United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EPA-450/1-90-0043
September 1990
&EPA
CANCER RISK
FROM OUTDOOR
EXPOSURE TO
AIR TOXICS
Volume I
Final Report
-------
-------
EPA-450/l-90-004a
September 1990
CANCER RISK FROM OUTDOOR EXPOSURE
TO AIR TOXICS
VOLUME I
FINAL REPORT
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
-------
-------
ACKNOWLEDGEMENTS
This report was prepared under the guidance of Joseph Padgett,
U.S. Environmental Protection Agency, Office of Air Quality Planning and
Standards, Research Triangle Park, North Carolina. Key individuals
assisting Mr. Padgett in the development of this report were Bob Faoro
(Monitoring and Reports Branch), Fred Hauchman (Pollutant Assessment
Branch), and Tom Lahre (Pollutant Characterization Branch).
Several offices within EPA provided technical guidance and
support. Special recognition is due to Penny Carey (Office of Mobile
Sources) for her assistance on the mobile sources portion of the report.
Others who contributed welcomed support include Brenda Riddle, Warren
Peters, Bob Lucas, Scott Voorhees, Beth Hassett-Sipple, K.C. Hustvedt,
Joellen Lewtas, Ila Cote, Shiva Garg, and James Hardin.
Finally, the analysis for this study was conducted by Pacific
Environmental Services, Inc., Durham, North Carolina. Special
recognition is due to Ken Meardon, who performed the analysis and
authored the report. Assistance to Mr. Meardon was provided by Karin
Gschwandtner.
-------
-------
TABLE OF CONTENTS
ACKNOWLEDGEMENTS
GLOSSARY
ACRONYMS
iii
xi
xix
CHAPTER ES EXECUTIVE SUMMARY
Technical Approach
Sources of Information
Methodology
Pollutant and Sources Not Evaluated
Additive Risk
Results
Magnitude of the Problem
Nature of the Cancer Risk
Comparison with 1985 Six-Month Study
ES-3
ES-3
ES-3
ES-6
ES-6
ES-6
ES-7
ES-10
ES-13
CHAPTER 1.0 INTRODUCTION
Background
Purpose of Current Study
Other Studies or Reports on Air Toxics
Indoor Air Pollution
Noncancer Health Risk Study
-SARA Title III
Outline of the Report
1-1
1-3
1-7
1-7
1-9
1-9
1-11
CHAPTER 2.0 SCOPE OF STUDY AND ANALYSES
Data Base
Annual Cancer Incidence Analysis
Methodology
Limitations and Uncertainties
Limitations
'Uncertainties
2-1
2-10
2-13
2-36
2-36
2-40
CHAPTER 3.0 THE MAGNITUDE AND NATURE OF THE CANCER RISK
Magnitude of the Cancer Risk Problem
Annual Cancer Cases
Lifetime Individual Risk
3-2
3-2
3-7
-------
TABLE OF CONTENTS (concluded)
CHAPTER 3 continued
Nature of the Cancer Risk Problem
Individual Pollutants
Source Categories
Geographic Variation
Comparison with the Results from the 1985
Six-Month Study
Magnitude of the Problem
Nature of the Problem
3-21
3-21
3-24
3-31
3-50
3-50
3-55
CHAPTER 4.0 SUMMARY AND CONCLUSIONS
Magnitude of the Cancer Risk
Annual Cancer Incidence
Lifetime Individual Risks
Nature of the Cancer Risk
Individual Pollutants
Sources
Geographic Variability
4-1
4-1
4-3
4-4
4-4
4-6
4-9
APPENDIX A COMMENTS RECEIVED ON THE EXTERNAL REVIEW DRAFT
APPENDIX B CANCER RISK REDUCTION ANALYSIS FOR SELECTED POLLUTANTS
APPENDIX C SUMMARIES OF POLLUTANT-SPECIFIC AND SOURCE-SPECIFIC
STUDIES (Including Noncancer Health Risk Project on Air
Toxics)
-------
LIST OF TABLES
1-1
2-1
2-2
2-3
2-4
2-5
2-6
2-7
2-8
2-9
2-10
2-11
3-1
3-2
3-3
3-4
3-5
3-6
3-7
Summary of Estimated Nationwide Annual
Cancer Cases by Pollutant
Main Conclusions of the 1985 Six-Month Study
List of Reports Used in Study
EPA Source Category and Pollutant Studies
Number of Pollutants Included in Cancer Incidence
Estimates, by Study
Number of Studies that Included Specific
Pollutant in Cancer Risk Estimate, by Pollutant
Distribution of Source Categories by Number
of Studies
Unit Risk Factors Used to Compare Cancer Risks
Unit Risk Factors Used to Estimate Cancer Risk
from PIC
Pollutants with Unit Risk Factors Different
from Those Used in this Report
Effect of Changes in Unit Risk Factors Used
in this Report on Original Estimates of
Annual Cancer Cases
Effect of Unit Risk Factors on Estimated Annual
Cancer Cases: The South Coast Study
Selected Limitations of Modeled and Ambient-
Measured Concentrations for Estimating
Cancer Risk
Summary of Estimated Nationwide Annual
Cancer Cases by Pollutant
Summary of Maximum Individual Risks of Cancer
as Reported in the Various Studies
Distribution of Maximum Lifetime Individual
Cancer Risks to the Most Exposed Individual
from Hazardous Waste Combustors-Boilers and
Furnaces
Maximum Lifetime Individual Cancer Risks from
Coke Oven Emissions
Distribution of Maximum Individual Cancer Risk
at 22 Drinking Water Aerators
Areawide Lifetime Individual Cancer Risks for
Selected Cities
Summary of Lifetime Individual Cancer Risks
for Selected Cities
Page
ES-8
1-4
2-3
2-5
2-7
2-8
2-11
2-19
2-23
2-26
2-27
2-28
2-31
3-3
3-10
3-16
3-17
3-18
3-19
3-20
vn
-------
3-8
3-9
3-10
3-11
3-12
3-13
3-14
3-15
3-16
3-17
3-18
3-19
3-20
3-21
3-22
3-23
3-24
3-25
3-26
3-27
3-28
LIST OF TABLES .(continued)
Relative Contribution of Individual Pollutants
to Total Estimated Cancer Cases 3-22
Summary of Estimated Cancer Cases Based on
Modeled Ambient Concentrations, by Source
Category 3-25
Distribution of Estimated Cancer Cases from
Secondary Formaldehyde Formation Among Source
Categories 3-30
Contribution of Area vs. Point Sources to
Nationwide Annual Cancer Cases 3-32
Comparison of Measured Ambient Concentrations
of Selected Pollutants in Selected Cities 3-33
Intracity Comparison of Ambient Concentrations
Ug/nr) for Selected Pollutants in Three Cities,
by Location 3.34
Variation in Annual Cancer Cases and Cancer
Rates Due to Exposure to Outdoor Air Toxics
by Geographic Locales 3-35
Variation in Maximum Lifetime Individual Risk,
by Location 3.37
Estimates of Maximum Lifetime Individual
Cancer Risks in Neighborhoods Surrounding
Facilities in the Kanawha Valley 3-40
City-to-City Variation in Relative Contribution
of Selected Pollutants to Total Annual
Cancer Incidence 3.41
Maximum Lifetime Individual Cancer Risks in
Baltimore by Individual Pollutant 3-43
Relative Contribution of Individual Pollutants to
Maximum Lifetime Individual Risk of Cancer in
the Southeast Chicago Area 3-44
Areawide Lifetime Individual Risks of Cancer:
Monitored vs. Modeled Ambient Air Concentra-
tions in Philadelphia 3-45
Estimates of Multi-Pollutant Lifetime Cancer
Risks to the Most Exposed Individual
to Various Sources in Philadelphia 3-46
Estimated Cancer Risk to Maximum Exposed
Individuals to Organic Gases in Santa Clara
for Selected Sources 3-47
Areawide Individual Risk of Cancer from Lifetime
Exposure to Organic Gases in Santa Clara 3-48
Estimates of Areawide Lifetime Individual Risks
' of Cancer Across Area and Point Sources in the
Kanawha Valley 3.49
Comparison of Annual Cancer Cases Per Million
Population with 1985 Six-Month Study 3-51
Comparison of Annual Cancer Cases with 1985
Six-Month Study 3-52
Comparison of Unit Risk Factors 3-54
Contribution of Sources to Estimated Annual
Cancer Cases and Areawide Lifetime Individual
Risks in Southeast Chicago 3-59
-------
LIST OF FIGURES
Figure No. Title Page
ES-1 Relative Contribution to Total Estimated
Nationwide Cancer Cases Per Year, by
Pollutant ES-11
ES-2 Relative Contribution by Source Categories
to Total Estimated Nationwide Cancer
Cases Per Year ES-12
1-1 Relationship of this Study to Other Air
Toxic Risk Studies 1-8
2-1 Illustrative Relationship of Pollutants
and Source Categories Covered by Five
Hypothetical Studies 2-15
3-1 Relative Contribution to Total Estimated
Nationwide Cancer Cases Per Year, by Pollutant 3-23
3-2 Relative Contribution by Source Categories to Total
Estimated Nationwide Cancer Cases Per Year 3-27
_
-------
-------
GLOSSARY
Ambient fair) monitoring. The collection of ambient air samples and the
analysis thereof for air pollutant concentrations.
Acute exposure. One or a series of short-term exposures generally lasting
less than 24 hours.
Additivltv. A pharmacologic or toxicologic interaction in which the
combined effect of two or more chemicals is approximately equal to the
sum of the effect of each chemical alone. (Compare with: antagonism,
synergism.)
Adverse effect. A biochemical change, functional impairment, or
pathological lesion that either singly or in combination adversely affects
the performance of the whole organism, or reduces an organism's ability
to respond to an additional environmental challenge.
Aggregate risk. The sum of individual increased risks of an adverse
health effect in an exposed population.
Annual incidence. The number of new cases of a disease occurring or
predicted to occur in a population over a year.
Antagonism. A pharmacologic or toxicologic interaction in which the
combined effect of two chemicals is less than the sum of the effect of
each chemical alone; the chemicals either interfere with each other s
actions, or one interferes with the action of the other. (Compare with:
additivity, synergism.)
Areawide average Individual risk. Average individual risk to everyone in
an area (but not necessarily the actual risk to anyone). May be computed
by dividing lifetime aggregate incidence by the population within the
area.
Areawide incidence. Incidence over a broad area, such as a city or
county, rather than at a particular location, such as an individual grid
cell.
Background. A term used in dispersion modeling representing the
contribution to ambient concentrations from sources not specifically
modeled.in the analysis, including natural and manmade sources.
Bioassav. A test conducted in living organisms to determine the hazard
of potency of a chemical by its effect on animals, isolated tissues, or
microorganisms.
Box model. A simplified modeling technique that assumes uniform emissions
within an urban area and uniformly mixed concentrations within a specified
mixing depth.
Cancer. A malignant new growth. Cancers are divided into two broad
categories: carcinoma and sarcoma.
XI
-------
Carcinogenic. Able to produce malignant tumor growth. Operationally most
benign tumors are usually included also.
Carcinogenic process. A series of stages at the cellular level after
which cancer will develop in an organism. Some believe there are at least
3 stages, initiation, promotion, and progression. While hypothesized as
staged process, little is known about specific mechanisms of action.
Chronic exposure. Long-term exposure usually lasting six months to a
lifetime.
Comparative potency factor. A cancer unit risk factor for a complex
substance or mixture that is extrapolated from human risk data for a
reference substance and the ratio of short term bioassay responses of the
complex substance to the reference substance. The EPA is developing
comparative potency factors for various classes of POM.
Confidence limit. The confidence interval is a range of values that has
a specified probability (e.g., 95 percent) of containing a given parameter
or characteristic. The confidence limit referees to the upper value of
the range (e.g., upper confidence limit).
Criteria pollutants. Pollutants defined pursuant to Section 108 of the
Clean Air Act and for which national ambient air quality standards are
prescribed. Current criteria pollutants include particulate matter, SO ,
NOX, ozone, CO and lead. x
Dispersion modeling. A means of estimating ambient concentrations at
locations (receptors) downwind of a source, or an array of sources, based
on emission rates, release specifications and meteorological factors such
as wind speed, wind direction, atmospheric stability, mixing height and
ambient temperature.
Dose-response relationship. A relationship between: (1) the dose, often
actually based on "administered dose" (i.e., exposure) rather than
absorbed dose, and (2) the extent of toxic injury produced by that
chemical. Response can be expressed either as the severity of injury or
proportion of exposed subjects affected. A dose-response assessment is
one of the four steps in a risk assessment.
Excess risk. An increased risk of disease above the normal background
rate.
Exposure. Contact of an organism with a chemical, physical, or biological
agent. Exposure is quantified as the amount of the agent available at the
exchange boundaries of the organism (e.g., skin, lungs, digestive tract)
and available for absorption.
Exposure assessment. Measurement or estimation of the magnitude,
frequency, duration and route of exposure of animals or ecological
components to substances in the environment. The exposure assessment also
describes the nature of exposure and the size and nature of the exposed
populations, and is one of four steps in risk assessment.
xn
-------
Human Exposure Model (HEM). A mathematical model used in exposure
assessments for toxic air pollutants to quantify the number of people
exposed to pollutants emitted by stationary sources and the pollutant
concentrations they are exposed to. Input data include plant
characteristics such as location, emission, parameters, etc. as well as
Bureau of Census data used in the estimation of persons exposed and
appropriate meteorological data.
Incidence. The number of new cases of a disease within a specified time
period. It is frequently presented as the number of new cases per 1,000,
10,000, or 100,000. The incidence ra.te is a direct estimate of the
probability or risk of developing a disease during a specified time
period.
Individual risk. The increased risk for a person exposed to a specific
concentration of a toxicant. May be expressed as a lifetime individual
risk or as an annual individual risk, the latter usually computed as 1/70
of the lifetime risk.
Lifetime. Covering the lifespan of an organism (generally considered 70
years for humans).
