not
included this chapter. (The study authors note that their comprehensive
approach offers an opportunity to evaluate the impact of LEW in a single
metric. Readers may wish to consult their work, depending on the goals of
their morbidity valuation.)
Lewit et al. (1995) used a number of sources for their study. Estimates for
LEW babies during the first year of life were obtained by combining data
from two other studies, as well as the data from the CIGNA Corporation's
national employer-based business survey. The data were obtained from a
private source; the birth weight-specific cost ratios were therefore adjusted
to match the actual distribution of birth weights in the 1988 U.S. birth
cohort. The population-weighted relative cost ratios were applied to the
estimated national expenditures for infant health care in 1988 to determine
the allocation of expenditures between LEW and normal birth weight
infants.
To estimate the incremental increase in costs for LEW children beyond the
first year of life, the 1988 Child Health Supplement of the National Health
Interview Survey and the 1991 National Household Education Survey were
used. The large number of children involved in the surveys allowed for
evaluation of age-specific resource utilization. Data regarding birth weight,
Chapter III.2 III.2-9 Cost of Low Birth Weight
-------
hospitalization costs, and other relevant data were used. Multivariate
statistical analyses were conducted to control for the effects of potential
confounding variables, such as income and the mother's educational level.
Estimates of health care costs are also largely based on a previous study
that used the 1987 National Medical Expenditure Survey (USDHHS,
1988).
The incremental costs of special education and of repeating a grade were
estimated by Lewit et al. for children through the age of 15 years.
Inclusion in special education classes and grade repetition both occur with
greater frequency among LEW children than among non-LBW children.
Although they are not a part of medical costs, they are included as
supplemental information and may be considered to be a part of special
services related to a medical condition.
The Lewit et al. (1995) analysis of children's medical costs after the first
year of life includes only the hospital and medical fees associated with
hospitalization. They evaluated the different hospitalization rates for LEW
and non-LBW children and used this ratio, with the average annual hospital
costs per child for all children, to estimate the incremental annual
hospitalization costs for LEW children. This approach will underestimate
total medical costs because it does not include office visits,
Pharmaceuticals, therapy, and other medical services that are not a part of a
hospitalization incident. For example, annual medical costs for asthmatic
children may be high, even when they are not admitted to the hospital.
(Chapter IV.2 contains a discussion of costs associated with asthma.)
LEW children have an increased rate of asthma occurrence and would be
expected to have this type of non-hospitalization cost. This cost is not
included in the Lewit cost estimate because it is not a hospitalization cost.
There are likely to be many other medical services required for LEW
children that do not require hospitalization; the costs of services were not
calculated by Lewit et al.
This handbook addressed the lack of cost data for medical services
unrelated to an inpatient visit (hospital stay) for children over the age of
one by applying an inpatient/outpatient factor obtained from another study
to estimate the costs of outpatient services. A number of data sources
were considered for this factor, including data obtained from costs of
treating birth defects (Waitzman et al., 1996 data as discussed in Chapters
III.3 through III.8) and the Health Care Financing Administration (HCFA)
data. A prominent consideration in selecting the source of the factor was
the anticipated medical services required for LEW children after the age of
one year. It was assumed that much of the care would be for chronic
health problems and that surgical intervention would be minimal. These
assumptions would lead to an allocation of services more heavily weighted
Chapter III.2 III.2-10 Cost of Low Birth Weight
Link to Chapter IV.2
-------
toward outpatient care than would be the case for treatment of diseases
that require substantial surgical intervention, such as treatment of many of
the birth defects discussed in Chapters III.l through III. 8 of this handbook.
Consequently, costs from those chapters were not considered optimal.
Likewise, the HCFA data, which includes surgical and other acute care as a
major component, was not considered appropriate for establishing the ratio
for chronic care treatment of LEW children.
Due to its focus on chronic care after initial diagnosis, cost data for the
treatment of asthma, discussed in Chapter IV.2 of this handbook, was
selected as having the most analogous type of care and cost requirement to
the chronic care provided for LEW children. The inpatient/outpatient cost
ratio can be calculated from the data provided in Chapter IV.2. The costs
for young children (ages 4 to 5) shown in Table IV.2-15 were used to
estimate the ratio. The sum of the office visit, drug therapy and emergency
room costs is $605.37 per year, and the hospitalization cost is $105.79 per
year. The outpatient/inpatient ratio was calculated to be 5.72. This value
was used to estimate the outpatient costs for LEW children by multiplying
the inpatient costs times 5.72. This value is a source of considerable
uncertainty in the cost estimate. However, it was felt that it would be
better to provide an estimate than to omit the costs for outpatient care
altogether.
Link to Chapter IV.2, Table IV.2-15
Additional data will be sought on this issue and the value may be updated
in the future. Results obtained from application of this factor are shown in
the "Results" section below.
III.2.B.1.2 Supplementary Cost Estimates
Lewit et al. (1995) did not estimate incremental medical costs for ages 1 to
2 years and for children over 11 years of age. An estimate of the costs for
the missing years was made for this handbook, to provide a more
comprehensive estimate. Costs were estimated for the missing years as
follows:
1 -2 years. To estimate costs for this age group, the two closest age
groups were considered as sources of information. It was assumed that the
medical and related costs of LEW are highest in infancy, and that these
costs should drop off quite quickly after the first year of life.
Consequently, the incremental cost during the first year of life, of
approximately $25,000, was not considered an appropriate data source.
The cost per year for three- to five-year-olds was assumed to be the most
representative of the one- to two-year age group, although some of the
substantial medical costs seen in infancy may persist into this age group for
those children most seriously affected (i.e., the very low birth weight
infants). Despite this drawback, the three- to five-year-olds' cost per year
Chapter III.2 III.2-11 Cost of Low Birth Weight
-------
I.2.B.2 Results
was assumed to be the best option for the estimate and was used for ages
one and two.
11-15 years. Although incremental educational costs were estimated for
this group by Lewit et al., the researchers did not estimate the costs for
medical services for this age group. The costs were extrapolated from the
costs per year for six- to ten-year-olds because it was assumed that the
medical costs would be very similar. LEW children often have health
problems (as noted in Section 11.2. A, above) that are chronic and continue
through the life of the individual (Lewit et al., 1995). For example, Section
II.2. A noted that LEW children are more likely to have neurological
damage, asthma, and gastrointestinal disorders. These disorders are not
age-limited. Although many very serious acute health problems related to
LEW are addressed in the first year of life, those that persist into school
age are chronic health problems that are likely to recur over the lifetime of
the individual. Consequently, the increased costs seen during childhood in
the post-acute treatment period (e.g., ages five to ten years) are a
reasonable estimate of the costs that may be incurred later in childhood, as
well as in adulthood (see below).
15 - 75 years. LEW children have higher hospitalization rates and require
more medical services in both the preschool- and school-age years than
non-LBW children (Corman and Chaikind, 1993; Corman, 1994). Due to
the inability to track many children over the age of 15 years (as a result of
their dropping out of school, etc.) and corresponding lack of good tracking
data, the analysis was not conducted on children over the age of 15 by
Lewit et al. Increased medical and related costs are likely to continue for
these individuals, due to persistent medical and, in some cases, social and
educational costs. To address this factor, an estimate was made of the
potential lifetime incremental medical costs of LEW, assuming that the
increased yearly costs seen in the six- to ten-year-old age cohort would be
representative of the costs over a lifetime. As noted above, in all cases the
cost data are for increased rates of hospitalization only and do not include
outpatient services, long-term care facilities, pharmaceuticals, etc. The
values are underestimates of the actual total incremental medical costs
associated with LEW.
This section contains both comparative information regarding the costs for
LEW infants versus normal weight infants, and costs associated with LEW
over the childhood years.
III.2.B.2.1 Aggregated First-Year Costs for All Infants
Table III. 2-2 shows the total and incremental costs for LEW infants during
the first year of life, in contrast with normal birth weight infants. The LEW
costs, based on 1988 data from Lewit et al. (1995), are weighted averages
Chapter III.2
1.2-12
Cost of Low Birth Weight
-------
based on the costs for all LEW categories and the proportion of infants in
those categories. The costs are updated using the Consumer Price Index
(CPI) for medical care from 1988 to 1996 dollars (1996:1988=1.6465).
The total costs and costs for all births and normal weight infants are
provided for comparative purposes.
Table III. 2-2. Comparative Medical Costs for LBW and Normal Birth Weight Infants During the
First Year of Life in 1996 Dollars.
Birth Weight Group
(in grams)
all births
normal weight
(2,500+)
LBW
(<2500)
extremely LBW (<1 ,000) or
having respiratory distress
syndrome
other LBW (1,000 -2500
without respiratory distress
syndrome)
Number of
Births in
1988 in the
U.S.
3,871,000
3,600,000
271,000
57,000
214,000
Incremental
Cost per Child
in 1996 Dollars
(1st year)
3,128
1,554
24697
52,688
16,465
Total Costs in
1996 Dollars
(billions)
12.2
5.6
6.6
3.0
3.6
Percentage
of Total
Health
Care Costs
for Infants
100%
65%
35%
16%
19%
Source: Lewit et al., 1995 adjusted for inflation using CPI (1996:1988=1.6465)
a. Average for all LBW infants.
III.2.B.2.2 Annual Per Capita Costs for LBWInfants
Table III.2-3 shows the incremental direct costs per year estimated by
Lewit et al. (1995) for each age group for the average LBW infant. It also
shows outpatient cost estimates derived in this handbook from the ratio of
outpatient to inpatient costs (see methodology section for detail).
As noted above, Lewit et al. (1995) estimated health care costs for ages
through ten years only and did not estimate any costs for ages 1 and 2
years. As noted above, these gaps were addressed by using costs for the 3
to 5 year old age group as representative of the costs for ages 1 and 2
years. In addition, the medical costs for the 11 to 15 year olds and the 16
to 75 year olds were extrapolated from the costs per year for 6 to 10 year
olds.
Chapter III.2
1.2-13
Cost of Low Birth Weight
-------
Table III. 2-3. Annual Incremental direct costs per LBW child (1996$)
Age Group
Infancy
1 to 2 years
3 to 5 years
6 to 10 years
6 to 15 years
1 1 to 1 5 years
1 1 to 1 5 years
16 to 75 years
Cost Type
Health Care (all medical costs)
Health Care (inpatient only)
Health Care (inpatient only)
Health Care (inpatient only)
Special Education
Health Care (inpatient only)
Grade Repetition
Health Care (inpatient only)
Mean Cost
per Year per
LBW Child
$24,6971
$4772
$4771
$7741
$2471
$7742
$742
$7742
Outpatient
Cost
Estimate
Based on
Ratio3
(not calculated
separately )
$2,728
$2,728
$4,427
N/A
$4,427
N/A
$4,427
1 Costs from Lewit et al. (1995), adjusted for inflation, as shown in Table III. 2-2.
2 Costs based on Lewit et al., estimated and modified as discussed in the Methods section of this chapter.
3 Outpatient costs were estimated based on an outpatient/inpatient ratio derived from treatment of asthma
patients. See Methods section of text for discussion.
Link to Table III. 2-2
Lewit et al. estimated that there were approximately four million LBW
children from ages birth to 15 years in 1988 and that their total incremental
direct costs were $5.4 trillion in 1988 dollars. Using the Consumer Price
Index (CPI) for medical care from 1988 to 1996 dollars
(1996:1988=1.6465), the total cost in 1996 dollars (assuming a population
the same size as in 1988) is $8.91 billion.
III.2.B.2.3 Childhood Costs
The direct medical costs for ages 0 to 15 years are shown in Table III.2-4
at four discount rates (zero, three, five, and seven percent). These are
average values; as Table III.2-2 demonstrated, the costs are much higher
for some children, born with extreme low birth weight and/or respiratory
distress.
The total estimated cost for special education, grade repetition, and
medical care (as itemized in Table III.2-3) for ages 0 to 15 years is $85,447
(undiscounted in 1996$).
Chapter III.2
1.2-14
Cost of Low Birth Weight
-------
Table III. 2-4: Discounted childhood medical care costs (Age 0-15) for Low Birth Weight in
1996$
Inpatient and Estimated
Outpatient Medical Care
0%
$82,607
3%
$69,765
5%
$63,292
7%
$58,052
III.2.B.2.4 Lifetime Costs
Medical and other costs for LEW individuals persist into adulthood. They
often have health problems (as noted in Section III.2.A, above) that are
chronic and continue through the life of the individual. To address this
factor, an estimate was made of the potential lifetime incremental medical
costs of LEW, as described in Section III.2.B.1, above. Table III.2-5
below provides an estimate of lifetime incremental medical costs that may
be incurred.
In reality, it is most likely that those individuals most seriously affected by
LEW will have substantially increased medical costs that exceed the
estimate given below (e.g., those with serious brain damage, respiratory
insufficiency), and that many individuals with moderate LEW will have
minimal incremental costs associated with their birth condition. The values
below are an estimate of the likely average costs that may be incurred.
Link to Section III.2.B. 1
The total estimated cost for special education, grade repetition, and
medical care (as itemized in Table III.2-3) is $436,514 for a full lifetime of
75 years (undiscounted in 1996$).
Table III. 2-5: Discounted lifetime medical costs (Age 0-75) for Low Birth Weight in 1996$
Inpatient and Estimated Outpatient
Medical Care
0%
$348,227
3%
$148,406
5%
$103,601
7%
$80,578
III.2.B.2.5 Limitations
The data provided in the Lewit et al. (1995) study have several limitations,
discussed briefly below, with regard to determining average costs.
The data used by Lewit et al. had very few observations of very low birth
weight children. The cost estimates may be underestimated because very
LEW children incur the highest costs.
The authors noted that the costs may be underestimated due to limited
access to data on charges in the neonatal intensive care unit (NICU, where
the most significant charges are incurred) and on subsequent hospital stays.
Chapter III.2
1.2-15
Cost of Low Birth Weight
-------
The data used by Lewit et al. were from one geographic area. There are
likely to be some regional differences in practices and costs.
Some aspects of the cost estimate relied on parental surveys, which
introduce uncertainty.
Data regarding some age groups are lacking. Best estimates of the costs
for age cohorts 11 to 15 years and 16 to 75 years were made based on
reasonable use of the existing data. Actual cost data would be preferable.
The lack of data for the one to two years age group is particularly
problematic. Because the one- and two-year-old child is still very close in
age to the birth experience and attendant medical problems, they are more
likely to require continuing medical and developmental services than they
would in at ages three to five. The three- to five-year age group was the
source of the cost data extrapolated to the one to two year age group. The
costs for the three- to five-year-olds are likely to underestimate the medical
costs for the 1 to 2 year age group, especially among the very low birth
weight children.
Finally, the most serious limitation is the lack of data, aside from
hospitalization costs, on non-hospital costs after the first year of life.
Hospitalization costs are clearly a significant factor in lifetime costs, but
they underestimate the total incremental medical costs by a factor that is
unknown. This uncertainty is offset partly by the comprehensive medical
costing that was carried out for the first year of life by Lewit et al. The
first year is the most expensive for LEW individuals, especially for those
with the most costly medical treatment. Accurate accounting during this
critical period provides a good basis for the overall cost estimate for LEW.
The outpatient cost estimates based on the experience of asthmatics
provides a reasonable estimate of outpatient costs for LEW individuals
after the first year of life. The lack of comprehensive medical cost data for
latter years introduces additional uncertainty into the cost estimate and will
result in an underestimate of total costs.
As with all cost estimates in this handbook, changes in technology create
uncertainty regarding the costs and practices that are currently relevant.
III.2.B.3 Other Studies
McCormick et al. (1991) provides another estimate of LEW costs, but only
for the first year of life. The researchers analyzed the costs incurred by
parents of Very low birth weight babies after the initial neonatal
hospitalization. Costs were obtained from a study of Very low birth weight
infants discharged from the Infant Intensive Care Unit of the Children's
Hospital of Philadelphia from July 1983 to October 1984. The study
population consisted of 32 infants. Incremental costs were evaluated using
Chapter III.2 III.2-16 Cost of Low Birth Weight
-------
control group of 34 infants born with normal birth weight. Costs were
estimated only for the first year of life and obtained solely for the one
hospital's birth cohort.
The costs incurred by the family after leaving the NICU were recorded and
include hospital care, visits to doctors, diagnostic tests, prescription and
nonprescription medication, medical supplies and equipment, and special
infant formulas. Other costs reported by McCormack et al. are: hotel
charges associated with rehospitalization, renovations to the house, special
diets, travel to obtain health care, and the cost of in-home care or day care.
While important, they are not discussed in this chapter, since they do not
fall within the scope of the handbook.
Total per capita direct costs estimated by McCormick et al., over the first
year are presented in Table III.2-6. The estimates are updated from 1984
to 1996 dollars using the Consumer Price Index for medical costs
(1996:1984=2.14). The costs were calculated after subtracting the costs
for medical care for the control group (the non-LBW children) to obtain
incremental costs of LEW.
Table III. 2-6: First year medical costs for LBW infants in 1996$
(McCormick et al., 1991)
Type of Costs
Hospital Care
Physician Visits
Other Direct Medical Expenditures
Total
Total Year
$15,705
$709
$2,667
$19,145
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
below
Link to inflation factors
Additional studies conducted in the 1980s provide some information on
this topic (OTA, 1988; Boyle et al., 1983). These studies, however, have
several drawbacks. They do not provide complete cost estimates, use data
that are not well-suited to this analysis (either Canadian costs, or costs
primarily for moderate LBW rather than including very or extremely LBW
infants), and the ages of the studies are not optimal, due to rapidly
changing medical practices in this field. Additional studies are available on
specific population groups, but do not provide representative data on costs
(Hacketal., 1995).
Chapter III.2
1.2-17
Cost of Low Birth Weight
-------
.2.B.4 Conclusions
The Lewit et al. (1995) study is the most comprehensive and uses a number
of credible databases and assumptions to derive incremental cost estimates.
In addition, costs were extrapolated for ages not covered by the Lewit et
al. study; these extrapolations were reasonable and provide an estimate of
average costs. The incremental costs for very LEW and extremely LEW
infants are likely to be much higher throughout their lifetime due to
persistent serious medical conditions. The McCormick et al. study looks
only at the first year of life, while the Lewit et al., (1995) study estimates
health care costs over the first ten years, and other costs up to age 15.
Because of this restriction, cost estimates based on the Lewit et al. results
are recommended for use in LEW valuation. As discussed in the
"limitations" section above, the Lewit et al. data are likely to underestimate
the actual incremental medical costs for LEW.
Chapter III.2 III.2-18 Cost of Low Birth Weight
-------
CHAPTER III.3. COST OF CLEFT LIP AND PALATE
Clicking on the sections below will take you to the relevant text.
III.3.A Background
III.3.A. 1 Description
III. 3. A. 2 Concurrent Effects
III.3.A.3 Causality
III.3.A.4 Treatment and Services
III. 3. A. 5 Prognosis
III.3.B Costs of Medical Treatment and Other Services
III.3.B.1 Methodology
III.3.B.2 Results
Chapter III.3 III.3-1 Cost of Cleft Lip and Palate
-------
CHAPTER III.3. COST OF CLEFT LIP AND PALATE
III.3.A Background
This chapter contains a discussion of the methods used and the results of
estimating the direct medical costs incurred by individuals with cleft lip
and/or palate, and the results of the analysis.1 It does not include
information on elements such as indirect medical costs, pain and suffering,
lost time of unpaid caregivers, etc. The reader is referred to Chapter I.I
for a discussion of the cost estimation methods and cost elements that are
relevant to all benefits estimates. In addition, Chapter III. 1 contains
information regarding the special characteristics of developmental defects,
and a list of chemicals that may cause developmental abnormalities.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and III. 1
Link to inflation factors
I.3.A.1 Description
Cleft lip and palate occur when structures in the nose and mouth fail to
close during embryonic development. These birth defects appear as
openings or incomplete structures in the centerline of the face and mouth.
They often occur concurrently (approximately 50 percent of the time), due
to the mechanism of damage that leads to these defects (Fraser, 1970).
Cleft lip and palate may involve a small or large portion of the facial
structures associated with the nose and mouth. Without treatment, infants
may have problems sucking and swallowing and may aspirate food into
their lungs. Other problems arising from this deformity include delayed and
distorted speech, ear and sinus infections, reading disabilities, crossbite,
and problems with social interactions (Waitzman et al. 1996).
The incidence of cleft lip and palate varies across ethnic groups. Cleft lip,
with or without cleft palate occurs in approximately one in 1,000 births
among whites, 0.3 in 1,000 among blacks, 3.5 in 1,000 among Native
Americans, 1.7 in 1,000 among Chinese, and 2.5 in 1,000 among Japanese
(Oski, 1994). Although there are over 300 recognized cleft syndromes,
these represent a small percentage of the total cleft cases (Oski, 1994).
1 "Costs" in this chapter refer to direct incremental per capita medical costs, unless otherwise
noted.
Chapter III.3 III.3-2 Cost of Cleft Lip and Palate
-------
I.3.A.2 Concurrent Effects
.3.A.3 Causality
More than 50 percent of infants born with cleft lip and/or palate have other
anomalies of the face, heart, and head (microcephaly). Cleft lip and/or
palate occur with slightly greater probability in individuals with Down
syndrome than in the general population (Waitzman et al. 1996). Middle
ear infections are a common occurrence in children with cleft palate (Oski,
1994).
Cleft lip and palate occur early in development, when the basic skeletal
structures are being formed. The latest time that this anomaly can be
induced is approximately 36 days post-conception (Bennett and Plum,
1996). Consequently, factors that cause this disorder must occur before
the pregnancy begins, or in its very early stages.
Both genetic and environmental factors may be responsible for cleft lip and
palate, inhibiting the normal flow of cells during development. A long list
of teratogenic substances can cause clefting in rodents, including
corticosteroids, folic acid inhibitors, Vitamin A, and phenytoin (Oski,
1994).
Table III. 1-1 in Chapter III. 1 lists numerous chemicals associated with
developmental abnormalities in human and/or animal studies. Many of
these have identified structural and anatomical defects. Cleft lip and palate
fall into this category of defects. Little human data exist on developmental
effects of chemical exposures, as discussed in Chapter III.l, even though
birth defects are relatively common occurrences.
Link to Table III 1-1 in Chapter III. 1
Twin data suggest some genetic component to the occurrence of cleft lip
and palate; however, only 35 percent concordance in the occurrence of the
disorder is seen in identical twins. A simple genetic inheritance pattern is
therefore not responsible for the disorder in most cases (Oski, 1994), or all
identical twins (who have the same genetic composition) would have
identical patterns of occurrence (i.e., either both or neither would have the
anomalies).
A recently completed study of the offspring of pesticide application
workers found that their incidence of skeletal anomalies (as well as
respiratory system, circulatory system, and urogenital anomalies) were
significantly greater than those of the general population in the same area
of the United States (Garry et al., 1996). Musculoskeletal anomalies
include any abnormality in the size, shape, or function of part of the
skeletal system, muscles, and related tissues (e.g., cartilage). They include
Chapter III.3 III.3-3 Cost of Cleft Lip and Palate
-------
the absence or shortening of limbs (as discussed in Chapter III.4) and the
abnormal formation of part of the skeleton or related soft tissue and
cartilage, as is the case in cleft lip and palate. The chemicals evaluated in
the Garry et al. study that were associated with the birth defects were
trifluralin, triazine herbicides (including atrazine, a very common well
contaminant in agricultural areas), and chlorophenoxy herbicides (including
MCPA and 2,4-D, a pesticide with very high usage). There was also a
significant increase in birth defects among infants conceived in the spring
(i.e., during peak chlorophenoxy use) compared to infants conceived
during other periods of the year (Garry et al., 1996).
As indicated in Chapter III.l, the timing of exposure is often a critical
determinant of whether and what type of effects will occur. Generally, the
earlier during a pregnancy that damage occurs, the more serious the effect
will be because all cells developing from the damaged cells may also be
damaged, and basic structures are being formed during the first trimester
(three months). Although Garry et al. do not provide detailed information
on the specific nature of the anomalies, this information can be obtained
from the authors of the study or from EPA (the work was funded by U.S.
EPA).
Link to Chapter III. 1
III.3.A.4 Treatment and Services
Cleft lip and palate are typically corrected during infancy or early
childhood. They are usually treated surgically shortly after birth to close
the lip. Surgical correction of the palate typically occurs at 6 to 18 months.
Remaining defects are corrected during adolescence. Speech and
orthodontic services are frequently required (Waitzman et al., 1996). A
medical team consisting of a maxilofacial surgeon, audiologist, speech
pathologist, prosthodontist, otolaryngologist, pediodontist, and geneticist
are often involved in the treatment of this disorder over an extended period
during childhood.
.3.A.5 Prognosis
Long-term survival and quality of life after the first year of life are good, in
the absence of other medical problems. As noted under concurrent effects
above, other anomalies that frequently accompany cleft lip and palate may
shorten life and diminish abilities (Waitzman et al., 1996). The mortality
experience in California of individuals born with cleft lip and/or palate as
reported by Waitzman et al. is discussed in Section III.3.B.
Chapter III.3 III.3-4 Cost of Cleft Lip and Palate
-------
III.3.B Costs of Medical Treatment and Other Services
Cleft lip and palate are typically corrected during infancy or early
childhood; consequently, their cost evaluation is simpler than many of the
other developmental effects discussed in this handbook. Cleft lip and
palate correction focuses on surgical and other medical care that occurs
over a relatively short time period. The medical procedures often fully
address the problem, and there are not medical costs occurring over the
lifespan of the individual, as there often are with effects such as spina bifida
and cerebral palsy (Waitzman et al., 1996).
As noted above, cleft lip and palate are very often associated with other
birth defects. They also occur with slightly greater probability in
individuals with Down syndrome than in the general population.
Consequently, it may be necessary to evaluate the costs of multiple effects
in addition to cleft lip and palate (Waitzman et al., 1996). The need for this
type of analysis is supported by the occurrence of multiple birth defects in
toxicological tests of environmental chemical exposures in animals.
Multiple effects, which often occur from exposure to chemicals, would all
be considered in a comprehensive cost estimate.
III.3.B.1 Methodology
This chapter and chapters III.4 through III.8 in this section use cost of
illness estimates developed by primarily Waitzman et al. (1996). The
methodology and relevant considerations are discussed in detail in the
following sections. The results and elements related specifically to each
condition are then discussed in each individual chapter.
To estimate the lifetime medical costs incurred by an individual with a birth
defect, Waitzman et al. estimated the average lifetime medical costs for an
individual with the birth defect. From this value, the authors subtracted the
average lifetime medical costs for an individual without the birth defect.
This method has two important implications. First, unlike the costs
reported for many of the diseases in this handbook, cost estimates based on
Waitzman et al. include the costs of concurrent effects. These estimates
yield a more comprehensive assessment of total costs than would be
obtained if only individual effects were evaluated. This is of particular use
in valuing the avoidance of birth defects because they very frequently occur
in clusters within an individual. Second, the Waitzman et al. method
estimates the incremental costs for individuals with birth defects — that is,
the costs above and beyond the average costs that would be incurred by
individuals without the birth defect.
The Waitzman et al. cost estimates are well-researched and up-to-date.
These cost estimates are based on ongoing costs of birth defects in
California across many ages and the occurrence of birth defects in a large
Chapter III.3 III.3-5 Cost of Cleft Lip and Palate
-------
cohort of children born in California in 1988. The state of California has
spent considerable resources to evaluate causes and occurrences of birth
defects and has an ongoing birth defects monitoring program.
Consequently, the use of this state data provides excellent sources of
information on occurrences, costs, and related information. California's
large size and diversity makes it a good heterogenous source of cost data.
Waitzman et al. used multiple databases, including the California Birth
Defects Monitoring Program incidence, birth, and death records, the
National Health Interview survey, The California Office of Statewide
Health Planning and Development hospital discharge data, MediCal claims
files, California Department of Development Services data, the National
Longitudinal Study of Special Education Students, California Special
Education Enrollment and Expenditure data, and the Survey of Income and
Program Participation. The variable-specific data sources and limitations
will be discussed in greater detail in Section III.3.B.1.6.
III.3.B. 1.1 Direct and Indirect Costs
Cost of illness studies typically involve analysis of two types of costs:
direct and indirect. Direct costs are associated with resources used to
provide medical care to people with a particular illness. These costs
include medical costs such as inpatient and outpatient costs, and
nonmedical costs, such as the cost of special education. Indirect costs, on
the other hand, are associated with resources lost to society as a result of
premature mortality and/or morbidity in patients. These costs are related
to activity limitations and include foregone earnings.
Waitzman et al. (1996) estimated three categories of costs: direct medical
costs, direct nonmedical costs, and indirect costs. Direct nonmedical and
indirect costs are reported at the end of this chapter without discussion for
the reader's convenience. Direct medical costs, specifically inpatient care,
outpatient care, pharmaceuticals, laboratory tests, X-rays, appliances, and
long-term care are included in the direct medical cost estimates shown in
this chapter. For more information on methods used to derive these costs,
the reader may wish to consult Waitzman et al. (1996).