Limited evidence. According to the USEPA carcinogen risk assessment
guidelines, limited evidence is a collections of facts and accepted
scientific inferences that suggests the agent may be causing an effect
but the suggestion is not strong enough to be an established fact.
Lowest-observed-adverse-effect level (LOAED. The 11owest dose or exposure
level of a chemical in a study at which there is a statistically or
biologically significant increase in the frequency or severity of an
adverse effect in the exposed population as compared with an appropriate^
unexpected control group.
Lowest-observed effect level fLOEU. In a study, the lowest dose or
exposure level at which a statistically or biologically significant effect
is observed in the exposed population compared with an appropriate
unexposed control group.
Malignant. A condition of a neoplasm (tumor) in which it has escaped
normal growth regulation and has demonstrated the ability to invade local
or distance structures, thereby disrupting the normal architecture or
functional relationship of the tissue system.
Maximum individual risk (MIR). The increased risk for a person exposed
to the highest measured or predicted concentration of a toxicant.
Maximum likelihood estimate fMLE^. A statistical best estimate of the
value of a parameter from a given data set.
Mobile source. Any motorized vehicle, such as cars, trucks, airplanes,
trains. Sometimes refers specifically to highway vehicle sources.
xm
-------
Monitoring. The collection and analysis of ambient air samples.
Sometimes refers specifically to just sampling and not to analysis. Can
also refer to source (stack) sampling.
Motor vehicle. On-road or off-road cars, trucks or motorcycles.
Multistage model. A mathematical function used to extrapolate the
probability of incidence of disease from a bioassay in animals using high
doses, to that expected to be observed at the low doses that are likely
to be found in chronic human exposure. This model is commonly used in
quantitative carcinogenic risk assessments where the chemical agent is
assumed to be a complete carcinogen and the risk is assumed to be
proportional to the dose in the low region.
Hutaoenlc. Ability to cause a permanent change in the structure of DNA.
More specific than, but often used interchangeably with, genotoxic.
Noncancer risk. Risk of a health effect other than cancer.
Nonthreshold toxicant. An agent considered to produce a toxic effect from
any dose; any level of exposure is deemed to involve some risk. Usually*
used only in regard to carcinogenesis.
Nontradltional sources. Sources not usually included in an emission
inventory, such as wastewater treatment plants, groundwater aeration
facilities, hazardous waste combustors, landfills, which are air emitters
due to intermedia transfer from water or solid waste.
No-observed-adverse-effect level (NOAEU. The highest experimental dose
at which there is no statistically or biologically significant increases
in frequency or severity of adverse health effects, as seen in the exposed
population compared with an appropriate, unexposed population. Effects
may be produced at this level, but they are not considered to be adverse.
No-observed-effect level fNOEn. The highest experiment dose at.which
there is no statistically or biologically significant increases in
frequency or severity of toxic effects seen in the exposed compared with
an appropriate, unexposed population.
Normalized modeling. Modeling of unit weights (e.g., 1 Mg/yr) of
emissions from each source, rather than modeling of actual emissions, and
displaying incremental receptor concentrations or receptor coefficients.
Thereafter, the resulting normalized receptor coefficients are adjusted
by actual emission rates to simulate different emission scenarios rather
than re-running the model over and over with different emissions totals.
This process assumes linearity between emissions and modeled ambient air
concentrations, which does not always hold if stack and exhaust parameters
change.
Photochemlcallv formed pollutant. A secondarily formed pollutant due to
atmospheric photochemistry. Some examples are formaldehyde and PAN.
Potency. A comparative expression of chemical or drug activity measured
in terms of the relationship between the incidence or intensity of a
xiv
-------
particular effect and the associated dose of a chemical, to a given or
implied standard or reference.
Receptor. A particular point in space where a monitor is located or where
an exposure or risk is modeled.
Receptor grid. An array of receptors. Generally synonymous with network.
Receptor modeling. A technique for inferring source culpability at a
receptor(s) by analysis of the ambient sample composition. There are
various receptor models employing microscopic and chemical methods for
analysis.
Reference dose (RfD). An estimate (with uncertainty spanning perhaps an
order of magnitude) of the daily exposure to the human population
(including sensitive subpopulations) that is likely to be without
deleterious effects during a lifetime. The RfD is reported in units of
mg of substance/kg body weight/day for oral exposures, or mg of
substance/m3 of air breathed for inhalation exposures.
Risk. The probability of injury, disease, or death under specific
circumstances. In quantitative terms, risk is expressed in values ranging
from zero (representing the certainty that harm will not occur) to one
(representing the certainty that harm will occur).
Risk assessment. The scientific activity of evaluating the toxic
properties of a chemical and the conditions of human exposure to it in
order both to ascertain the likelihood that exposed humans will be
adversely affected, and to characterize the nature of the effects that
they may experience. May contain some or all of the following four steps:
Hazard identification. The determination of whether a particular
chemical is or is not causally linked to particular health
effect(s).
Dose-response assessment. The determination of the relation between
the magnitude of exposure and the probability of occurrence of the
health effects in question.
Exposure assessment.
exposure.
The determination of the . extent of human
Risk characterization. The description of the nature and often the
magnitude of human risk, including attendant uncertainty.
Risk characterization. The final step of a risk assessment, which is a
description of the nature and often the magnitude of human risk, including
attendant uncertainty.
Risk management. The.decision-making process that uses the results of
risk assessment to produce a decision about environmental action. Risk
management includes consideration of technical, scientific, social,
economic,, and political information.
xv
-------
Route of exposure. The means by which toxic agents gain access to an
organism (e.g., ingestion, inhalation, dermal exposure, intravenous,
subcutaneous, intramuscular, intraperitoneal administration).
Scoping study. Also known as screening study. An assessment of analysis
using tentative or preliminary data whose results are not accepted as
absolute indicators of risk or exposures, but rather, are taken as an
indication of the relative importance of various sources, pollutants and
control measures. Most urban air toxics assessments conducted to date
have been considered to be scoping studies, useful for pointing out where
more detailed work is needed prior to regulation.
Species profile. A set of apportioning factors that allow one to
subdivided VOC or PM emission totals into individual chemicals or chemical
classes. Generally, species profiles are multiplicative in nature.
Subchronic exposure. Exposure to a substance spanning approximately 10
percent of the lifetime of organism.
Synergism. A pharmacologic or toxicologic interaction in which the
combined effect of two or more chemicals is greater than the sum of the
effect of each chemical alone. (Compare with: additivity, antagonism.)
Threshold Limit Value fTLV). The concentration of a substance below which
no adverse health effects are expected to occur for workers assuming
exposure for 8 hours per day, 40 hours per week. TLVs are published by
the American Conference of Governmental Industrial Hygienists (ACGIH).
This listing may be useful in identifying substances used in the workplace
and having the potential to be emitted into the ambient air.
Threshold toxicant. A substance showing an apparent level of effect that
is a minimally effective dose, above which a response occurs; below that
dose no response is expected.
Transformation. The conversion, through chemical or physical processes,
of one compound or several compounds into other compounds as a result of
aging and irradiation in the atmosphere.
Transport. The movement of pollutants by wind flow. Transport is
characterized for modeling purposes by wind speed and wind direction.
Unit cancer risk. A measure of the probability of an individual's
developing cancer as a result of exposure to a specified unit ambient
concentration. For example, an inhalation unit cancer risk of 3.0 x
10 near a point source implies that if 10,000 people breathe a given
concentration of a carcinogenic agent (e.g., 1 /ig/m3) for 70 years, three
of the 10,000 will develop cancer as a result of this exposure. In water
the exposure unit is usually 1 /tg/1, while in air it is 1
Wei ght-of -evi dence . The extent to which the available biomedical data
support the hypothesis that a substance causes an effect in humans. For
example, the following factors increase the weight-of-evidence that a
chemical poses a hazard to humans; an increase in the number of tissue
xvi
-------
sites affected by the agent; an increase in the number of animal species,
strains, sexes, and number of experiments and doses showing a response;
the occurrence of a clear-cut dose-response relationship as well as a high
level of statistical significance in the occurrence of the adverse effect
in treated subjects compared with untreated controls; a dose related
shortening of the time of occurrence of the adverse effect; etc.
xvi i
-------
-------
ACRONYMS
ATERI'S
BaP
BID
CAG
CDD/CDF
DOE
EDB
EDC
EPA
HEM
IACP
IEMP
LOAEL
LOD
ME I
MIR
MLE
MWC
NESHAP
NRC
OAQPS
PCB
PIC
POHC
POM
Air Toxic Exposure and Risk Information System
Benzo(a)pyrene
Background information document
Carcinogen Assessment Group
Chlorinated dibenzo-p-dioxins and chlorinated
dibenzofurans
Department of Energy
Ethylene dibromide
Ethylene dichloride
Environmental Protection Agency
Human exposure model
Integrated Air Cancer Program
Integrated Environmental Management Project
Lowest-observed-adverse-effect -1 eve!
Limit of detection
Maximum exposed individual
Maximum individual risk
Maximum likelihood estimate
Municipal waste combustor
National emission standard for hazardous air
pollutants
Nuclear Regulatory Commission
Office of Air Quality Planning and Standards
Polychlorinated biphenyl
Products of incomplete combustion
Principal organic hazardous constituents
Polycyclic organic matter
xnx
-------
POTW
PUL
RCRA
SARA
SHED
STAPPA/ALAPCO
STB
TCCD
TSDF
TSP
VOC
g/mi 1 e
Publicly owned treatment works
Plausible upper limit
Resource Conservation and Recovery Act
Superfund Amendments and Reauthorization Act
SAI Human Exposure Dosage Model
State and Territorial Air Pollution Program
Administrators and the Association of Local Air
Pollution Control Officials
Science and Technology Branch
Tetrachlorinated dibenzodioxin
Treatment, storage, and disposal facilities (for
hazardous waste)
Total suspended particulates
Volatile organic compound
microgram per cubic meter
grams per mile
xx
-------
EXECUTIVE SUMMARY
This report presents an analysis of cancer risks in the United
States from outdoor exposures to airborne toxic pollutants. It is
intended to provide updated information to suggest priorities for air
toxics control. This study is an update of an EPA report issued in 1985
entitled The Air Toxics Problem in the United States: An Analysis of
Cancer Risks for Selected Pollutants (EPA-450/1-85-001, May 1985), known
as the "Six-Month Study."
This analysis is based primarily on information derived from
recent studies and reports. Results are expressed as cancer risk from .
individual pollutants and source categories in terms of excess lifetime
individual cancer risks1 and nationwide annual cancer cases.
Health risks due to indoor exposure and noncancer health effects
resulting from outdoor exposure are not included in this analysis, but
are addressed in separate studies.2 Risks from indoor exposures to
certain pollutants can be significant because of higher indoor
concentrations and the fact that most people spend much of their time
indoors. Noncancer risks from outdoor exposure also may be significant,
but more information is needed to adequately quantify these risks.
About 90 toxic air pollutants and 60 source categories were
addressed in one or more of the studies examined. Additional risks
associated with other pollutants and sources are not characterized. Of
particular concern is the absence of information on pollutants
secondarily formed in the atmosphere. Only one (formaldehyde) is
considered in this analysis.
Significant uncertainties are associated with estimating risk.
These are due to both data limitations and assumptions inherent in our
current risk assessment methodology and the methodology required to
combine and extrapolate information from individual studies to develop .
national estimates.
Assumptions about cancer potencies of various chemicals or
chemical mixtures are generally considered to overestimate the risk, as
do some assumptions about exposures. Uncertainties such as those due to
missing pollutants, uncharacterized sources, long-range transport of
1 "Lifetime individual risk" is a measure of the probability that
an individual will develop cancer as a result of exposure to an air
pollutant over a lifetime (i.e., a 70-year period).
2 See Report to Congress on Indoor Air Quality (EPA-400/1-89-001,
August, 1989) for current estimates of cancer public health risks from
exposure to indoor air toxics. EPA also is evaluating the noncancer
public health risks resulting from short-term and long-term outdoor
exposures to toxic air pollutants. This latter study is discussed in
Appendix C of this report.
ES-1
-------
pollutants, and pollutant transformation in the atmosphere will
underestimate the risk.
Major findings on national cancer incidence and lifetime
individual risk, which are subject to uncertainties and data limitations
as noted above, are highlighted below.
Cancer Incidence
• Based on the pollutants and source categories examined, total
excess cancer cases were estimated to be between 1,700 and
2,700 per year nationwide. This is equivalent to between 7 and
11 cancer cases per year per million population.
• Of the approximately 90 pollutants evaluated, 12 accounted for
over 90% of total annual cancer incidence. Of these, PIC
(products of incomplete combustion) were responsible for about
35% of the total. Other major contributors include 1,3-
butadiene, hexavalent chromium, benzene, formaldehyde, and
chloroform.
• Motor vehicles accounted for almost 60% of total cancer
incidence. Other area sources accounted for approximately 15%
of the total. Point sources accounted for the remaining 25% of
the total annual cancer incidence.
Lifetime Individual Risk
• Maximum lifetime individual risks exceeding 10~4 (exceeding 1
chance in 10,000 of contracting cancer) from multi-pollutant
exposures were reported in almost all studies. Risks of 10"3
or greater from individual pollutants were reported adjacent to
various types of sources.
• The relative contribution of pollutants and sources to risk in
a specific urban area can vary significantly. However, the
areawide lifetime individual risks in urban areas from the
combined exposure to many pollutants generally are in the 10"4
range, but varied from 10~s to 10"3. These levels result from
exposure to emissions from mobile and stationary sources
combined.
The numerical estimates presented in this report should be viewed
only as rough indications of the potential for cancer risk caused by a
limited group of pollutants found in the ambient air. Many of the risks
cited in this report are almost certainly inaccurate in an absolute
sense. The best use of the risk estimates is in describing the broad
nature of cancer risk posed by these toxic air pollutants and by making
relative comparisons of risks between pollutants and sources.
The technical approach for this study, including a description of
the methodology and a discussion of uncertainties and assumptions, is
presented in the next section. Additional information on major findings
Is provided under Results, and a comparison with the findings of the
ES-2
-------
1985 Six-Month Study is presented under Comparison With 1985 Six-Month
Study.
TECHNICAL APPROACH
Sources of Information
This study is based on information contained in 10 area-specific
or national air quality based risk-related reports on air toxics, 14 EPA
source category and pollutant-specific studies, risk assessments per-
formed for the development of National Emission Standards for Hazardous
Air Pollutants (NESHAP), and source specific risk data contained in the
EPA Air Toxic Exposure and Risk Information System (ATERIS) data base.
These reports and studies are described in Chapter 2 of this report.
They represent a much larger data base and more comprehensive coverage
than used for the 1985 Six-Month Study.
Additional information on air toxics emissions data is being
collected under Title III of the Superfund Amendments and
Reauthorization Act (SARA). However, in their present form, these data
can not be used to estimate risks. Therefore, this study does not
present risk estimates based on the SARA Title III emissions data.
Methodology
Estimates of annual cancer incidence were derived by first
developing estimates of the annual cancer cases per million population
for each pollutant/source category combination (e.g., 1,3-butadiene
emissions from mobile sources) reported in the data sources. These were
modified as necessary to reflect updated unit risk and emission factors.
Estimates of total nationwide annual incidence then were calculated, in
most instances, by multiplying the annual cancer cases per million
population by the total U.S. population and then summing across all
pollutant/source categories. Lifetime individual cancer risk estimates
either were obtained directly from each study or modified based on
updated information.
Because studies were of varying quality and most were concerned
with specific geographic areas, source categories, and/or pollutants, a
number of factors had to be examined to evaluate study results before
they could be combined and extrapolated to obtain national cancer
incidence estimates. These include the geographic scope of the study,
source category definitions, unit risk factors, method of estimating
ambient concentrations (modeled vs. monitored), and emission estimates.
These factors are discussed below.
Geographic Scope of Studies. Cancer rates for a pollutant and
source category were extrapolated to nationwide estimates based on the
geographic scope of each study examined. Most pollutant/source
categories were included in at least one study that was nationwide in
scope and this permitted a direct extrapolation to total nationwide
estimates.
ES-3 .
-------
A few pollutants and source categories were included only in a
study of limited geographic scope. In such instances, it was determined
whether the pollutant/source category might be unusually concentrated in
the area studied or was fairly common across the United States. This
information was used to determine how study results could be
extrapolated to obtain total nationwide estimates.
Source Category Definitions. Source category definitions in each
study were examined to minimize the possibility of double-counting.
This was especially difficult for the heating/combustion source category
because the various studies used different terminology and not all
reports clearly indicated what was or was not covered.
Unit Risk Factors. The unit risk factor is defined as an estimate
of the probability that an individual will develop cancer when exposed
to a pollutant at an ambient concentration of one microgram per cubic
meter (/tg/nr) for 70 years. These are either upper-bound values or
maximum likelihood values. The estimate of cancer risk for each
pollutant, considering the unit risk factor alone, is conservative; that
is, while the actual risk may be higher, it is more likely to be lower
and may even be as low as zero. The weight-of-evidence that a pollutant
causes cancer varies from proven human carcinogen (e.g., benzene) to
probable human carcinogen (e.g., 1,3-butadiene) to possible human
carcinogen (e.g., vinylidene chloride). All were included in this
analysis as carcinogens.
The cancer rates presented in the studies were updated, as
necessary, based on common unit risk factors used by EPA. With one
exception, this adjustment generally had little effect on the magnitude
of the total risk estimated by the various studies. The exception was
the South Coast study where the estimated cancer risk was 10 times
higher than the adjusted estimate based on EPA factors.
Although the'unit risk factors used in this report come from EPA
studies, not all of them have been officially approved by EPA. In
addition, many of the unit risk factors remain uncertain and are subject
to change as further evidence of carcinogenicity is obtained. For many
substances, this factor probably has the greatest potential for error in
estimating cancer risk. This is a significant issue and affects
pollutants such as formaldehyde, vinyl chloride, products of incomplete
combustion (PIC4), and diesel particulate (which is included with PIC).
Of particular concern are the unit risk factors for PIC mixtures
since these mixtures are responsible for about one-third of the cancer
cases estimated in this study. While many unit risk factors used in
this study have been approved by EPA, PIC is an important exception.
3 "Maximum likelihood estimate" refers to a statistical best
estimate of the risk.
4 "PIC" is primarily composed of "polycyclic organic matter" (POM).
Benzo(a)pyrene (BaP), which is used as a surrogate for PIC exposure, is
a component of POM.
ES-4
-------
There are no current EPA-approved unit risk factors for these mixtures
or for individual PIC components, although unapproved unit risk factors
are available for some of the compounds (e.g., benzo(a)pyrene). A
number of methods have been used to estimate the aggregate carcinogenic
potency of various PIC mixtures. The values and method selected for
this study were based on a review of ten EPA studies that included risk
estimates for PIC. In view of the potentially high risks associated
with PIC, there is a need for more thorough review to establish an EPA-
approved risk methodology and unit risk factor(s) for PIC mixtures that
can be used in future studies of this type.
Modeled vs. Measured Ambient Concentrations. Cancer risk
estimates can be derived from either modeled or measured ambient
concentrations. Each method has advantages and limitations. Both
ambient and modeled estimates were given equal weight in estimating
cancer risk unless there were clear reasons to prefer one estimate to
another.
A limitation of dispersion models is the need for accurate
estimates of pollutant emissions. A limitation of monitoring for many
analyses is the difficulty of monitoring at enough locations to
characterize the variability of ambient concentrations. This is true
primarily for point source analyses. In urban areas where the object is
to estimate average individual risk over a wide area, the specific
location of the monitor may not be as critical. In this case, the use
of measured ambient concentrations for risk estimation should provide
more credible and reliable results than reliance on modeled estimates.
Modeling and monitoring produced similar risk estimates for some
pollutants, such as for cadmium, methylene chloride, and trichloro-
ethylene. For others, such as chloroform, ethylene dibromide, and
formaldehyde, risk estimates based on measured ambient concentrations
were greater than model-based estimates.
For formaldehyde, the difference in results probably is due to the
fact that this pollutant is formed primarily in the atmosphere from
other volatile organic compounds. Models are not yet available which
can properly account for this, whereas ambient measurements do. The
reasons for the different results for chloroform and ethylene dibromide
are not clear, but a likely possibility is that the modeled results do
not include all sources of emissions of these pollutants.
Emission Estimates. Modeled ambient concentrations, and therefore
cancer risk estimates, are directly proportional to source emissions.
The quality of emissions data can vary significantly. Three pollutants
for which large uncertainties are associated with emissions estimates
are dioxin, gasoline vapors, and hexavalent chromium. These uncer-
tainties are recognized by reporting the risk from these pollutants as a
range.
The uncertainty in dioxin emissions estimates is associated with
emissions from hazardous waste treatment, storage, and disposal facili-
ties (TSDFs). The range of emission estimates for gasoline vapors is
ES-5
-------
due to the uncertainty as to the fraction of vapors that is carcino-
genic.
With respect to chromium, only hexavalent chromium is known to be
a carcinogen, but only total chromium is measured. Information has been
developed on the percent of total chromium emissions that is hexavalent
for specific emission sources, such as cooling towers, and this has been
used in modeling studies. Studies based on measured ambient concentra-
tions assume that a fraction of the measured chromium is hexavalent, but
this is not well-defined.
Pollutants and Sources Not Evaluated
Although approximately 90 different toxic air pollutants and over
60 source categories were addressed in one or more of the studies used
in this report, there are thousands of airborne chemicals that are
potentially toxic, but have neither adequate exposure nor health effects
data. Also, reliable quantitative emission estimates remain unavailable
for many potentially important source categories. The lack of data for
these pollutant and source categories could result in a significant
underestimate of risk.
There also is a lack of information on risks associated with
pollutants formed photochemically in the atmosphere (i.e., secondary
formation). There is evidence that the mutagenicity of mixtures of some
pollutants increases greatly as they undergo transformation in the
atmosphere, but insufficient data are available to derive cancer risk
estimates. Data on only one secondarily formed pollutant (formaldehyde)
are included in this study.
Additive Risk
Total nationwide annual incidence was calculated by summing the
risks for all pollutants and source categories. In addition, additive
lifetime individual risks were obtained by summing risks for different
pollutants at the same geographic location. This is the accepted
approach and was used in all of the studies reviewed.
It should be noted that the assumption of additivity can lead to
substantial errors in risk estimates if synergistic or antagonistic
interactions occur. Although dose additivity has been shown to predict
the acute toxicity of many mixtures of similar and dissimilar compounds,
some marked exceptions have been identified. In some cases, risks would
be greatly overestimated and, in other cases, greatly underestimated.
The available data are insufficient for estimating the magnitude of
these errors.
RESULTS
From the foregoing discussion, it is clear that there are numerous
assumptions and significant uncertainties associated with the risk
estimates in this study. In addition, potential sources of error are
important to recognize and are discussed in detail in this report. In
ES-6
-------
spite of these potential sources of error, it was concluded that point
estimates would be a more useful way to compare risks among various
pollutants and sources than by expressing broad ranges of risks.
Nevertheless, .for reasons discussed below, ranges were expressed for
several pollutants.
Magnitude of the Problem
For the pollutants and source categories examined, the total
nationwide cancer incidence due to outdoor concentrations of air toxics
in the U.S. was estimated to range from approximately 1,700 to 2,700
excess cancer cases per year (see Table ES-1). This estimate is based
on the most recent available unit risk factors and, in general, 1986
emissions data. It is roughly equivalent to between 7 and 11 annual
cancer cases per million population (1986 population of 240 million).
The number of deaths resulting from these projected cancer cases
is unknown. By way of comparison, the American Cancer Society has
estimated total cancer deaths in the U.S. in 1989 to be 500,000.
The range of estimated excess cancer cases per year in this study
is due primarily to the following uncertainties: (1) the unit risk
factor for diesel particulate (which is included in the estimated cancer
risk from PIC); (2) dioxin emissions from TSDFs; (3) the cancer-causing
portion of gasoline vapors; and (4) the fraction of total chromium that
is hexavalent.
Maximum lifetime individual risks of 1 x 10"4 (1 chance in 10,000
of contracting cancer) or greater were reported in almost all of the
studies examined. Maximum lifetime individual risk levels exceeding 1 x
10"4 were reported for multi-pollutant exposures from such sources as
major chemical manufacturers, waste oil incinerators, hazardous waste
incinerators, municipal landfills, publicly owned treatment works
(POTWs), and TSDFs.
Maximum individual risks of 1 x.10"4 or greater were reported
adjacent to at least one source for each of 16 pollutants included in
the NESHAP/ATERIS data base. Twelve of these pollutants were estimated
to be responsible for maximum individual risks of 1 x 10 or greater.
For the urban areas studied, areawide lifetime individual risks
from all pollutants for point and area sources combined generally were
in the 10"4 range, and ranged from 10"5 to 10"3. Lifetime individual
risks in four urban areas6 due to multi-pollutant exposure (9 to 16
pollutants) ranged from 10"4 to 10"3 based on measured ambient
concentrations. The contribution of specific area and point sources to
5 Acetaldehyde, acrylonitrile, arsenic, benzene, 1,3-butadiene,
carbon tetrachloride, chloroform, hexavalent chromium, coke oven
emissions, ethylene dichloride, epichlorohydrin, ethylene oxide,
methylene chloride,.p-dichlorobenzene, styrene, and vinylidene chloride.
6 Los Angeles, Baton Rouge, Boston, and Chicago.
ES-7
-------
TABLE ES-1
SUMMARY OF ESTIMATED NATIONWIDE ANNUAL CANCER CASES BY POLLUTANT
POLLUTANT
1. Acrylonitrile
2. Arsenic
3. Asbestos
4. Benzene
5. 1,3-Butadiene
6. Cadmium
7. Carbon tetrachloride
8. Chloroform
9. Chromium (hexavalent)
10. Coke Oven Emissions
11. Dioxin
12. Ethyl ene di bromide
13. Ethyl ene di chloride
14. Ethyl ene oxide
15. Formaldehyde
16. Gasoline vapors
17. Hexachlorobutadiene
18. Hydrazine
19. Methyl ene chloride
20. Perchloroethylene
21. PIC
22. Radionuclides0
23. Radon0
24. Trichloroethylene
25. Vinyl chloride
26. Vinyl idene chloride
27. Miscellaneous11
Total s
EPA
CLASSIFICATION8
Bl
A
A
A
B2
Bl
B2
B2
A
A
B2
B2
B2
B1-B2
Bl
B2
C
B2
B2
B2
_b-
A
A
B2
A
C
ESTIMATED ANNUAL
CANCER CASES
13
68
88
181
266
10
41
115
147-265
7
2-125
68
45
6
124'
19-76
9
6
5
6
438-1120
3
2
7
25
10
15
1,726 - 2,706
NOTE: Values in this figure are not absolute predictions of cancer-
occurrence and are intended to be used in a relative sense only.
The dose-response relationships and exposure assumptions have a
conservative bias, but omissions due to uncharacterized pollutants
(either directly emitted or secondarily formed) and emission
sources, the long-range transport of pollutants, and the lack of
knowledge of total risk from multi-pollutant exposures will offset
this bias to an unknown extent.
ES-8
-------
FOOTNOTES TO TABLE ES-1
a For a discussion of how EPA evaluates suspect carcinogens and more
information on these classifications, refer to "Guidelines for
Carcinogen Risk Assessment" (51 Federal Register 33992).