III.3.B.1.2 Prevalence versus Incidence
Cost of illness studies usually employ either a prevalence or an incidence
approach. The prevalence of an illness at a given point in time is the
number of individuals who have the illness at that time. The prevalent
population is defined as a cross section of the population with the disease
of interest at a given time. This group may be subdivided into age-specific
prevalent populations. Incidence refers to the newly diagnosed cases of an
illness. The incidence of an illness in a given year, for example, is the
number of cases of the illness that were newly diagnosed in that year.
Chapter III.3 III.3-6 Cost of Cleft Lip and Palate
-------
Prevalence approaches are more useful for assessing treatment strategies.
Incidence approaches, on the other hand, are more useful for assessing
prevention strategies, because preventing the occurrence of an illness
avoids the entire stream of costs that would have resulted from the illness
from its inception to the death of the individual. Part of the benefit of
preventing an illness is the benefit of avoiding the costs associated with the
illness, measured as the present discounted value of the stream of costs that
would be incurred over the entire course of the illness (Waitzman et al.,
1996).
Waitzman et al. use an incidence-based approach, but use both incidence
and prevalence numbers to arrive at the final incidence-based estimates.
Cross-sectional cost data from the larger population group of all California
residents diagnosed with a condition (i.e., the prevalent population) were
divided by the total number of people in this group to obtain estimates of
the costs for the cohort of interest (i.e., those born in 1988). These costs
were adjusted based on the average cost for a healthy individual, in order
to obtain incremental costs (discussed below). The per-capita incremental
costs were then multiplied by estimates of the size of the 1988 cohort at
each age (i.e., the incident population) to obtain a total incidence-based,
age-specific cost estimate for the 1988 cohort. Finally, the total costs were
divided by the size of the cohort at birth with each defect to obtain the cost
per case.
Table III.3-1 provides California prevalence data for a point in time (July 1,
1988) and California incidence data for infants born in 1988 for cleft lip
and palate, along with other commonly observed developmental defects.
This table shows the differences between prevalence and incidence, as well
as the variation across defects in each. Although no exact relationship
between the prevalence of these conditions on July 1, 1988 and their
incidence during 1988 exists, the two are highly correlated. For example,
the two conditions with the highest incidences in 1988 (cleft lip or palate
and cerebral palsy) are also the two most prevalent conditions on July 1,
1988. The third most prevalent condition on July 1, Down syndrome, also
has the third greatest incidence during 1988. The least prevalent condition,
truncus arteriosis, has the lowest incidence.
Chapter III.3 III.3-7 Cost of Cleft Lip and Palate
-------
Table III. 3-1: Prevalence and incidence of the most frequent birth defects in California:
1988*
Condition
Spina bifida (Ch. III.6)
Truncus arteriosis (Ch. III. 5)
Transposition (Ch. III. 5)
Tetralogy of fallot (Ch. III.5)
Single ventricle (Ch. III. 5)
Cleft lip or palate
Upper limb reduction (Ch.
III.4)
Lower limb reduction (Ch.
III.4)
Down syndrome (Ch. III. 8)
Cerebral palsy (Ch. III. 7)
Prevalence of the Birth
Defect on July 1, 1988**
8,859
1,591
7,469
5,336
1,932
24,956
7,895
3,856
14,095
28,745
Incidence of the Birth
Defect Among Infants Born
in California in 1988***
226
56
263
187
68
944
234
114
558
656
* Numbers are from Tables 3-1 and 3-4 in Waitzman et al., 1996.
**The prevalence of a birth defect at a given time is the number of individuals (of all ages) who have the
birth defect at that time.
***The incidence of a birth defect in a given year is the number of infants born with the birth defect during
that year. For example, whereas there were 8,859 individuals (of all ages) in California on July 1, 1988
who had spina bifida, 226 babies were born with spina bifida in California during 1988. Those infants born
with spina bifida in California in 1988 on or before July 1 and still alive on July 1 would be counted among
the prevalent population on July 1, 1988.
Waitzman et al. estimated the lifetime costs of birth defects in a cohort and
therefore used an incidence-based approach. Ideally, they would have
tracked the costs of the cohort members over time, until the death of the
last cohort member. Because the members of the cohort were born in
1988, however, this was not possible. Instead, estimates of the costs
incurred at each age were based on estimates of per capita costs in the
prevalent population of that age. The method is described more fully in
Section III.3.B.I.5.
III.3.B.1.3 Incremental Costs
Waitzman et al. emphasize that their cost estimates are "cost estimates for
individuals with birth defects rather than the costs of the birth defects per
se" (Waitzman et al., 1996). They are, moreover, estimates of incremental
Chapter III.3
Cost of Cleft Lip and Palate
-------
costs. Because medical costs are incurred by the population as a whole, the
costs incurred by an individual with a birth defect must be adjusted to
reflect these baseline costs. The per capita incremental cost of a birth
defect is the cost incurred above and beyond the cost incurred by an
"average child" without the birth defect. Waitzman et al. attempted to
isolate those costs specifically related to the condition of interest and
associated anomalies. In order to do this, the costs for each individual with
a particular effect (e.g., cleft lip) were tracked. Costs for the average non-
affected person were subtracted from those for the average person with the
effect.
III.3.B.1.4 Costs of Concurrent Effects
Many of the defects discussed in the following chapters are associated with
other defects. By their nature, Waitzman et al.'s cost estimates include the
costs of concurrent effects. As noted above, Waitzman et al. estimated the
costs incurred by individuals with birth defects, including all medical costs
incurred, rather than the cost of the birth defect per se. For example, the
mean per-capita cost incurred by a person with Down Syndrome would
include the costs associated with other defects. This approach has
advantages and disadvantages (as discussed in detail by Waitzman et al.).
If two defects are dependent, then it makes sense to include costs related to
both conditions; the benefits of preventing the first would be equal to the
costs of both, since prevention of one leads to prevention of the other.
Alternatively, if two defects are independent of each other but show up in
the same person, this methodology would yield an overestimate of costs.2
Little is known about the pathogenesis of birth defects, and the assumption
is that most defects fall somewhere in between total dependence and
independence. Given the relative rarity of many severe birth defects, the
probability that they would occur in the same person, without any linkage
in causation, is very small.
As Waitzman et al. note, the costs of associated anomalies are included as
part of the estimate of the costs incurred by an individual with a given birth
defect. For this reason, their cost estimates cannot be aggregated across
birth defects because of the possibility of double counting.
Given the large size of the California databases used by Waitzman et al.,
the combination and frequency of concurrent effects in the authors' sample
are likely to be representative of those in the larger United States
population, and therefore appropriate for a benefits assessment.
2 Waitzman et al. address this situation when they extrapolate the per-capita mean costs to the
total mean costs per disease. Rather than including a single person in the incidence totals for both Down
syndrome and cleft palate, for instance, they include a person with both defects in the incidence total of the
more costly defect. When the per-capita costs are multiplied by the total number of people with the defect,
each person is counted only once. The costs in this handbook focus on the per-capita costs only.
Consequently, the adjustment described above is not applicable to the costs presented here.
Chapter III.3 III.3-9 Cost of Cleft Lip and Palate
-------
III.3.B.1.5 Analysis
The estimation of the average per capita lifetime cost associated with a
birth defect is based on the method of Waitzman et al. (1996). Waitzman
et al. estimate both the total lifetime cost for a cohort of individuals and the
average lifetime cost per case — i.e., for a single individual in the cohort.
(The average cost per case is obtained by simply dividing the total cost for
the cohort by the original number of individuals in the cohort.) The
Waitzman et al. method is discussed briefly below.
Waitzman et al. estimate the present discounted value of lifetime costs
associated with a birth defect for a cohort born with the defect in 1988 in
California. This value is the sum of costs over all years in which cohort
members are alive, with future costs discounted appropriately. The total
costs incurred during the first year after birth, denoted TCl3 are the costs
incurred by all members of the cohort who survive to one year of age; these
costs are discounted back to the year of birth by dividing TCj by the
discount factor for the first year, (1 + r), where r is the discount rate. The
total costs incurred during the second year after birth, denoted TC2, are
the costs incurred by all members of the cohort who survive to two years
of age; these costs are discounted back to the year of birth by dividing TC2
by the discount factor for the second year, (1 + r)2. In general, the costs
incurred during the rth year after birth, TQ, are the costs incurred by those
members of the cohort who survive to age /'. These costs are discounted by
dividing by the discount factor for the rth year, (1 + r )'.
The total cost of the birth defect, COBD, is the sum of these discounted
age-specific total costs:
COBD =£;
As discussed in Section III.3.B.1.3, the total costs incurred at a given age
are incremental costs — that is, those costs above and beyond the costs
incurred by the average child of that age.
Link to Section III. 3.B. 1.3
One way to estimate TQ, the total incremental costs during the rth year
after birth (i.e., at age /'), might be to estimate the incremental costs
incurred by those members of the cohort who survive to age /'. For a
cohort born in 1988, however, this calculation would be possible only
through age nine, because members of that cohort would reach that age in
the present year, 1997. Waitzman et al. instead base their estimates of total
incremental costs for a given age on costs in the prevalent population of
that age. For example, the average per capita total incremental cost among
15-year-olds is estimated from data on individuals with the birth defect
who are 15 years old (none of whom can be members of a cohort born in
Chapter III.3 III.3-10 Cost of Cleft Lip and Palate
-------
1988). This method assumes that the real costs associated with the birth
defect for individuals of a given age will not change appreciably over time
— for example, that the real costs for a 15-year-old in the year 2003, when
the surviving members of the cohort born in 1988 are 15, will be the same
as the real costs for an individual who is 15 years old in the prevalent
population examined (e.g., in 1988). Waitzman et al. note that this
conclusion in turn rests on the assumption that future treatment patterns
will resemble current treatment patterns. An estimate of the per capita
cost based on the prevalent population of age /', PCPREV;, multiplied by an
estimate of the number of individuals in the 1988 cohort who are expected
to survive to age /', S;, yields an estimate of the total incremental costs of
those cohort members who survive to age /':
TQ = (S; ) x (PCPREV; ) .
Waitzman et al. estimate per capita costs in the prevalent population of age
/', PCPREV;, in two different ways, depending on data availability. One
method is to simply divide the total costs in the prevalent population of age
/' by the number of individuals in that population. If the necessary
information is not available for this method, an alternative method is used.
Not all individuals with birth defects incur incremental costs at each age.
The alternative method estimates the proportion of the prevalent
population at age /' who do incur incremental costs, and multiplies that
proportion by the average per capita costs incurred by the individuals who
incur costs. This second method is used when estimates of the total or per
capita costs for the prevalent population of age /' are not available, but
estimates of the average per capita cost for those incurring costs are
available.
The focus in this handbook is on the expected lifetime incremental costs for
an individual with the birth defect — i.e., the average lifetime cost per case.
As noted above, Waitzman et al. obtain this value by simply dividing the
total cost for the cohort by the original number of individuals in the cohort.
This method is equivalent to following the calculation outlined above, with
one alteration: now, the probability of surviving to age /' among those
individuals born with the birth defect, ps;, is used, rather than S;, the
number surviving to age /'.3 The expected per capita cost at age /', PCQ, of
an individual born with the birth defect would then be:
PCQ = (ps;) x (PCPREV;) .
3 To estimate the number of cohort members surviving to a given age, Waitzman et al. multiplied
the number in the original cohort by an estimate of the probability of surviving to that age. The data
necessary to estimate survival numbers therefore include the data necessary to estimate survival
probabilities.
Chapter III.3 III.3-11 Cost of Cleft Lip and Palate
-------
The present discounted value of expected per capita lifetime costs of the
birth defect, PCCOBD, is just the sum of these expected age-specific per
capita costs, appropriately discounted:
PCCOBD
III.3.B.1.6 Variable-Specific Data Sources
Table III. 3. 2 lists the data sources used for each variable necessary for the
calculations described above. It also briefly describes the limitations
associated with each of these sources, and the methodology used to derive
the data from the relevant sources.
Table III. 3-2: Variable specific data sources and limitations
Variable
Number
surviving to a
given age,
S = |xm
I (Incident Cases)
(not including Cerebral
Palsy)
I (Cerebral Palsy)
m (age-
specific
survival
proportion)
Age 0-1
Quadrant IV
conditions
Mortality
Source
CBDMP 1
CA Cerebral Palsy
Project
CBDMP linked with
death certificates
MEDLINE search for
each condition
Review of clinical lit,
panel of clinical
advisors
Data Limitations
1 . Actual numbers were not collected for three
major counties, but were estimated for the study.
These estimates could be biased.
2. Only live-born children are included.
3. Certain internal defects are not apparent
during the first year, and therefore prevalence
may be underestimated.
4. Does not account for birth defects treated
exclusively on an outpatient basis.
1 . Exceptionally stringent criteria for including
cases. May underestimate.
1 . Assumes that mortality at time of database is
same at time of study — overestimates costs.
2. Assumes normal mortality beyond a specific
age — underestimates costs.
1. Assumption of normal mortality beyond a
specific age — may overestimate the number of
survivors.
Chapter III.3
1.3-12
Cost of Cleft Lip and Palate
-------
Table III. 3-2: Variable specific data sources and limitations
Variable
Size of the
prevalent
population
Total
incremental
costs in the
prevalent
population
(for each
age)
PCAFF
TC — special
education
Age 0-1
After age 1
Inpatient/Outpatient
Long-term (Down
syndrome and cerebral
palsy)
Long-term (others)
Other
Incremental Costs
Developmental Costs
fc — proportion of the
affected population
p — proportion receiving
special ed
PCFC — per-capita
special education costs
CALPROP-the distribution
of students
Source
CBDMP and incidence
estimate (see above)
NHIS2
OSHPD3 (adjusted
using MediCal and
cost-to-charge ratios)
DOS file 4
MediCal
Shriners' hospitals
(add'l data not
reported in OSHPD or
MediCal)
NMCES5
DOS file
SRI6
NHIS
CA State Department
of Education
CA Department of
Education
Data Limitations
1. Applied 1983-86 estimates to 1988 —
assumes that these numbers are similar.
2. Underestimates # of children who actually
used services, i.e, if a child died before the one
year cut-off date they would not be included, and
yet they may have received care.
3. Assumed first-year mortality was equal
among males and females.
4. Net migration not accounted for — may
underestimate prevalence if families migrate to
CA b/c of the availability of specialized care.
1. Assumes no net migration and no mortality
over time. These two may actually cancel each
other out.
2. Assumes that the prevalence of each defect
has been constant overtime.
1. Cost-to-charge ratio may be too drastic.
1. Assumed that all costs were borne by
Medical.
2. Assumed that expenditures for such care
were at cost.
1. These data may include persons with defects
and therefore might be slightly overestimated,
making incremental costs slightly
underestimated.
1 . Unlikely that all underlying etiologies were
recorded — may have underestimated # of
people w/ birth defects receiving DOS services.
1. Assumed the national distribution applied to
CA.
1. May underestimate proportion receiving
special ed.
1 California Birth Defects Monitoring Program
2 National Health Interview Survey
3 1988 California Office of Statewide Health Planning and Development hospital discharge file
4 California Department of Developmental Services Masterfile, 1988-89
5 National Medical Care Expenditure Survey, 1987
6 National Longitudinal Transition Study of Special Education Students, 1990
Chapter III.3
1.3-13
Cost of Cleft Lip and Palate
-------
.3.B.2 Results
III.3.B.2.1 Annual Direct Medical Costs
Waitzman et al.'s (1996) estimates of the total lifetime medical costs of
cleft lip and palate are outlined in the following tables. They are updated
from 1988 to 1996 dollars based on the medical care cost component of
the Consumer Price Index (1996:1988=1.6465). Table III.3-3 shows
annual per capita medical costs incurred by individuals with cleft lip and
palate by age group.
Table III. 3-3: Annual Per-Capita Medical Costs of Cleft Lip and Palate by Age Group (1996$)
Condition
Cleft lip and palate
Age 0-1
$11,186
Age 2-4
$2,025
Age 5-1 7
$1,510
Age 18+
$1,421
III.3.B.2.2 Incremental Lifetime Direct Medical Costs
The medical cost of the average population was then subtracted from these
costs to obtain incremental costs. Waitzman et al. (1996) discounted these
costs using three different discount rates: two percent, five percent and ten
percent. Although these discount rates do not match the standard EPA
rates used in many other chapters in this handbook (zero percent, three
percent, five percent and seven percent), there is insufficient information
provided in Waitzman et al. (1996) to allow a conversion to discounted
costs using standard EPA discount rates. This problem exists in all
chapters based on the Waitzman et al. data (Chapters III.3 through III.8).
The present discounted values of average per capita lifetime incremental
costs, using discount rates of two percent, five percent, and seven percent,
are listed in Table III.3-4 below.
Table III. 3-4: Per-Capita Incremental Medical Costs, Nonmedical Costs, and Total Costs of
Cleft Lip and Palate (1996$)
Cost Element
Direct Medical Costs
0%
2%
$19,758
5%
$18,111
10%
$16,465
Direct Non-Medical Costs:
Developmental
Services
Special Education
Total Direct Costs
$688
$5,218
$25,664
$649
$3,866
$22,626
$589
$2,453
$19,507
The costs presented in this chapter were current in the year the chapter was written. They can be updated using
inflation factors accessible by clicking below.
Link to inflation factors
Chapter III.3
1.3-14
Cost of Cleft Lip and Palate
-------
III.3.B.2.3 Other Costs
Categories of costs that are not usually included in this handbook were
available from Waitzman et al. and so are reported in Table III.3-4. For
information on their estimation methods see Waitzman et al. (1996).
III.3.B.2.4 Limitations
Reliance on the Waitzman et al. (1996) study has a number of limitations.
The first concerns the reliance solely on California data. Medical practices
and costs vary across the country, and an ideal data set would contain cost
information from a representative sample of the entire United States. As a
large state with varied areas (urban, rural, suburban, high, moderate, and
low income populations), California is likely to reflect some of the diversity
seen throughout the country. The California data are also especially
accurate because of the birth defects monitoring program that the state has
in place.
Another limitation involves the basis of the Waitzman et al. cost estimates.
Waitzman et al. base their estimates of costs on actual costs incurred. To
the extent that services were not readily available to or obtained by parents
of children with birth defects in their sample, the Waitzman et al. estimates
may understate the costs necessary for children to receive the level of care
considered by doctors to be adequate.
What is considered "adequate" may, of course, vary from one individual to
another (just as the willingness to pay to avoid the occurrence of a birth
defect may vary from one individual to another). An alternative approach
would be to delineate the collection of medical services and treatments that
constitute "proper care" for a given birth defect or illness, as defined by
consensus among doctors, and estimate the costs associated with that
"bundle" of services and treatments. This theoretical cost methodology has
the advantage of avoiding the possibility that the cost estimates may reflect
the unwillingness or inability of parents to pay for adequate medical care
rather than the cost of the care itself. These "ideal" costs are the costs of
the care that, according to doctors, should be received. They are thus
likely to be closer to the benefits to society of avoiding an illness. Actual
costs, on the other hand, reflect the care that actually is received, and are
therefore presumably a more accurate representation of what occurs.
The degree to which actual costs understate "ideal" costs will depend on
the illness. Children with a life-threatening heart anomaly (See Chapter
III.5) will typically be diagnosed properly at birth and require treatment,
whereas children with Down syndrome may receive varying degrees of
services to address their needs, depending on parental and other social
factors and economic resources.
Chapter III.3 III.3-15 Cost of Cleft Lip and Palate
-------
Waitzman et al. point out other limitations of their study, several of which
are likely to lead to underestimates of cost. For example, they
underestimate developmental services costs because the Department of
Developmental Services file included only public expenditures, not private,
out-of-pocket spending. In addition, the databases do not necessarily
include complete longitudinal profiles of costs for all individuals because it
was not always possible to link an individual's files across the entire
lifespan. This omission would also result in underestimates of total lifetime
costs. Finally, costs could be either under- or overestimated due to
changes in technologies. Because the study attempts to estimate lifetime
costs, changes in medical care technologies or policies could be important.
These changes could either increase or decrease costs.
A final limitation warrants mention. Only select costs are presented in this
chapter. There are other important categories of costs that are not
included here. For example, a person born with a birth defect often has a
higher probability of early death (see Section III.3.A.5, above); the cost of
premature mortality is a real cost associated with many birth defects and
illnesses. Individuals with illnesses or birth defects often experience a
decrease in productivity, resulting in a loss to society of goods and
services. In addition, there are the indirect costs of the pain and suffering
of the individual and of family members, as well as the lost productivity of
family members. These costs can be considerable (e.g., the direct cost of
lost work time calculated for spina bifida by Waitzman et al. was $656,879
in 1996 dollars). These costs, while not presented for most illnesses in this
handbook, are valid components of the total economic costs associated
with the illnesses and birth defects discussed in this handbook.
Link to Section III 3. A. 5
Chapter III.3 III.3-16 Cost of Cleft Lip and Palate
-------
CHAPTER III.4. COST OF LIMB REDUCTIONS
Clicking on the sections below will take you to the relevant text.
III.4.A Background
III.4.A.1 Description
III.4. A.2 Concurrent Effects
III. 4. A. 3 Causality
III.4. A.4 Treatment and Services
III. 4. A. 5 Prognosis
III.4.B Costs of Treatment and Services
III.4.B.1 Methodology
III.4.B.2 Results
Chapter III.4 III.4-1 Cost of Limb Reductions
-------
CHAPTER III.4. COST OF LIMB REDUCTIONS
I.4.A Background
This chapter contains a discussion of the methods used and the results of
estimating the direct medical costs incurred by individuals with limb
reductions, and the results of the analysis.1 It does not include information
on elements such as indirect medical costs, pain and suffering, lost time of
unpaid caregivers, etc. The reader is referred to Chapter I.I for a
discussion of the cost estimation methods and cost elements that are
relevant to all benefits estimates. In addition, Chapter III. 1 contains
information regarding the special characteristics of developmental defects,
and a list of chemicals that may cause developmental abnormalities.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and III. 1
Link to inflation factors
I.4.A.1 Description
Limb reductions are the partial (meromelia) or complete (amelia) absence
of arms or legs. They vary with respect to the bones, muscles, and other
structures affected.
I.4.A.2 Concurrent Effects
.4.A.3 Causality
Children with limb reductions frequently have other birth defects. In 30 to
53 percent of affected children, other malformations are present, including
anomalies of the heart, kidney, anus, abdominal walls, esophagus,
vertebrae, and palate. Webbing between digits and spina bifida are also
associated with this defect (Waitzman et al., 1996).
Table III. 1-1 in Chapter III. 1 lists numerous chemicals associated with
developmental abnormalities in human and/or animal studies. Many of
these chemicals have caused structural and anatomical defects. Limb
reductions fall into this category of defects.
Link to Table III. 1-1 in Chapter III. 1
1 "Costs" in this chapter refer to direct incremental per capita medical costs, unless otherwise
noted.
Chapter III.4 III.4-2 Cost of Limb Reductions
-------
A recently completed study of the offspring of pesticide application
workers found that their incidence of skeletal anomalies, which includes
limb reductions, was significantly greater than that of the general
population in the same area of the United States (Garry et al., 1996).
Respiratory system, circulatory system, and urogenital anomalies also
occurred at increased rates. Musculoskeletal anomalies include any
abnormality in the size, shape, or function of part of the skeletal system,
muscles, and related tissues (e.g., cartilage). They include the absence or
shortening of limbs (as discussed in this chapter) and the abnormal
formation of part of the skeleton or related soft tissue and cartilage, (as
discussed in Chapter III.3). The chemicals evaluated in the study that were
associated with the birth defects were trifluralin, triazine herbicides
(including atrazine, a very common well contaminant in agricultural areas),
and chlorophenoxy herbicides (including MCPA and 2,4-D, a pesticide
with very high usage). There was also a significant increase in birth
defects among infants conceived in the spring (i.e., during peak
chlorophenoxy use) compared to infants conceived during other periods of
the year (Garry et al., 1996). As indicated in the introductory
developmental effects chapter (III.l), the timing of exposure is often a
critical determinant of whether and what type of effects will occur.
Links to Chapters III. 3 and III.l
Generally, the earlier during a pregnancy that damage occurs, the more
serious the effect will be, because all cells developing from the damaged
cells may also be damaged or eliminated, and basic structures are being
formed during the first trimester (three months). Although Garry et al.
(1996) does not provide detailed information on the nature of the skeletal
and other anomalies, information can be obtained from the author or from
EPA (which funded the work).
III.4.A.4 Treatment and Services
Individuals with limb reductions are treated using a variety of surgical
techniques, are fitted with prostheses, and usually require physical and/or
occupational therapy (Mason, 1991). Limb reductions may occur at any
point from the joint which attaches the limb to the body (i.e., shoulder or
hip) to the distal point of that limb, and may affect some or all of the bones,
muscles, cartilage, and other structures comprising the limbs. The
structures involved and the types of surgical approaches used are too
diverse and numerous to describe in this chapter. Depending on the nature
of the limb reduction, this type of birth defect may require multiple
complex surgical corrections. Following the initial surgical and related
treatments, prostheses are usually replaced each year up to five years.
They are replaced biennially after that, up to twelve years, and every two to
five years after that (Waitzman et al., 1996).
Chapter III.4 III.4-3 Cost of Limb Reductions
-------
.4.A.5 Prognosis
Twelve to twenty percent of infants with limb reductions die during
infancy, primarily due to other anomalies (as noted above) (Froster-
Iskenius and Baird, 1990). Individuals with limb reductions who have
received appropriate medical treatment usually have good functional
capabilities (in the absence of other unrelated serious medical problems),
although their physical actions may be slowed by their disabilities. There
are significant psychosocial implications of these permanent disabilities;
these may require medical or other treatment (Waitzman et al., 1996).
I.4.B Costs of Treatment and Services
I.4.B.1 Methodology
Chapters III.3 through III. 8 of this handbook use cost of illness estimates
developed primarily by Waitzman et al. (1996). Waitzman et al. used the
same methodology to estimate the costs incurred by individuals with limb
reductions as for all the birth defects for which they estimated costs. The
methodology and relevant considerations are detailed in Chapter III.3,
including discussions of direct and indirect costs, prevalence versus
incidence, incremental costs, and concurrent effects. The analytic method,
the sources of data, and the limitations of the Waitzman et al. method are
also discussed in Chapter III.3. The methodology is outlined briefly here.
Link to Chapter III. 3.
To estimate the lifetime medical costs incurred by an individual with a birth
defect, Waitzman et al. estimated the average lifetime medical costs for an
individual with the birth defect. From this value, the authors subtracted the
average lifetime medical costs for an individual without the birth defect.
This yielded the incremental costs associated with the birth defect.
Because they estimated lifetime costs, they used an incidence-based
approach. Ideally, they would have tracked the costs of the cohort
members over time, until the death of the last cohort member. Because the
members of the cohort were born in 1988, however, this tracking was not
possible. Instead, estimates of the costs incurred at each age were based
on estimates of per capita costs in the prevalent population of that age (see
Chapter III.3, Section III.3.B.1.2).
Link to Chapter III. 3, Section III. 3.B. 1.2
This method has two important implications. First, Waitzman et al.
estimated the costs incurred by individuals with birth defects, including all
medical costs incurred, rather than the cost of the birth defect per se.
Chapter III.4 III.4-4 Cost of Limb Reductions
-------
These cost estimates therefore include the costs of concurrent effects
(unlike the costs reported for many of the diseases in this handbook). This
method yields a more comprehensive assessment of total costs than would
be obtained if only individual effects were evaluated. This method is of
particular use in valuing the avoidance of birth defects because they very
frequently occur in clusters within an individual. As Waitzman et al. note,
however, the costs of associated anomalies are included as part of the
estimate of the costs incurred by an individual with a given birth defect.
These cost estimates therefore cannot be aggregated across birth defects
because of the possibility of double counting.
Second, the Waitzman et al. method estimates the incremental costs for
individuals with birth defects — that is, the costs above and beyond the
average costs that would be incurred by individuals without the birth
defect.
Waitzman et al. (1996) estimated three categories of costs incurred by
individuals with limb reductions: direct medical costs, direct nonmedical
costs, and indirect costs.2 Direct medical costs, specifically inpatient care,
outpatient care, pharmaceuticals, laboratory tests, X-rays, appliances, and
long-term care are included in the cost estimates shown in this and other
chapters (Chapters III.3 through III. 8) based on the work of Waitzman et
al. Nonmedical direct costs, specifically developmental services, and
special education are also included in this handbook.
The Waitzman estimates of the costs incurred by individuals with limb
reductions are based on the costs of this birth defect in California across
many ages, and its occurrence in a large cohort of children born in
California in 1988. California's ongoing birth defects monitoring program
provides an excellent source of data. The California data sets were linked
with other national data sets so that Waitzman et al. could estimate the
incremental costs associated with limb reductions.