The EPA classifications used in this report are:
A = proven human carcinogen; B = probable human carcinogen (Bl
indicates limited evidence from human studies and sufficient
evidence from animal studies; B2 indicates sufficient evidence
from animal studies, but inadequate evidence from human studies);
C = possible human carcinogen
b EPA has not developed a classification for the group of pollutants
that compose products of incomplete combustion (PIC), although EPA
has developed a classification for some components, such as
benzo(a)pyrene (BaP), which is a B2 pollutant.
c From sources emitting significant amounts of radionuclides (and
radon) to outdoor air. Does not include exposure to indoor
concentrations of radon due to radon in soil gases entering homes
through foundations and cellars.
d Includes approximately 68 other individual pollutants, primarily from
the TSDF study and the Sewage Sludge Incinerator study.
ES-9
-------
these levels of risk generally could not be identified, but the
relatively narrow range of the areawide lifetime individual risks in the
urban areas studied suggests that other large urban populations may be
subject to similar risk levels.
Nature of the Cancer Risk
Individual Pollutants. Available information suggests that 177 of
the approximately 90 pollutants included in the data sources may each
account for risks of at least 10 cancer cases per year. Of these, 13
pollutants may each account for 40 or more cases per year. The
relative contributions of these pollutants to the total annual cancer
cases are shown in Figure ES-1.
' - •.
The pollutants found in this study to be the primary contributors
to annual cancer incidence also were frequently associated with high
maximum individual risks. Other individual compounds, such as
epichlorohydrin and styrene, which account for smaller aggregate cancer
incidence, also are associated with high individual risks (greater than
1 x lO'4).
Source Categories. Many types of sources contribute to aggregate
incidence and lifetime individual risk. Figure ES-2 illustrates the
relative contribution to total annual cancer incidence for each of the
source categories evaluated.
On an individual source category basis, motor vehicles were the
largest contributor to nationwide annual incidence, contributing
approximately 58% of the total [including approximately 35% of the total
contribution for which secondarily formed formaldehyde (shown as a
separate category) is responsible]. Electroplating (6%) was another
large contributor as a result of chromium emissions. Other major
contributors are TSDFs (5%); woodsmoke (5%); asbestos, demolition (4%);
gasoline marketing (3%); cooling towers (3%); and solvent use/degreasing
(3%).
A significant portion of the cancer risk from most sources usually
was due to a few pollutants, even where a source emitted many different
pollutants. For example, over 70 pollutants were included in the
analysis of hazardous waste combustors, but only two pollutants (cadmium
and hexavalent chromium) were estimated to be responsible for almost 90
percent of the estimated cancer cases in this source category.
Similarly, three pollutants (cadmium, hexavalent chromium, and arsenic)
were responsible for almost 90 percent of the estimated cancer cases
from hazardous waste boilers and furnaces.
Acrylonitrile, arsenic, asbestos, benzene, 1,3-butadiene,
cadmium, carbon tetrachloride, chloroform, hexavalent chromium, dioxin,
ethylene dibromide, ethylene dichloride, formaldehyde, gasoline vapor,
PIC, trichloroethylene, and vinyl chloride.
8 Arsenic, asbestos, benzene, 1,3-butadiene, carbon tetrachloride,
chloroform, hexavalent chromium, dioxin, ethylene dibromide, ethylene
dichloride, formaldehyde, gasoline vapor, and PIC.
ES-10
-------
o
a.
NOTE: Values in this figure are not absolute predictions of cancer occurrence
• and are intended to be used in a relative sense only. The dose-response
relationships and exposure assumptions have a conservative bias, but
omissions due to uncharacterized pollutants (either directly emitted or
secondarily formed) and emission sources, the long-range transport
of pollutants, and the lack of knowledge of total risk from multi-
pollutant exposures will offset this bias to an unknown extent
o
CO
0) W
.2 .2
S •
'
-S
9
42
i
3
o
CM
3-Butadiene
um, hexavalent
40
30
2
0
PERCENT CONTRIBUTION
O
CD
5
c
O
to
T3
CD
es
.is
w «
LU j|
lo "o
2 °-
•" ^
2 ^.
C CO
O CD
s t
^2 CD
5 Q.
5 w
O CD
0 %
| O
'I 8
CD e
CC co
. O
ES
o>
LL
iNvirmod ivnaiAiaNi
ES-ll
-------
CO
CO
•g CO
o P
o>>-
CD *-
CD CD
O O>
c §
§0
II
•fe C
o .2
O to
CD °^
& E
CO
LU
I
D)
U_
ES-12
-------
Comparing aggregate source categories, mobile sources were
estimated to contribute approximately 58 percent and stationary sources
approximately 42 percent of the total annual incidence. Area sources
were responsible for approximately 75 percent and point sources 25
percent of the total annual incidence associated with outdoor exposure
to air toxics.
Geographic Variability. Ambient concentrations of individual air
toxics vary on a city-to-city basis as well as on an intra-city basis.
For the cities included in this study, the variation among cities ranged
from a factor of 2 for benzene to almost 20 for chloroform. Similar
variations were found within cities. Many factors could account for
this. These include meteorological conditions, the location of sources
relative.to the population, and, where cancer risks were estimated from
measured ambient concentrations, the location of the monitors.
COMPARISON WITH 1985 SIX-MONTH STUDY
The present study shows approximately 500 to 900 more cancer cases
per year than reported in the 1985 Six-Month Study published in May,
1985. This apparent increase is due primarily to the inclusion of more
pollutants, a better accounting of emission sources, and, in some cases,
substantial increases in unit risk estimates.
The present study shows additive lifetime individual risks to be
similar to those estimated in the 1985 Six-Month Study. However, the
broader scope of the present study has identified additional source
types (e.g., TSDFs, POTWs) that can cause high lifetime individual
risks.
The individual compounds found in the present study to be the most
important contributors to cancer risk are, for the most part, the same
as those identified in the 1985 Six-Month Study. The most important
addition is 1,3-butadiene. Dioxin also may be an important contributor,
but the uncertainty associated primarily with estimates of dioxin
exposure from TSDFs makes it difficult to conclude this at the present
time. Several pollutants (asbestos, ethylene oxide, and
trichloroethylene) appear to be somewhat less of a factor in terms of
aggregate cancer risk, but not necessarily in terms of maximum
individual risk.
The 1985 Six-Month Study found that area sources accounted for
over 75 percent and point sources accounted for less than 25 percent of
the total annual cancer incidence. This finding was essentially
confirmed by the results of the present study. Findings in the present
study on the geographic variability of risk also are reasonably
consistent with those in the 1985 Six-Month Study.
ES-13
-------
-------
1.0 INTRODUCTION
Background
The U.S. Environmental Protection Agency (EPA) initiated a broad
"scoping" study in November, 1983, with a goal of gaining a better
understanding of the size and causes of the health problems caused by
outdoor exposure to air toxics. This broad scoping study, often
referred to as the Six-Month Study1, was published in May, 1985, and is
hereafter referred to in this report as the 1985 Six-Month Study.
The objective of the 1985 Six-Month Study was to assess the
magnitude and nature of the air toxics problem by developing
quantitative estimates of the cancer risks posed by selected air
pollutants and their sources from a national and regional perspective.
It was designed to answer four basic questions:
1. What is the approximate magnitude of the air toxics problem,
as measured by the estimated cancer risks associated with
air pollution?
2. What is the nature of the air toxics problem; that is, what
pollutants and sources appear to increase the incidence of
cancer and what are their relative importance?
3. Does the cancer risk problem vary geographically and, if so,
in what ways?
4. Are current air toxics data bases adequate for assessing the
cancer risk from air toxics? If not, what are the
significant data gaps?
1 Haemisegger, E. et. al_. The Air Toxics Problem in the United
States: An Analysis of Cancer Risks for Selected Pollutants. EPA-450/1-
85-001. May 1985. ,
1-1
-------
These questions were answered primarily by conducting three
analyses to estimate cancer incidence (i.e., cancer cases per year) and
lifetime individual risks.2 One analysis estimated national exposure
and risks from about 40 pollutants being considered for listing under
Section 112 of the Clean Air Act.3 The risks estimated in this
analysis, which is referred to as the NESHAP (National Emission
Standards for Hazardous Air Pollutants) Study, were national in scope
and considered emissions obtained from traditional air pollution
inventories. The emphasis of the NESHAP Study was on large point
sources, but both mobile and area sources were also covered. The second
analysis provided a more detailed estimate and analysis of exposure and
risk in 35 counties for approximately 20 pollutants.4 This second
analysis, which is referred to as the 35-County Study, was designed to
examine risk from air toxics on a more local perspective than the NESHAP
Study. The analysis in the 35-County Study included sources not usually
considered in previous studies, such as publicly owned treatment plants
(POTWs) and waste oil combustors. The third analysis, which is referred
to as the 1985 Ambient Air Quality Study, estimated cancer risks based
on ambient air quality data for fourteen pollutants.5 Quantitative risk
"Lifetime individual risk" is a measure of probability that an
individual will develop cancer as a result of exposure to the ambient
concentration of an air pollutant over a lifetime (i.e., a 70-year
period).
3 Schell, R.M. Estimation of the Public Health Risks Associated with
Exposure to Ambient Concentrations of 87 Substances. OAQPS, U.S. EPA,
July 1984. Revised February 1985.
4 Versar; American Management System, Inc. Hazardous Air Pollutants:
An Exposure and Risk Assessment for 35 Counties. U.S. EPA Contract No.
68-01-6115, September 1984.
5 Hunt, Bill, et..al. Estimated Cancer Incidence Rates from Selected
Toxic Air Pollutants Using Ambient Data. U.S. EPA, revised March 1985.
-------
assessments available from other EPA activities for asbestos, radio-
nuclides, and gasoline marketing supplemented these three analyses in
the 1985 Six-Month Study. Information available on several source
categories for which data at that time were insufficient to perform a
quantitative risk assessment were also analyzed and summarized in the
study. The main conclusions reached in the 1985 Six-Month Study are
summarized in Table 1-1.
Purpose of Current Study
The primary objective of the current study is to evaluate the
magnitude and nature of the cancer problem associated with outdoor
concentrations of air toxics in the United States. The magnitude of the
cancer problem is addressed in terms of annual cancer incidence (i.e.,
the number of cancer cases per year nationwide) and lifetime individual
risk (i.e., areawide and maximum individual risk6). The nature of the
cancer problem is addressed primarily by examining the relative
contributions of pollutants and sources to annual cancer incidence and
the geographic variability of cancer risk and important contributors to
that risk. In addition, the results of this study are compared with
those of the 1985 Six-Month Study. Finally, while the current study
does not include a reevaluation of EPA's air toxics control strategy,
the study seeks to present information on the magnitude and nature of
the air toxics problem due to outdoor exposure that may be used to help
set priorities for the control of air toxics and to better define
research and data needed to support a more effective control program.
6 "Areawide" individual risk refers to the average lifetime individual
risk to everyone in an area. "Maximum" individual risk refers to the
maximum level of risk to which a person could be exposed, and is located
at a specific point within an area.
1-3
-------
TABLE 1-1
MAIN CONCLUSIONS OF THE 1985 SIX-MONTH STUDY
Nationwide annual cancer cases were estimated as 1,300 to 1,700 (5
to 7.4 cases of cancer per year per million population) for the 15
to 45 pollutants examined in each analysis.
Maximum lifetime individual risks of 1 x 10"4 (1 in ,10,000) or
greater in the vicinity of major point sources were estimated for
21 pollutants, about half of those that were studied. Maximum
lifetime individual risks of 1 x 10"3 (1 in 1,000) or greater were
estimated for 13 pollutants.
-3
Additive lifetime individual risks in urban areas due to
simultaneous exposure to 10 to 15 pollutants ranged from 1 x 10
to 1 x 10 . These risks, which were calculated from monitoring
data, did not appear to be directly related to specific point
sources. Instead, they represent a portion of total risks
associated with the complex pollutant mixtures typical of urban
ambient air.
Thirteen specific pollutants8 were identified as possibly
important contributors to aggregate cancer cases from air toxics.
Although little aggregate cancer incidence (less than 1 cancer
case per year total) was found for 21 low production organic
chemicals, some of these compounds appear to be associated with
high individual risks. The low aggregate incidence for these
compounds may be due in part to the' lack of data concerning their
emissions and toxicity.
A wide variety of sources was found to contribute to cancer risk
from air toxics, with combustion/incineration probably the largest
single source of risk. Among this wide variety of sources were
sources,that have not historically been part of emission
inventories, such as publicly owned treatment works (POTWs) and
hazardous waste treatment, storage, and disposal facilities
(TSDFs), which were found to possibly pose important risks in some
locations.
Both point sources (major industrial sources) and area sources
(smaller sources that may be widespread across a given area, such
as solvent usage and motor vehicles) appear to contribute
significantly to cancer risk from air toxics. Large point sources
tended to be associated with many high individual risks, while
area sources appeared to be responsible for the majority of
aggregate cancer cases.
*
Where it could be analyzed, large city-to-city and neighborhood-
to-neighborhood variation in pollutant levels and sources was
found.
1-4
-------
TABLE 1-1 (concluded)
MAIN CONCLUSIONS OF THE 1985 SIX-MONTH STUDY
Major weaknesses and data gaps in the air toxics data bases at the
Federal, State, and local levels made it impossible to accurately
characterize most local air toxics problems. Problems identified
with the few available air toxics data bases were inconsistencies
and anomalies in the emission inventories, lack of sufficient data
to develop population exposure estimates, and lack of compounds
for which adequate health effects tests have been performed.
EPA's criteria pollutant15 program appears to have reduced air
toxics levels. One analysis estimated the cancer rate from 16 air
toxics in 1980 was less than half that for 1970 (6.8 vs. 17.5
cancer cases per year per million population).
SOURCE: Haemisegger, E. et. al_. (1985) pp. 94-96.
a The thirteen pollutants were: chromium, arsenic, asbestos, products of
incomplete combustion (PIC), formaldehyde, benzene, ethylene oxide, gasoline
vapors, chloroform, carbon tetrachloride, perch!oroethylene,
trichloroethylene, and vinylidene chloride.
b EPA's criteria pollutants are: carbon monoxide, ozone, lead, total
suspended particulate, oxides of nitrogen, and sulfur dioxide.
1-5
_
-------
One of the key analyses in this study is based on combining the
risk assessments from a number of information sources and developing an
estimate of total cancer incidence from all of the pollutants and
sources included in the various reports and studies. Ideally, this
analysis would lead to a point estimate-of total cancer incidence.
While a point estimate seemed reasonable for a large number of
individual pollutants, certain aspects of the risk assessment
methodology for other pollutants did not allow for identifying a point
estimate. For these other pollutants, only ranges could be identified.
The analysis then tried to narrow the range as much as possible.
Existing information from various reports and studies available
since the 1985 Six-Month Study was released has been gathered,
organized, and evaluated in order to accomplish these objectives. Some
of the quantitative estimates of risk used in this study to help
identify high risk air toxics and sources come from studies that are
part of the regulatory decision making process (e.g., background
documents in support of NESHAPs under Section 112 of the Clean Air Act).
Other quantitative risk estimates come from reports or studies that are
of a general "scoping" nature and are not,of the level of detail
necessary for regulatory decisions. In addition, the risk estimates
contained in these studies and reports are based on an uneven level of
quality, which affects the certainty that one can attach to the risk
estimates. For these reasons, the quantitative risk estimates can not
be used to support regulatory decisions on source regulation. These
results should, nevertheless, be useful in developing and evaluating air
toxics control strategies and in establishing priorities within these
strategies. Since there are limitations in most risk analyses, it is
important for the reader using this report to consider the caveats and
1-6
-------
assumptions associated with the analyses in order to interpret and use
the results properly.
Other Studies or Reports on Air Toxics
Health risk from air toxics encompasses both cancer and rioncancer
effects, and results from both indoor exposure as well as outdoor
exposure to air toxics. This report examines cancer risk from outdoor
concentrations of air toxics. The risk estimates presented in this
study are associated with just one part of the total risk from air
toxics (see Figure 1-1). Health risks from indoor exposure to air
toxics and the noncancer risks from outdoor exposure to air toxics are
the subjects of separate studies, which are discussed below. Also
discussed are air toxics emissions data recently released under Title
III of the Superfund Amendments and Reauthorization Act (SARA) of 1986.
Indoor Air Pollution
Under the Radon Gas and Indoor Air Quality Research Act of 1986
(Title IV of SARA), EPA is establishing a research program on all
aspects of indoor air quality. , As part of this program, EPA is seeking
to identify high risk pollutant sources and characterize the exposures
and health risks of various populations to those sources. Source
categories that have been identified are: environmental tobacco smoke,
combustion appliances, materials and furnishings, biological
contaminants, consumer products (e.g., hair spray, paint solvents,
cleaning fluids), outdoor sources (e.g., infiltration of radon, vehicle
exhaust, pesticides), and nonionizing radiation. The indoor air program
also addresses generic research activities. Generic research needs
emphasize the concept of limiting total exposures and include develop-
ment of standard measurement protocols, establishment of emission
reduction baselines, identification of mitigation techniques, and
1-7
-------
EXPOSURE
Indoor Outdoor
x
EC
<
o
O
This study
/y/// Indoor study program
/ScS< Noncancer study
"CANCER RISK FROM \
•,
OUTDOOR EXPOSURE V
TO AIR TOXICS"
Figure 1 -1. Relationship of this Study to Other Air Toxic Risk Studies
1-8
-------
dissemination of information to the public. The EPA's ultimate goals in
addressing indoor air quality problems are to characterize and
understand the risks to human health that pollutants pose and to reduce
those risks by reducing exposures. A report to Congress has been
prepared that estimates cancer risk from indoor air toxics.7
Noncancer Health Risk Study8
In a separate study, EPA is evaluating the noncancer public health
risks resulting from short-term and long-term outdoor exposure to toxic
air pollutants. Noncancer effects range from subtle biochemical,
physiological, or pathological effects to gross effects, including
death. The main focus of the noncancer study is on the evaluation of
risk from exposure to air toxics that are routinely emitted from
industrial or commercial sources. Excluded from the noncancer analysis
is the consideration of occupational exposures, indoor air pollutants,
criteria air pollutants, secondary atmospheric reaction products, and
accidental releases. The Executive Summary from the Noncancer Health
Risk study is presented in Appendix C of this report.
SARA Title III
Under Title III of SARA, EPA is collecting air toxics emissions
data from industrial and manufacturing sources that are covered by
certain Standard Industrial Classification (SIC) codes, have 10 or more
employees, and handle listed chemicals above threshold amounts. These
data are collected as part of the Toxic Release Inventory mandated under
7 See Report to Congress on Indoor Air Quality' (EPA-400/1-89-001,
August 1989) for current estimates of cancer public health risks from
exposure to indoor air toxics.
8 U.S. Environmental Protection Agency, OAQPS. Toxic Air Pollutants
and Noncancer Risks. Summary of Screening Study. External Review Draft,
September, 1990.
1-9
-------
Section 313 of SARA. Based on data contained in EPA's Toxic Release
Inventory Data (1989), approximately 320 toxic chemicals were identified
as being released to the environment and over 2.3 billion pounds of
toxic chemicals were identified as being released to the air from the
reporting facilities in 1987. Limiting the number of employees and
specifying threshold amounts resulted in excluding smaller producers and
facilities.
The following items highlight the relationship between the air
toxic emissions data collected under SARA Title III and the risk
estimates presented in this report.
- SARA Title III data concern only estimates of air toxics
emissions and not estimates of cancer risk. This report
focuses on estimates of cancer risk from exposure to air
toxics. This report does not estimate emissions of air toxics,
although the studies upon which the risk estimates are drawn
had to estimate such emissions.
• SARA Title III emissions data are limited to industrial and
manufacturing sources covered by SIC codes 20 through 39. This
report is not limited to these sources, but includes
additional emission sources such as mobile vehicles, treatment,
storage, and disposal facilities for hazardous wastes (TSDFs),
and dry cleaners.
• Generally, SARA Title III covers many more air toxics than
this report, which focuses on the subset of pollutants for
which cancer is the health concern and for which unit risk
factors are available.
• The only information source used in this report that is similar
to the SARA Title III effort is the Air Toxic Exposure and Risk
Information System (ATERIS) data base, which includes nation-
wide emission estimates of many pollutants covered by SARA
Title III.
The emission data submitted under SARA Title III were not used in
this study to estimate cancer risk. The SARA Title III emission data
are not reported in a form that allows estimation of risk. Thus, these
data could not be used to estimate cancer risk for this study. However,
1-10
-------
the information on source emissions gathered under SARA Title III may be
useful in identifying sources of concern for future risk assessments.
Outline of the Report • •_
This report is divided into two volumes. Volume I contains a
glossary of key terms; a list of acronyms; the Executive Summary;
Chapter 1, Introduction; Chapter 2, Scope of Study and Analyses; Chapter
3, The Magnitude and Nature of the Cancer Risk; and Chapter 4, Summary
and Conclusions. Volume II contains several appendices. The following
paragraphs describe the remaining chapters of Volume I. This is then
followed by a brief description of the material contained in Volume II.
In Chapter 2, "Scope of Study and Analyses," the various reports
and information used, the analytical methodology used to develop
estimates of annual cancer incidence, and major limitations and
uncertainties associated with the risk estimates presented in the study
are discussed.
The results of the study are presented in Chapter 3, "The
Magnitude and Nature of the Cancer Risk." The magnitude of risk
estimated, in terms of both estimated annual incidence and lifetime
individual risk, is presented first. The nature of the cancer risk, in
terms of individual pollutants, source categories, and geographic
variability, is then presented. The results are then compared with
those reported in the 1985 Six-Month Study.
Chapter 4, "Summary and Conclusions," summarizes the results of
the study and presents the conclusions drawn from it with regard to
the magnitude and nature of the cancer risk from outdoor air toxics.
As noted above, Volume II of this report contains the appendices.
Appendix A lists the individuals who commented on the external review
draft of this report and a summary of their comments. Appendix B
1-11
-------
provides detailed summaries of the analyses conducted for determining
the estimates of cancer cases per year per million population that would
be used in estimating total nationwide annual cancer incidence and the
resulting estimates of total nationwide cancer incidence for each of
those pollutants initially identified as possibly resulting in at least
ten cancer cases per year nationwide. Appendix C provides summaries of
the 14 EPA studies that focused on individual pollutants and source
categories which formed part of the data base. As noted earlier, the
Executive Summary to the Noncancer Health Risk study is also provided in
Appendix C.
1-12
-------
2.0 SCOPE OF STUDY AND ANALYSES
The purpose of this chapter is to provide the reader with an
overview of the scope of the study, the analyses performed in estimating
the cancer risk from air toxics, and an understanding of the limits and
uncertainties associated with it. The scope is described by a
discussion of the data base used. This discussion identifies for the
reader the various reports and studies included and the pollutants and
source categories examined. Next, the methodology used to derive the
nationwide estimates of annual cancer incidence is described. This
description gives the reader an understanding of the major components of
the annual cancer incidence analysis, as well as some of its boundaries.
Following the description of this analysis, limits and uncertainties
associated with the risk estimates presented in this report are
identified. By keeping in mind the scope of the study and the limits
and uncertainties associated with these risk estimates, the reader will
be able to more properly interpret and use the results of the study.
Data Base
A number of reports dealing with air toxics have been completed by
EPA or other agencies since the 1985 Six-Month Study was published. A
list of these reports was compiled and circulated to EPA Regional
Offices, the State and Territorial Air Pollution Program Administrators
and the Association of Local Air Pollution Control Officials (STAPPA/
2-1
-------
ALAPCO), and others to identify any additional reports that might be
included in this study. In addition to the reports, information from 14
individual source category- and pollutant-specific studies being
conducted by EPA was included in this study. Two of these studies (the
Superfund study and the Woodstove study) did not provide estimates of
cancer risk that could be used to estimate nationwide cancer risk. Risk
estimates based on the NESHAP (National Emission Standards for Hazardous
Air Pollutants) analysis used for the 1985 Six-Month Study and
supplemented by data contained in the Air Toxic Exposure and Risk
Information System (ATERIS) data base developed by the EPA's Office of
Air Quality Planning and Standards (OAQPS)1 were also used in the
analysis. As a result, the magnitude and nature of the cancer risk
posed by air toxics were evaluated based upon information contained in
ten reports, twelve source category- or pollutant-specific studies, and
the NESHAP/ATERIS data base. The ten reports are listed in Table 2-1
and the fourteen source categories and pollutants for which information
was obtained from other EPA studies are listed in Table 2-2.
The purposes of these reports and studies vary. Some were under-
taken as general scoping studies to estimate cancer risk from air toxics
in a specific locale (e.g., the"Integrated Environmental Management
Project (IEMP) studies, the South Coast Air Basin study) or on a
national basis (e.g., the Mobile Source study, the Ambient Air Quality
study). Some studies were undertaken to estimate cancer risk from a
specific source category (e.g., publicly owned treatment works (POTWs),
sewage sludge incinerators, mobile sources) or a specific pollutant
1 This is referred to in this study as the NESHAP/ATERIS data base.
The NESHAP risk estimates from the 1985 Six-Month Study were updated by
applying new unit risk factors for those pollutants whose unit risk
factors had changed since the original analysis.
2-2
-------
TABLE 2-1
LIST OF REPORTS USED IN STUDY
1. U.S. EPA, Region III. Kanawha Valley Toxics Screening Study,
Final Report. July 1987. (lEMP-Kanawha Valley)
2. U.S. EPA, OPPE. Santa Clara Valley Integrated Environmental
Management Project: Revised Stage I Report. May 30, 1986.
(lEMP-Santa Clara)
3. U.S. EPA, OPPE. Baltimore Integrated Environmental Management
Pro.iect: Phase I Report. May 1987. (lEMP-Baltimore)
4. U.S. EPA, Region V. Estimation and Evaluation of Cancer Risk
Attributable to Air Pollution in Southeast Chicago (Draft).
January 1989. (Southeast Chicago)
5. U.S. EPA, OAQPS. Analysis of Air Toxics Emissions, Exposures,
Cancer Risks and Controllability in Five Urban Areas. Volume I,
Base Year Analysis and Results. EPA-450/2-89-012a. July 1989.
(5 City)
6. U.S. EPA, OAQPS. Updated Estimated Cancer Incidence for Selected
Toxic Air Pollutants Based on Ambient Air Pollution Data. August
1989. (Ambient Air Quality)3
7. South Coast Air Quality Management District. The Magnitude of
'Ambient Air Toxics Impacts from Existing Sources in the South
Coast Air Basin. 1987 Air Quality Management Plan Revision
Working Paper No. 3. June 1987. (South Coast)b
8. U.S. EPA, OPPE. Final Report of the Philadelphia Integrated
Environmental Management Pro.iect. December 1986. (IEMP-
Philadelphia)
9. American Management Systems. Updated 35-Countv Study. March
1988. (35-County) This report was prepared under contract to the
U.S. EPA.
10. U.S. EPA, Office of Mobile Sources. Air Toxics Emissions from
Motor Vehicles. EPA-AA-TSS-PA-86-5. (Mobile Sources)0
2-3
-------
FOOTNOTES TO TABLE 2-1
a The VOC data used in this study was obtained from either: (1) J.J.
Shah and E. K. Heyerdahl, National Ambient Volatile Organic Compounds
(VOC's) Data Base Update, U.S. EPA, Atmospheric Sciences Research
Laboratory, Research Triangle Park, NC, February 1988, or (2) A.