The method of calculating the expected lifetime incremental costs for an
individual with a birth defect — i.e., the average lifetime cost per case —
is the same for all the birth defects considered by Waitzman et al. The
expected per capita cost at age /', PCQ, for an individual born with the birth
defect is the probability of surviving to age / (among those individuals born
with the birth defect), ps;, times the per capita cost among individuals who
do survive to age / (PCPREV;, measured in the prevalent population):
PCQ = (ps;) x (PCPREV;) .
2 Indirect costs are not generally discussed in this handbook and so are not included in this chapter.
The reader may wish to consult Waitzman et al. (1996) for information on these costs.
Chapter III.4 III.4-5 Cost of Limb Reductions
-------
Link to Chapter III. 3
I.4.B.2 Results
Waitzman et al. estimate per capita costs in the prevalent population of age
/', PCPREV;, in two different ways, depending on data availability (see
Chapter III.3).
The present discounted value of expected per capita lifetime costs of the
birth defect, PCCOBD, is just the sum of these expected age-specific per
capita costs, appropriately discounted (as explained more fully in Chapter
III.3):
PCCOBD =£; PCC/O+ry .
Waitzman et al. (1996) estimate the medical costs of two different groups
of limb reductions: upper limb reductions and lower limb reductions. The
following tables outline the various costs associated with both types. They
are updated from 1988 to 1996 dollars based on the medical care cost
component of the Consumer Price Index (1996:1988=1.6465).
Table III.4-1 shows annual per capita medical costs incurred by individuals
with limb reductions by age group. Upper limb reductions tend to be
correctable at birth, and thus the costs decrease over time. The costs of
lower limb reductions, on the other hand, tend to increase with time
beyond the initial cash outlay, because lower limb reductions are less
correctable and require ongoing medical services.
Table 111.4-1: Annual Per-Capita Medical Costs of Limb Reductions by Age Group (1996$)
Condition
Upper Limb Reduction
Lower Limb Reduction
Age 0-1
$8,985
$11,888
Age 2-4
$659
$1,215
Age 5-1 7
$456
$1,786
Age 18+
$280
$3,492
The medical cost of the average population was then subtracted from these
costs to obtain incremental costs. Waitzman et al. (1996) discounted these
costs using three different discount rates: two percent, five percent, and ten
percent. Although these discount rates do not match the standard EPA
rates used in many other chapters in this handbook (zero percent, three
percent, five percent, and seven percent), there is insufficient information
provided in Waitzman et al. (1996) to allow a conversion to discounted
costs using standard EPA discount rates. This problem exists in all
chapters based on the Waitzman et al. data (i.e., Chapters III.3 through
III. 8).
Chapter 111.4
1.4-6
Cost of Limb Reductions
-------
The present discounted values of average per capita lifetime incremental
costs, using discount rates of two percent, five percent, and seven percent,
are listed in Table III.4-2 below. Direct medical costs and direct non-
medical costs are listed separately. The sum of per-capita direct medical
and nonmedical costs provides an estimate of the total per-capita costs
incurred by individuals with limb reductions.
Table III. 4-2: Per-Capita Net Medical Costs, Nonmedical Costs, and Total Costs of Limb
Reductions (1996$)
Cost Element
2%
5%
10%
Upper Limb Reduction
Net medical costs
Net nonmedical costs
Total costs
$8,232
$28,372
$36,604
$8,232
$21,574
$29,806
$8,232
$14,342
$22,574
Lower Limb Reduction
Net medical costs
Net nonmedical costs
Total costs
$39,515
$28,332
$67,847
$26,343
$21,537
$47,880
$18,111
$14,311
$32,422
The costs presented in this chapter were current in the year the chapter was written. They can be updated using
inflation factors accessible by clicking below.
Link to inflation factors
The costs associated with lower limb reductions are significantly higher
than those associated with upper limb reductions. This difference is due to
the weight bearing issues associated with lower limb reductions that require
more extensive surgical therapy. Because the legs are necessary for proper
balance, lower limb reductions often produce more complications, and thus
require more surgery than reductions in the arms.
Chapter III.4
I.4-7
Cost of Limb Reductions
-------
CHAPTER III.5. COST OF CARDIAC ABNORMALITIES
Clicking on the sections below will take you to the relevant text.
III. 5. A Background
III. 5. A.I Description
III. 5. A. 2 Causality
III.5.A.3 Truncus Arteriosus
III.5.A.4 Transposition of the Great Arteries
III.5.A.5 Double-outlet Right Ventricle
III.5.A.6 Single Ventricle
III.5.A.7 Tetralogy of Fallot
III.5.B Costs of Treatment and Services
III. 5.B.I Methodology
III.5.B.2 Results
III.5.B.3 Other Studies
III.5.C Conclusions
Chapter III.5 III.5-1 Cost of Cardiac Abnormalities
-------
CHAPTER III.5. COST OF CARDIAC ABNORMALITIES
III.5.A Background
This chapter contains a discussion of the methods used and the results of
estimating the direct medical costs incurred by individuals with cardiac
abnormalities and the results of the analysis.1 It does not include
information on elements such as indirect medical costs, pain and suffering,
lost time of unpaid caregivers, etc. The reader is referred to Chapter I.I
for a discussion of the cost estimation methods and cost elements that are
relevant to all benefits estimates. In addition, Chapter III. 1 contains
information regarding the special characteristics of developmental defects,
and a list of chemicals that may cause developmental abnormalities.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and III. 1
Link to inflation factors
I.5.A.1 Description
A number of cardiac anomalies occur at birth or in early infancy, and are
quite varied. These are structural defects in the development of the heart,
arteries, and associated tissues. They arise when the function, movement
and relationships among cardiac cells fail to progress normally. Five
defects, all conotruncal heart anomalies, are discussed in this chapter. The
specific anomalies include: truncus arteriosus, transposition of the great
arteries, double-outlet right ventricle (DORV), single ventricle, and
tetralogy of Fallot. They were selected from among the cardiac
abnormalities for which cost data are available because they occur with
greater frequency than other defects and are not usually fatal when treated.
Each anomaly is described below in turn in Section A, followed by the cost
data in Section B.2
1 "Costs" in this chapter refers to direct incremental per capita medical costs, unless otherwise
noted.
2 Certain forms of DORV are grouped together for the cost analysis because they resemble
transposition and actually fall under the same International Classification of Diseases ICD-9 codes.
Chapter III.5 III.5-2 Cost of Cardiac Abnormalities
-------
.5.A.2 Causality
Information on causality is discussed for each specific cardiac anomaly
described below. Many anomalies occur concurrently with other cardiac or
non-cardiac anomalies. Although there appears to be a genetic component
in some anomalies, none of the anomalies described below has a strong
clear hereditary pattern. The causation of the anomalies is therefore for the
most part unknown. These anomalies may be due to a variety of factors,
including maternal health, heredity, and environmental factors.
A recently completed study of the offspring of pesticide application
workers found that their incidence of circulatory system anomalies was
significantly greater than that of the general population in the same area of
the United States (Garryet al., 1996). Respiratory system, musculoskeletal,
and urogenital anomalies also occurred at increased rates. The chemicals
evaluated in the study that were associated with the birth defects were
trifluralin, triazine herbicides (including atrazine, a very common well
contaminant in agricultural areas), and chlorophenoxy herbicides (including
MCPA and 2,4-D, a pesticide with very high usage). There was also a
significant increase in birth defects among infants conceived in the spring
(i.e., during peak chlorophenoxy use), compared to infants conceived
during other periods of the year (Garry et al., 1996). As indicated in the
introductory developmental effects chapter (III.l), the timing of exposure
is often a critical determinant of whether and what type of effects will
occur.
Generally, the earlier during a pregnancy that damage occurs, the more
serious the effect will be, because all cells developing from the damaged
cells may also be damaged or eliminated, and basic structures are being
formed during the first trimester (three months). Although Gerry et al.
(1996) do not provide detailed information on the nature of the circulatory
and other anomalies, information can be obtained from the author or from
EPA (which funded the work).
Table III. 1-1 in Chapter III.l lists numerous chemicals associated with
developmental abnormalities in human and/or animal studies. Many of
these studies have identified structural and anatomical defects. Cardiac
abnormalities fall into this category of defects.
Link to Table III. 1-1 in Chapter III. 1
Chapter III.5 III.5-3 Cost of Cardiac Abnormalities
-------
.5.A.3 Truncus Arteriosus
III.5.A.3.1 Description
Truncus arteriosus occurs when abnormalities in major arteries and valves
lead to a mixing of oxygenated and de-oxygenated blood. This mixing
results in an insufficiently oxygenated supply of blood to tissues. This
disorder ultimately leads to congestive heart failure if uncorrected. In the
absence of surgery, 75 percent of children die within the first year, and the
survivors have severely limited abilities due to heart disease (Oldham et al.,
1972). Truncus arteriosus occurs in approximately two percent of children
with congenital heart defects (Oski, 1994).
III.5.A.3.2 Concurrent Effects
Children with this defect also frequently have underdevelopment of the
aortic arch, displacement or stenosis at the origin of the coronary arteries,
and absence of the main pulmonary artery supplying the lungs. This and
other cardiac anomalies occur with greater frequency among Down
syndrome children (statistics on occurrence were not available) (Waitzman
etal., 1996).
III.5.A.3.3 Treatment and Services
Although previously considered an inoperable defect, substantial
improvements in surgical and other treatments have been made in recent
years. The surgical methods and medical follow-up continue to be refined
(Oski, 1994). This defect is usually surgically treated at 6 to 12 weeks
through constructive surgery and insertion of valves. Follow-up surgical
repair or valve replacement is often required within the first five years after
surgery (15 to 50 percent). All valves are replaced at twelve years due to
growth. Follow-up surgery (i.e., dilations) is not uncommon (Waitzman et
al., 1996).
III.5.A.3.4 Prognosis
Based on a small sample (106 infants) there is a six percent mortality rate
prior to surgery and a ten percent mortality rate during the surgical period
(Ebert et al., 1984; Bove et al., 1989). Complications may result from
follow-up medical treatments (Waitzman et al., 1996).
.5.A.4 Transposition of the Great Arteries
III.5.A.4.1 Description
Transposition of the great arteries occurs when the major arteries are
transposed, leading to insufficient oxygen flow to tissues and lungs. This
condition causes widespread cyanosis (Oski, 1994) and is fatal without
surgical intervention (Waitzman et al., 1996). Transposition of the great
arteries is a common cardiac abnormality, occurring in approximately five
percent of all patients with congenital heart disease (Oski, 1994).
Chapter III.5 III.5-4 Cost of Cardiac Abnormalities
-------
III.5.A.4.2 Concurrent Effects
Various associated cardiac abnormalities involving multiple structures
within the heart complicate the treatment of this disorder (for a detailed
description see Oski, 1994). Ventricular septal defect also occurs in 20
percent of children with this anomaly (Kirklin, 1991).
III.5.A.4.3 Treatment and Services
Surgery to correct this defect is performed shortly after birth. Ten percent
of affected individuals require additional surgery within one year
(Waitzman et al., 1996). As noted under concurrent effects above,
concurrent heart defects often occur, and the constellation of effects is
considered when determining the appropriate medical treatment.
III.5.A.4.4 Prognosis
During the immediate post-surgical period, the mortality rate is
approximately 14 to 22 percent, depending on the specific nature of the
defects and procedures used. Between 30 and 80 percent of children with
this defect who have received medical treatment develop cardiac
arrhythmias within ten years, again depending on the defects and surgical
approaches. Three to ten percent of these children require insertions of
pacemakers (Waitzman et al., 1996).
Cardiac dysfunction occurs in approximately ten percent of children within
ten years. Arrhythmias and heart failure lead to death in approximately 11
percent of children between 30 days and 10 years after surgery (Morris and
Menashe, 1991). Depending on the type of procedures done to correct
this disorder, various medical problems may arise, some requiring surgery
at a later date. For a detailed description of the common long-term
sequelae, see Oski (1994).
.5.A.5 Double-outlet Right Ventricle
III.5.A.5.1 Description
Double-outlet right ventricle refers to a diverse group of heart defects that
occur when both the aorta and pulmonary artery originate from the right
ventricle, and a septal defect is present (Oski, 1994). This condition leads
to insufficient blood flow to critical tissues, and may cause cyanosis,
exhaustion with exercise, heart failure, and pulmonary vascular disease
(Waitzman et al., 1996). This disorder occurs in approximately two
percent of congenital heart defects. Although sometimes associated with
trisomy-18 and material diabetes, most cases occur with no other
congenital anomalies (Oski, 1994). The symptoms of this anomaly vary
with the specific structural defects and may include cyanosis, congestive
heart failure, and related effects.
Chapter III.5 III.5-5 Cost of Cardiac Abnormalities
-------
III.5.A.5.2 Concurrent Effects
Complex lesions of the atrial or ventricular septum, valve defects,
mislocation of the heart, and ductus arteriosus have also been observed
with this defect (Waitzman et al., 1996).
III.5.A.5.3 Treatment and Services
Surgery is usually performed at 6 to 24 months. Arrhythmias in children
are common, even with surgery, and may require follow-up care
(Waitzman et al., 1996).
III.5.A.5.4 Prognosis
Various mortality rates have been reported in the literature. A 15 percent
post-surgical mortality rate in infants was reported by Judson et al., (1983).
Shen et al. (1990) reported that approximately 25 percent of children who
have received appropriate medical attention die during childhood, probably
due to arrhythmias. More recent reports indicate a much better prognosis
with a 90 to 95 percent post-surgical survival rate and excellent functional
status following surgery (Oski, 1994).
III.5.A.6 Single Ventricle
III.5.A.6.1 Description
Single ventricle occurs when one, rather than the usual two ventricles are
present. One may dominate (they are usually balanced) and the other may
be very small, or there may be only one present. In either case, there is one
that is primarily functional (Oski, 1994). This anomaly causes insufficient
blood flow to tissues and can lead to cyanosis, heart failure, and pulmonary
vascular disease. There is a 70 percent mortality rate in childhood in the
absence of surgery (Waitzman et al., 1996). This disorder is found in
approximately one percent of children with congenital anomalies.
III.5.A.6.2 Concurrent Effects
Multiple concurrent cardiac defects occur with this anomaly. Twenty to 40
percent of children also have other non-cardiac problems, including
scoliosis and lack of a spleen (Waitzman et al., 1996).
III.5.A.6.3 Treatment and Services
Correction of this defect commonly involves two surgical interventions,
one shortly after birth, and one at 18 to 36 months. Additional surgery is
required in 14 percent of children (Waitzman et al., 1996).
III.5.A.6.4 Prognosis
With surgery, survival varies considerably. The mortality rate is 29 to 43
percent within ten years. Among survivors, ten percent have limited
activity and three percent were severely limited (Waitzman et al., 1996).
Most patients exhibit exercise intolerance and cyanosis. Causes of death
include dysrhythmia, congestive heart failure, brain abscess, pancreatitis,
Chapter III.5 III.5-6 Cost of Cardiac Abnormalities
-------
cerebral infarction, cerebral embolus, and hemorrhage, and pulmonary
embolus and valve occlusion (Oski, 1994). Due to the long-term medical
problems associated with this anomaly, research into improvements in
surgical and other medical treatments is being carried out. There is not a
single currently accepted treatment method at this time, and various
approaches are used. (Oski, 1994).
III.5.A.7 Tetrology of Fallot
III.5.A.7.1 Description
Tetrology of Fallot refers to a group of abnormalities of the heart that
have the common characteristics of unrestrictive ventricular septal defects
and an obstruction of the right ventricular outflow (Oski, 1994). These
abnormalities also involve malpositioning of the aorta and thickening of the
right ventricular wall (Pinsky and Arciniegas, 1990). The severity varies
considerably from a heart murmur to life-threatening hypoxia. If it is not
detected during infancy it may lead to a misshapen (boot-shaped) heart
(Oski, 1994). This disorder occurs in approximately six percent of
children with congenital heart defects (Oski, 1994).
III.5.A.7.2 Concurrent Effects
Multiple additional cardiac defects are associated with this anomaly
Waitzman et al., 1996). Non-cardiac abnormalities are associated with
this disorder in approximately 16 percent of cases, more so than with most
other cardiac defects. The children are also more likely to have concurrent
effects that are more serious than those found with other cardiac defects,
including cleft lip and palate, hypospadias (reproductive organ abnormality
in males), and skeletal malformations (Oski, 1994).
III.5.A.7.3 Treatment and Services
When Tetralogy of Fallot is detected shortly after birth, surgery to correct
the defects usually occurs at three to twelve months of age. Depending on
the specific nature of the defects, surgery may also be required shortly after
birth. Five to fifteen percent of children require additional surgery within
thirteen years (Waitzman et al., 1996). Treatment of arrhythmias may be
required in some patients long after surgery, and bacterial endocarditis may
also occur and require treatment later in childhood or adulthood (Oski,
1994).
III.5.A.7.4 Prognosis
The rate of post-surgical mortality is three to eight percent in infants
(Touati et al., 1990; Walsh et al., 1988). At ten years the survival rate is
87 to 90 percent, and 85 percent survive into their twenties. The prognosis
for normal function and activity is good in surviving children (Waitzman et
al., 1996). As noted under the treatment section above, a number of
problems may be anticipated in the long-term follow-up of children with
this disorder (e.g., bacterial endocarditis, arrhythmias) and effects of
Chapter III.5 III.5-7 Cost of Cardiac Abnormalities
-------
coronary artery disease at a more advance age may be more severe in
people with this disorder (Oski, 1994).
I.5.B Costs of Treatment and Services
I.5.B.1 Methodology
Link to Chapter III. 3
Waitzman et al. (1996) provide an estimate of the direct medical and non-
medical costs of treating cardiac abnormalities, specifically for the five
types listed above. For the purpose of the cost analysis, transposition and
double-outlet right ventricle (DORV) were grouped together. Waitzman et
al. used the same methodology to estimate the costs incurred by individuals
with each type of cardiac abnormality as for all the birth defects for which
they estimated costs. The methodology and relevant considerations are
detailed in Chapter III.3, including discussions of direct and indirect costs,
prevalence versus incidence, incremental costs, and concurrent effects.
The analytic method, the sources of data, and the limitations of the
Waitzman method are also discussed in Chapter III.3. The methodology is
outlined briefly here.
To estimate the lifetime medical costs incurred by an individual with a birth
defect, Waitzman et al. estimated the average lifetime medical costs for an
individual with the birth defect. From this value, the authors subtracted the
average lifetime medical costs for an individual without the birth defect.
Because they estimated lifetime costs, they used an incidence-based
approach. Ideally, they would have tracked the costs of the cohort
members over time, until the death of the last cohort member. Because the
members of the cohort were born in 1988, however, this tracking was not
possible. Instead, estimates of the costs incurred at each age were based
on estimates of per capita costs in the prevalent population of that age (see
Chapter III.3, Section III.3.B.1.2).
Link to Chapter HI. 3, Section III.3.B. 1.2
This method has two important implications. First, Waitzman et al.
estimated the costs incurred by individuals with birth defects, including all
medical costs incurred, rather than the cost of the birth defect per se.
These cost estimates therefore include the costs of concurrent effects
(unlike the costs reported for many of the diseases in this handbook). This
method yields a more comprehensive assessment of total costs than would
be obtained if only individual effects were evaluated. This method is of
particular use in valuing the avoidance of birth defects because they very
frequently occur in clusters within an individual. As Waitzman et al. note,
however, the costs of associated anomalies are included as part of the
Chapter III.5 III.5-8 Cost of Cardiac Abnormalities
-------
Link to Chapter III. 3
estimate of the costs incurred by an individual with a given birth defect.
These cost estimates therefore cannot be aggregated across birth defects
because of the possibility of double counting.
Second, the Waitzman et al. method estimates the incremental costs for
individuals with birth defects — that is, the costs above and beyond the
average costs that would be incurred by individuals without the birth
defect.
Waitzman et al. (1996) estimated three categories of costs incurred by
individuals with limb reductions: direct medical costs, direct nonmedical
costs, and indirect costs.3 Direct medical costs, specifically inpatient care,
outpatient care, pharmaceuticals, laboratory tests, X-rays, appliances, and
long-term care are included in the cost estimates shown in this and other
chapters (Chapters III.3 through III. 8) based on the work of Waitzman et
al. Nonmedical direct costs, specifically developmental services, and
special education are also included in this handbook.
The Waitzman estimates of the costs incurred by individuals with limb
reductions are based on the costs of this birth defect in California across
many ages, and its occurrence in a large cohort of children born in
California in 1988. California's ongoing birth defects monitoring program
provides an excellent source of data. The California data sets were linked
with other national data sets so that Waitzman et al. could estimate the
incremental costs associated with each type of cardiac abnormality.
The method of calculating the expected lifetime incremental costs for an
individual with a birth defect — i.e., the average lifetime cost per case —
is the same for all the birth defects considered by Waitzman et al. The
expected per capita cost at age /', PCQ, for an individual born with the birth
defect is the probability of surviving to age /' (among those individuals born
with the birth defect), ps;, times the per capita cost among individuals who
do survive to age /' (PCPREV;, measured in the prevalent population):
PCQ = (ps;) x (PCPREV;) .
Waitzman et al. estimate per capita costs in the prevalent population of age
/', PCPREV;, in two different ways, depending on data availability (see
Chapter III.3).
The present discounted value of expected per capita lifetime costs of the
birth defect, PCCOBD, is just the sum of these expected age-specific per
3 Indirect costs are not generally discussed in this handbook and so are not included in this chapter.
The reader may wish to consult Waitzman et al. (1996) for information on these costs.
Chapter III.5 III.5-9 Cost of Cardiac Abnormalities
-------
I.5.B.2 Results
capita costs, appropriately discounted (as explained more fully in Chapter
III.3):
PCCOBD =£; PCC/O+ry .
Waitzman et al (1996) estimate the total lifetime medical costs of each of
the cardiac defects according to the methodology outlined above. As
outlined in Table III.3-1 in Chapter III.3, the prevalence on July 1, 1988
and the incidence in 1988 of cardiac abnormalities in the state of California
was as follows:truncus arteriosus: 1,591 and 56, respectively;
transposedon/DORV: 7,469 and 263, respectively; tetralogy of Fallot:
5,336 and 187, respectively; and single ventricle: 1,932 and 68,
respectively. The occurrence data provide an indication of the relative rate
of yearly occurrence of the different anomalies discussed in this handbook.
Table III 3-1 in Chapter III. 3
Table IE.5-1 shows the annual per capita medical costs incurred by
individuals with each type of cardiac anomaly by age group with costs
updated from 1988 to 1996 dollars based on the medical care cost
component of the Consumer Price Index (1996:1988=1.6465).
Table III. 5-1: Annual Per-Capita Medical Costs of Heart Defects Patients by Age Group (1996$)
Condition
Truncus arteriosus
Transposition/DORV
Tetralogy of Fallot
Single ventricle
Age 0-1
$162,463
$56,296
$70,729
$57,119
Age 2-4
$104,225
$15,790
$18,952
$15,727
Age 5-1 7
$7,197
$3,071
$5,939
$12,867
Age 18+
$5,139
$1,744
$1,251
$10,071
The medical cost of the average population was then subtracted from these
costs to obtain incremental costs. Waitzman et al. (1996) discounted these
costs using three different discount rates: two percent, five percent, and ten
percent. Although these discount rates do not match the standard EPA
rates used in many other chapters in this handbook (zero percent, three
percent, five percent, and seven percent), there is insufficient information
provided in Waitzman et al. (1996) to allow a conversion to discounted
costs using standard EPA discount rates. This problem exists in all
chapters based on the Waitzman et al. data (i.e., Chapters III.3 through
III. 8).
Chapter 111.5
1.5-10
Cost of Cardiac Abnormalities
-------
The present discounted values of average per capita lifetime incremental
costs, using discount rates of two percent, five percent, and seven percent,
are listed in Table III. 5-2 below. Direct medical costs and direct non-
medical costs are listed separately. The sum of per-capita direct medical
and nonmedical costs provides an estimate of the total per-capita costs
incurred by individuals with each type of cardiac abnormality.
Children with some heart anomalies have a much lower probability of
survival (see the "prognosis" descriptions in part III.5.A, under the specific
descriptions above) than children with most cardiac or other anomalies
discussed in this section of the handbook. This report focuses on medical
costs, whereas other research often includes estimates of the value that a
person would pay to save his or her life, termed "value of life" estimates.
In the case of heart anomalies, survivorship is an issue for some of these
children. The cost estimates presented here could therefore be dwarfed by
the additional value of life associated with each estimate.
Chapter III.5 III.5-11 Cost of Cardiac Abnormalities
-------
Table 111.5-2: Per-Capita Net Medical Costs, Nonmedical Costs, and Total Costs of Heart
Defects (1996$)
Cost Element
2%
5%
10%
Truncus arteriosus
Net medical costs
Net nonmedical costs
Total costs
$375,394
$2,918
$378,312
$344,111
$2,228
$346,339
$316,121
$1,492
$317,613
Transposition/DORV
Net medical costs
Net nonmedical costs
Total costs
$120,192
$4,562
$124,754
$113,606
$3,479
$117,085
$107,020
$2,323
$109,343
Tetralogy of Fallot
Net medical costs
Net nonmedical costs
Total costs
$194,283
$5,800
$200,083
$179,465
$4,422
$183,887
$161,354
$2,954
$164,308
Single ventricle
Net medical costs
Net nonmedical costs
Total costs
$227,212
$3,507
$230,719
$163,000
$2,674
$165,674
$123,485
$1,786
$125,271
The costs presented in this chapter were current in the year the chapter was written. They can be updated using
inflation factors accessible by clicking below.
Link to inflation factors
.5.B.3 Other Studies
Several other studies have been conducted on the costs of cardiac
abnormalities. In particular, two recent studies are especially useful for
comparison. The first, by Pearson et al. (1991), looks at hospital use and
inpatient charges for cardiac disease patients in the first year of life. The
second, by Silberbach et al. (1993), predicts the hospital charges for
congenital heart disease surgery. These two studies are easiest to compare
to Waitzman et al. because they examine similar types of direct medical
costs. Although these studies vary slightly in their methodologies, the
results tend to corroborate the Waitzman et al. results. No recent studies
examining the direct non-medical costs associated with cardiac
abnormalities were identified.
Chapter 111.5
1.5-12
Cost of Cardiac Abnormalities
-------
III.5.B.3.1 Pearson et al. (1991)
The Pearson et al. study analyzes the inpatient charges for infants with
cardiac disease in the first year of life. Infants admitted to The Johns
Hopkins Hospital Children's Center in Maryland in 1988 were identified
for the study. Complete data for 93 of these infants were available. The
infants were subdivided into three groups: those with complex diseases,
those with extra-cardiac anomalies, and those with both. Hospital charges
were recorded, including all routine care charges, laboratory charges,
medical and surgical supplies, physical therapy, operating room time and
supplies, radiologic procedures, pharmacy supplies, and blood and blood
related products.
Pearson et al. estimated hospital charges per infant. These charges have
been adjusted to 1996 dollars using the Consumer Price Index for medical
care (1996:1988=1.6465) to facilitate comparison across studies. Average
hospital charges per infant were estimated at $60,506. The number is
larger for infants with complex cardiac diseases.
Pearson et al.'s results are similar, in the first year of life, to those reported
by Waitzman et al., particularly for transposition/DORV, tetralogy of Fallot
and single ventricle (see Table IE.5-1 for Waitzman et al.'s cost estimates).
Waitzman et al.'s higher cost estimate for truncus arteriosus may be
reflected in the upper range of costs that Pearson et al. estimated. They
identified a range of $1,651 to $770,177 per patient.
III.5.B.3.2 Silberbach et al. (1993)
The Silberbach et al. study is also useful for comparison because it looks at
the hospital charges for congenital heart disease surgery. Although the
study is not strictly limited to children, and is on a per-surgery basis, 49
percent of the surgeries examined occurred in children less than 12 months
old. The study is also useful because it breaks out costs according to ten
different types of congenital heart disease, facilitating direct comparison
with the Waitzman et al. study. The study was conducted between 1985
and 1989. A conservative adjustment to 1996 dollars assumes that all
reported costs in the Silberbach et al. study were in 1989 dollars
(Consumer Price Index for medical care (1996:1989=1.53)).
Silberbach et al. predicted the average hospital charge for a patient with
congenital heart disease at $41,699. More specifically, they estimated that
tetralogy of Fallot surgery costs $56,997 and surgery for
transposition/DORV of the great arteries costs $44,789.
The Silberbach et al. study found similar costs to those reported in
Waitzman et al., particularly in their finding that surgery for
transposition/DORV is less expensive than for tetralogy of Fallot. The
Silberbach et al. estimates are a bit lower. This difference in relation to
Waitzman et al. and Pearson et al. is probably due to the limited nature of
Chapter III.5 III.5-13 Cost of Cardiac Abnormalities
-------
the Silberbach et al. study; they calculate hospital charges for a specific
surgical procedure and do not include other costs that would be incurred
during the rest of the year.
.5.C Conclusions
Because the Waitzman et al. study provides cost estimates past the first
year of life, it is more appropriate for benefits evaluation than other studies
reviewed. The other studies, discussed above, generally support the
estimates provided by Waitzman et al.