Pollack, Systems Applications, Inc., Updated Report on the Interim
Data Base for State and Local Air Toxic Volatile Organic Chemical
Measurements, prepared for Bob Faoro, U.S. EPA, OAQPS, Research
Triangle Park, NC, August 1988. The trace metal data were obtained
from the Aerometric Information Retrieval System, U.S. EPA, OAQPS,
Research Triangle Park, NC, March 1988, and the benzo(a)pyrene (BaP)
data from J. Bumgarner, Environmental Monitoring and Systems
Laboratory, U.S. EPA, Research Triangle Park, NC, September 1988.
b Reprinted by the U.S. Environmental Protection Agency as Multiple Air
Toxics Exposure Study, Working Paper No. 3, South Coast Air Basin,
EPA-450/4-88-013, November 1988.
c Information in this study has been updated in this report using "Air
Toxics Emissions from Motor Vehicles," prepared by Penny Carey and
Joseph Somers. This paper was presented at the 81st Meeting of APCA,
Dallas, Texas, June 19-24, 1988.
2-4
-------
TABLE 2-2
EPA SOURCE CATEGORY AND POLLUTANT STUDIES
Source Categories
1. Coal and oil combustion
2. Drinking water aerators
3. Gasoline marketing
4. Hazardous waste combustors
5. Municipal incinerators
6. Municipal solid waste landfills
7. Publicly owned treatment works (POTWs)
8. Sewage sludge incinerators
9. Superfund sites
10. Treatment, storage, and disposal facilities for hazardous waste
(TSDFs)
11. Waste oil combustors
12. Woodstoves
Pollutants
13. Asbestos
14. Radionuclides
NOTE: The references used to obtain risk estimates from these source
category and pollutant studies are identified in Appendix C.
2-5
-------
(e.g., radionuclides, asbestos). The ATERIS data base contains
information generated from assessments of potentially toxic air
pollutants performed by OAQPS. The ATERIS contains data from all stages
of air toxics analyses, from the very preliminary to the more detailed
analyses. The data contained in the ATERIS are intended for the
relative comparison and ranking of source categories and pollutants on a
nationwide basis. The information in ATERIS is not considered an
authoritative source for verified estimates of risk attributable to
individual point sources.
The number of pollutants and source categories included in the
individual studies varied. As shown in Table 2-3, the number of
pollutants contributing to the estimated cancer risk in a study varied
from one (the Asbestos study) to 74 (the Hazardous Waste Combustor
study). The study for the treatment, storage, and disposal facilities
for hazardous waste (TSDFs) used an initial list of 84 potential air
pollutants, 74 of which were identified as being emitted. Of these 74
pollutants, risk estimates for 42 were made on the basis of available
EPA unit risk factors. Most studies included 9 to 20 individual
pollutants in their risk estimates.
A total of 90 different pollutants2 were included in the 22
studies and reports (see Table 2-4). Forty-eight of the pollutants were
included in one or two studies. Most of these 48 pollutants were found
in the NESHAP/ATERIS data base, the Hazardous Waste Combustor study, the
Sewage Sludge Incinerator study, or the TSDF study. Another 22
pollutants were found in three to six studies. Twenty pollutants were
included in more than six studies.
2 Not all of these pollutants, however, have EPA-derived unit risk
factors, as shown in Tables 2-6 and 2-7 and as indicated in Table 2-4.
2-6
-------
TABLE 2-3
NUMBER OF POLLUTANTS INCLUDED IN CANCER INCIDENCE ESTIMATES, BY STUDY
STUDY
NUMBER OF POLLUTANTS
INCLUDED IN RISK ESTIMATE
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
NESHAP/ATERIS Data Base
Ambient Air Quality
35-Coimty
5-City
I EMP- Baltimore
lEMP-Kanawha Valley
lEMP-Philadelphia
lEMP-Santa Clara
South Coast
Southeast Chicago
Mobile Sources
Asbestos
Coal and Oil Combustion
Drinking Water Aerators
Gasoline Marketing
Hazardous Waste Combustors
Municipal Waste Combustors
POTWs
Radionuclides
Sewage Sludge Incinerators
TSDFs
Waste Oil Combustors
45
20
19
19
9
18
9
14
15
30
9
1
9
10
4
74
10
7
2
33
42
8
NOTE: The Municipal Solid Waste Landfills, Superfund Sites, and
Woodstove studies do not include estimates of cancer
incidence.
2-7
-------
TABLE 2-4
NUMBER OF STUDIES THAT INCLUDED SPECIFIC POLLUTANT IN
CANCER RISK ESTIMATE, BY POLLUTANT
POLLUTANT
NUMBER OF
STUDIES
POLLUTANT
NUMBER OF
STUDIES
1. Acetaldehyde 3
2. Acrolein3 1
3. Acrylamide 4
4. Acrylonitrile 8
5. Aldrin 3
6. Ally! chloride 3
7. Aniline 2
8. Arsenic 13
9. Asbestos 4
10. Benz(a)anthracene 2
11. Benzene 17
12. Benzidine 2
13, Benzo(a)pyrene (BaP) 6
14. Benzyl chloride3 2
15. Beryllium 12
16. Bis(2-chloroethyl)
ether 1
17. Bis(chloromethyl)
ether 2
18, Bis(2-ethyhexyl)
phthalate 1
19. 1,3-Butadiene 9
20. Cadmium 15
21. Carbon tetrachloride 15
22. Chlordane 3
23. Chloroform 15
24. Chloromethane 2
25. Chlorophenolsb 4
26. Chromium (VI) 13
27, Coke Oven Emissions 3
28. DDT 2
29. Dibenz(a,h)anthracene 3
30. l,2-Dibromo-3-chloro-
propane 4
31. p-dichlorobenzene 1
32. 1,2-Dichloropropane 3
33. Dieldrin 2
34. Diethylstilbestrol 2
35. Diethanolamine8 1
36. Dimethylnitrosamine 1
37. 2,4-Dinitrotoluene 2
38. Dioctyl phthalate3 1
39. 1,4-Dioxane 2
40. Dioxin 6
41. 1,2-Diphenyl
hydrazine 2
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
Epichlorohydrin
Ethyl acrylate3
Ethylene dibromide0
Ethylene dichlorided
Ethylene oxide
Formaldehyde
Gasoline vapors
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Hexachlorobutadi ene
gamma-Hexachlorocy-
clohexane (lindane)
Hexachloroethane
Hydrazi ne/Hydrazi ne
sulfate
4,4 Isopropylidene
diphenol3
Methyl chloride
3-Methylchloanthrene
Methyl hydrazine
Methylene chloride
4,4-Methylene
dianiline3
Nickel (subsulfide)
Nitrobenzene3
2-Nitropropane
n-Nitroso-n-
butylamine
n-Ni tro-n-methylurea
n-Ni trosodi ethyl ami ne
Nitrosomorpholine3
N-Ni trosopyrroli di ne
Pentachloronitro-
benzene
Perchloroethylene
PICe
PCBf
Pronamide
Propylene dichloride3
Propylene oxide
Radionuclides
Radon
Reserpine
Styrene
Terephthalic acida
4
2
12
14
7
11
8
2
2
4
2
2
2
1
3
3
1
13
1
6
2
3
1
1
2
1
1
1
16
8
7
1
2
2
2
1
2
4
1
2-8
-------
TABLE 2-4 (concluded)
NUMBER OF STUDIES THAT INCLUDED SPECIFIC POLLUTANT IN
CANCER RISK ESTIMATE, BY POLLUTANT
NUMBER OF
POLLUTANT STUDIES
82,
83.
84.
85.
1,1,2,2-Tetrachloro-
ethane
Thiourea
Titanium dioxide3
Toxaphene
3
2
1
2
86.
87.
88.
89.
90.
, NUMBER OF
POLLUTANT STUDIES
1,1,1 -Tri chl oroethane3
Trichloroethane
Trichloroethylene
Vinyl chloride
Vinyl idene chloride
1
2
16
11
6
a EPA-derived unit risk factors not available.
b Includes pentachlorophenol and trichlorophenol.
factor for trichlorophenol is available.
c 1,2-dibromoethane.
d 1,2-dichloroethane.
e PIC = products of incomplete combustion.
/
f PCB = polychlorinated biphenyls.
Only an EPA unit risk
2-9
-------
As shown in Table 2-5, 65 source categories were identified from
among the studies and reports. Two of the source categories are general
in their coverage. These are: (1) chemical manufacturing (unspecified)
and (2) unspecified sources. Forty-five of the source categories were
identified as being in only one study. Most of these source categories
were identified in the NESHAP/ATERIS data base. Nine of the source
categories were included in four or more studies, with gasoline
marketing included in the most (nine) studies. It is likely that some
of the specified so'urce categories are included in the "chemical
manufacturing (unspecified)" and the "unspecified" categories.
Annual Cancer Incidence Analysis
The total nationwide estimate of cancer incidence was based on the
estimated cancer incidence from all pollutants for which unit risk
factors have been developed by EPA and from all source categories
covered by the studies and reports in the data base. It is important to
understand that not all of the unit risk factors developed by EPA have
undergone the same level of scrutiny. In general, many of the unit risk
factors (e.g., those for benzene and carbon tetrachloride) have been
"verified" by the Agency, having undergone review by an Agency work
group, the Carcinogen Risk Assessment Verification Endeavor. Such unit
risk factors are identified in Table 2-6 by reference to the Integrated
Risk Information System. Most of the unit risk factors, however, have
not been Agency-verified. The non-verified unit risk factors have
undergone various levels of review. Some have received review by the
Office of Health and Environmental Assessment. Others have received
little review. Among the least reviewed unit risk factors are those
estimated for the group of compounds referred to in this study as
2-10
-------
TABLE 2-5 ,u, ',-,
DISTRIBUTION OF SOURCE CATEGORIES BY NUMBER OF STUDIES
SOURCE CATEGORY
NUMBER OF
STUDIES
SOURCE CATEGORY
-NUMBER OF i;r
.STUB-IE$-'\.
1. ABS/SAN production8 1
2. Acrylic fiber production 1 34.
3. Acrylonitrile monomer 1 35.
4. Asbestos, demolition 1 36.
5. Asbestos, fabrication 1 37.
6. Asbestos, manufacturing 1 38.
7. Asbestos, milling 1 39.
8. Asbestos, renovation 1
9. Benzene fugitives 1 40.
10. Benzene storage I, - 41.
11. Benzene usage 1 42.
12. 1,3-Butadiene production 1
13. 1,4-Butanediol 1 43.
14. Cadmium pigment mfg. 1
15. Cadmium stabilizer mfg. 1 44.
16. Carbon tetrachloride
production 1 45.
17. Chemical manufacturing 46.
(unspecified) 6 47.
18. Chlorinated drinking
water 1 48.
19. Chlorine production 1 49.
20. Chlorinated hydrocarbon 50.
production 1
21. Chloroflourocarbon 51.
production .1 52.
22. Chlorinated hydrocarbon 1 53.
23. Chlorinated hydrocarbon 54.
users 1
24. Coal and Oil Combustion/ 55.
Heating 5 56.
25. Commercial Sterilization/ 57.
Hospitals 2 58.
26. Cooling Towers 3 59.
27. Drinking Water Aerators 2
28. Drycleaning 4 60,
29. EBS production13 1 61.
30. EDB manufacturing0 " 1 62,
31. Electroplating 4 63,
32. ETO production01 1 64,
33. Formaldehyde production 1 65.
r:v
>\»
'*'..' '*'
v.V/.tfff
4^-T.i.^ •
;V-. ••> '•*'•;
•- v*?'*
v^i' .•"
Gasoline Marketing ••j.-jj_
Glass mfg. ' T . ';
Hexamethyl enetetram: mfg.',
Ind. solvent coatings .
Iron and Steel mfgi'v "
Mel ami ne fjormal dehyde '.-..;
resin •'•••••• , V
4,4-Methylenedianiline L
Motor vehicles . . ";.-;. 8:
Municipal solid waste - ;•,
landfills ' -. . .:": :• -•;-v',2;
Municipal waste ... ',...,.-•- •
combustors :./ .. .•:*, 4
Nitrile elastomer - ,;*,'>7-y
evaporation. ' f. , :'.'..,,1,
Other organic evaporation, ;.; 1
Pentaerythitol production .-I,
Pesticide Production/ ., '; ;
Usage . ,•-.,,.'•.' ,2
Petroleum Refineries, ,.': - 2
Pharmaceutical mfg. ''.:•!
Phenol formaldehyde. : 4- '
resins ' ; ' /;;,"!.-.-w
Phthalic anhydride " ; l:-":-i
Polyacetal resins ; 1;U::
Polybutadiene production ''!,.;
Publicly owned treatment .>.;•/•;•';*'.
works : i;/ ; :6 .,"
Pulp and paper mfg; " ':T-..-"-i'f;-:':
Sewage sludge incinerators 2 >;
Solvent Use/Decreasing , 5' .!
SBR production ; 1 .v
Stripping (paint, photo--; ";..';""
resist) h-,'-••r-1^,
TSDFsf '. --..^;-..-.- ^2-r-r,
Trimethylopropane ' " :Vvi^:;;;
Unspecified • 7"T^;'i
Urea Formaldehyde .,„" T'"^
Waste Oil Burning ' ,: 3..'•.' •
Woodsmoke '„ 1' :'-7'
2-11
••:.-,• , ., ;&. -•••'.•
• • • . ^ -" ./v1-;^'--' *,
-------
FOOTNOTES TO TABLE 2-5
a ABS/SAN - acrylonitrile butadiene styrene/styrene acrylonitrile
b EBS = ethyl benzene styrene
0 EDB = ethylene dibromide
d ETO - ethylene oxide
e SBR - styrene butadiene rubber
f TSDF - treatment, storage, and disposal facilities for hazardous waste
2-12
-------
products of incomplete combustion (PIC).3 Finally, many of the unit
risk factors remain uncertain and are subject to change as further
evidence of carcinogenicity is obtained.
Primarily due to the limited time and resources available for
the report, the annual cancer incidence analysis was limited in two
aspects:
. The analysis did not try to verify the results of the various
studies. Any errors that might be contained in the studies
would, therefore, be carried over into this study. In a few
instances, some information was double-checked .as calculations
suggested a possible error or two. Double checking of
information was the exception, however, and not the rule.
• The initial analysis was carried out on the basis of readily
available background documents and reports. In some instances,
the documents and reports did not provide all of the necessary
level of detail that would have been preferred. This left a
level of uncertainty in trying to compare data and resolve
differences. In general, these instances have been identified
in the pollutant-by-pollutant analysis summaries, which are
found in Appendix B.