Chapter III.5 III.5-14 Cost of Cardiac Abnormalities
-------
CHAPTER III.6. COST OF SPINA BIFIDA
Clicking on the sections below will take you to the relevant text.
III.6.A Background
III.6.A. 1 Description
III.6.A.2 Concurrent Effects
III.6.A.3 Causality
III.6.A.4 Treatment and Services
III. 6. A. 5 Prognosis
III.6.B Costs of Treatment and Services
III.6.B.1 Methodology
III.6.B.2 Results
III.6.B.3 Other Studies
III.6.C Conclusions
Chapter III.6 III.6-1 Cost of Spina Bifida
-------
CHAPTER III.6. COST OF SPINA BIFIDA
I.6.A Background
This chapter contains a discussion of the methods used and the results of
estimating the direct medical costs incurred by individuals with spina bifida
and the results of the analysis.1 It does not include information on elements
such as indirect medical costs, pain and suffering, lost time of unpaid
caregivers, etc. The reader is referred to Chapter I.I for a discussion of the
cost estimation methods and cost elements that are relevant to all benefits
estimates. In addition, Chapter III. 1 contains information regarding the
special characteristics of developmental defects, and a list of chemicals that
may cause developmental abnormalities.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and III. 1
Link to inflation factors
I.6.A.1 Description
Spina bifida occurs when the neural tube, from which the brain and spinal
cord develop (central nervous system), fails to close properly. Depending
on where closure fails to occur, portions of the brain, spinal cord, and
nerves connected to them will not function properly. If failure to close
occurs on the lower portion of the spinal cord, then the bowel, bladder or
sexual organs will be affected. Failure at mid-level may cause paralysis or
malfunction of the arms and legs. Anomalies at higher levels may affect the
brain. In most spina bifida cases, the normal flow of cerebrospinal fluid is
also blocked (Arnold-Chiari malformation), which would result in
hydrocephalus unless treated (Waitzman et al., 1996).
Some common disabilities and medical problems associated with spina
bifida are:
• sight problems, including atrophy of the optic nerve in 17 percent
of children and strabismus in 42 percent of children (Gaston, 1985);
• dysfunction in the arms in 45 percent of children (Turner, 1986);
1 "Costs" in this chapter refer to direct incremental per capita medical costs, unless otherwise
noted.
Chapter III.6 III.6-2 Cost of Spina Bifida
-------
epilepsy in 20 to 30 percent of children (Bartoshesky et al., 1985);
and
• bladder dysfunction in most children (McLone, 1983).
.6.A.2 Concurrent Effects
Approximately six percent of children with this anomaly have
malformations outside the central nervous system. These commonly affect
the diaphragm, esophagus, and kidneys. Cleft lip and palate are also
associated with spina bifida (Waitzman et al., 1996). Additional
malformations outside of the central nervous system include:
• pressure sores and skin problems in 70 percent of adolescents and
young adults (Blum et al., 1992);
curvature of the spine in 15 percent of children (Samuelsson and
Eklof, 1988); and
• weight substantially over the norm in 30 to 50 percent of children
(Thomas et al., 1987).
.6.A.3 Causality
Spina bifida is associated with prenatal exposure to sulfonamides and
antihistamines, maternal folate deficiency, and maternal diabetes.
Chemicals that interfere with the development of the neural crest and fold
during embryogenesis may cause spina bifida (Waitzman et al., 1996).
Table III. 1-1 in Chapter III.l lists numerous chemicals associated with
developmental abnormalities in human and/or animal studies.
Link to Chapter III. 1, Table III. 1-1
III.6.A.4 Treatment and Services
Treatment may begin before birth if diagnosis occurs during pregnancy.
Lab tests are now available that provide an indication of whether spina
bifida is likely. Ultrasound may be used to confirm the test results.
Cesarean delivery may be used to prevent damage to the infant. The spinal
canal defect is usually closed surgically shortly after birth. Approximately
90 percent of infants receive a ventriculaoperitoneal shunt to carry fluid
from the head to the abdominal cavity (to prevent the occurrence of
hydrocephalus noted above) (Waitzman et al., 1996).
Concurrent disorders are treated as needed, including dialysis or kidney
transplant for patients with severe kidney disease, and surgery to fuse the
vertebrae for patients with scoliosis (Waitzman, et al., 1996).
Chapter III.6 III.6-3 Cost of Spina Bifida
-------
.6.A.5 Prognosis
Spina bifida usually results in permanent disabilities of some type; these
may be severe (e.g., paralysis). Social isolation is common, and
employment prospects for adults are reduced. Even with appropriate
medical treatment, children with spina bifida have a shortened lifespan.
Ongoing medical treatment may be required, including replacement of
shunts and treatment of kidney and urinary problems (Waitzman et al.,
1996).
1.6.B Costs of Treatment and Services
I.6.B.1 Methodology
Link to Chapter III. 3
Chapters III.3 through III. 8 of this handbook use cost of illness estimates
developed primarily by Waitzman et al. (1996). Waitzman et al. used the
same methodology to estimate the costs incurred by individuals with spina
bifida as for all the birth defects for which they estimated costs. The
methodology and relevant considerations are detailed in Chapter III.3,
including discussions of direct and indirect costs, prevalence versus
incidence, incremental costs, and concurrent effects. The analytic method,
the sources of data, and the limitations of the Waitzman method are also
discussed in Chapter III.3. The methodology is outlined briefly here.
To estimate the lifetime medical costs incurred by an individual with a birth
defect, Waitzman et al. estimated the average lifetime medical costs for an
individual with the birth defect. From this value, the authors subtracted the
average lifetime medical costs for an individual without the birth defect.
Because they estimated lifetime costs, they used an incidence-based
approach. Ideally, they would have tracked the costs of the cohort
members over time, until the death of the last cohort member. Because the
members of the cohort were born in 1988, however, this tracking was not
possible. Instead, estimates of the costs incurred at each age were based
on estimates of per capita costs in the prevalent population of that age (see
Chapter III.3, Section III.3.B.1.2).
Link to Chapter III.3, Section III.3.B.I.2
This method has two important implications. First, Waitzman et al.
estimated the costs incurred by individuals with birth defects, including all
medical costs incurred, rather than the cost of the birth defect per se.
These cost estimates therefore include the costs of concurrent effects
(unlike the costs reported for many of the diseases in this handbook). This
Chapter III.6 III.6-4 Cost of Spina Bifida
-------
method yields a more comprehensive assessment of total costs than would
be obtained if only individual effects were evaluated. This method is of
particular use in valuing the avoidance of birth defects because they very
frequently occur in clusters within an individual. As Waitzman et al. note,
however, the costs of associated anomalies are included as part of the
estimate of the costs incurred by an individual with a given birth defect.
These cost estimates therefore cannot be aggregated across birth defects
because of the possibility of double counting.
Second, the Waitzman et al. method estimates the incremental costs for
individuals with birth defects — that is, the costs above and beyond the
average costs that would be incurred by individuals without the birth
defect.
Waitzman et al. (1996) estimated three categories of costs incurred by
individuals with limb reductions: direct medical costs, direct nonmedical
costs, and indirect costs.2 Direct medical costs, specifically inpatient care,
outpatient care, pharmaceuticals, laboratory tests, X-rays, appliances, and
long-term care are included in the cost estimates shown in this and other
chapters (Chapters III.3 through III. 8) based on the work of Waitzman et
al. Nonmedical direct costs, specifically developmental services, and
special education are also included in this handbook.
The Waitzman estimates of the costs incurred by individuals with limb
reductions are based on the costs of this birth defect in California across
many ages, and its occurrence in a large cohort of children born in
California in 1988. California's ongoing birth defects monitoring program
provides an excellent source of data. The California data sets were linked
with other national data sets so that Waitzman et al. could estimate the
incremental costs associated with spina bifida.
The method of calculating the expected lifetime incremental costs for an
individual with a birth defect — i.e., the average lifetime cost per case —
is the same for all the birth defects considered by Waitzman et al. The
expected per capita cost at age /', PCQ, for an individual born with the birth
defect is the probability of surviving to age /' (among those individuals born
with the birth defect), ps;, times the per capita cost among individuals who
do survive to age /' (PCPREV;, measured in the prevalent population):
PCQ = (ps;) x (PCPREV;) .
Waitzman et al. estimate per capita costs in the prevalent population of age
/', PCPREV;, in two different ways, depending on data availability (see
Chapter III.3).
2 Indirect costs are not generally discussed in this handbook and so are not included in this chapter.
The reader may wish to consult Waitzman et al. (1996) for information on these costs.
Chapter III.6 III.6-5 Cost of Spina Bifida
-------
Link to Chapter III. 3
I.6.B.2 Results
The present discounted value of expected per capita lifetime costs of the
birth defect, PCCOBD, is just the sum of these expected age-specific per
capita costs, appropriately discounted (as explained more fully in Chapter
III.3):
PCCOBD
Waitzman et al (1996) estimate the total lifetime medical costs incurred by
individuals with spina bifida according to the methodology outlined above.
The following tables outline the various costs, updated from 1988 to 1996
dollars based on the medical care cost component of the Consumer Price
Index (1996:1988=1.6465). Table III.6-1 shows the annual per capita
medical costs incurred by individuals with spina bifida by age group.
Table III. 6-1: Annual Per-Capita Medical Costs of Spina Bifida by Age Group (1996$)
Condition
Spina Bifida
Age 0-1
$34,013
Age 2-4
$14,924
Age 5-1 7
$13,208
Age 18+
$4,194
The medical cost of the average population was then subtracted from these
costs to obtain incremental costs. Waitzman et al. (1996) discounted these
costs using three different discount rates: two percent, five percent, and ten
percent. Although these discount rates do not match the standard EPA
rates used in many other chapters in this handbook (zero percent, three
percent, five percent, and seven percent), there is insufficient information
provided in Waitzman et al. (1996) to allow a conversion to discounted
costs using standard EPA discount rates. This problem exists in all
chapters based on the Waitzman et al. data (i.e., Chapters III.3 through
III. 8).
The present discounted values of average per capita lifetime incremental
costs, using discount rates of two percent, five percent, and seven percent,
are listed in Table III.6-2 below. Direct medical costs and direct non-
medical costs, including developmental services costs and special education
costs, are listed separately. The sum of per-capita direct medical and
nonmedical costs provides an estimate of the total per-capita costs incurred
by individuals with spina bifida.
Chapter III.6
1.6-6
Cost of Spina Bifida
-------
Table III. 6-2: Per-Capita Net Medical Costs, Nonmedical Costs, and Total Costs of Spina Bifida
(1996$)
Cost Element
Net direct medical costs
2%
$210,747
5%
$163,000
10%
$123,485
Net direct nonmedical costs
Developmental Costs
Special Education Costs
Total Costs
$2,694
$50,719
$264,160
$1,635
$38,298
$202,933
$1,004
$25,155
$149,644
The costs presented in this chapter were current in the year the chapter was written. They can be updated using
inflation factors accessible by clicking below.
Link to inflation factors
.6.B.3 Other Studies
Waitzman et al. present a study by Lipscomb (1986) that used an incidence
approach to estimate the total lifetime costs per individual of spina bifida
based on data gathered from individual reviews of clinical records in North
Carolina in 1985. Waitzman et al. adjust the Lipscomb estimates to
account for differences in medical care prices. They deflate their California
estimates by the difference in the Employee Compensation Index between
California and the nation. To account for differences in the base year they
adjusted the North Carolina estimates to 1988 dollars. While there is some
variability in the cost distribution across age groups, the total cost
estimates are strikingly similar; Waitzman et al. estimated total lifetime
costs per individual of $36,529 and Lipscomb estimated total costs of
$34,949 in the U.S. in 1988.
.6.C Conclusions
The results based on the Waitzman et al (1996) work are recommended for
use in benefits valuation. The Lipscomb work uses a smaller database and
older data. The similarity of the two independently researched results lends
credibility to the cost estimates.
Chapter III.6
1.6-7
Cost of Spina Bifida
-------
CHAPTER III.7. COST OF CEREBRAL PALSY
Clicking on the sections below will take you to the relevant text.
III.7.A Background
III.7.A. 1 Description
III.7.A.2 Concurrent Effects
III.7.A.3 Causality
III.7.A.4 Treatment and Services
III. 7. A. 5 Prognosis
III.7.B Costs of Treatment and Services
III.7.B.1 Methodology
III.7.B.2 Results
III.7.B.3 Other Studies
III.7.C Conclusions
Chapter III.7 III.7-1 Cost of Cerebral Palsy
-------
CHAPTER III.7. COST OF CEREBRAL PALSY
I.7.A Background
This chapter contains a discussion of the methods used and the results of
estimating the direct medical costs incurred by individuals with cerebral
palsy and the results of the analysis.1 It does not include information on
elements such as indirect medical costs, pain and suffering, lost time of
unpaid caregivers, etc. The reader is referred to Chapter I.I for a
discussion of the cost estimation methods and cost elements that are
relevant to all benefits estimates. In addition, Chapter III. 1 contains
information regarding the special characteristics of developmental defects,
and a list of chemicals that may cause developmental abnormalities.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and III. 1
Link to inflation factors
I.7.A.1 Description
Cerebral palsy is a motor disorder appearing in early childhood that is
caused by brain damage (Waitzman et al., 1996).2 It is the most common
movement disorder of childhood and affects approximately one to six
children per 1,000 births. The estimate varies considerably because mild
cases may not be determined in early childhood, and all cases may be
obscured by other developmental disabilities, such as seizures and mental
retardation. The most severe cases may result in rapid death and not be
detected. When estimates of the incidence of cerebral palsy are based on
evaluations in the neonatal period, the occurrence will be underestimated.
It is very difficult to identify cerebral palsy during this period by clinical
methods, due to the relative immaturity of the nervous system of newborn
infants (Oski, 1994).
Both muscle tone and the control of movement are affected in cerebral
palsy (Oski, 1994), which leads to a complex array of movement, posture,
and communication problems. Various types of cerebral palsy include
athetoid (15 percent of cases), spastic (70 percent), and ataxic (5 percent).
Approximately 15 percent of cases are mixed (Wollack et al., 1991).
1 "Costs" in this chapter refer to direct incremental per capita medical costs, unless otherwise
noted.
2 Cerebral palsy is often used synonymously with static encephalopathy (Oski, 1994).
Chapter III.7 III.7-2 Cost of Cerebral Palsy
-------
Cerebral palsy may affect one or more limbs. Athetoid cerebral palsy
patients exhibit uncoordinated and uncontrolled movements. Spastic
cerebral palsy patients exhibit exaggerated reflexes, increased muscle tone,
weakness, and joint contractures (spasms). Ataxic cerebral palsy patients
often have difficulty performing repetitive movements and have a wide-
based gait (Waitzman et al., 1996).
Individuals affected with this disorder are among the most handicapped in
our society. Although the incidence declined in the 1960s due to better
pre- and perinatal care, the incidence has increased in recent years due to
the improvement in survival among low birth weight infants. Cerebral palsy
usually originates in the pre- or perinatal period; however, it can be
brought on in childhood by infection, trauma, and other causes (Oski,
1994).
.7.A.2 Concurrent Effects
.7.A.3 Causality
Approximately 21 percent of children with cerebral palsy have
malformations unrelated to cerebral palsy; however, no specific occurrence
patterns have been reported for major malformations. Occurrences of hip
dislocation, curvature of the spine, and poorly developed dental enamel
have been frequently reported (Waitzman et al., 1996). About 50 percent
of children with cerebral palsy also have strabismus. Depending on the
nature of the cerebral palsy, children may have severe retardation, frequent
seizures, and blindness, or have a normal intellect with minimal non-motor
effects (Oski, 1994).
Other central nervous system (CNS) disorders are common in children with
cerebral palsy. Epilepsy occurs in 22 to 50 percent of children with
cerebral palsy, increasing with the number of limbs affected.
Approximately 25 percent of the children have severe or profound mental
retardation, with the degree of retardation generally related to the severity
of cerebral palsy. Most children with all limbs affected are severely
retarded. Other CNS disorders include severe hearing and visual
impairments and speech disorders (Waitzman et al., 1996).
Risk factors for cerebral palsy include the presence of other malformations,
maternal retardation, premature separation, low weight of the placenta,
breech birth, and low birth weight (Waitzman et al., 1996), although these
were more prominent causes in the past, when management of pregnancy
and delivery were less refined. Today most cerebral palsy occurs without
Chapter III.7 III.7-3 Cost of Cerebral Palsy
-------
identifiable risk factors (Oski, 1994). Table III. 1-1 in Chapter III.l lists
numerous chemicals associated with developmental abnormalities in human
and/or animal studies.
Link to Chapter III. 1, Table III. 1-1
III.7.A.4 Treatment and Services
Treatment of cerebral palsy focuses on improving the quality of life and
function of the individual. There is no "cure". Surgery on muscles, limbs,
and associated sites may be used to restore balance. Brain surgery may be
used to reduce spasticity, but may have undesirable consequences.
Appliances may include provision of wheelchairs, walkers, casts, or splints.
In some cases, medication is used (Waitzman et al., 1996).
Treatment objectives change as the individual ages (Oski, 1994). In early
childhood, improving communication abilities is often the focus because
these abilities are more closely related than motor function to long-term
outcome. In some cases, when speech is not possible, other supportive
materials are provided to facilitate communication. Other areas of
significant effort are development of the ability to perform basic daily living
activities and improving motor skills (Oski, 1994). Treatment of cerebral
palsy is rapidly changing and many options are usually offered, often
involving multiple medical and educational specialists. Treatment plans are
often complex and may require extensive non-medical care, including
considerable assistance in day-to-day living activities. These are not a part
of the direct medical costs, but they may contribute substantially to the
overall costs associated with this disorder.
.7.A.5 Prognosis
Among those with average or above intelligence (IQ > 80), the limitations
are not generally severe. Approximately 39 percent are able to function
independently as adults, and 48 percent are partially dependent. Of those
with lower intelligence, only 1 percent are independent and 17 percent are
partially dependent (Cohen and Kohn, 1979). The balance of both the
average and below-average intelligence groups are totally dependent.
Other issues related to this disorder are low self-esteem and difficulty with
sexual relationships. Among those with severe retardation (IQ < 50) there
is a mortality rate of 28 percent during the first 20 years of life. This is
compared to 2 percent among those with higher IQs (Hutton et al., 1994).
Chapter III.7 III.7-4 Cost of Cerebral Palsy
-------
1.7.B Costs of Treatment and Services
I.7.B.1 Methodology
Link to Chapter III. 3
Chapters III.3 through III. 8 of this handbook use cost of illness estimates
developed primarily by Waitzman et al. (1996). Waitzman et al. used the
same methodology to estimate the costs incurred by individuals with
cerebral palsy as for all the birth defects for which they estimated costs.
The methodology and relevant considerations are detailed in Chapter III.3,
including discussions of direct and indirect costs, prevalence versus
incidence, incremental costs, and concurrent effects. The analytic method,
the sources of data, and the limitations of the Waitzman et al. method are
also discussed in Chapter III.3. The methodology is outlined briefly here.
To estimate the lifetime medical costs incurred by an individual with a birth
defect, Waitzman et al. estimated the average lifetime medical costs for an
individual with the birth defect. From this value, the authors subtracted the
average lifetime medical costs for an individual without the birth defect.
Because they estimated lifetime costs, they used an incidence-based
approach. Ideally, they would have tracked the costs of the cohort
members over time, until the death of the last cohort member. Because the
members of the cohort were born in 1988, however, this tracking was not
possible. Instead, estimates of the costs incurred at each age were based
on estimates of per capita costs in the prevalent population of that age (see
Chapter III.3, Section III.3.B.1.2).
Link to Chapter III.3, Section III.3.B.I.2
This method has two important implications. First, Waitzman et al.
estimated the costs incurred by individuals with birth defects, including all
medical costs incurred, rather than the cost of the birth defect per se.
These cost estimates therefore include the costs of concurrent effects
(unlike the costs reported for many of the diseases in this handbook). This
method yields a more comprehensive assessment of total costs than would
be obtained if only individual effects were evaluated. This method is of
particular use in valuing the avoidance of birth defects because they very
frequently occur in clusters within an individual. As Waitzman et al. note,
however, the costs of associated anomalies are included as part of the
estimate of the costs incurred by an individual with a given birth defect.
These cost estimates therefore cannot be aggregated across birth defects
because of the possibility of double counting.
Chapter III.7 III.7-5 Cost of Cerebral Palsy
-------
Link to Chapter III. 3
Second, the Waitzman et al. method estimates the incremental costs for
individuals with birth defects — that is, the costs above and beyond the
average costs that would be incurred by individuals without the birth
defect.
Waitzman et al. (1996) estimated three categories of costs incurred by
individuals with limb reductions: direct medical costs, direct nonmedical
costs, and indirect costs.3 Direct medical costs, specifically inpatient care,
outpatient care, pharmaceuticals, laboratory tests, X-rays, appliances, and
long-term care are included in the cost estimates shown in this and other
chapters (Chapters III.3 through III. 8) based on the work of Waitzman et
al. Nonmedical direct costs, specifically developmental services, and
special education are also included in this handbook.
The Waitzman estimates of the costs incurred by individuals with limb
reductions are based on the costs of this birth defect in California across
many ages, and its occurrence in a large cohort of children born in
California in 1988. California's ongoing birth defects monitoring program
provides an excellent source of data. The California data sets were linked
with other national data sets so that Waitzman et al. could estimate the
incremental costs associated with cerebral palsy.
The method of calculating the expected lifetime incremental costs for an
individual with a birth defect — i.e., the average lifetime cost per case —
is the same for all the birth defects considered by Waitzman et al. The
expected per capita cost at age /', PCQ, for an individual born with the birth
defect is the probability of surviving to age /' (among those individuals born
with the birth defect), ps;, times the per capita cost among individuals who
do survive to age /' (PCPREV;, measured in the prevalent population):
PCQ = (ps;) x (PCPREV;) .
Waitzman et al. estimate per capita costs in the prevalent population of age
/', PCPREV;, in two different ways, depending on data availability (see
Chapter III.3).
The present discounted value of expected per capita lifetime costs of the
birth defect, PCCOBD, is just the sum of these expected age-specific per
capita costs, appropriately discounted (as explained more fully in Chapter
III.3):
PCCOBD =£; PCC/O+ry .
3 Indirect costs are not generally discussed in this handbook and so are not included in this chapter.
The reader may wish to consult Waitzman et al. (1996) for information on these costs.
Chapter III.7 III.7-6 Cost of Cerebral Palsy
-------
.7.B.2 Results
Waitzman et al.'s (1996) estimates of the total lifetime medical costs of
cerebral palsy are outlined in the following tables. Estimates are updated
from 1988 to 1996 dollars based on the medical care cost component of
the Consumer Price Index (1996:1988=1.6465). Table III.7-1 shows the
annual per capita medical costs incurred by individuals with cerebral palsy
by age group.
Table III. 7-1: Annual Per-Capita Medical Costs of Cerebral Palsy by Age Group (1996$)
Condition
Cerebral Palsy
Age 0-1
$16,236
Age 2-4
$9,848
Age 5-1 7
$11,451
Age 18+
$19,349
The medical cost of the average population was then subtracted from these
costs to obtain incremental costs. Waitzman et al. (1996) discounted these
costs using three different discount rates: two percent, five percent, and ten
percent. Although these discount rates do not match the standard EPA
rates used in many other chapters in this handbook (zero percent, three
percent, five percent, and seven percent), there is insufficient information
provided in Waitzman et al. (1996) to allow a conversion to discounted
costs using standard EPA discount rates. This problem exists in all
chapters based on the Waitzman et al. data (i.e., Chapters III.3 through
III. 8).
The present discounted values of average per capita lifetime incremental
costs, using discount rates of two percent, five percent, and seven percent,
are listed in Table III.7-2 below. Direct medical costs and direct non-
medical costs are listed separately. The sum of per-capita direct medical
and nonmedical costs provides an estimate of the total per-capita costs
incurred by individuals with cerebral palsy.
Chapter III.7
1.7-7
Cost of Cerebral Palsy
-------
Table III. 7-2: Per-Capita Net Medical Costs, Nonmedical Costs, and Total Costs of Cerebral
Palsy (1996$)
Cost Element
Net direct medical costs
2%
$461,010
5%
$233,798
10%
$116,899
Net direct nonmedical costs
Developmental Services
Special Education
Total Costs
$145,106
$94,454
$700,570
$68,979
$71,771
$374,548
$30,853
$47,634
$195,386
The costs presented in this chapter were current in the year the chapter was written. They can be updated using
inflation factors accessible by clicking below.
Link to inflation factors
I.7.B.3 Other Studies
Waitzman et al. (1996) present a study conducted in 1991 by The National
Foundation for Brain Research (NFBR), using a prevalence approach, on
the cost of cerebral palsy. Rather than using the primary diagnosis,
however, to construct costs (as in Rice et al., discussed in Chapter III.3),
cost estimates were made for cerebral palsy appearing as the first through
fifth diagnosis for inpatient stays, and the first through third diagnosis for
physician visits. The NFBR also estimated indirect family costs based on
family income reported in the 1989 National Health Interview Survey.
Waitzman et al. compare their estimate for direct medical costs of $172
million, adjusted to the nation and 1991 dollars, to the NFBR estimate for
inpatient stays of $318 million. They identify two reasons for the
difference in these two estimates. First, an incidence approach, such as the
approach used by Waitzman et al., uses discounting, whereas a prevalence
approach, such as that used by NFBR, does not. For example, Waitzman
et al. calculate that the 1991 estimate of $172 million discounted at five
percent would increase by over $100 million to $282 million if a two
percent discount rate was used instead. Second, Waitzman et al. estimate
incremental costs by subtracting the average costs of an individual without
cerebral palsy, whereas the NFBR study does not subtract average costs.
Waitzman et al. show that had they not subtracted the average costs, their
estimate for cerebral palsy in 1991 at a five percent discount rate would be
$330 million, an estimate much closer to the one produced by the NFBR
study.
Link to Chapter III. 3
Chapter III.7
1.7-8
Cost of Cerebral Palsy
-------
.7.C Conclusions
The cost estimates based on the research by Waitzman et al. (1996) are
recommended for use in benefits valuation. The NFBR study does not
include many relevant medical costs (including all non-hospital costs).
They also do not use an incremental approach. Consequently, the NFBR
cost estimates are not as closely matched to the direct medical costs which
are reported in this handbook as the Waitzman et al. results.
Chapter III.7 III.7-9 Cost of Cerebral Palsy
-------
CHAPTER III.8. COST OF DOWN SYNDROME
Clicking on the sections below will take you to the relevant text.
III.8.A Background
III.8.A. 1 Description
III.8.A.2 Concurrent Effects
III.8.A.3 Causality
III.8.A.4 Treatment and Services
III. 8. A. 5 Prognosis
III.8.B Costs of Treatment and Services
III. 8.B.I Methodology
III.8.B.2 Results
Chapter III.8 III.8-1 Cost of Down Syndrome
-------
CHAPTER III.8. COST OF DOWN SYNDROME
I.8.A Background
This chapter contains a discussion of the methods used and the results of
estimating the direct medical costs incurred by individuals with Down
Syndrome and the results of the analysis.1 It does not include information
on elements such as indirect medical costs, pain and suffering, lost time of
unpaid caregivers, etc. The reader is referred to Chapter I.I for a
discussion of the cost estimation methods and cost elements that are
relevant to all benefits estimates. In addition, Chapter III. 1 contains
information regarding the special characteristics of developmental defects,
and a list of chemicals that may cause developmental abnormalities.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and III. 1
Link to inflation factors
I.8.A.1 Description
Down syndrome occurs as a result of having three, rather than two, copies
of chromosome 21 (hence the name "trisomy 21"). Mental retardation and
a group of physical characteristics are commonly associated with Down
syndrome. In addition, a number of serious defects in critical organs (e.g.,
heart, digestive system) are also commonly found in people with Down
syndrome. The syndrome involves clusters of external physical anomalies,
learning disabilities, and organ system anomalies. Physical anomalies and
their prevalence (given in parenthesis) among Down syndrome children are:
unusually small head (50 percent), excess skin folds on eyelids (50-70
percent), speckled irises in eyes (30-80 percent), narrow and short palate
(60-90 percent), protruding tongue (40-60 percent), broad hands (70
percent), and other relatively minor changes in physical appearance. Down
syndrome children usually have a flattened nose bridge, additional skin on
the back of the neck, small or anomalous ears, abnormally formed fingers,
and "simian" hand creases (Pueschel and Rynders, 1982).
The severity of mental retardation associated with this disorder varies
considerably. Between 3 and 12 percent of Down syndrome children are
profoundly retarded. Many of these individuals are unable to walk, talk, or
eat without assistance. Approximately 25 percent are severely retarded, 55
1 "Costs" in this chapter refer to direct incremental per capita medical costs, unless otherwise
noted.
Chapter III.8 III.8-2 Cost of Down Syndrome
-------
percent are moderately retarded, and 14 percent are mildly retarded.
Seizures occur in five to nine percent of children, increasing with age. In
adulthood, memory loss and reduced cognitive abilities are usually evident
after age 45 in people with Down syndrome. Pathology studies have found
structural changes in the brain similar to those found in Alzheimers disease
patients (Waitzman et al., 1996).
Children with Down syndrome often have very early onset Alzheimer-type
changes in cognitive ability. These changes may affect their functional
abilities, and are observed when comparing the abilities of Down syndrome
children with other children over the range of school ages. By young
adulthood, the IQ scores of Down syndrome children have progressed from
the range of low-mild to high-moderate retardation seen in early school
years to a range that includes severe retardation (Oski, 1994).
Down syndrome is the most common chromosomal disorder observed in
the newborn period and accounts for approximately one third of all
chromosomal abnormalities. Overall, chromosomal abnormalities occur in
approximately 1 in 200 live births and account for a sizable fraction of the
malformations and neonatal deaths that occur. The frequency of Down
syndrome is estimated to range from one in 700 to one in 1,000 (Oski,
1994). With the availability of prenatal testing, not all Down syndrome
fetuses are brought to term. Although the total medical costs of Down
syndrome and its incidence (which is measured in terms of live births) are
thereby reduced, the individual and societal costs may be very substantial.
.8.A.2 Concurrent Effects
Associated organ system anomalies include congenital heart disease,
duodenal atresia or stenosis, cleft lip or palate, fused digits,
tracheoesophageal fistula, imperforate anus, failure of the testicles to
descend, and incomplete development of neck vertebrae. Of these, the
most common and serious are heart defects, which occur in approximately
29 percent of children with Down syndrome. Approximately 50 percent of
adolescents and adults have mitral valve prolapse in the absence of
symptoms of heart disease. The children usually also have poor muscle
tone, increased joint flexibility, instability between the first two neck
vertebrae, and an abnormally developed pelvis. Their musculoskeletal
abnormalities often lead to degenerative changes in the joints. Thyroid
function is often decreased (Waitzman et al., 1996).
Hearing loss affects approximately 60 percent of Down syndrome children,
and 15 to 26 percent have moderate to profound hearing loss. Cataracts
affect approximately 12 to 20 percent of children; strabismus affects 20 to
45 percent. Severe nearsightedness and deformity of the cornea are also
common (Waitzman et al., 1996).
Chapter III.8 III.8-3 Cost of Down Syndrome
-------
Down syndrome children have an increased risk of skin and lung infections,
acute leukemia, and an impaired immune system function (Waitzman et al.,
1996), and increased rates of testicular cancer and retinoblastoma (a
childhood cancer) (Oski, 1994).
.8.A.3 Causality
Down syndrome occurs when the egg or sperm receives an extra copy of
chromosome 21. This condition is hereditary in some cases and is also
associated with maternal age. Chemicals that cause chromosomal
abnormalities, especially interference with normal chromosomal replication
and disjunction, may cause this type of genetic abnormality (Waitzman et
al., 1996). Table III. 1-2 in Chapter III.l lists chemicals associated with
genotoxic effects. A review of material in the references listed in Chapter
III. 1 will provide information regarding those chemicals that have been
shown to cause non-disjunction, and other genotoxic effects that may
specifically cause an abnormal number of chromosomes.
Link to Chapter III. 1, Table III. 1-2
III.8.A.4 Treatment and Services
A variety of medical treatments and special services are required for Down
syndrome patients. These vary widely, depending on the specific cluster of
abnormalities occurring in an individual. The numerous structural
abnormalities may require immediate attention in the postnatal period, or
be corrected later in childhood if they are not immediately life-threatening.
Some structural problems related to bone structure, muscles, and joints
may require specialized equipment and physical therapy. Hearing problems
and recurrent ear infections are often treated with pressure-equalization
tubes and antibiotics. Due to the retardation associated with this
syndrome, special education services are usually required. Many Down
syndrome patients require lifelong services in the form of specialized
housing and medical care.
III.8.A.5 Prognosis
Down syndrome involves a cluster of effects. While some effects, such as
facial appearance, do not change over time, others become more severe
over time. For example, a deterioration in memory and other cognitive
functions is seen in most Down syndrome patients over the age of 45
(Waitzman et al., 1996). Laxity in the ligaments often leads to degenerative
changes in the joints, especially in the knees and spine (Semine et al., 1978;
Mendez et al., 1988). The prevalence of seizures, which occur in
approximately five to nine percent of Down syndrome patients (Waitzman
et al, 1996), increases with age (McVicker et al., 1994). In addition to
Chapter III.8 III.8-4 Cost of Down Syndrome
-------
increased morbidity over time, the lifespan of Down syndrome patients is
likely to be shortened, due to the multitude of serious effects associated
with this disease.
1.8.B Costs of Treatment and Services
I.8.B.1 Methodology
Link to Chapter III. 3
Waitzman et al. (1996) provide an estimate of the direct medical and non-
medical costs of treating Down syndrome. They used the same
methodology to estimate the costs incurred by individuals with Down
syndrome as for all the birth defects for which they estimated costs. The
methodology and relevant considerations are detailed in Chapter III.3,
including discussions of direct and indirect costs, prevalence versus
incidence, incremental costs, and concurrent effects. The analytic method,
the sources of data and the limitations of the Waitzman method are also
discussed in Chapter III.3. The methodology is outlined briefly here.
To estimate the lifetime medical costs incurred by an individual with a birth
defect, Waitzman et al. estimated the average lifetime medical costs for an
individual with the birth defect. From this value, the authors subtracted the
average lifetime medical costs for an individual without the birth defect.
Because they estimated lifetime costs, they used an incidence-based
approach. Ideally, they would have tracked the costs of the cohort
members over time, until the death of the last cohort member. Because the
members of the cohort were born in 1988, however, this tracking was not
possible. Instead, estimates of the costs incurred at each age were based
on estimates of per capita costs in the prevalent population of that age (see
Chapter III.3, Section III.3.B.1.2).
Link to Chapter III.3, Section III.3.B.I.2
This method has two important implications. First, Waitzman et al.
estimated the costs incurred by individuals with birth defects, including all
medical costs incurred, rather than the cost of the birth defect per se.
These cost estimates therefore include the costs of concurrent effects
(unlike the costs reported for many of the diseases in this handbook). This
method yields a more comprehensive assessment of total costs than would
be obtained if only individual effects were evaluated. This method is of
particular use in valuing the avoidance of birth defects because they very
frequently occur in clusters within an individual. As Waitzman et al. note,
however, the costs of associated anomalies are included as part of the
estimate of the costs incurred by an individual with a given birth defect.
These cost estimates therefore cannot be aggregated across birth defects
because of the possibility of double counting.
Chapter III.8 III.8-5 Cost of Down Syndrome
-------
Second, the Waitzman et al. method estimates the incremental costs for
individuals with birth defects — that is, the costs above and beyond the
average costs that would be incurred by individuals without the birth
defect.
Waitzman et al. (1996) estimated three categories of costs incurred by
individuals with Down syndrome: direct medical costs, direct nonmedical
costs, and indirect costs.2 Direct medical costs, specifically inpatient care,
outpatient care, pharmaceuticals, laboratory tests, X-rays, appliances, and
long-term care are included in the cost estimates shown in this and other
chapters (Chapters III.3 through III. 8) based on the work of Waitzman et
al. Nonmedical direct costs, specifically developmental services, and
special education are also included in this handbook.
The Waitzman estimates of the costs incurred by individuals with Down
syndrome are based on the costs of this birth defect in California across
many ages, and its occurrence in a large cohort of children born in
California in 1988. California's ongoing birth defects monitoring program
provides an excellent source of data. The California data sets were linked
with other national data sets so that Waitzman et al. could estimate the
incremental costs associated with Down syndrome.
The method of calculating the expected lifetime incremental costs for an
individual with a birth defect — i.e., the average lifetime cost per case —
is the same for all the birth defects considered by Waitzman et al. The
expected per capita cost at age /', PCQ, for an individual born with the birth
defect is the probability of surviving to age /' (among those individuals born
with the birth defect), ps;, times the per capita cost among individuals who
do survive to age /' (PCPREV;, measured in the prevalent population):
PCQ = (ps;) x (PCPREV;) .
Waitzman et al. estimate per capita costs in the prevalent population of age
/', PCPREV;, in two different ways, depending on data availability (see
Chapter III.3).
Link to Chapter III. 3
2 Indirect costs are not generally discussed in this handbook and so are not included in this chapter.
The reader may wish to consult Waitzman et al. (1996) for information on these costs.
Chapter III.8 III.8-6 Cost of Down Syndrome
-------
.8.B.2 Results
The present discounted value of expected per capita lifetime costs of the
birth defect, PCCOBD, is just the sum of these expected age-specific per
capita costs, appropriately discounted (as explained more fully in Chapter
III.3):
PCCOBD =£; PCC/O+ry .
Waitzman et al. (1996) estimate the total lifetime medical costs of Down
syndrome as outlined in the following tables, updated from 1988 to 1996
dollars based on the medical care cost component of the Consumer Price
Index (1996:1988=1.6465). Table III.8-1 shows the annual per capita
medical costs associated with Down syndrome by age group.
Table III. 8-1 : Annual Per-Capita Medical Costs of Down syndrome by Age Group (1996$)
Condition
Down syndrome
Age 0-1
$27,265
Age 2-4
$5,577
Age 5-1 7
$2,231
Age 18+
$7,529
The medical cost of the average population was then subtracted from these
costs to obtain incremental costs. Waitzman et al. (1996) discounted these
costs using three different discount rates: two percent, five percent, and ten
percent. Although these discount rates do not match the standard EPA
rates used in many other chapters in this handbook (zero percent, three
percent, five percent, and seven percent), there is insufficient information
provided in Waitzman et al. (1996) to allow a conversion to discounted
costs using standard EPA discount rates. This problem exists in all
chapters based on the Waitzman et al. data (i.e., Chapters III.3 through
III. 8).
The present discounted values of average per capita lifetime incremental
costs, using discount rates of two percent, five percent, and seven percent,
are listed in Table III. 8-2 below. Direct medical costs and direct non-
medical costs, including developmental services costs and special education
costs, are listed separately. The sum of per-capita direct medical and
nonmedical costs provides an estimate of the total per-capita costs incurred
by individuals with Down syndrome.
Chapter III.8
1.8-7
Cost of Down Syndrome
-------
Table III. 8-2: Per-Capita Net Medical Costs, Nonmedical Costs, and Total Costs of Down
syndrome (1996$)
Condition
Net direct medical costs
2%
$141,596
5%
$90,556
10%
$64,212
Net direct nonmedical costs
Developmental services
Special education
Total Costs
$67,645
$144,138
$353,379
$35,439
$109,622
$235,617
$19,089
$72,854
$156,155
The costs presented in this chapter were current in the year the chapter was written. They can be updated using
inflation factors accessible by clicking below.
Link to inflation factors
Chapter III.8
I.8-8
Cost of Down Syndrome
-------
CHAPTER III.9 COST OF REDUCING HIGH BLOOD LEAD LEVELS IN
CHILDREN
Clicking on the sections below will take you to the relevant text.
III.9.A Background
III.9.A. 1 Description
III.9.A.2 Concurrent Effects
III.9.A.3 Causality and Special Susceptibilities
III.9.A.4 Treatments and Services
III. 9. A. 5 Prognosis
III.9.B Costs of Treatments and Services
III.9.B.1 Methodology
III.9.B.2 Treatment Profile and Costs by Risk Level
III.9.B.3 Average Treatment Costs
III.9.B.4 Survival
III.9.B.5 Present Value Costs
III.9.B.6 Limitations
III.9 III.9-1 Cost of Reducing High Blood-Lead Levels
-------
CHAPTER III.9 COST OF REDUCING HIGH BLOOD LEAD LEVELS IN
CHILDREN
III.9.A Background
This analysis focuses solely on the medical costs associated with efforts to
reduce blood lead (PbB) levels in children under the age of six. Information
is provided regarding treatment, source reduction and education prescribed
in response to elevated PbB levels. The chapter does not include medical
costs of treating health effects that result from lead exposure. Cost
estimates may be developed for lead-induced effects in the future
(concurrent effects are discussed in more detail below). The chapter also
does not include information on elements such as indirect medical costs,
pain and suffering, lost time of unpaid caregivers, etc. The reader is
referred to Chapter I.I for a discussion of the cost estimation methods and
cost elements that are relevant to all benefits estimates. In addition,
Chapter III. 1 contains information regarding the special characteristics of
developmental defects, and a list of chemicals that may cause
developmental abnormalities.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and III. 1
Link to inflation factors
I.9.A.1 Description
Elevated PbB levels in young children occur when children are exposed to
lead via any media (i.e., air, water, food, soil). Elevated PbB in children is
a considerable public health concern, due to the potential adverse effects of
lead on multiple organ systems and the particular susceptibility of young
children to many of these effects, including neurological damage. Lead is
toxic to the kidneys and is associated with low birth weight, male sterility,
cancer, and a wide array of neurological disorders. Elevated PbB may lead
to neurological impairment, behavioral abnormalities, and damage to the
cardiovascular, kidney, liver, gastrointestinal, blood-forming, reproductive,
and endocrine systems. Lead is also a suspected carcinogen and mutagen
(EPA, 1987), and impairs the immune system, causing increased
susceptibility to infectious agents (ATSDR, 1997).J
1 This analysis used data provided in EPA, 1985 and 1987, which contain a cost-benefit analysis of
reducing lead in gasoline.
III.9 III.9-2 Cost of Reducing High Blood-Lead Levels
-------
Children are particular susceptibility to neurological impairment. Lead
damages the developing neurons in the brain by damaging the protective
coating of myelin on nerve cells; it is suspected of causing irreversible
limitations in brain function. A relationship between a decrease in
cognitive functioning (as measured by IQ tests) and lead exposure in young
children has been reported in numerous studies.
Elevated PbB levels are used to evaluate the level of risk and determine
treatment. The Centers for Disease Control (CDC) have developed a
classification system for risks to children, based on their PbB levels and a
related measure of lead exposure, erythrocyte protoporphyrin (EP). Their
classification system was modified in EPA (1987) and is shown in Table
III.9-1. This system is used in Section B to determine treatment costs.
Table III. 9-1
Condensed Version of CDC Risk Classification Table
Blood Lead
Level
(ug/dl)
0-20
21-40
>40
Erythrocyte
(EP)
(UQ/dl)
0-32
I
Ib
*
Protoporphyrin Level
33-53 >53
la la
II III
III IV
* = not generally observed
Source: U.S. EPA (1987)
.9.A.2 Concurrent Effects
As noted above, lead exposure may lead to a variety of adverse health
effects. The occurrence of effects will depend on the levels of PbB (as
reflective of exposure throughout the body), the health status of the
individual (e.g., poor nutritional status is especially problematic), and
individual factors. Most children in the U.S. today do not experience
severe adverse health effects that are measurable. Damage to the nervous
system is very difficult to quantify, especially in young children. Aside
from IQ loss, this damage not usually identified except in severe lead
poisoning cases. Severe lead poisoning may lead to coma and death. The
form of lead is important; and organic lead, such as tetraethyl lead (TEL),
has caused deaths in children.
III.9
1.9-3 Cost of Reducing High Blood-Lead Levels
-------
.9.A.3 Causality and Special Susceptibilities
Children absorb considerably more lead than adults when ingesting the
same contaminated media. Adults absorb five to ten percent of dietary lead
and retain little of it; young children absorb 40 to 50 percent of dietary lead
and retain 20 to 25 percent of it (Oski et al., 1994). Demographic and
cultural factors affecting nutrition and dietary patterns may be considered
when evaluating risks and costs. As noted above, the occurrence of
adverse effects depends, in part, on the health status of the individual and
individual factors. Both the uptake of lead into the body (which impacts the
PbB levels) and the severity of health impacts may be exacerbated by
various factors including poor health and nutritional status. Elevated EP
(an indicator of lead poisoning) is often associated with iron deficiency
(EPA, 1987). Diets high in fat and low in calcium, magnesium, iron, zinc,
and copper increase the absorption of lead (Oski et al., 1994). Lead is
stored in the body primarily in bone and may be released, causing toxicity,
over many decades.
.9.A.4 Treatments and Services
The treatments and services provided for children with elevated PbB
depend on their risk classification, as shown in Table III.9-1. Detailed
treatment descriptions are provided in Section B of this chapter along with
cost data, and so are not presented here.
Link to Table III. 9-1
III.9.A.5 Prognosis
Elevated PbB levels can always be brought down over time. The prognosis
for health effects related to the elevated levels is more serious. The
prognosis for full and unimpaired recovery depends on the degree of lead
poisoning that occurred (e.g., the risk classification), the amount of time
during which the PbB levels were elevated, the age of the child, general
health and nutrition status, the degree of intervention (including special
education strategies provided), and individual factors. It is not possible to
predict the outcomes for individual children, due to the variety of factors
which impact the final outcome. High risk children (as determined by the
CDC classification system) are generally more likely to experience
permanent damage than children with moderate or low risk levels.
PbB levels greater than 10 to 15 ug/dL sustained during early childhood
carry a substantial risk for long-lasting but subtle injury to the nervous
system, even if no clinical symptoms are detected. Attention deficits and
reading disabilities have been observed in cohorts of young children with
elevated PbB levels. In adults with elevated PbB as children, increased
III.9 III.9-4 Cost of Reducing High Blood-Lead Levels
-------
rates of dropping out and having long-term reading disabilities have been
observed. When levels are very high and encephalopathy has resulted,
serious sequelae may occur in later years that include seizure disorders,
mental disorders, and (in some rare cases) blindness and hemiparesis. In
some cases, residential care is required (Oski et al., 1994).
1.9.B Costs of Treatments and Services
I.9.B.1 Methodology
To estimate the average costs of testing and treating children with high
PbB levels, this analysis relies heavily on methodologies developed for the
benefit-cost analysis of reducing lead in gasoline (U.S. EPA, 1985) and
later applied by the EPA's Office of Air Quality Planning and Standards
(U.S. EPA, 1987).
The average direct cost per child with high PbB levels was calculated in
four steps:
a typical treatment profile was developed for each risk level,
the costs of relevant treatments were determined,
the costs of treatments were combined with the treatment profiles
to produce an estimate of the average cost per child, and
• the costs were reduced to reflect the fact that only a portion of
children with high PbB levels will be screened.
I.9.B.2 Treatment Profile and Costs by Risk Level
The costs of relevant treatment elements (e.g., chelation therapy,
neuropsychological evaluation, family education) were estimated by U.S.
EPA (1985) and adjusted in U.S. EPA (1987). These costs are based on
the CDC's (1978) recommended clinical management program and are
linked to the risk level by determining PbB and EP levels (See Table
III.9-1).
Link to Table III.9-1
The treatment profile is developed at the age at which the child is first
screened. This analysis considers the cost of follow-up through age five
regardless of the age of the child at the initial screening. Consequently, the
actual costs for children initially screened at older ages may differ from
those estimated in this chapter. They may be greater or lesser, depending
on the medical consequences of later screening. In addition, treatment is
III.9 III.9-5 Cost of Reducing High Blood-Lead Levels
-------
still likely to take place when screening occurs later in childhood, so overall
costs may not be reduced regardless of the screening results. It was
assumed that the distribution of the ages of children screened is uniform
over the ages one through five. The cost of follow-up tests are adjusted
accordingly.
Regardless of the age of initial screening, follow-up testing of children is
assumed to continue through age five. A child initially screened at age one
will therefore have follow-up tests for five years (ages one to five,
including the initial screening year), while a child initially screened at age
five will have follow-up tests only in the year of screening. The cost
estimates presented below, given in 1996$, are averages for all screened
children in the risk group regardless of the age at which they were initially
screened.
Risk Level I. Children in this risk group are considered to be at low risk.
Because screening is concentrated on children in areas with high-risk
factors, follow-up treatment and family education are recommended. The
estimate of the average (undiscounted) cost per child at this risk is $522.
Risk Level la. Children in this risk group do not have elevated PbB levels.
They do have elevated EP levels, which may be indicative of iron
deficiency or other medical problems. In addition to initial testing, all
children at this risk level are tested for iron deficiency, at a cost of $39. The
cost of medication for anemia is estimated as $127. With a mean
population PbB of 20, 23.8 percent of those classified in risk level la will
receive the medication, resulting in an average cost of $29 per child. The
percentage of the children testing positive for anemia will vary with the
mean population PbB concentration; a higher mean population would
increase the percentage of children testing positive, and would therefore
increase the average treatment costs per child. Health education, stressing
nutritional needs, is assumed to be provided to 50 percent of the families
with children in this risk group. The average (undiscounted) costs per
screened child at this risk level is $692.
Risk Level Ib. In addition to the initial screening, children in risk category
Ib receive periodic follow-up tests and limited family education. Table
III.9-2. shows an estimated average (undiscounted) cost for this risk group
of$623.
Risk Level II. As can be seen from Table III.9-2, children at moderate
risk receive no chelation therapy. Also, no neuropsychological evaluation
is performed and family education is limited. The major cost for this group
comes from follow-up tests. The estimated average (undiscounted) cost
for screened children in risk level II is $1,205.
III.9 III.9-6 Cost of Reducing High Blood-Lead Levels
-------
Risk Level III. The treatment for children in risk level III is similar to
treatment for children in risk level IV. For risk level III, however, a
CaNa2-EDTA provocation test is included. This test is used to evaluate the
responsiveness of the child to chelation therapy. Also, the costs of
chelation treatment fall sharply compared with children in risk level IV, due
to the decreased percentage of children estimated to need this therapy.
Only 0.43 percent of risk level III children are assumed to require this
therapy, as compared with 37 percent of children in risk level IV. The cost
of follow-up tests also decreases, since the frequency of follow-up tests is
higher for children who have received chelation therapy than for children
who have not. The estimated average (undiscounted) cost for screened
children who are in risk level III is $2,632.
Risk Level IV. As can be seen from the table, the major cost element for
children in the urgent risk group is chelation therapy. The probability of
requiring chelation therapy increases with PbB concentrations. EPA
(1987) estimated that 37 percent of the children in this risk group would
require chelation therapy, based on a mean population PbB level of 20
Hg/dl. Note that the cost of treatment for children in this risk group
depends on this assumption regarding mean population PbB levels. A
higher mean population PbB level would result in a greater percentage of
children in this risk group requiring chelation therapy. Cost estimates for
this risk group would therefore increase.
EPA reports that in some cases, an initial chelation therapy may be
followed by a rebound in PbB levels as the body attempts to equilibrate
between lead in soft tissue and lead in blood. A second and sometimes a
third chelation treatment is therefore necessary. Fifty percent of children
who receive one chelation treatment are assumed to require a second
treatment. Fifty percent of children requiring a second chelation treatment
are assumed to require a third. The total average (undiscounted) cost of
treating a child in risk level IV is estimated to be $5,200.
Typical treatment for children in each risk level is shown in Table III.9-2,
along with the associated average costs per screened child in each risk
level.
III.9 III.9-7 Cost of Reducing High Blood-Lead Levels
-------
Table III. 9-2
Average Direct Cost of Treatment per Screened Child
Cost Per Child (1996$)
Year
Year of
screening
2nd year
3rd year
4th year
5th year
TOTAL
Totals may
Sources: U
Treatment
Chelation, inpatient
2nd chelation, inpatient
3rd chelation, inpatient
Initial lab test
CaNa2 provocation test
Iron deficiency test
Medication for anemia
Follow-up tests
Neuropsychological
evaluation
Family education
Total
Follow-up tests
Follow-up tests
Follow-up tests
Follow-up tests
not add due to rounding.
.S. EPA, 1987, Bureau of Labor
Level IV Level III
1,486
743
372
61
0
0
0
570
1,204
401
4,837
145
108
72
36
5,200
Statistics.
13
5
3
61
221
0
0
481
1,204
401
2,390
97
72
48
25
2,632
Level II Level Ib Level la Level I
0
0
0
61
221
0
0
481
0
200
964
97
72
48
25
1,205
0
0
0
61
0
0
0
120
0
200
382
97
72
48
25
623
0
0
0
61
0
39
29
120
0
200
450
97
72
48
25
692
0
0
0
61
0
0
0
120
0
101
281
97
72
48
25
522
.9.B.3 Average Treatment Costs
To determine the average costs per child it was first necessary to estimate
what percent of children in each risk category are screened. The percent of
children who are placed in each risk category based on the screening results
were then estimated. The risk level was used to determine the percent of
patients receiving each type of treatment.
For children in risk levels III and IV, it was assumed that the lead exposure
would result in behavioral symptoms; therefore, all children in these risk
levels would be screened. The average cost per screened child in risk
levels III and IV was assumed to equal the average cost per child in these
risk groups.
Children in the lower risk categories may not develop obvious symptoms
and may not receive treatment. U.S. EPA (1987) has estimated that 20
percent of children are screened for blood lead levels in any one year. A
probability analysis could be performed to estimate the likelihood of a child
being treated between disease onset and the age of six. Costs could then
be adjusted based on the predicted average years past onset that screening
occurs, but the level of detail of the information available on frequency of
screening does not warrant this detailed analysis. As a simple and
approximate method of accounting for the fact that not all children with
high PbB levels are screened, the treatment costs for children in risk
categories II, I, IA, and IB are multiplied by 0.2. The estimated costs for
each risk level are shown in Table III.9-3.
III.9 III.9-8 Cost of Reducing High Blood-Lead Levels
-------
Table 111.9-3
Average Direct Costs per Child With High Blooc
Risk Level Cost per Disease Year (1996 $)
Per Child 1 2
IV 4,837 145
III 2,390 97
llb 193 19
lbb 77 19
lab 90 19
lb 56 19
Lead Levels-All Risk Groups3 (1996 $)
3
108
72
14
14
14
14
a Assuming uniform distribution of the tested children over ages 1
b As explained in the text, cost estimates for these risk categories
percentage of children expected to be screened. If it is assumec
them, then the values reported in Table III. 9-3 for Levels I and II
Source: U.S. EPA, 1987.
4
72
48
10
10
10
10
5
36
25
5
5
5
5
Total Cost
(1996 $)
5,200
2,632
241
125
138
105
through 5.
have been multiplied by 0.20 to reflect the
that all children who need services will receive
should be used.
I.9.B.4 Survival
No data linking survival rates to elevated blood lead levels were found.
This analysis assumes, therefore, that survival rates match typical survival
rates in the entire U.S. population. The calculated survival rates for disease
years one through five for children ages one to five, however, are 100
percent, as reported in Vital Statistics of the United States (U.S.
Department of Health and Human Services, 1985). Survival rates therefore
have no impact on the cost estimates.
.9.B.5 Present Value Costs
Table III.9-4 shows the discounted direct cost estimates in Table III.9-3.
Estimates of the present value costs per child with high PbB levels are
shown in Table III.9-4, using discount rates of zero, three, five, and seven
percent. Using a five percent discount rate, estimates of the net present
value of medical costs per child with high blood levels are $5,162 for risk
level IV, $2,610 for risk level III, $237 for risk level II, $120 for risk level
Ib, $134 for risk level la, and $101 for risk level I. The discount rates have
only a minor impact on the present value cost estimates, due to the fact that
the majority of costs are incurred in the year of screening and are therefore
unaffected by the discount rates.2
2 Screening may not be coincident with onset of high PbB levels; no information was located
regarding a typical time period between onset of high PbB levels and screening. If such information was
available, then the costs of the disorder could be discounted over this lag period.
III.9 III.9-9 Cost of Reducing High Blood-Lead Levels
-------
Table 111.9-4
Discounted Average Direct Costs per Child Age One to Five Years with
High Blood Lead Levels (1996 $)
Risk Level Discounted Per Patient Costs
IV
III
II
Ib
la
I
Sources: U.S. EPA, 1987
The costs presented in this
inflation factors accessible
Link to inflation factors
0
5200
2632
241
125
138
105
Bureau of Labor Statistics
chapter were current in the
by clicking below.
(Discount Rate %)
3
5185
2623
240
123
137
104
year the chapter was written.
5
5162
2610
237
120
134
101
They can
7
5135
2591
234
117
132
97
be updated using
Lead exposure analyses can produce estimates of the distribution of PbB
levels and EP levels among the population, and therefore the distribution of
the population among the risk levels shown in Table III.9-1. This
distribution can be compared to levels of PbB and EP in the population
under various regulatory options to estimate the change in the number of
individuals in each of the risk categories. By multiplying the change in the
number of individuals in each risk category by the costs of screening and
treating an individual in that risk category and summing over all risk
categories, the benefits of the regulations may be calculated.
Link to Table III.9-1
III.9.B.6 Limitations
There are numerous limitations to the cost estimates provided in this
chapter. Foremost is the restriction of the information to those costs
associated directly with the reduction of lead levels in blood. As discussed
in Section III.9.A, numerous serious concurrent effects occur as a result of
elevated blood lead levels. These effects are the reason that time and
money are spent to reduce blood lead level in children. The costs of their
treatment may be substantial. Some children may not experience
measurable (or any) consequences of elevated blood lead levels,
particularly if the elevations are very small, the children are healthy, and
they have a good nutritional status.