Methodology
The annual cancer incidence analysis began by assembling the
annual cancer incidence estimates for each pollutant by source category
from each of the 22 studies. Because the 22 studies varied in
geographic scope and thus population exposed, the annual cancer
incidence estimates were of limited value by themselves, especially
where the study was of limited geographic scope. Therefore, an attempt
was made to correct for geographic scope by calculating the cancer
incidence per year per million population for each pollutant in each
3 In this study, PIC refers to the large number of primarily
particulate compounds that result from incomplete combustion. PIC is
composed primarily of "polycyclic organic matter" (POM). Some studies
use the term POM when estimating the risk from this class of compounds.
In addition, some studies use benzo(a)pyrene (BaP), which is a component
of POM, as a surrogate to estimate risk from PIC.
2-13
-------
source category for each study. For the smaller, localized studies, the
population reported in the studies was used to calculate the annual
cancer incidence per million population. For each nationwide study, a
1986 population of 240 million was used rather than trying to determine
the base year for each nationwide study.4
The various pollutants and source categories frequently
"overlapped" between reports; that is, the same pollutant/source
category combination (e.g., 1,3-butadiene emissions from motor vehicles)
was included in'more than one study or report. Figure 2-1 illustrates
this overlap in a simplified diagram for five hypothetical studies.
Studies No. 1 and 2 represent some of the larger studies, such as the
35-County study or the 5-City study. Study No. 3 represents a source
category specific study. Studies No. 4 and 5 represent pollutant
specific studies. For example, Study No. 2 is seen in Figure 2-1 to
cover three of the same pollutants for two source categories as Study
No. 1, and the same three pollutants for one source category as Study
No. 3. Study No. 5 overlaps one pollutant/source category combination
with Study No. 3. Study No. 4 covers some of the same source categories
found in Studies No. 1 and 2, but for a different pollutant.
Where overlaps of pollutants and source categories occurred, the
estimates of annual cancer incidences per million population from each
study were compared. If the estimates were the same (or essentially the
same) for a pollutant/source category across all studies, additional
analysis to identify potential causes for differences was obviously
The risk estimates in all of the studies and reports used in this
study are based on 1980 to 1987 data (i.e., emission inventories, ambient
measurements, populations, etc.). For purposes of this study, these data
were treated as applying to the same time frame. The risk estimates can
be considered as mid-1980 numbers, or 1986 estimates.
2-14
-------
0 CO
CO CD
•a T3
C 3
I 8
I "
O ^
i^_ O-
•f 5
g iZ
"CD "S
-> O
i= co
CO CD
3. "v—
= O
D)
CD
3
D)
O
2-15
-------
unnecessary. If differences in the estimates of annual cancer incidence
per million population were found within a pollutant/source category
combination, a reduction analysis5 (as discussed below) was conducted to
resolve the differences and develop a point estimate6 of the annual
cancer incidence per million population for that pollutant/source
category combination. If a pollutant/source categpry combination was
unique to an individual study, the estimate of annual cancer incidence
per million population for that pollutant/source category was considered
the best available estimate. Once the point estimates of annual cancer
incidence per million population were identified, they were adjusted, as
necessary, to common unit risk factors for each pollutant. (This
adjustment is discussed later in this chapter under the Reduction
Analysis section.)
In extrapolating the estimates of annual cancer incidence per
million population to total nationwide annual incidence, the geographic
scope of the study was considered. Most pollutants and source
categories were in at least one study that was nationwide in scope.
This enabled, in most instances, a direct extrapolation to total
nationwide estimates (i.e., multiplying the cancer rate by the total
U.S. population of 240 million). A few pollutants and source categories
were included only in a study of limited geographic scope. In such
instances, an attempt was made to determine whether the pollutant/source
5 This type of reduction analysis was not undertaken for individual
risks because individual risks are site-specific numbers that cannot be
extrapolated to a nationwide estimate of individual risk. Instead, the
study presents the estimates of individual risk as found in each of the
studies used in the data base for this study.
6 In some instances, it was not possible to develop a point estimate.
In such cases, the range of estimates for the cancer rate was narrowed as
much as possible.
2-16
-------
category was unique to the geographic area, unusually concentrated in
the area, or fairly common across the United States. If it was unique
to the area or appeared to be unusually concentrated in the area, then
generally only the cancer incidence estimated in the study for that -.
category was included in the total nationwide estimate. If the
pollutant/source category appeared to be fairly widespread, the estimate
.of annual cancer incidence per million population was extrapolated to a
total nationwide estimate (i.e., multiplied by 240 million population).
Once this was done, the estimates of risk for each pollutant/
source category combination were summed to calculate the nationwide
estimate of annual cancer incidence.
Reduction Analysis. As noted previously, a large number of
pollutant/source category combinations with discrepant estimates of
annual cancer incidence per million population were identified. An
analysis was undertaken in an attempt to derive a single estimate of the
annual cancer incidence per million population.
A decision was made to limit the number of pollutant/source
category combinations for which the reduction analysis would be
conducted. It was decided to analyze the estimates of annual cancer
incidence per million population of pollutant/source category
combinations for those pollutants that could potentially result in 10 or
more cancer cases per year nationwide based on information in any one
study. These pollutants were identified in one of two ways:
(1) by the total number of annual cancer cases estimated for them
in studies that were nationwide in scope (e.g., the Ambient
Air Quality study, the Mobile Source study); or
(2) by the calculated number of cancer cases per year per million
population which when extrapolated nationwide might result in
10 or more cancer cases per year for the smaller geographic
studies (e.g., the four IEMP studies).
2-17
-------
A total of 23 pollutants were identified. It is these 23 pollutants
that are presented in Appendix B.
The reduction analysis looked to identify and reduce the
discrepancies by analyzing the following set of factors:
• unit risk factors
• emission factors
• modeled vs. ambient-measured concentrations
• source category definition and coverage
• geographic scope of the study
• study specific considerations
Each of these factors are discussed below as to how they were used and
considered in the reduction analysis.
Unit Risk Factors.7 Perhaps the most obvious reason that two estimates
of annual cancer incidence per million population would differ is that a
different unit risk factor had been used. Unit risk factors have
changed in the past and may change in the future. Thus, the first step
in the analysis was to put these estimates on the same "footing"; that
is, making sure the risk estimates are compared using the same unit risk
factors. The unit risk factors used in each study8 were compared to
those identified in Table 2-6 and Table 2-7. Table 2-7 shows the unit
risk factors used to estimate the cancer risk from PIC. The pollutants
7 The unit risk factor is a quantitative estimate of the
carcinogenicity potency of a pollutant. It is often expressed as the
chance of contracting cancer from a 70-year lifetime continuous exposure
to a concentration of one microgram per cubic meter (1 /*g/m3) of a given
pollutant. For example, benzene has a unit risk factor of 8.3x10
Ug/m3)"1. In a population of 100,000 people exposed to 10 ng/m of
benzene for 70 years, the upper-bound estimate of cancer cases is
calculated to be 8.3 cancer cases over 70 years (10 ^g/m3 x 100,000 people
x 8.3xlO"6 (/ig/m3)"1 = 8.3 cancer cases over 70 years).
8 Unit risk factors used in the Municipal Waste Combustor study were
not in the available reports, and were assumed to be the same as those in
Table 2-6. Unit factors for radionuclides and radon were accepted "as is"
in the reports.
2-18
-------
TABLE 2-6
UNIT RISK FACTORS USED TO COMPARE CANCER RISK
EPA
POLLUTANT CLASSIFICATION
(CAS NO.)
•1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
Acetaldehyde (75-07-0)
Acryl amide (79-06-1)
Acrylonitrile (107-13-1)
Aldrin (309-00-2)
ATlyl chloride (107-05-1)
Aniline (62-53-3)
Arsenic (7440-38-2)
Asbestos (1332-21-4)
Benz (a) anthracene
(56-55-3)
Benzene (71-43-2)
Benzidine (92-87-5)
Beryllium (7440-41-7)
Bi s (2-chl oroethyl ) ether
(111-44-4)
Bi s (chl oromethyl ) ether
(542-88-1)
Bi s (2-ethyl hexyl )phthal ate
(117-81-7)
1,3-Butadiene (106-99-0)
Cadmium (7440-43-9)
Carbon tetrachloride
(56-23-5)
Chlordane (12789-03-6)
Chloroform (67-66-3)
Chloromethane (74-87-3)
Chromium (VI) (7440-47-3)
Coke Oven Emissions
DDT (50-29-3)
Dibenz (a, h) anthracene
(53-70-3)
l,2-Dibromo-3-chloro-
propane (96-12-8)
1 , 2-Di chl oropropane
(78-87-5)
Dieldrin (60-57-1)
Di ethyl sti 1 besterol
(56-53-1)
Dimethyl ni trosami ne
(62-75-9)
2,4-Dinitrotoluene
(121-14-2)
1,4-Dioxane (123-91-1)
Dioxin (1746-01-6)
B2
B2
BI
B2
B2
B2
A
A
B2
A
A
B2
B2
A
B2
B2
BI
B2
B2
B2
--
A
A
B2
B2
B2
C
B2
--
B2
B2
B2
UNIT RISK
a FACTORS
(tta/m3r1
2.2xlO'6
l.lxlO'3
6.8xlO'5
4.9xlO'3
5.5xlO'8
7.4xlO'6
4.3xlO'3
7.6xlO'3
8.9xlO'4
f
8.3xlO"6
6.7xlO"2
2.4xlO'3
3.3xlO'4
2.7xlO"3
*7
2.4xlO'7
/
2.8xlO"4
1.8xlO"3
1.5xlO"5
3.7xlO"4
2.3xlO'5
3.6xlO'6
1.2xlO'2
6.2xlO'4
S.OxlO'4
1.4xlO'2
6.3xlO'3
1.8xlO'5
"2
4.6xlO'3
l.4x!0'1
o "
1.4xlO'2
c
8.8xlO'5
£
1.4xlO'6
3.3xl01
REFERENCE
1
2
1
1
3
2
1
lb
2
1
1
1
1
2°
4
1
1
1
1
1
4
1
1
2
2
2
5
2
4
1
2
2
2
2-19
-------
TABLE 2-6 (continued)
UNIT RISK FACTORS USED TO COMPARE CANCER RISK
EPA UNIT RISK
POLLUTANT CLASSIFICATION3 FACTORS
(CAS NO.) (Ua/m3r1
34. 1,2-Diphenyl hydrazi ne
(122-66-7)
35. Epichlorohydrin (106-89-8)
36. Ethyl ene di bromide
(106-93-4)
37. Ethyl ene di chloride
(107-06-2)
B2
B2
B2
B2
38. Ethylene oxide (75-21-8) B1-B2
39. Formaldehyde (50-00-0)
40. Gasoline vapors
(8006-61-9)
41. Heptachlor (76-44-8)
42. Heptachlor epoxide
(1024-57-3)
43. Hexachl orobenzene
(118-74-1)
44. Hexachl orobutadi ene
(87-68-3)
45. gamma-Hexachloro-
cyclohexane
(lindane) (58-89-9)
46. Hexachl oroethane
(67-72-1)
47. Hydrazine (302-01-2)
48. Methyl chloride (74-87-3)
49. 3-Methylchloanthrene
(56-49-5)
50. Methyl hydrazi ne (60-34-4)
51. Methyl ene chloride
(75-09-2)
52. Nickel (subsulfide)
(12035-72-2)
53. 2-Nitropropane (79-46-9)
54. n-Nitrosodi-n-
butylamine (924-16-3)
55. n-Nitrosodiethylamine
(55-18-5)
56. n-Nitroso-n-methylurea
(684-93-5)
57. n-Nitrosopyrrolidine
(930-55-2)
58. Pentachloronitro-
benzene (82-68-8)
59. Perch! oroethyl ene
(127-18-4)
Bl
B2
B2
B2
B2
C
C
C
B2
C
B2
B2
B2
A
B2
B2
B2
B2
B2
C
B2
2.2xlO'4
1.2xlO"6
2.2xlO'4
2.6xlO'5
l.OxlO'4
l.SxlO'5
6.6xlO'7
l.SxlO'3
2.6xlO'3
4.9xlO"4
2.2xlO'5
3.8xlO'4
4.0xlO'6
2.9xlO"3
3.6xlO'6
2.7X10'3
S.lxlO'4
4.7xlO"7
4.8X10'4
2.7xlO'3
1.6xlO'3
4.3xlO"2
8.6xlO"2
6.1xlO'4
7.3xlO'5
5.8xlO'7
REFERENCE
1
1
1
1
2
1
2
1
1
4
~
1
2
1
4d
2
2
2
2
1
2
1
2
2
1
2
2
2-20
-------
TABLE 2-6 (concluded)
UNIT RISK FACTORS USED TO COMPARE CANCER RISK
EPA
POLLUTANT CLASSIFICATION
(CAS NO.)
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
PCB's (1336-36-3)
Pronamide (23950-58-5)
Propylene oxide (75-56-9)
Reserpine (50-55-5)
Styrene (100-42-5)
1,1,2,2-Tetrachloro-
ethane (79-34-5)
Thiourea (62-56-6)
Toxaphene (8001-35-2)
1 , 1 , 2-Tri chl oroethane
(79-00-5)
Tri chl oroethyl ene
(79-01-6)
2,4,6-Trichlorophenol
(88-06-2)
Vinyl chloride (75-01-4)
Vinyl idene chloride
(75-35-4)
B2
C
B2
B2
B2
C
B2
B2
C
B2
B2
A
C
UNIT RISK
a FACTORS
(tta/ni3r1
1.2xlO'3
4.6xlO'6
3.7xlO"6
S.OxlO'3
5.7xlO'7
5.8xlO'5
5.5xlO'4
3.2xlO"3
1.6xlO'5
£.