Another major limitation is the assumption that only 20 percent of children
are tested, valued when the chapter was originally written in 1993. There
has since been an increased awareness of the risks associated with lead
poisoning in children and a concurrent increase in the testing of young
III.9 III.9-10 Cost of Reducing High Blood-Lead Levels
-------
children, which will lead to an increase in both testing and treatment costs.
The costs presented in this chapter are therefore an underestimate of actual
costs.
Another limitation is the lack of data regarding the age at which testing is
performed. As stated in the text, testing may not be done at the age of one
year. If it is done later, the costs of treatment may be greater or less than
those estimated here. Costs would be lower if only minimal treatment were
required because they would extend over a shorter period. If a delay in
testing led to higher levels requiring more complex treatment and long-
term impairment, then the costs could be much higher.
Finally, the cost estimates in this chapter are based largely on a
methodology and data that were collected in the 1980s. These data may be
outdated if new protocols are in place, and the costs may therefore differ
from those presented here.
III.9 III.9-11 Cost of Reducing High Blood-Lead Levels
-------
CHAPTER IV.1 INTRODUCTION TO THE COST OF RESPIRATORY
ILLNESS AND SYMPTOMS
Clicking on the sections below will take you to the relevant text.
IV. 1.1 Overview
IV. 1.2 Illness Definition and Duration
IV.1 IV.1-1 Cost of Respiratory Illness and Symptoms
-------
CHAPTER IV.1 INTRODUCTION TO THE COST OF RESPIRATORY
ILLNESS AND SYMPTOMS
IV. 1.1 Overview
Section IV of the Cost of Illness Handbook provides direct medical cost
estimates for respiratory illness and symptoms, excluding cancers of the
respiratory system. Cancers are discussed in Section II, and at the time of
this writing includes one respiratory illness, lung cancer.
This introductory chapter contains a brief summary of respiratory illness
definitions and duration that are relevant to estimating the costs of
respiratory illness and symptoms (referred to subsequently as "illness" for
simplicity).1 The illnesses included in this section are those that were being
evaluated by EPA in a manner that required cost estimates (e.g., for policy
evaluations, benefits assessments, comparative illness reduction strategies).
Although the specific applications of the cost data vary, basic cost
information is provided for each illness. There is considerable variation in
both the level of detail that was required for the illness in this section and
the type of information that was available for the illness.
IV.1.2 Illness Definition and Duration
Respiratory illnesses involve the upper or lower respiratory system, which
usually includes the nose, tonsils, throat, mouth, trachea (wind pipe), and
all the structures of the lungs (bronchi, alveoli, etc.). Respiratory illnesses
also are usually defined to include ear infections, sinusitis, and related
illnesses (Oski et al., 1994). Often the illnesses involve multiple parts of
the respiratory system.
A characteristic of the respiratory illnesses that are discussed in this section
is that they are commonly treated by primary care physicians. This is in
contrast to the illnesses discussed in Sections II and III, which were more
likely to be treated by a physician or team of physicians with
subspecializations, such as oncology and pediatric surgery, neonatal
pediatrics, etc. The illnesses in this section may be very serious and even
life-threatening. They usually are not, however, and are often treated with
medications and minimally invasive procedures. Most illnesses of this type
can be treated on an outpatient basis, rather than requiring hospitalization.
Consequently, the costs of the illnesses, which are estimates of the average
costs, are generally much lower, on an annual basis, than costs for illnesses
that require more intensive treatment.
^'Cost" refers to direct medical costs unless otherwise noted.
IV.1 IV.1-2 Cost of Respiratory Illness and Symptoms
-------
All direct medical costs presented in the Handbook are estimates of the
"average" cost for patients, so that the extremely high and low cost
treatments that may be required in a small number of cases have been
averaged into the overall distribution of care costs to yield an average cost.
For most illnesses, the circumstances that would lead to unusually high or
low costs are noted in the individual illness chapters.
This section includes cost estimates for both acute and chronic illnesses.
An acute illness generally has a rapid onset (hours or days) and is
sufficiently intense to cause someone to seek medical treatment. A chronic
illness, such as asthma, is also assumed to require medical attention, but
may continue over months and years, with a requirement for continuous or
intermittent medical supervision and care. Symptoms are observed
pathological responses in the body that may be associated with numerous
causes, including illnesses or other stimuli (e.g., a cough in response to
smoke).
IV.1 IV.1-3 Cost of Respiratory Illness and Symptoms
-------
CHAPTER IV.2: COST OF ASTHMA
Clicking on the sections below will take you to the relevant text.
IV.2.A Background
IV.2. A.I Description
IV.2.A.2 Concurrent Effects
IV.2.A.3 Causality and Special Susceptibilities
IV.2.A.4 Treatment and Services
IV.2.A.5 Prognosis
IV.2.B Costs of Treatment and Services
IV.2.B.1 Methodology Summary
IV.2.B.2 Diagnosis
IV.2.B.3 Long-term Management
IV.2.B.4 Acute Care
IV.2.B.5 Annual Cost of Treatment and Services
IV.2.B.6 Summary of Lifetime Costs
IV.2.C Sensitivity Analysis
IV.2.D Uncertainty Analysis
Appendix IV.2-A Chemicals Associated with Asthma
Appendix IV.2-B Drug Therapies Recommended by NHLBI
Chapter IV.2 IV.2-1 Cost of Asthma
-------
CHAPTER IV.2: COST OF ASTHMA
IV.2.A. BACKGROUND
This chapter contains a discussion of the lifetime incremental direct medical
costs incurred by asthma patients, and methods used to estimate those
costs. Asthma is of particular concern to EPA because there are many
pollutants that may cause or exacerbate this disease. Regulation of primary
and toxic air pollutants may result in a reduced number of cases of chronic
airway diseases, including asthma. Programs to reduce indoor air
pollutants, such as environmental tobacco smoke (ETS), are also crucial in
reducing the impacts of this disease. The benefits of such activities can be
estimated in part, by evaluating the direct medical costs avoided. A full
measure of the costs of asthma would also include direct non-medical costs
and indirect costs. This chapter does not include information on elements
of willingness-to-pay (WTP), such as indirect medical costs, pain and
suffering, lost time of unpaid caregivers, lost productivity of patients, etc.
The direct medical costs presented in this chapter may be useful in
providing a lower-bound measure of WTP. The reader is referred to
Chapter I.I for a discussion of direct cost estimation methods and cost
elements that are relevant to all benefits estimates. In addition, Chapter
IV. 1 contains general information regarding respiratory illnesses.
The costs presented in this chapter were current in the year the chapter was
written. They can be updated using inflation factors accessible by clicking
on the sidebar at left.
Link to Chapters 1.1 and IV. 1
Link to inflation factors
This chapter uses the Guidelines for Diagnosis and Management of Asthma
developed by the National Heart, Lung, and Blood Institute (NHLBI,
1997), which provide disease definition and clinical practice guidelines.
The guidelines were developed by clinicians and researchers and provide
information on current approaches to treating asthma. Current Medicare
reimbursements and journal articles were used to obtain cost estimates.
The medical and economics literature was consulted to obtain supporting
information. A cost estimate was developed for the average patient and an
upper bound value was estimated based on patients with moderate or
severe asthma who had a high use rate of acute care services.
IV.2.A.1 Description
IV.2.A.1.1 Definition
Asthma is a chronic inflammatory disorder of the airways, designated as
ICD9-CM-493 in the International Classification of Diseases (ICD-9).
Chapter IV.2 IV.2-2 Cost of Asthma
-------
Airway inflammation contributes to airway hyperresponsiveness, airflow
limitation, respiratory symptoms, and the chronic nature of the disease.
Airflow limitation and the narrowing of airways can be manifested as acute
bronchoconstriction, airway edema, mucus plug formation, and remodeling
of the airway walls. In susceptible individuals, this inflammation causes
recurrent episodes of wheezing, breathlessness, chest tightness, and
coughing. Asthma patients are usually categorized as having mild
persistent, mild intermittent, moderate persistent, or severe persistent
asthma, based on symptoms and the results of diagnostic tests. Their care
depends, to some extent, on this categorization (NHLBI, 1997).
Asthma episodes (also referred to as exacerbations or acute symptoms) are
periods when bronchial constriction restricts airflow and causes the
symptoms described above. The episodes can be triggered by allergens,
irritants such as cigarette smoke, odors, pollution, sulfite preservatives,
weather changes, and emotions. Prolonged severe attacks may be
precipitated by common colds (e.g., influenza, rhinovirus). Some drugs
cause short severe attacks (e.g., aspirin, nonsteroidal anti-inflammatory
drugs). Responses are typically triggered within a few hours and may
persist as a hypersensitive response to stimuli for two to three weeks
(Eggleston, 1994).
In recent years the importance of inflammation in asthma has been further
substantiated by research. When inflammation occurs, it is usually
associated with airflow obstruction that is often reversible spontaneously or
with treatment. The inflammation also causes an increase in the existing
bronchial hyperresponsiveness to the triggers listed above. At the
physiological level, asthma results from complex interactions among
inflammatory cells, mediators, and other cells and tissues in the airways
(NHLBI, 1997).
Although asthma may affect individuals throughout their life, the disease
has certain age-specific characteristics and differential diagnosis issues.
These issues need to be considered in the etiology and treatment of asthma
(NHLBI, 1997). This chapter focuses on asthma that is diagnosed in
childhood because that is the most common period of diagnosis, and much
of the care provided for asthmatics is provided during childhood. In
addition, the Agency has a particular interest in costs associated with
childhood asthma. Most of the information provided is relevant to patients
of any age, however, and information is provided to allow the reader to
calculate costs associated with asthma onset at any age.
Asthma is a leading cause of morbidity among children and is the most
commonly cited reason for school absenteeism, accounting for one-third of
all school days lost. It is the most common cause for hospitalization of
children. The median age of onset of asthma is four years; however, more
than 20 percent of children who are diagnosed with asthma develop
symptoms during the first year of life (Eggleston, 1994).
Chapter IV.2 IV.2-3 Cost of Asthma
-------
IV.2.A. 1.2 Sources of Health Statistics Data
Although asthma is one of the most common chronic diseases in the U.S.
and has increased in importance over the past 20 years, surveillance of
asthma trends was very limited until recently. CDC provided summary
data in 1998 for the overall period 1960 to 1995 that described some of the
trends in asthma occurrence (CDC, 1998a). These data include asthma
prevalence(for the years 1980-1994).l They also include asthma office
visits (1975-1995), emergency room visits (1992-1995), hospitalizations
(1979-1984), and deaths (1960-1995). The CDC report noted that the
overall trend has been toward an increase in asthma prevalence and asthma
deaths, with substantial differences in death rates within a single geographic
region. Asthma hospitalizations have increased in some areas and
decreased in others. Surveillance data are not available at the state or local
levels, with the exception of asthma mortality (CDC, 1998a).
The National Center for Health Statistics (NCHS) collected much of the
basic information on which the 1998 CDC surveillance summary was based
(CDC, 1998a). They used the National Health Interview Survey which
provides data on prevalence supplied by patients.2 They also used the
National Ambulatory Medical Care Survey which provides physician office
visit data; the National Hospital Ambulatory Medical Care Survey, which
provides emergency room visit data; and the National Hospital Discharge
Survey, which provides data on in-hospital stays. Mortality data were also
collected from each state (CDC, 1998a). Information is expressed both as
the absolute value (e.g., number of cases) and the rate in the population
(e.g., 2 per 10,000 people).3 Age groupings used in the CDC surveillance
summary are 0 to 4, 5 to 14, 15 to 34, 35 to 64, and >64 years (CDC,
1998a).
Mortality data from CDC (1998a) were limited, so additional information
was also obtained from NCHS's FASTATS (CDC, 1999a). Multiple areas
within this large database were used; they can be accessed through the
website listed in the References section.
1 Prevalence is a measure of how many people have a disease, rather than how many were newly diagnosed
in a particular year.
2 The prevalence of a chronic disease is often determined through patient surveys.
3 It is useful to consider both values because neither value alone can fully express trends and potential
impacts. Rates provide information on changes, standardized to a specific size population. Absolute
values reflect a combination of changes in the rate, along with changes in the underlying population size.
For example, a rate could stay the same, but if the population increased by 20 percent, then the absolute
number of cases would need to increase by 20 percent to maintain the same rate.
Chapter IV.2 IV.2-4 Cost of Asthma
-------
CDC also published a forecast estimate of self-reported asthma prevalence
for the U.S. in 1998 (CDC, 1998b). This estimate was developed using
1995 survey data, 1998 census data, and linear extrapolations of region-
specific increases in asthma prevalence over previous years (CDC, 1998b).
Another report by CDC contains an evaluation of the number and rate of
ambulatory care visits for various diagnostic categories for the period 1993
to 1995 for children under 15 years of age (CDC, 1998c). The National
Hospital Ambulatory Medical Care Survey provided hospital outpatient
data used in this analysis. There are age restrictions in the data (only ages
0 to 14 are covered); however, other data sources were not located for
hospital outpatient visits. The survey data are therefore used both as the
source of the rate for those patients aged 0 to 14, and to estimate the visits
for patients 15 years and older.
Data from the various government statistical summaries are provided
below. In addition, some statistics are calculated using existing data. Most
calculated statistics are for 1994 because this is the most current year for
which prevalence data are available. Prevalence is important because it is a
commonly used denominator in calculating statistics in this analysis, such as
the rates of various services per asthmatic.
An important consideration when reviewing the asthma statistics is the way
the rates are presented. Data on rates are provided in the CDC reports
using the entire general population in the group of interest as a
denominator (e.g., all people, all people aged zero to four, all blacks, etc).
This division gives a somewhat distorted sense of the utilization of services
by asthmatics because both the prevalence of asthma in the population and
the size of the overall population have increased. The rate of service use
could therefore appear to increase in the general population, while it
actually decreased among asthma patients. Consequently, many statistics
are presented using two types of rates for this discussion:
• rate per number in the population — denominator is total number
of people in group of interest (e.g., children aged zero to four), and
rate per number of asthmatics — denominator is number of
asthmatics in the group of interest.
The rates are important because they are used to estimate the proportion of
asthmatics using various services in the cost section of this chapter, which
follows. Because the rates are not generally provided in the CDC reports,
most rates per asthmatic were calculated for this analysis using the CDC
statistics.
Chapter IV.2 IV.2-5 Cost of Asthma
-------
IV.2.A.1.3 Prevalence of Asthma
The self-reported prevalence of asthma increased 75 percent from 1980 to
1994 to 13.7 million people during the period from 1993 to 1994, and the
trend was observed among all races, sexes, and age groups (CDC, 1998a).
Estimates of asthma prevalence in 1998 by CDC, based on population
trends and the pattern of increasing incidence, yielded a CDC-estimated
total of 17,229,000 asthmatics of all ages in the U.S. (CDC, 1998b). This
section focuses on the actual reported values (CDC, 1998a), rather than on
the estimated prevalence (CDC, 1998b).
The increase in asthma prevalence was the greatest for children aged 0 to 4
(160 percent) with a rate increase during the period 1980 to 1994 from
22.2 per 1,000 (2.2 percent) to 57.8 per 1,000 (5.8 percent).4 The current
rate indicates that more than 1 in every 20 very young children have
asthma, with approximately 1.3 million children in this age group having
asthma during the most recent survey period (1993-4), in contrast with
360,000 in 1980 (CDC, 1998a). As noted above, the average age of
diagnosis is four years, so half of all children will be diagnosed with asthma
after the age of four. There was a 74 percent increase in prevalence in the
age group 5 to 14 from 1980 to 1994. By the age of 18, the prevalence is
7.5 percent. This is in contrasts to the general population (all ages), with a
prevalence of 5.7 percent. The prevalence rates are near or below 5
percent among age groups over 18 (CDC, 1998b). Among adults 35 years
and older, the asthma rate was 44.6 per 1,000 (4.5 percent) in 1993-4.
Asthma prevalence is 14 percent higher among blacks than whites, with
rates of 57.8 versus 50.8 per 1,000 (5.8 versus 5.1 percent) during the
most recent survey period (CDC, 1998a).
Asthma is most commonly diagnosed in children by the age of five and in
adults in their thirties, although onset of symptoms can occur at any age.
About ten percent of patients are first diagnosed with asthma after age 64
(AAFA, 1999).
Family income is related to asthma incidence, with those under the age of
45 having a prevalence of eight percent when family incomes were less
than $10,000 compared to six percent for those with incomes greater than
$35,000 (CDC, 1999a).5
4 All rates are expressed in terms of the entire population rather than just asthma patients, unless otherwise
stated. For example, 10 per 1,000 indicates 10 cases among all the population, of whom some are asthma
patients. This standard is used for all rates expressed in this section, unless otherwise noted.
5 NCHS does not provide much detail on the age group of most interest to this analysis in their summary
statistics (they group all those under the age of 45 together for many statistics). Consequently, the data are
reported for this group, although those under 18 are of most interest.
Chapter IV.2 IV.2-6 Cost of Asthma
-------
Many studies have reported an association between urban residence and
asthma. The NCHS aggregate statistics (not provided by individual age
group) support this association, with a rate of 5.8 percent in metropolitan
statistical areas compared with 5.1 percent in non-metropolitan areas. The
rate is slightly higher in central city areas compared to non-central city
areas within a metropolitan area (CDC, 1999a). Some areas have reported
much higher rates, including Chicago, Bronx, and an area in Louisiana
(CDC, 1998b).
There is substantial variation in asthma rates among states, although data
are limited. In three states where surveys of adults were recently carried
out, self-reported medically diagnosed active asthma in the previous year
occurred in 6.6 percent and 7.4 percent of the population in Oregon in the
years 1995 and 1996 respectively. Of those, 9 percent received emergency
care for asthma in the preceding year. In New Hampshire, 11 percent of
respondents reported having medically-diagnosed active asthma, and 19.9
percent of males and 44.6 percent of females had used medication
(CDC,1998a). In Washington state, 10.8 percent of adults reported having
asthma at some point in their life and 12.8 percent of children had asthma.
Family incomes below $20,000 per year were associated with an
approximately two-fold increase in asthma prevalence among children
(MMWR, 1999). More state data will be available in the future, allowing
national patterns to be evaluated.
IV.2.A.1.4 Office Visits
Use of office visits has increased in approximate correspondence to the
prevalence of asthma. From 1975 to 1995, the estimated annual number of
office visits for asthma increased from 4.6 to 10.4 million. The lowest rate
of office visits was among people aged 15 to 34 years. The annual rate of
office visits for asthma during the period 1993 to 1995 was 39.6 for whites
and 43.8 for blacks per 1,000. The rate was higher for pediatric cases:
50.3 for ages 0 to 4 and 51.5 for ages 5 to 14 (CDC, 1998a).
The rates of office visits among asthmatics was calculated for this analysis
using the prevalence statistics and the number of office visits reported by
CDC (1998a). In 1994 there were 1,024,000 office visits for asthma
among children aged 0 to 4 and a population of 1,300,000 children aged 0
to 4 with asthma. These statistics yield a rate of 780 per 1,000 asthmatic
children for age 0 to 4, yielding an average annual office visit rate of 0.78
per patient. Using the same calculation for the age group 5 to 14, the rate
was 720 per 1,000, yielding an annual average visit rate of 0.72 per patient.
As with all services discussed in this section (e.g., hospitalization,
emergency room use), it was not possible to determine what percentage of
patients had more than one office visit more in a year, nor the distribution
of visits among patients. However, it was possible to estimate the average
number of visits per patient (CDC,1998a). For costing purposes, the rates
Chapter IV.2 IV.2-7 Cost of Asthma
-------
make it possible to calculate an average cost per patient, even though some
patients will have much higher or lower costs, depending on actual
utilization of medical services.
IV.2.A. 1.5 Hospital Outpatient Visits
There is a trend toward increasing provision of outpatient services by
hospitals. These may function as clinics, or be more similar in practice to a
group medical practice within a hospital. These outpatient visits provide
the same kind of care that is provided in a physician's office; it is therefore
anticipated that the costs will be the same (discussed in Section IV.2.B,
below). It is important to determine the number of these visits that occur
annually, to estimate the total number of medical visits and to determine
the number of visits per year by a patient.
The CDC report "Ambulatory Health Care Visits by Children: Principal
Diagnosis and Place of Visit" provides information on the number of
children and rate per 100 children for ambulatory visits to physicians'
offices, hospital outpatient departments, and emergency rooms (CDC,
1998c). Other sources of information for physicians' visits and emergency
room visits are preferable for most statistics because they cover all ages,
while this report covers patients only up to age 14. This report is useful,
however, for evaluating patient utilization of hospital outpatient services,
which the other sources do not describe.
CDC (1998c) reports an outpatient visit rate of 0.8 per 100 children aged 0
to 14 per year, during the 1993 to 1995 period, with 469,000 visits per
year. Dividing the number of visits by the total number of asthmatic
children in this age group of 4,090,000 children yields an average annual
visit rate of 0.115 per asthmatic child. To evaluate the use of outpatient
hospital care among those over 14 years of age, it was assumed that the
relationship between this type of care and physician's office visits would be
the same across ages. The ratio of hospital outpatient visits to office visits
among children aged 5 to 14 is 0.115/0.72 = 0.159. The rate of office
visits among all asthmatics in 1994 was estimated to be 0.607 per year
(including adults and children). By applying the outpatient/office visit
ratio determined for children (0.159) to the office visit rate for all ages
(0.607), the rate of outpatient use per patient can be estimated as: 0.159 x
0.607 = 0.10.
IV.2.A. 1.6 Combined Medical Visit Rate
When the hospital outpatient visit rate is added to the office visit rate, the
total annual physician visits per patient can be estimated. As described
above, the per patient rate of office visits among all asthmatics in 1994 was
estimated to be 0.607 per year and the rate of outpatient use was estimated
to be 0.10. Summing these two values yields an overall patient visit rate of
0.707 for the average patient. In subsequent discussions, the outpatient
and physicians' office visits are discussed together as physician's office
Chapter IV.2 IV.2-8 Cost of Asthma
-------
visits, due to their similarity and to the fact that a physician is usually seen
in either location.
An estimate of the medical visit rate for children can also be calculated.
CDC (1998c) reports a physician visit rate of 5.3 per 100 children (similar
to that reported in CDC 1998a), with 3,029,000 visits per year.
Combining the hospital outpatient visits with the office visits yields
3,498,000 visits per year. This total yields an overall rate of 0.855 per
patient aged 0 to 14 (2,498,0000/4,090,000). An annual visit rate can be
calculated for the two childhood age divisions. For zero- to four-year-olds,
the office visit rate listed above was 0.78 per patient. Adding the rate of
0.115 from above yields 0.895 per year. For five- to fourteen-year-olds,
the office visit rate of 0.72 per patient was combined with the rate of 0.115
to obtain a rate of 83.5 per patient (taken directly from or calculated from
statistics provided in CDC, 1998a).
IV.2.A. 1.7 Emergency Room Use
Only recent data (1992-1995) are available on emergency room use, so
longitudinal trend analysis is limited.6 In 1994 the overall rate was 6.3 per
1,000 and 0.117 per asthmatic. There was not a statistically significant
change in use over the four-year period among all patients (CDC, 1998a),
but the rate of use by black asthmatics increased by 50.1 percent. Among
white asthmatics the increase was 4.1 percent (calculated from statistics
provided in CDC, 1998a and CDC, 1998c). Blacks had consistently higher
rates of use than whites with a rate in 1995 (the most recent year with data)
of 22.9 per 1,000 (population) versus 4.9 in whites. In 1994, the likelihood
of an asthma patient visiting an emergency room was 0.337 for blacks and
0.087 for whites (taken directly from or calculated from statistics provided
in CDC, 1998a).
Rates were higher for younger than for older patients. In 1994, the rate for
children aged 0 to 4 was 14.5 per 1,000 people. For ages 5 to 14 the rate
was 8.0 per 1,000 people. Rates of emergency room use in 1994 among
asthmatic children were 239 per 1,000 (23.9 percent) for ages 0 to 4, and
112 per 1,000 (11.2 percent) for ages 5 to 14. Rates per 1,000 in the
general population were very similar among the age groups within the span
from 15 to 64 (approximately 7 per 1,000) and much lower for the elderly
(65+ years) at 3.0 per 1,000. Rates among asthmatics in those age groups
were lower than the rates for children (taken directly from or calculated
from statistics provided in CDC, 1998a).
As discussed above under office visits, it was not possible to determine
what percentage of patients used an emergency room more than once in a
given year, so the values provided are an average across all patients. As
6 Use is defined by CDC as admission for asthma listed as first diagnosis.
Chapter IV.2 IV.2-9 Cost of Asthma
-------
discussed in Section IV.2.B below, some patients have a much higher rate
of emergency room use and hospitalization than the average patient.
IV.2.A.1.8 Hospitalization
Asthma is the ninth leading cause of hospitalization nationally (CDC,
1998b). Although the number of hospitalizations has increased
substantially during the period 1980 to 1994 from 386,000 to 466,000, the
rate of hospitalizations has not changed significantly (17.6 to 18.1 per
10,000 people). When the increase in asthma prevalence is considered, the
hospitalization rate among asthma patients has actually decreased over the
years considered. The hospitalization rate among asthmatics in 1994 was
0.034 per patient (calculated from data on prevalence and hospitalization
provided in CDC, 1998a)
There have been dramatic shifts in hospitalization usage by whites and
blacks during the period of study (1980 to 1994). While overall
hospitalization rates for asthma patients have decreased by approximately
25 percent during that period, hospitalization among blacks has increased
by 37 percent from 26.0 to 35.5 per 10,000. During the most recent
observation period (1994) blacks had a hospitalization rate that was more
than three times greater than whites, even though blacks have only a 14
percent greater prevalence of the disease (see prevalence discussion above)
(CDC,1998a).
Hospitalization of patients of different ages has also changed during the
observation period. Hospitalization of patients aged 35 and older has
declined, but the number (not rate) of hospitalizations of very young
children has increased dramatically. Among zero- to four-year-olds there
has been a 45 percent increase in the hospitalization rate, from 34.3 to 49,7
per 10,000. This rate should be compared with an overall rate increase in
the prevalence of asthma among this age group (discussed above) of 160
percent, which is substantially greater than the rate of hospitalization. (The
total number of children in this age group with the disease in 1980 was
360,000 versus 1,280,000 in 1994). Using these numbers with the number
of hospitalizations in 1980 (56,000) and 1994 (97,000), the hospitalization
rates were estimated. In 1980, the rate of hospitalization among young
asthma patients was approximately 16 percent, and in 1994 it was
approximately 8 percent.7 The hospitalization rate among asthma patients
in this age group is therefore approximately one-half of what it was in 1980
(taken directly from or calculated from statistics provided in CDC, 1998a).
There has been a very modest increase among older children and young
adults in the hospitalization rate per 10,000 people. For ages 5 to 14, the
7 There is no way to determine which patients were hospitalized more than once in a given year, but these
overlaps are not expected to occur more frequently in 1994 than in 1980. Due to this uncertainty, the
percents provided are approximate values.
Chapter IV.2 IV.2-10 Cost of Asthma
-------
increase has been from 15.9 to 18.0 per 10,000; for the age group 15 to 34,
the increase has been from 8.7 to 10.0 per 10,000 (taken directly from or
calculated from statistics provided in CDC, 1998a). As with very young
children discussed above, the overall asthma hospitalization rate among
asthma patients in this age group has declined substantially.
Evaluating the reasons behind trends observed in hospitalization of asthma
patients is difficult because many factors impact hospitalization, including:
• changes in the health status of patients (e.g., severity of the
disease, management of the disease),
• changes in guidelines for admission to hospitals due to
managed care,
• cost containment efforts or other considerations,
• a shift to emergency room use without admission rather
than admitting patients,
patient preferences,
• more self-administered therapy, or
other factors.
The reduction in hospitalizations should not be interpreted as an
improvement in the health status of patients, without additional information
on the causes of the decreases in admissions.
There are regional differences in hospitalization; however, these are not
discussed in this analysis. The differences may be reviewed in the summary
provided by CDC (CDC,1998a)
IV.2.A.1.9 Mortality
Mortality from asthma is relatively rare and directly impacts the medical
costs only by reducing the medical costs for those patients who die.
Mortality is a sentinel event, however, as infant mortality is, in expressing
the overall health or sickness of a population or population subgroup.
Mortality is usually preceded by considerable medical treatment for illness
(although not in every case for asthma). As such, it is useful to evaluate
the patterns of mortality to gain some insight into medical care. Because
mortality has little direct effect on the cost of medical care for asthma,
limited data are provided below.
The CDC has provided a summary of mortality numbers and rates, by five-
year increments over the past two decades (CDC, 1998a). They note that
changes in the ICD codes and diagnosis during that period complicate the
analysis of trends (CDC,1998a). Asthma-related deaths vary considerably
by age group, with 85 percent of deaths occurring among people over 34
years of age. This rate may be due to the overlap of asthma with chronic
Chapter IV.2 IV.2-11 Cost of Asthma
-------
obstructive pulmonary disease (COPD), which usually occurs in older
individuals. Decreases in airway function, which is reversible in asthma, is
not reversible with COPD (CDC,1998a).
During the 1993 to 1995 observation period, the annual death rate from
asthma was 17.9 per 1,000,000 population (CDC, 1999b). Using the total
asthma population of 13,690,000, and 5,429 deaths (both from 1994), the
death rate among asthmatics was 397 per 1,000,000 (an annual probability
of dying of 0.0004). Among whites the mortality rate was 15.1 per
1,000,000 general population; among blacks the rate was 38.5 per
1,000,000 general population (CDC, 1999b). In light of the fact that the
prevalence rate of asthma among blacks and whites is very similar (50.8
versus 57.8 per 1,000 general population, respectively (CDC, 1998a)), the
difference in death rates is striking.
In addition to the source used above, CDC also provides a more detailed
age-, race-, and sex-specific summary of mortality related to asthma, as
well as other diseases (CDC, 1999b). These data can be used to determine
the age- and race-specific mortality rates. There is no rate listed (numbers
are too small to estimate reliably) for the first year of life. For ages 1 to 4
and 5 to 9, the rate is 0.2 per 100,000 general population. For ages 10 to
14 and 15 to 19, the rate is 0.4 per 100,000. The rate very gradually
increases as age increases for subsequent ages (taken directly from or
calculated from statistics provided in CDC, 1998a).
There are clear racial differences in deaths due to asthma among children.
Among blacks, the numbers are too small to estimate reliably under age
five. From that age forward, the rates are substantially higher than for
whites. The black/white ratio for children is as follows:
Age Black Rate/White Rate
5 to 9 0.8 per thousand / too small to estimate
10 to 14 1.2/0.2 = 6-fold difference
15 to 19 1.0/0.3= >3-fold difference
The differences in mortality persist through adulthood. The overall
mortality rate due to asthma is 3.86 x 10 (-4) among whites and 6.31 x 10
(-4) among blacks with asthma.
A number of studies have evaluated differences in medical care,
hospitalization, emergency room use, and mortality between blacks and
whites. A discussion of the hypotheses offered for these differences is
beyond the scope of this chapter, which is focused on direct medical costs.
One frequently offered observation is that better patient outreach and
education reduces severe episodes and resulting emergency room visits and
Chapter IV.2 IV.2-12 Cost of Asthma
-------
hospitalizations by improving control of the disease. This improvement is
likely to have an impact on mortality. (A subset of these patients is
discussed below as "high-use" patients).
IV.2.A.1.10 Summary of Asthma Statistics
Table IV.2-1 contains information on prevalence and some aspects of
medical care for asthma. Most notable statistics in the table were discussed
in preceding sections. Statistics that are used later in cost calculations are
bolded with an * at the beginning of the entry.
Chapter IV.2 IV.2-13 Cost of Asthma
-------
Table IV. 2-1 : Asthma Statistics: Data on overall population statistics, age, and racial
characteristics a- bi Ci d
Characteristic
Statistic
Source
Prevalence
number 1994
number forecast 1998
number of blacks 1994
number of whites, 1994
overall rate per 1 ,000
overall rate under 18 years per 1 ,000
number of children aged 0 to 4 in 1 980
number of children aged 0 to 4 in 1 994
rate for ages 0 to 4 per 1 ,000 in 1980
rate for ages 0 to 4 per 1 ,000 in 1994
increase in prevalence rate in children 0 to 4
from 1980 to 1994
number of children aged 5 to 14 in 1980
number of children aged 5 to 14 in 1994
increase in prevalence rate in children 5 to 14
from 1980 to 1994
13,700,000
17,200,000
1,880,000
10,700,000
57 (5.7%)
75 (7.5%)
360,000
1,300,000
22.2 (2.2%)
57.8 (5.8%)
160%
1,520,000
2,790,000
74%
CDC, 1998a
CDC, 1998b
CDC, 1998a
CDC, 1998a
CDC, 1998a
CDC, 1998a
CDC, 1998a
CDC, 1998a
CDC, 1998a
CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
CDC, 1998a
CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Office Visits e
rate 1994 all ages per 1,000
rate 1994 per asthmatic patient
rate 1 994 ages 0 to 4 per 1 ,000
rate 1994 per asthmatic patient
rate 1 994 ages 5 to 1 4 per 1 ,000
rate 1994 ages 5 to 14 per asthmatic patient
rate 1994 among blacks per 1,000
rate 1 994 among whites per 1 ,000
34.1
0.607
50.3
0.78
51.5
0.72
43.8
39.6
CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
CDC, 1998a
CDC, 1998a
Chapter IV.2
IV.2-14
Cost of Asthma
-------
Table IV. 2-1 : Asthma Statistics: Data on overall population statistics, age, and racial
characteristics a- bi Ci d
Characteristic
Statistic
Source
Outpatient Hospital Visits
rate 1994 among children aged 0 to 14 per 100
children
rate 1994 among children aged 0-14 per
asthmatic child
rate 1994 extrapolated to patients over age 14
(see text for method)
0.8
0.115
0.10
CDC, 1998c
Calculated from data provided in
CDC, 1998cand 1998a
Calculated from data provided in
CDC, 1998cand 1998a
Combined Office/Outpatient Visits
rate 1994 for children 0 to 4 per asthmatic
patient
*rate 1994 extrapolated to all asthmatics
expressed per asthmatic patient (see text
for method)
0.855
0.707
Calculated from data provided
in CDC, 1998c and 1998a
Calculated from data provided
in CDC, 1998cand1998a
Emergency Room (ER) Visits (no historical data were available) '
rate 1994 all ages per 1,000
*rate 1994 of ER visits per asthmatic
rate 1 994 ages 0 to 4 per 1 ,000
rate 1 994 ages 5 to 1 4 per 1 ,000
rate 1994 among blacks per 1,000
rate 1 994 among whites per 1 ,000
rate of admissions among black asthmatics
1994 (635,000 admissions per 1,880,000
cases)
rate of admissions among white asthmatics
1994 (927,000 admissions per 10,700,000
cases)
increase in rate among blacks during period
1992 to 1995 (4 years) per 1,000 general
population (228.9 in 1995/151.9 in 1992)h
increase in rate among whites during period
1992 to 1995 (4 years) per 1,000 general
population (48.8 in 1995 versus 46.8 in 1992)
6.3
0.117
14.5
8.0
19.1
4.6
0.337
0.087
50.1%
4.2%
CDC, 1998a
Calculated from data provided
in CDC, 1998cand1998a
CDC, 1998a
CDC, 1998a
CDC, 1998a
CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Hospitalization (described as number of discharges) 9
number 1980
number 1994
386,000
466,000
CDC, 1998a
CDC, 1998a
Chapter IV.2
IV.2-15
Cost of Asthma
-------
Table IV. 2-1 : Asthma Statistics: Data on overall population statistics, age, and racial
characteristics a- bi Ci d
Characteristic
rate per 1,000 in 1980
rate per 1,000 in 1994
*rate of hospitalization among asthmatics
all ages in 1994
change in rate per asthma patient for all
patients
rate of hospitalization among asthmatic
children aged 0 to 4 in 1980
rate of hospitalization among asthmatic
children aged 0 to 4 in 1994
change in rate among asthmatic children aged
0 to 4 from 1980 to 1994
change in rate per 10,000 for children aged 0 to
4 from 1980 to 1994
change in rate per 1 ,000 for blacks from 1980
to 1994
Statistic
1.76
1.81
3.4%
-25%
16%
8%
-50%
+45%
+37%
Source
CDC, 1998a
CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Calculated from statistics
provided in CDC, 1998a
Mortality
rate 1994 per 1,000 general population
rate 1994 among asthmatics per patient
rate 1994 among whites per 1000
rate 1994 among blacks per 1,000
rate 1994 among asthmatic blacks per patient
rate 1994 among asthmatic whites per patient
0.0179
0.0004
0.0151
0.0385
6.31 x 10(-4'
3.86 x 10(-4'
CDC, 1999b
Calculated from data provided in
CDC, 1998a
CDC, 1999b
CDC, 1999b
Calculated from data provided in
CDC, 1998aand 1999b
Calculated from data provided in
CDC, 1998aand 1999b
Chapter IV.2
IV.2-16
Cost of Asthma
-------
Characteristic
Statistic
Source
Table IV.2-1: Asthma Statistics: Data on overall population statistics, age, and racial
characteristics a-bi Ci d
a. All numbers and rates are annual. See text for additional detail.
b. All rates are per person in the general population unless otherwise noted. When rates are expressed per
asthmatic, the number of asthmatics is the 1993-4 value, unless otherwise noted.
c. The study period 1993-1994 is listed as 1994, while the study period 1979-1980 is listed as 1980.
d. Age- and race-related statistics are presented only when there are differences across the ages or races. If
the values are very similar (e.g., as with overall prevalence of the disease among whites and blacks), no data
are listed. Data are not provided on the racial designation "other," which comprises a very small portion of
asthmatic patients and represents a diverse group of individuals who are categorized as neither black or white
(e.g., Hispanic, Native American, Asian).
e. It was not possible to determine the percentage of patients with more than one office visit per year. The
statistics presented do not indicate the total number of persons treated, but rather the number of visits that
occurred.
f. It was not possible to determine the percentage of patients with more than one visit to the ER. The statistics
presented do not indicate the total number of persons treated, but rather the number of visits that occurred.
g. It is important to note the differences in rates when expressed per population versus per asthmatic patient.
The overall number of asthmatic patients has increased substantially, as well as the rate of asthma in the
population. Increases in use of services (e.g., office visits, ER visits, hospitalizations) must therefore be
considered in light of the number of asthmatics, rather than just the overall population.
h. The percentage change is calculated as: {(y2 - y1)/y1} x 100, where y1 = the number in earlier years and y2
= the number in the most current year considered.
* Statistics that are bolded (but not italicized) with an * are used in calculating costs later in this chapter.
Chapter IV.2 IV.2-17 Cost of Asthma
-------
IV.2.A. 1.11 Variations in the Management and Use of Medical
Services Among Asthmatic Patients.
Variations in disease management and the use of medical services occur on
the basis of individual patient characteristics, including the severity of the
disease. Considerable variation also results from the degree to which a
patient has, and follows, an adequate asthma management plan. Under
optimal circumstances, most patients would have a long-term asthma
management plan that managed the symptoms of asthma sufficiently well
that they did not have asthma episodes requiring medical intervention in
emergency rooms or as inpatients in hospitals. Theoretically, the
management plan would control inflammation sufficiently well with long-
acting therapies so that short-acting therapies in response to airway
restrictions were not routinely required. NHLBI has issued guidelines
designed to address this issue. In practice, many patients either do not
receive drug plans, or do not follow drug plans that manage asthma at this
level of control.
Considerable evidence supports the contention that many, if not most,
patients do not receive or follow treatment plans that are consistent with
NHLBI guidelines, although the degree of compliance varies. Recent
studies suggest that many patients who are seen in emergency rooms report
not having treatment plans or not taking anti-inflammatory medication on a
regular (proactive) basis. It has also been estimated that over one half of
all people who use inhalers do not use them properly (AAFA, 1999).
It has been suggested, anecdotally, that it will take some time for NHLBI
guidelines to become familiar and comfortable to most primary care
physicians. In addition, there is not consensus among all physicians that
they should prescribe according to NHLBI guidelines. Another factor is
constraints on medical providers' time for patient education (due to cost
containment or other reasons). This factor may be an important
contributor to the seemingly high percentage of patients who do not use
drug therapies that are optimal. Patients may also resist or not remember
treatment plans. Regardless of the reasons behind the noncompliance with
NHLBI guidelines, the fact of its occurrence is a reality. This
noncompliance poses problems in evaluating both treatment of and costs
for asthma.
One approach to this problem is to evaluate the studies that have been done
on patients before and after training in optimal asthma management. Such
an evaluation can provide information that can be used to estimate
variations in management and services that occur and their costs. Studies
of educational interventions have focused mainly on populations that have
had a high level of acute care (e.g., emergency room use or hospitalization)
in the recent past. Although these populations are not representative of the
cross-section of asthma patients, the studies provide insight into both 1) the
Chapter IV.2 IV.2-18 Cost of Asthma
-------
high level of costs incurred by some patients, and 2) the savings in medical
services and costs that may be gained through patient compliance with
asthma management plans.
It has been estimated that the costs of hospitalizations and emergency room
(ER) visits for asthma together comprise about 73 percent of the total
direct expenditures on asthma for children (age 17 and under) in the United
States (Weiss et al., 1992). It is also generally believed that most
hospitalizations and ER visits for asthma could be avoided by following an
asthma management protocol, such as NHLBI guidelines (see, for example,
Weiss et al., 1992; Coventry et al., 1996; and Higgins et al., 1998).
Although there are no national statistics on the costs of asthma for those
asthmatics who do not follow such guidelines, several studies of the cost
savings resulting from asthma management education programs support the
hypothesis that the asthma-related medical costs for those asthmatics who
do not follow an asthma management protocol are likely to be substantially
greater than the costs for those who do follow such a protocol - largely
due to increased utilization of hospitals and emergency rooms, which are
relatively more costly than physician and pharmaceutical services.
Several prospective studies have compared utilization and the
corresponding costs of medical services by asthma patients for a period of
time (usually a year) before an intervention program to those for the same
period of time after the program. These studies focused on patients who
had used acute care services (e.g., hospitalizations or emergency room
visits) recently, and most evaluated patients who had a relatively high rate
of utilization and a moderate or severe form of asthma. These studies
therefore do not represent the experiences of the average patient. The
data from these studies can be used to evaluate the costs of patient services
before and after intervention for patients with a high use of services at the
outset. By multiplying the average cost of utilization (e.g., for
hospitalization) by the number of occurrences of utilization, before and
after an intervention, and for each type of medical service considered, the
average per-patient cost before the intervention can be compared to the
average per-patient cost after the intervention. Each study thus affords an
estimate of the extent to which complying with an asthma management
protocol may reduce asthma-related medical costs — or, conversely, the
extent to which failure to comply may increase these costs for some
patients.
There are several factors, including the exact nature of the intervention, the
utilization categories considered, and the severity of asthma in the study
subjects, that would likely affect the ratio of before-intervention costs to
after-intervention costs; these factors vary from study to study. Moreover,
because these studies (some of which are pilot studies) were often based on
relatively small samples and may not necessarily be representative of the
population of asthmatic children as a whole, their results should not be
Chapter IV.2 IV.2-19 Cost of Asthma
-------
taken as definitive. They do, however, afford a rough idea of the
magnitude and range of the extent to which the asthma-related costs of the
"noncompliant" population of asthmatic children may exceed those of the
"compliant" population of asthmatic children. The results of these studies
are summarized in Table IV.2-2.
Table IV. 2-2: Summary of Asthma Intervention Studies
Higgins et al.,
1998
Westley et al.,
1997
Greineder et
al., 1995
Gaioni et al.,
1996
Utilization categories included in study:
Hospitalization
ER visits
Office (clinic) visits
Chest radiographs
Inhaled anti-inflammatory
drugs
Beta-2 agonists
Considered children
separately?
Number of subjects
(children)
Were the estimated savings
net of the costs of the
intervention?
Ratio of before-intervention
to after-intervention costs
X
X
X
X
X
X
yes
61
no
7.62
X
X
X
yes
43
no
3.60
X
X
yes
53
yes
3.84
X
X
no
207
yes
2.01
These studies suggest that the asthma-related medical costs of children in
the "noncompliant" population could be from twice to over seven times the
costs of children in the "compliant" population. As noted above, however,
several factors are likely to affect these ratios. First, the intervention
programs differed from one study to another. Second, some studies took
into account the cost of the intervention program itself, and considered the
savings net of those costs, while other studies did not. Third, these studies
generally selected asthmatics whose asthma was more severe than average,
focusing primarily on patients with considerable hospitalizations and
emergency room use prior to the study. Both their "before intervention"
and "after intervention" acute care use rates are much higher than the
national average. This frequency suggests that they would be categorized
as patients with moderate or severe asthma. Fourth, these patients may
have had much poorer compliance with management plans (or a lack of a
plan) than other patients, leading to their very high acute care use rate at
the outset of study participation.
Chapter IV.2
IV.2-20
Cost of Asthma
-------
The results of these studies shed light on the cost impacts that the
variations in treatment may have. No single study fully illustrates national
patterns of treatment, but the information in these and other studies is used
later in this analysis as the basis for estimating medical costs for some
patients. Additional detail is provided on the studies in Section IV.2.B.
IV.2.A.2. Concurrent Effects
Asthma usually occurs concurrently with allergies (as described above),
although this is not always the case with adult-onset asthma. In the elderly
it is often associated with other respiratory diseases such as chronic
obstructive pulmonary disease (COPD). It has also been linked to other
upper and lower respiratory tract diseases (e.g., sinusitis, rhinitis). It is not
clear in most cases whether one disease triggered another.
As with most pharmaceutical s, those that are used to treat asthma have
adverse side effects. Long-term use of corticosterioids have been
associated with growth retardation in children and other serious health
effects. There are also potential health risks listed for most other drugs
used to control asthma. It is beyond the scope of this analysis to evaluate
the health risks and potential direct medical costs of these secondary
illnesses. These illnesses, however, are likely to occur in some patients and
lead to an underestimate of medical costs when not considered.
Asthma may exacerbate cardiac and other problems. Asthma often leads to
restricted activities in patients and a sedentary lifestyle that has been
associated with numerous health problems. Asthma is a leading cause of
activity restriction in children (NCHS web site, 1999). Perhaps most
significantly, asthma during childhood has been associated with permanent
structural changes in the lungs that are associated with adult respiratory
diseases. This is one reason for the current focus in medical services on
reducing episodes of asthma and the severity of the disease. Quantitative
data were not obtained on the concurrent effects that result from the
adverse physiological impacts of asthma. Due to the lack of information on
these impacts, however, the cost estimates provided in this chapter, which
focus solely on costs associated directly with asthma treatment, will
underestimate total medical costs.
IV.2.A.3 Causality and Special Susceptibilities
Asthma usually begins in childhood and is often associated with atopy —
the genetic susceptibility to produce IgE (an immune response) in response
to exposure to common environmental allergens. Atopy occurs in 30 to 50
percent of the asthmatic population, and occurs in the absence of asthma in
some individuals. It is the strongest predisposing factor for the
Chapter IV.2 IV.2-21 Cost of Asthma
-------
development of asthma. In young children who wheeze in response to viral
infections, the presence of allergy in the child or their family is very
strongly associated with asthma throughout childhood (NHLBI, 1997).
Other risk factors are neonatal lung disease, especially in infants with
reduced lung volumes, and respiratory infections. Respiratory syncytial
virus (RSV) has been particularly highlighted in association with asthma.
Approximately half of all children with RSV bronchiolitis develop chronic
asthma (Eggleston, 1994).
In adults asthma occurs both with and without IgE responsiveness.
Without it, asthma is often associated with sinusitis, nasal polyps, and
sensitivity to aspirin. Although the genesis may differ, the inflammatory
process is similar to that seen in atopic asthma. Workplace exposures to
some materials can also cause clinical signs of asthma, which may persist
after the workplace exposure has ceased (NHLBI, 1997). Because asthma
is often associated with atopy in response to exposure to common
environmental allergens (NHLBI, 1997), these allergens are important
asthma triggers. In adults asthma occurs both with and without IgE
responsiveness and has been linked to numerous environmental agents.8
Numerous studies have identified air pollutants (e.g., particulate matter and
ozone) as contributors to asthma (EPA, 1996). Hospitalizations for asthma
have been shown to increase during air pollution episodes. Passive
cigarette smoke (ETS) is also a strong instigator of asthma, with multiple
studies showing associations between parental smoking and childhood
asthma. EPA has provided a detailed summary of literature related to ETS
through 1992 (EPA, 1992).
There has been an increase, as shown in the statistics presented earlier, in
the incidence of asthma, asthma hospitalizations, and asthma deaths in
recent decades. The specific causes of this increase are not known,
although it has been hypothesized that air pollution and smoking in the
home may be contributing factors.9
Appendix IV.2-A contains a list of many of the chemicals known or
suspected of causing asthma that have been reported in the EPA Hazardous
Substances Data Base (HSDB). High quality human dose-response studies
are uncommon because non-pharmaceutical chemicals are rarely tested on
8 The medical literature on asthma usually refers to external agents that can trigger asthma as
"environmental factors," although they do not necessary mean "environmental" in the sense in which it is
used in this Handbook. The environmental factors referred to in the literature include any aspects of the
environment that are external to the body and include, but are not limited to, environmental pollutants.
9 Maternal versus paternal smoking is a stronger risk factor for childhood asthma, and more women smoke
now than in the first half of the century.
Chapter IV.2 IV.2-22 Cost of Asthma
-------
humans, which limits the information that can be obtained directly from
human studies. Animal studies provide an additional source of information
for links between a pollutant and disease. Potential causality of asthma as
indicated by the HSDB toxicity excerpts was the only requirement for
including a chemical in this table.
Regarding special susceptibility, it is difficult to determine whether an
increase in the occurrence of a disease within a population group is due to
genetic susceptibilities of the group, or a common characteristic that is
unrelated to their genetic predispositions. That is the case with asthma,
which affects a disproportionate number of socioeconomically
disadvantaged minority children who often live in urban areas. It is likely
that a combination of factors are at work in this case, including nutritional
status, the presence of numerous pollutants in the environment, the greater
likelihood that someone in the home smokes, less prenatal care, and other
factors.
In addition to a greater likelihood of the occurrence of asthma, there is a
greater likelihood that the same populations will be hospitalized for asthma,
and that they will die of the disease (as shown in Table IV.2-1). Both of
these endpoints indicate a much more severe case of asthma than what is
routinely observed and can be treated through periodic office visits and
self-medication with physician-prescribed treatments. At present there are
clearly higher risks of the disease associated with the characteristics
described above. These have environmental justice implications for
benefits evaluations that focus on asthma.
Although socioeconomic and geographic factors may be the primary
factors in the observed differences in asthma rates in the national
population, genetic differences may also play an important role. This has
recently been found to the case with some types of cardiovascular disease;
African-Americans appear to have higher risks in some studies even when
socioeconomic and other factors are carefully controlled. The rapid
increase in asthma incidence and serious consequences has made careful
evaluation of the risk factors for this disease difficult, and it has yet to be
determined whether there are large differences within the U.S. population
in the inherent (genetic) susceptibility to the disease (NHLBI, 1997).
IV.2.A.4 Treatment and Services
This chapter provides cost estimates that are based on a description of
specific treatments and services provided to asthma patients. It is unlike
chapters that use cost estimates obtained from journal articles or other
sources. Because the cost estimates are generated for each specific service,
very detailed treatment information is provided along with the costs of each
Chapter IV.2 IV.2-23 Cost of Asthma
-------
service. Consequently, treatment is more appropriately discussed in
Section IV.2.B on cost. This section provides only a brief description of
asthma treatment and services.
Treatment of asthma involves three phases: 1) initial diagnosis of the
disease with efforts directed at immediate stabilization of the patient, 2)
ongoing care to minimize episodes and maximize the quality of life of the
patient, and 3) acute care for treatment of asthma episodes.
Depending on the severity of the asthma episode at initial diagnosis, a
variety of treatment strategies are employed and may be provided on either
an inpatient or outpatient basis. Follow-up care and long-term
management of the disease involves multiple strategies. A major goal of
asthma treatment is to enable patients to control their symptoms through
the use of medications that they self-administer, and to control their
exposure to situations that trigger their asthma episodes. This patient-
centered focus requires determining the best long-term disease
management approach and considerable training by medical personnel to
achieve the objective of enabling the patient to live a life as free from
asthma symptoms as possible (NHLBI, 1997).
Most asthma is managed through the use of drugs and avoidance of asthma
triggers (e.g., allergens, irritants, behavior such as exercise in cold
environments). When management is not adequate, asthma episodes occur
that require more aggressive care, either provided at home or by a medical
professional. These two scenarios, long-term management and acute
episodic treatment, are reflected in the drug therapies offered to patients.
Two major types of drugs are used to address these two circumstances: 1)
long-term control agents: anti-inflammatory drugs that are self-
administered;10 and 2) acute control agents: bronchodilators that relieve
acute symptoms (also self-administered, except in crisis). The drug
therapies comprise the majority of costs in well-managed long-term asthma
care. When long-term management is inadequate, as it often is for some
groups of patients, episodes occur that require the use of acute control
agents (and in some cases hospitalization, emergency room use, and
mortality).
As the above paragraph indicates, this disease is largely a manageable
disease for most patients. In practice, there are large differences in how
well the disease is managed due to differences in treatment by medical care
providers and the patient. As a result of this variation, the cost section of
this chapter (Section IV.2.B) contains a discussion and cost evaluations
based on differences in the use of medications and resulting medical
services.
10
Self-administered includes administration by a parent or non-medical care provider.
Chapter IV.2 IV.2-24 Cost of Asthma
-------
IV.2.A.5 Prognosis
Asthma is a chronic disease that is commonly observed in childhood, but it
may be diagnosed or persistent at any age. Approximately 60 percent of
asthmatic children undergo remission in young adult life, but 50 percent of
those (30 percent of the original population) become symptomatic again as
young adults. Among those who are asymptomatic as adults, some studies
have shown that they retain airway hyperresponsiveness. Based on the
above statistic, it is assumed in this analysis that 30 percent of asthmatic
children will become permanently asymptomatic at age 18 and incur no
additional medical costs for asthma treatment. Tests of airway
hyperactivity show that the airways have not returned to normal among
asymptomatic young adults who were previous asthma patients. As a
general rule, children whose asthma is resolved are those with:
• less severe intermittent asthma;
few positive skin tests to inhalant allergens;
• no persistent wheezing or rhonchi; or
no heavy exposure to pollution, allergens, or cigarette
smoke (Eggleston, 1994).
The assumption that 30 percent of children will undergo remission is based
on observations made in the past. It is not known if the remission rate at
this time is the same as it was when the 30 percent statistic was observed.
It is also not known which percentage of asthmatics in any severity
category (discussed below), will undergo remission. In this analysis it was
assumed that the severity mix among asthmatics who became
asymptomatic was the same as the mix in the original asthma population.
This assumption introduces uncertainty to this cost analysis.
Although many children see a gradual lessening of symptoms as they
progress toward adulthood, this is not universally the case. In some cases,
the disease becomes more severe and in rare cases is fatal. As discussed
under Concurrent Effects above, damage that is done to lung tissue and
structure during asthma episodes may lead to adverse long-term or
permanent changes in the lung. Current treatment strategies are designed
to minimize the inflammatory processes that occur during an asthma
episode, in part to avoid long-term damage to the respiratory system
(NHLBI, 1997).
The long-term disease course and effects of asthma differ among
individuals. In some patients fibrosis of the subbasement membranes in the
respiratory system may occur, and this may contribute to persistent
abnormalities in lung function in these individuals (NHLBI, 1997). Due to
concerns regarding the irreversible damage that can occur, effort is being
Chapter IV.2 IV.2-25 Cost of Asthma
-------
Link to Table IV. 2-1
focused on early diagnosis, limiting the severity of the disease, and drug
therapy that minimizes the inflammation causing structural changes in the
respiratory system (NHLBI, 1997).
When considering the prognosis for asthma patients, it is necessary to take
into account both the recurrence of asthma episodes and the effects that
treatment of asthma may have on overall health. Inflammation is an early
and persistent component of asthma and therapy to suppress it is long term.
Greater asthma control achieved with high doses of inhaled corticosteroids
causes less airway inflammation. The drug therapies used to control
asthma episodes are not without side effects, as discussed previously.
There are a variety of products on the market and each has specific
advantages and disadvantages. There is some evidence that some of the
most effective asthma treatments also lead to growth retardation in
children. Consequently, the prognosis for this disease must take into
account both the long-term nature of asthma in many individuals and
potential damage to the patient that results from control of asthma
symptoms. Quantitative data on long-term prognosis is not available for
newly-diagnosed patients because many new drugs have been introduced
recently.
Urban residence and poverty are major risk factors for asthma morbidity
and mortality. There are areas of increased mortality in cities, especially
among people in lower socioeconomic groups (Eggleston, 1994).
Mortality due to asthma is a rare event. The mortality rate for asthmatics is
0.0004 (see Table IV.2-1 above).11 Some population subgroups have much
higher rates than average (for example, teenagers in lower socioeconomic
groups). Increased mortality among young and otherwise healthy patients
has been associated with an inability to understand or comply with the
sometimes demanding drug and activity regimens required to control
asthma episodes. They may also be less likely or able to obtain timely
health care for acute episodes.
As discussed in the section IV.2.A.4 above, there are differences in how
individual patients manage their disease. These differences can lead to
major differences in the prognosis among patients. Self- or parental
management of the disease is much more important for asthma than for
most diseases discussed in the Handbook. This distinction places a large
burden on the medical community regarding outreach, education, patient
tracking, and follow-up care. Ultimately, the prognosis for asthma patients
depends largely on their ability and interest in managing their disease
11 The value of a statistical life (VSL) can be computed for those patients who die of the disease when
asthma avoidance is being considered in a benefits analysis.
Chapter IV.2 IV.2-26 Cost of Asthma
-------
through complying with treatment plans. Asthma deaths have occurred
among patients with mild, moderate, and severe asthma, and so are not
strictly related to the severity of the disease. Both hospitalizations and
mortality are strongly related to the degree of control of inflammation and
a timely response to acute asthma episodes. Patient (or caregiver)
education, access to medical care, and appropriate use of drug therapies
are very strong determinants in the overall prognosis for asthma patients.
Link to Section IV.2.A.4
IV.2.B. Costs of Treatment and Services
This section contains a description of the methodology used to estimate
cost, followed by a description of the services provided for the three main
components of care: diagnosis, long-term management and acute care
management, and a presentation of the costs of these services. Costs are
presented on an annual and lifetime basis for both the average and high
services use patient.
IV.2.B.1. Methodology Summary
Estimates of the direct costs of asthma are constructed using national data
on asthma medical services, utilization, reimbursement by Medicare, and
national recommendations and private sector costs regarding drug therapy.
The probabilities of patients utilizing various services are multiplied by the
cost per service. The methodology by which estimates of the lifetime
incremental costs are developed proceeds in four steps:
1) Develop treatment descriptions and probabilities,
2) Estimate the cost of each treatment component,
3) Estimate the annual costs of treatment, and
4) Sum costs over the lifetime of an average patient and apply
discount rates.
After initial diagnosis, asthma management can be considered as two
separate activities: long-term management and management of acute
episodes. Long-term management includes the establishment of an
appropriate treatment program, education for the patient and for family
members to allow them to manage the disease, and office visits to evaluate
the patient's ongoing health and use of medications. It also includes the
drug therapy that is used to manage the disease. Acute episode
management focuses on reducing symptoms of an acute asthma attack and
preventing further damage to the patient's health. This section is organized
according to the treatment components listed above, with diagnosis
described first, followed by long-term management and then acute episode
management.
Chapter IV.2 IV.2-27 Cost of Asthma
-------
IV.2.B. 1.1 Treatments and Services Evaluated
The first step in estimating the cost of the disease is to model the typical
course of the disease and the corresponding treatment. Three main sources
were used to construct disease course and treatment profiles for asthma in
this chapter: NHLBI Guidelines (NHLBI, 1997 — primary source), data
from the literature, and, to a limited degree, data from a physician panel
convened in 1991.12
Asthma treatments and costs have three main components:
1) diagnosis,
2) long-term management, and
3) acute episode management.
Diagnosis is required for all patients, and the types of diagnostic tests and
evaluations conducted are assumed to be the same for all patients (although
differences may exist among care providers). Long-term management
consists of office or outpatient visits and self-administered drug therapy.
Acute care results from immediate health emergencies that require
emergency room use and/or hospitalization. These components of health
care are first described (in Sections IV.B.2, 3, and 4) and then the costs of
the services are presented and discussed (in Section IV.B.5).
IV.2.B. 1.2 Evaluation of Differences in Asthma
Management13
IV.2.B.1.2.1 Overview of Issues
As discussed in Section IV.2.A.1.11 above, under optimal circumstances
all patients would have a long-term asthma management plan that managed
the symptoms of asthma sufficiently well that they did not have asthma
episodes requiring medical intervention. In practice, many patients either
do not receive drug plans, or do not follow drug plans that manage asthma
at this level of control. The lack of adequate medication often leads to
higher costs for office visits and acute care (emergency room use and
hospitalization).
Link to Section IV.2.A. 1.11
12 A panel of three physicians (two pulmonologists and an internist) was convened in 1991 to determine a
standard treatment protocol and the percentage of patients receiving various types of treatments. The
purpose of the panel was to supply information that was used to develop the first asthma COI chapter in
1991. These physicians, all from New England, did not represent a cross-section of primary care
physicians who treat most asthma cases. Aware of these limitations, they reviewed Medicare treatment
records and other materials and attempted to provided useful information on overall treatment patterns.
The information they provided has been largely superseded by NHLBI guidelines and other information.
13 All referrals to asthma management acknowledge that this process involves both the medical community
and the patient, due to the importance of self-medication and other activities in asthma treatment.
Chapter IV.2 IV.2-28 Cost of Asthma
-------
Link to Table IV.2-1
As discussed in Section IV.2.A.1.11 above, some patients use services at a
much higher rate than the average patient, suggesting that substantial
differences exist in the utilization of medical services and the resulting costs
among asthma patients. In addition to obtaining a cost estimate for the
"average" patient, it would be useful to obtain cost estimate for patients
with two different patient profiles: one in compliance with treatment plans
(such as those specified by NHLBI guidelines), and one that does not have
well-managed asthma and has health problems related to the management
plan. There are limited data on the variations in treatment that preclude
providing highly reliable national estimates for these two patient groups,
but there are some studies of high-use patients (as discussed in Section
IV.2.A. 1.11). These studies were used to describe a hypothetical patient
who is not following an optimal treatment plan and incurs the associated
higher medical costs.
IV.2.B.1.2.2 Patient Types Evaluated in this Cost Analysis
Descriptions of treatments and costs are provided for two types of patients:
1) the average patient: a profile was developed using national statistics
on office and outpatient visits, emergency room use, and
hospitalization, as described in Section IV.2.A, Table IV.2-1. No
data are available on drug therapy use, so NHLBI guidelines are
assumed to be relevant. Issues related to drug therapy are
discussed below.
2) the high-use patient: a profile was developed using studies of
patients who have difficulty managing their asthma with a larger
number of acute episodes than average. The estimate of their
hospital and emergency room use is based on studies that evaluated
the experiences of patients before they had training on how to
follow their management plans. These patients tend to have
moderate or severe asthma. Their costs provide a type of
reasonable upper bound on costs.
Ideally, data would be available on patients with mild asthma who are in
compliance with treatment plans and have a resulting low utilization of
acute care services. These data, which could provide a lower-bound cost
estimate, are not available. The studies that evaluated "before" and "after"
intervention included mainly patients who had serious acute care problems
and were not likely to include mild asthma cases.14 Consequently, although
14 Mild asthma is estimated to comprise 70 percent of all asthma cases, as discussed below.
Chapter IV.2 IV.2-29 Cost of Asthma
-------
Link to Table IV.2-1
data are available on the impact of intervention on high-use patients, there
are no data on a cross section of asthma patients.
As discussed above, asthma treatments and costs have three main
components (diagnosis, long-term management, and acute episode
management). Diagnostic methods and costs are likely to be similar
regardless of the subsequent decisions of patients or their physicians.
Long-term management and acute episode services will vary, depending on
the factors discussed above. Although the number of hospitalizations and
emergency room visits are assumed to vary for the two patient types, the
level of services provided during hospitalization and emergency room visits
are not assumed to differ. (There is no basis on which to assume that the
services would differ.)
IV.2.B.1.2.3 Quantitative Data on Treatment and Services for
Two Patient Types
The Average Patient. National statistics listed in Section IV.2.A Table
IV.2-1 on treatment and services were used (office/outpatient visits,
emergency room use, and hospitalization).
The High-use Patient. Table IV.2-3 lists the frequency of use of various
services before and after the intervention of an asthma management
program. The "before" statistics have been used to develop a hypothetical
"high-use" patient profile, to provide an indication of the high costs that
can be associated with asthma in the absence of adequate disease
management. (The data in this table expand on information provided in
Table IV.2-2.) The "after" data are used in the sensitivity analysis
presented in Section IV.2.C.
Although studies indicate that inadequate disease management is a
relatively common problem, data were not located that adequately describe
the percentage of asthma patients with this problem or its extent.
Consequently, this analysis does not assign the "high-use" costs to a
specific proportion of asthma patients. It is likely that these patients' use
of acute care services comprises a substantial portion of the overall national
acute care services use reflected in the "average" patient costs referred to
above.
Chapter IV.2
IV.2-30
Cost of Asthma
-------
Table IV. 2-3: Acute Care Utilization Rates for "High-use" Asthmatics Before and After Participation in an Asthma Management Program
Study
Higgins et al.,
1998
Westley et al.,
1997
Greineder et al.,
1995
Gaioni et al.,
1996
Mayo etal., 1990
Average Across
Studies:
Average
Asthmatic (for
Comparison)
Utilization Rate Per "High-use" Asthmatic per Year
Hospitalization
Before
Intervention
0.149
0.530
0.660
0.758
1.560
0.732
After
Intervention
0.070
0.190
0.094
0.169
0.480
0.201
Ratio
2.13
2.79
7.00
4.49
3.25
3.93
0.034
Emergency Room Visits
Before
Intervention
0.498
3.670
1.358
1.126
1.663
After
Intervention
0.316
1.350
0.283
0.304
0.563
Ratio
1.58
2.72
4.80
3.70
3.20
0.117
Office (Clinic) Visits
Before
Intervention
2.587
4.395
3.491
After
Intervention
3.724
2.442
3.083
Ratio
0.70
1.80
1.25
0.707
Chapter IV.2
IV.2-31
Cost of Asthma
-------
IV.2.B. 1.3 Duration of Treatment and Services
Although patients can be diagnosed with asthma at any point in their lives,
most people are diagnosed as children. As noted earlier, the average age of
diagnosis is four years (Eggleston, 1994) and that is the age that is used as
the average onset of the disease for this analysis.. As noted above, asthma
patients are assumed to live a normal lifespan due to the minimal mortality
due to asthma. An average life expectancy of 75 years was assumed for
purposes of estimating care duration and cost.15 This lifespan is
recommended in EPA's Exposure Factors Handbook for general use (EPA,
1997), which yields an overall lifespan with asthma of 72 years (from age 4
through age 75). There is an assumption that 30 percent of patients with
mild asthma become asymptomatic and no longer require treatment at age
18. (See discussion in Section IV.2.A.5.).
Link to Section IV.2.A.5
Life expectancy increases as individuals survive through each year of their
life, so that by the average age of diagnosis (four years) life expectancy is
longer than 75 years for the average person in the United States. This
figure is balanced by the likelihood that people with chronic illnesses, such
as asthma, have a decreased life expectancy due to asthma and related
illnesses or the side-effects of treatment. The average duration of
treatment is uncertain for asthmatics, and the actual duration and
associated costs may be greater or lesser than the value estimated in this
analysis.
IV.2.B.2 Diagnosis
IV.2.B.2.1 Medical Evaluation During Diagnosis
The diagnosis of asthma usually occurs in response to the observation of
symptoms that prompt a visit to a physician's office or emergency room
(ER). It may also occur when a routine physical is scheduled, especially in
the case of children. Treatment in an ER does not imply a severe medical
condition; rather, it indicates an acute problem and may be the medical
location of choice for some patients who do not have a regular physician,
or when an asthma episode occurs outside of usual office hours. NHLBI
does not distinguish among diagnoses in different settings when describing
the diagnostic protocol (NHLBI, 1997).
Medical evaluation needed to diagnose asthma is not complex in most
cases. Asthma is indicated by an appropriate array of symptoms, an acute
15 There may be mortality due to therapy or other aspects of asthma that are not well-described at this
time. In the absence of information to the contrary, it was reasonable to assume a normal lifespan for
asthmatics.
Chapter IV.2 IV.2-32 Cost of Asthma
-------
reaction to asthma-related stimuli, and quick relief by appropriate therapy.
Other supporting evidence may include family history, an elevated IgE
level, and eosinophilia (Eggleston, 1994).
When considering a diagnosis of asthma, NHLBI recommends the
following:
• careful hi story-taking (specific questions can be viewed at
the NHLBI web site),
• physical examination, and
spirometry to evaluate air flow.
Beyond the above activities, NHLBI discusses some specific procedures
that may be dictated by the patient's characteristics and history, but that are
not required for most patients. These may include:
additional pulmonary function tests (lung volume,
inspiratory and expiratory flow loops, diffusion capacity test
(primarily in older patients));
• diurnal variation in PEF over one to two weeks;
• bronchoprovocation with methacholine, histamine, or
exercise;
chest X-ray;
• allergy testing;
screening for nasal polyps and sinus disease;
• evaluation of gastrointestinal reflux;
• evaluation of biomarkers of inflammation (in development);
and
other tests to rule various diseases suggested by patient
symptoms.
Many of the above additional evaluations have specific and narrow
indicators for their use and are not needed for most patients. It is not
possible to determine the percentage of patients that are evaluated for each
of the above. Some of the evaluations would be done during a second
diagnostic visit to a specialist (e.g., allergy testing) or may be
characteristics observed by the patient and reported back to the physician
(e.g., diurnal variation in PEF over one to two weeks). Second diagnostic
visits are considered below. The other evaluations listed above are beyond
the scope of this analysis, but their inclusion would increase costs. Thus,
their exclusion leads to an underestimate of costs.
Diagnosis of asthma is difficult in young children. They are often
diagnosed as having bronchitis, bronchiolitis, or pneumonia. In infants
there are two patterns of wheezing (a critical observation in asthma),
allergic and nonallergic. As children develop, the nonallergic infants no
longer wheeze during colds and other minor illnesses because their airways
Chapter IV.2 IV.2-33 Cost of Asthma
-------
enlarge. Allergic children continue to have episodes when their airways
constrict and may be diagnosed with asthma (AAFA, 1999).
Diagnoses of asthma are not frequently made in children under the age of
one year. When wheezing and related symptoms occur prior to that age,
they are generally diagnosed as bronchiolitis or other diseases that cause
wheezing, with the understanding that they may later be diagnosed as
asthma. The symptoms that occur prior to one year may be "asthma" in
the pragmatic, if not clinically defined, sense of the term. When asthma is
diagnosed in much older children, it may be because their symptom onset
occurs later or because there is a delay in diagnosis beyond the onset of
symptoms, which can lead to more severe forms of asthma when diagnosis
occurs. Numerous studies have suggested that an underdiagnosis of
asthma is a public health problem.
IV.2.B.2.2 Services Provided During the Diagnostic Visit
Based on NHLBI guidelines and the 1991 panel recommendations,
components of the diagnostic visit (that are costed individually for
Medicare reimbursement) are listed in Table IV.2-4. As discussed
previously, services will vary depending on patient characteristics, the
physician's practice methods, and other factors.16 X-rays were included in
this list because they appear as a possible procedure in NHLBI guidelines
and were listed by the 1991 physician panel as a diagnostic procedure.
Table IV.2-4: Diagnostic Procedures
Chest X-Ray, Two Views
Blood Gases: pH, pO2, pCO2
Automated Hemogram
Breathing Capacity Test
Office Visit, Level 5, New Patient
Drawing Blood for Specimen
Diagnosis may occur at a physician's office, in a hospital outpatient clinic,
or in an emergency room. It is assumed in this analysis that an outpatient
clinic would provide essentially the same services for the same cost as a
physician's office. When diagnosis occurs in an emergency room, it may be
that the patient is experiencing symptoms that require immediate attention
or that medical care is unavailable via other means (e.g., they do not have a
personal physician. Due to the potentially more serious nature of their
symptoms, there may be additional services provided. These services may
include: a complete blood count (CBC), theophylline administered
16 The NHLBI does not recommend different evaluations during diagnosis based on severity (severity
category is determined during diagnosis, not at the onset of the process).
Chapter IV.2 IV.2-34 Cost of Asthma
-------
intravenously, nebulizer therapy, or other services that are medically
required (as described by the 1991 physician panel). These added services
will lead to increased costs not considered in this analysis. (A description
of services and the costs for an emergency room visit are described below.)
Emergency room staff often do not have information available on the
medical history of a patient, so may treat each patient entering the
emergency room with asthma as a new diagnosis (Physician panel, 1991).
That was the assumption made when developing the description of services
and costs for the emergency room services discussed below. This
description and the associated costs may be relevant for both initial
diagnosis in an emergency room and visits to the emergency room that
occur after diagnosis. Whether this is the case will depend on the specific
hospital's practices, and whether the patient was referred to the emergency
room by a physician who provided the patient's medical history to the
hospital.
IV.2.B.2.3 Asthma Severity Level
When diagnosed with asthma, the severity of the patient's disease is
evaluated and a severity category is assigned. These categories are
subsequently used to develop an asthma management plan for the patient.
The National Heart, Lung, and Blood Institute (1997) has developed a
classification scheme for asthma by severity of disease. Patients are
diagnosed at their initial visit (or a subsequent visit to a specialist) with
either mild intermittent, mild persistent, moderate persistent, or severe
persistent asthma based on this classification scheme. Treatment plans and
costs vary depending on the severity of the asthma. The classification
scheme, shown in Table IV.2-5, shows symptoms and lung function
measures that are used to categorize asthma (NHLBI, 1997).
Recent information was not located in the literature regarding the
distribution of patients into the various categories listed above. Estimates
are available from the 1991 physician panel: approximately 70 percent of
asthmatics have mild asthma, 25 percent have moderate asthma, and 5
percent have severe asthma. In their 1997 publication, NHLBI separated
mild asthma into two components, intermittent and persistent (NHLBI,
1997). For purposes of this analysis, it is assumed that half of those with
mild asthma are in each diagnostic category.
It is also assumed that the 1991 panel's estimate of the percentage of
patients in each category is currently accurate. Due to increases in the
occurrence and acute services use by asthmatics in recent years, it is
possible that more patients are now in the moderate and severe asthma
categories than in the past. Uncertainly exists regarding this distribution
because data on the current severity distribution are not available. This
uncertainty does not impact service utilization, which is not based on
severity in this analysis (it is based on national statistics and the results of
Chapter IV.2 IV.2-35 Cost of Asthma
-------
studies of service utilization). Severity categories are used, however, to
estimate drug therapy use and cost. Consequently, if the severity
distribution among asthma patients is currently more severe than it was in
1991, the costs of drug therapy will be underestimated. As better
information becomes available and/or practices change, the percentages can
be modified to reflect current knowledge.
In this analysis, patients were assumed to retain the severity level that they
were originally diagnosed up to age 18. Beyond that age it was assumed
that 30 percent become symptom-free (Eggleston, 1994; see discussion in
Section IV.2.A.5). The remaining 70 percent continue to require
treatment at the severity level at which they were originally diagnosed. As
with the severity distribution, the percentage of patients becoming
symptom-free may differ from estimates made in the past.
Link to Section IV.2.A.5
Chapter IV.2 IV.2-36 Cost of Asthma
-------
Table IV. 2-5: Classification of Asthma Severity
Clinical Features Before Treatment*
STEP 4
Severe
Persistent
STEPS
Moderate
Persistent
STEP 2
Mild
Persistent
STEP1
Mild
Intermittent
Symptoms**
Continual symptoms
Limited physical activity
Frequent exacerbations
Daily symptoms
Daily use of inhaled short-
acting beta2-agonist
Exacerbations affect activity
Exacerbations > 2 times a
week; may last days
Symptoms > 2 times a week
but < 1 time a day
Exacerbations may affect
activity
Symptoms < 2 times a week
Asymptomatic and normal
PEF between exacerbations
Exacerbations brief (from a
few hours to a few days);
intensity may vary
Nighttime
Symptoms
Frequent
> 1 time a week
> 2 times a
month
< 2 times a
month
Lung Function
• FEV1 or PEF < 60%
predicted
• PEF variability > 30%
• FEVi or PEF > 60%-<80%
predicted
PEF variability > 30%
• FEV1 or PEF > 80%
predicted
• PEF variability 20-30%
• FEVi or PEF > 80%
predicted
PEF variability < 20%
* The presence of one of the features of severity is sufficient to place a patient in that category. An individual
should be assigned to the most severe grade in which any feature occurs. The characteristics noted in this
figure are general and may overlap because asthma is highly variable. Furthermore, an individual's
classification may change overtime.
** Patients at any level of severity can have mild, moderate, or severe exacerbations. Some patients with
intermittent asthma experience severe and life-threatening exacerbations separated by long periods of
normal lung function and no symptoms.
IV.2.B.2.4 Drug Therapy During Diagnosis
A patient is initially diagnosed with asthma may receive two types of drug
therapies that are consistent with the two aspects of managing asthma:
1) Long-term care. The patient will receive a management plan that
includes long-term drug therapy for control of their disease. If this
management plan adequately controls symptoms, the patient will be
maintained on that plan, with subsequent review to determine if dosages
can be "stepped down" and still maintain control over the disease.
2) Episodic care. A patient diagnosed in response to an asthma episode
will receive drugs during the visit to address current symptoms. These will
be short-acting drugs that are used in response to asthma episodes rather
than for long-term disease management.
Chapter IV.2
IV.2-37
Cost of Asthma
-------
In both cases, the drug regimens do not differ from those that would be
used under similar circumstances after initial diagnosis. Consequently, they
are discussed below under the two relevant headings: Long-term
Management and Management of Acute Episodes. The specific
therapeutic drugs that are recommended for use when asthma is initially
diagnosed vary, depending on the disease symptoms, patient
characteristics, and the physician's prescribing habits. There are two main
groups of drugs used to treat asthma, one for long-term management of the
disease and one for quick relief of symptoms.
In the cost section below, drug costs are assumed to begin when the
diagnosis is made at age four. The drugs prescribed or given in the
physician's office at diagnosis will be the same as those used during the
subsequent long-term management and quick-relief management.
Consequently, the specific drugs are discussed below under those two
headings. Separate drug costs are not assigned to the diagnostic process
because they are the same as those itemized under the Drug Therapies
section below.
IV.2.B.2.5 Referral to Specialists During Diagnosis
Most people with asthma are diagnosed by and receive care from a primary
care physician. NHLBI does not suggest that diagnostic procedures vary
depending on the severity category in which the patient may be ultimately
diagnosed. Consequently, it is assumed that all patients will undergo
similar diagnostic procedures, although NHLBI recommends referral to
specialists in asthma treatment (e.g., a pulmonologist) under specific
circumstances. The conditions leading to referal are related to ongoing
care and meeting treatment goals unrelated to the initial diagnosis;
consequently, they are not discussed in this section. Some criteria are
relevant at diagnosis, and one (criterion 5) is related to the severity
category of the patient. The criteria are:
1) signs and symptoms are atypical or there are problems in
differential diagnosis;
2) other conditions complicate asthma or its diagnosis;
3) additional diagnostic testing is indicated (e.g., allergy
testing, complete pulmonary function studies, rhinoscopy,
provocative challenge, bronchoscopy);
4) patient is being considered for immunotherapy;
5) patient has severe persistent asthma (Step 4, as shown in
Table IV.2-5), or is under age three and has moderate or
severe persistent asthma (Step 3 or 4) and in some cases
mild persistent asthma (Step 2); and
Chapter IV.2 IV.2-38 Cost of Asthma
-------
6) patient has significant psychiatric, psychosocial, or family
problems that interfere with asthma therapy. (NHLBI,
1997, refer to this source for additional detail)
Link to Table IV. 2-5
Referral to a specialist occurs when a patient's asthma or some other health
characteristic is complex and requires more sophisticated investigation
and/or treatment. This analysis does not attempt to speculate on what
specialized treatment would be provided because the reasons for referral
and the resulting treatments vary widely. The analysis assumes that the
cost of treatment by a specialist would be the same as the cost of the initial
office visit for diagnosis. This is a conservative assumption, because it is
likely that the treatment would be at least as complex (and costly) as the
initial office visit for diagnosis.
It is not known what percentage of patients may experience any of the
above, but an estimate can be made for criterion 5. As discussed
previously, the 1991 physician panel provided an estimate of the
percentage of individuals who fall into each severity category. Using this
information, the percentage of patients who would be referred to a
specialist based on criterion 5 can be estimated. It was estimated in 1991
that 70 percent of asthmatics have mild asthma (35 percent persistent, 35
percent intermittent), 25 percent moderate, and 5 percent severe asthma.
Using this information with criterion 5, an estimate of the number of
patients who will see a specialist following their initial visit with a primary
care physician was made as follows:
For severe asthma: the 5 percent of patients with severe asthma will all be
assumed to consult a specialist.
For moderate asthma: the criterion is age-related. There is not a
distribution of ages of diagnosis available; however, if four years is the
median age of diagnosis (50 percent above and 50 percent below), then it
will be assumed that half of all patients were diagnosed at age three or
younger. Thus, one half of the 25 percent of patients with moderate
asthma, or 12.5 percent of all asthma patients, will be assumed to consult
with a specialist.
For mild asthma: there is not quantitative guidance provided for those with
mild asthma. Based on the numerous criteria for referral to specialists, the
substantial number of very young children who have asthma (and are more
difficult to evaluate and treat) and the statement that some cases of mild
persistent asthma will require referral to a specialist, it was assumed that
one half of those with mild asthma, or 35 percent of all asthma patients,
will consult a specialist. Aggregating the three categories above, the total
percentage of patients who are assume to consult with a specialist
Chapter IV.2 IV.2-39 Cost of Asthma
-------
following initial consultation with a primary care physician is 52.5
percent.
It was assumed that patients meeting this criterion would meet with a
specialist after the first diagnostic visit, for additional diagnosis and
planning evaluations. As noted above, the specific additional evaluations
that are performed in this visit will vary depending on the reason for
referral and patient characteristics. Consequently, the specific activities
cannot be estimated. It is reasonable to assume that they will be at least as
complex (and as costly) as those provided by the primary care physician in
the first diagnostic visit.
IV.2.B.2.6 Level of Visit and Number of Visits for Diagnosis
To estimate the costs of office visits, it is necessary to specify the level of
visit (1 to 5), which is determined by the length of the visit. Visits for
asthma include a number of activities related to the initial diagnosis,
including the determination of the appropriate treatment plan through tests,
history taking, physical exam and evaluation, explanation of the disease and
management plan to the patient (or parent), and patient education
regarding drugs and equipment. A second visit follows shortly after
diagnosis to determine if the treatment plan is managing the disease
adequately; additional patient education may also be provided.
NHLBI has emphasized the importance of education and the management
of asthma by the patient and physician as key elements of care. Emphasis is
also placed on quality of life issues, which is closely tied to adequate
disease management (NHLBI, 1997). These foci require considerable time
be spent by medical staff on training and education of patients. This is
particularly true due to the daily need for self-medication for many
asthmatics and the use of inhalers and peak flow meters. The result of all
this is a requirement that more time be spent during office visits than is
spent for many other chronic diseases.17 Due to the time requirements, this
analysis assumes that a single maximum duration (level 5) office visit would
be used for the initial diagnostic consultation.
17 Consultation with one pediatrician who treats many asthma cases suggests that considerable time is
required for adequate patient education. He routinely spent one hour or more explaining the disease and
training the patient, which was followed by later consultations with nurse practitioners. It was felt that this
time investment was necessary to ensure that patients understood their treatment plans, how to avoid
asthma triggers, how to use inhalers and peak flow meters, and other aspects of their disease.
Chapter IV.2 IV.2-40 Cost of Asthma
-------
NHLBI guidelines indicate that the first diagnostic visit is followed shortly
by another visit to determine if the management plan is adequate for the
patient, and to make any necessary adjustments. Additional patient
education may also be provided.18 This analysis assumes that a second visit
will follow the first within a short time frame as part of the diagnostic
process. Within the set of those patients who require adjustment of their
treatment plans at this visit, there may be a subset that require another visit
to determine if the modified plan was optimal. The percentage of patients
requiring this third visit is not known. Consequently, this visit and any
subsequent planning visits related to initial diagnosis are not included. The
assumption that only two visits are required will cause an underestimate of
medical cost in this analysis.
IV.2.B.3 Long-term Management
Long-term management of asthma involves ongoing use of medications and
asthma trigger avoidance by patients. It also requires periodic assessments
and monitoring by medical care providers. For purposes of cost evaluation
the focus of this discussion is on:
1) the office visits or hospital outpatient visits that occur, and
2) the drug therapy regimen that is prescribed for the patient.
No skilled nursing or therapeutic services are typically provided to
asthmatics in their homes, so costs for such services are not included in this
analysis. It is assumed that severe asthmatics will use a nebulizer with
compressor in the home on an ongoing basis. This cost (which is minimal)
is listed with drug therapy, since it is self-administered and is more closely
related to drug therapy than to medical services.
As discussed above, the frequency and complexity of long-term care
depends substantially on the patient receiving an adequate treatment plan,
understanding it, and complying with it (particularly self-administered drug
regimens). Consequently, an estimate of the patient's use of treatment and
services and their cost is provided for an average patient, as well as for
high-use patients. When a patient is managing their exposures to asthma
triggers adequately and self-administers drugs that prevent asthmatic crises,
the course of their therapy and resulting costs for many patients may be
fairly predictable. When management is inadequate, there will be asthma
episodes requiring different self-administered drugs, more office and
emergency room visits, and potential hospitalization (discussed in the
following section). In either case, there will be office visits and self-
medication.
18 It has been observed that many |