1.7xlO"6
£
5.7xlO"6
4.1xlO'6
5.0xlO'5
REFERENCE
2
2
6
2
6
1
2
1
1
2
2e
£
f
1
a For a discussion of how EPA evaluates suspect carcinogens and more information on these
classifications, refer to "Guidelines for Carcinogen Risk Assessment" (51 Federal Register
33992). The EPA classifications used in this report are:
A = proven human carcinogen
B = probable human carcinogen (B1 indicates limited evidence from human studies and sufficient
evidence from animal studies; B2 indicates sufficient evidence from animal studies but
inadequate evidence from human studies)
C = possible human carcinogen
Derived from 2.3 x 10 per fibers per ml (millimeter), which is the unit risk factor reported in
_ IRIS (Integrated Risk Information System)
c IRIS currently reports a unit risk factor of 6.2x10"2 (jig/m3) .
j "? 3 -1
IRIS currently reports a unit risk factor of 4.9x10 (jig/m) .
e IRIS currently reports a unit risk factor of 3.1x10 (jig/m > •
An alternative unit risk factor of 4.2x10 (;tg/m ) has been developed by ORD. U.S.
Environmental Protection Agency, Office of Research and Development, Office of Health and
Environmental Assessment. Health Effects Assessment Summary Tables First Quarter FY89. January
1989.
2-21
-------
REFERENCES TO TABLE 2-6
1. U.S. EPA, Office of Health and Environmental Assessment. Cincinnati,
Ohio. Integrated Risk Information System (on-line data base).
2. U.S. Environmental Protection Agency. Hazardous Waste TSDF -
Background Information for Proposed RCRA Air Emission Standards.
Volume II - Appendices. Preliminary Draft. March 1988. pp. E-8
through E-13.
3. Schell, R.M. Estimation of the Public Health Risks Associated with
Exposure to Ambient Concentrations of 87 Substances. OAQPS, U.S.
EPA, July, 1984. Revised February 1985.
4. U.S. EPA, Office of Solid Waste. Draft Supplemental Rule for
Hazardous Waste Incinerators. Appendix B, Unit Risks for
Carcinogenic Constituents. January 16, 1989.
5. lEMP-Philadelphia. Developed from EPA's Drinking Water Criteria
Document. March 2, 1984.
6. U.S. Environmental Protection Agency, Office of Research and
Development, Office of Health and Environmental Assessment. Health
Effects Assessment Summary Tables Third Quarter FY90. July 1990.
7. U.S. Environmental Protection Agency, Office of Research and
Development, Office of Health and Environmental Assessment. Health
Effects Assessment Summary Tables First Quarter FY89. January 1989.
2-22
-------
TABLE 2-7
UNIT RISK FACTORS USED TO ESTIMATE
CANCER RISK FROM PIC
SOURCE
CATEGORY
Unspecified
Unspecified6
Coke Ovens0
Municipal Incinerators0
Industrial power plants,
oil0
Utility power plants, oil0
Industrial power plants,
coal0
Utility power plants, coal0
Residential Heating0
Oil
Coal
Wood
Gasoline vehicles
Diesel vehicles
Sewage Sludge Incinerators
Hazardous Waste Combustors
COMPONENT
BaP
PIC
POM
POM
POM
POM
POM
POM
POM
POM
POM
POM
POM
BaP
PIC.
UNIT RISK
FACTOR
Ug/m3)-1
1.7xlO"3a
4.2xlO'1
6.5xlO'5
8xlO'8
S.OxlO"7
S.OxlO"7
8xlO"8
8xlO"8
9xlO"6
l.OxlO"5
l.OxlO"5
2.5xlO"4
2.0x10"* to 10x10"*
3.7xlO"6
l.OxlO'5
REFERENCE
1
2
3
3
3
3
3
3
3
4
4
5
5
Based on inhalation study,
3.3xlO"3 (jig/m3)"1.
Oral study suggests a unit risk factor of
b This unit risk factor for products of incomplete combustion (PIC) was
based on relating lung cancer deaths to benzo(a)pyrene (BaP)
concentrations where BaP serves as a surrogate for the large category
of BaP-related pollutants referred to in the Six-Month Study as PIC.
For a more detailed explanation of its derivation, refer to pages 20
to 24a of the Six-Month Study.
0 These factors have been adjusted such that they are applied to the
total particulate concentration to estimate risk from the POM fraction
of the particulate matter.
2-23
-------
REFERENCES TO TABLE 2-7
1.
2.
3.
4.
5.
U.S. Environmental Protection Agency. Hazardous Waste TSDF -
Background Information for Proposed RCRA Air Emission Standards,
Volume II - Appendices. Preliminary Draft. March 1988. pp. E-8
through E-13.
U.S. EPA, Office of Policy, Planning, and Evaluation. The Air
Toxics Problem in the United States: An Analysis of Cancer Risks
for Selected Pollutants. EPA-450/1-85-001. May 1985.
U.S. EPA. Analysis of Air Toxics Emissions, Exposures, Cancer Risks
and Control!ability in Five Urban Areas. EPA-450/2-89-012a. July
1989.
U.S. EPA, Office of Mobile Sources.
Vehicles, September 1987.
Air Toxics Emissions From Motor
Memorandum. Shiva Garg, US EPA, Office of Solid Waste and Emergency
Response, to Joseph Padgett, US EPA, Office of Air Quality Planning
and Standards. Review of OAQPS Report on Six-Month Study of Impacts
of Air Toxics on Cancer Incidence. March 3, 1989.
2-24
-------
in each study that had different unit risk values than those shown in
Table 2-6 are identified in Table 2-8.
The estimates of cancer incidence for PIC reported in Chapter 3
and the Executive Summary are based primarily on the unit risk factors
specific to individual source categories that are shown in Table 2-7.
The unit risk factor for PIC of 4.2 x 10"1 Ug/m3)'1 for unspecified
sources was used only if a source-specific PIC unit risk factor was not
available. The method used to calculate this PIC unit risk factor was
unusual, and any risk estimate based on its use should be treated as a
very preliminary estimate. Some of the studies, such as the Ambient Air
Quality study, used this unit risk factor to estimate risk from PIC
using benzo(a)pyrene (BaP) ambient-measured concentrations as a
surrogate for PIC exposure. Some studies also used this unit risk
factor for purposes of comparing cancer incidence estimates using
various methodologies. For a discussion of these methodologies, please
refer to the section on PIC found in Appendix B.
If a pollutant's unit risk factor differed from that in Table 2-6,
the estimated annual cancer incidence was adjusted to reflect the unit
risk factor in Table 2-6. In general, there was little net effect on an
individual study's overall estimate of cancer cases as a result of this
modification (see Table 2-9). The one exception to this was the South
Coast study. The decrease in estimated annual cancer cases for the
South Coast study was due to large differences between the California
Department of Health Services (DOHS) unit risk factors used for several
pollutants in that study and EPA's unit risk factors for those
pollutants. As seen in Table 2-10, adjusting the South Coast study's
estimates of cancer cases by using the unit risk factors in Table 2-6
2-25
-------
TABLE 2-8
POLLUTANTS WITHIN IT RISK FACTORS DIFFERENT FROM
THOSE USED IN THIS REPORT
STUDY
POLLUTANTS WITH DIFFERENT
UNIT RISK FACTORS3
1. Ambient Air Quality
2. NESHAP/ATERIS
3. Asbestos
4. Coal and Oil Combustion
5. Drinking Water Aerators
6. Gasoline Marketing
7. Hazardous Waste Combustors
8. Mobile Sources
9. Municipal Waste Combustors
10. POTWs
11. Radionuclides
12. Sewage Sludge Incinerators
13. TSDFs
14. Waste Oil Combustors
15. 35-County
16. 5-City
17. lEMP-Baltimore
18. IEMP-Kanawha Valley
19. lEMP-Philadelphia
20. lEMP-Santa Clara
21. Southeast Chicago
22. South Coast
None
(see footnote b)
None
Beryllium, Formaldehyde
EDC, Perchloroethylene, TCE, Vinyl chloride
None
BaP, Methylene chloride.
Perchloroethylene, TCE, Vinyl chloride
Asbestos, Benzene, gasoline vapors, EDB, BaP
None
Methylene Chloride, TCE
BaP, Cadmium, PCBs, TCE
None
TCE, PCBs
Benzene, BaP, Methylene chloride, TCE
Benzene, Methylene chloride, TCE
Benzene, Perchloroethylene, TCE
Benzene, Perchloroethylene, Vinyl chloride,
Methylene chloride, TCE, BaP, Allyl chloride
EDC, TCE, Perchloroethylene
Benzene, Gasoline vapors, Methylene chloride,
Perchloroethylene, TCE, BaP
Acrylamide, 1,3-butadiene, PCB's,
Propylene oxide
Benzene, BaP, Chromium, EDB,
Methylene chloride, nickel, TCE
EDC * ethylene dichloride
PCBs = polychlorinated biphenyls
BaP = benzo(a)pyrene
TCE = trichloroethylene
Except for some methylene chloride source categories in the ATERIS data base, all of the unit
risk factors in the ATERIS data base are the same as those in Table 2-6. For the NESHAP study as
reported in the 1985 Six-Month Study, 21 unit risk factors have changed. The more important one
in terms of either annual cancer cases or percent change are: acrylamide, 1,3-butadiene,
ethylene dibromide, nickel subsulfide, trichloroethylene, and vinyl chloride. For a complete
listing, see Table 3-27.
2-26
-------
TABLE 2-9
EFFECT OF CHANGES IN UNIT RISK FACTORS USED IN THIS
REPORT ON ORIGINAL ESTIMATES OF ANNUAL CANCER CASES
ESTIMATED ANNUAL CANCER CASES
Using Risk Factors As Using Table 2-6
STUDY Reported in Study Unit Risk Factors
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
Ambient Air Quality
NESHAP/ATERIS
Asbestos
Coal and Oil Combustion
Drinking Water Aerators
Gasoline Marketing
Hazardous Waste
Combustors
Mobile Sources
Municipal Waste Combustors
POTWs
Radionuclides
Sewage Sludge
Incinerators
TSDFs
Waste Oil Combustors
35-County
5-City
lEMP-Baltimore
lEMP-Kanawha Valley
lEMP-Philadelphia
lEMP-Santa Clara
Southeast Chicago
South Coast
2,022
504a
82
11.1
0.021
24-75
0.3-9
628-1,874
1.7-2.3
1.5
16
13
140
0.10-0.56
469-553
92.6
2.8-7.0
1.8
0.37
2.2
1.21
162-221
2,022
496b
82
12.1
0.021°
24-75
0.3-9C
601-1,852
1.7-2.3
1,3
16
13
140
0.10-0.56
463-546
90.4
2.95-7.15
1.77
0.42
1.85
1.26
19-33
NOTE: The reports on Municipal Solid Waste landfills, Superfund sites,
and Woodstoves did not include estimates of annual cancer cases.
a Based on original NESHAP study as reported in the Six-Month Study.
b Incorporates revised NESHAP study estimates and ATERIS data base risk
estimates.
c The net effect of adjusting unit risk factors cannot be determined
as cancer risk attributable to individual organic compounds was not
available. The effect is expected to be small.
2-27
-------
TABLE 2-10
EFFECT OF UNIT RISK FACTORS ON ESTIMATED ANNUAL CANCER CASES:
THE SOUTH COAST STUDY
POLLUTANT
Benzene
Carbon tetrachloride
Chi orof orm
Ethyl ene di bromide
Ethyl ene di chloride
Methyl ene chloride
Perch! oroethyl ene
Tri chl oroethyl ene
Arsenic
Beryl 1 i urn
Cadmi urn
Chromium
Nickel
Total Annual Cancer Cases
South
Unit
Ambient
Measured0
99
1.4
1.3
0.37
-
8.0
0.59
0.33
1.5
0.09
0.49
108
0.37
221
ESTIMATES OF
ANNUAL CANCER CASES USING.
Coast Study
Risk Factors3
Model
Predicted0
55
0.001
0
0.007
0.007
3.4
0.43
-
0.0001
0.0003
0.96
102
0.09
162
EPA Unit
Ambient
Measured6
16
1.4
1.3
1.1
-
0.92
0.59
0.43
1.5
0.09
0.49
8.6
0.56
33
Risk Factors'3
Model
Predicted6
8.6
0.001
0
0.02
0.007
0.39
0.43
-
0.0001
0.0003
0.96
8.2
0.14
19
a The unit risk factors used in the South Coast study, which are California
Department of Health Services' unit risk factors, are found on page V-10 of
the South Coast study.
b The EPA unit risk factors used to adjust the estimates of annual cancer cases
are found in Table 2-6 of this report.
c Based on dividing estimated lifetime (70-year) cancer cases in Table VI-3, p.
VI-11, of the South Coast study by 70.
d For each pollutant, the annual cancer cases in this column were calculated as
follows: the estimate of annual cancer cases using the South Coast study's
unit risk factors was multiplied by the ratio of the EPA unit risk factor to
the California Department of Health Services unit risk factor for that
pollutant.
2-28
-------
reduced total estimated annual cancer cases by approximately 85 percent.
The issue of which unit risk factor, DOHS or EPA, more likely represents
actual risk is beyond the scope of this study. For purposes of this
study, all cancer risks are evaluated and reported"(unless otherwise
noted) on the basis of the unit risk factors presented in Table 2-6 and
Table 2-7; the unit risk factors from the South Coast study were not
used to estimate nationwide cancer risk.
Emission Factors. A second basic reason for different cancer risk
u
estimates is that different pollutant emission factors have been used."
Where emission factors could be compared, the most recent emission
factor was selected in the calculation of cancer risk. (This selection
assumes that the more recently developed emission factor is a better
(more accurate) factor than the previous emission factor.) In these
instances, appropriate adjustments were made to the cancer risks based
on "older" emission factors, and the "new" set of estimated annual
cancer incidences per million population were compared. Unfortunately,
except for motor vehicles, pollutant emission factors for most source
categories were either not readily available in that they were not
included in the final report or were reported in only one of the
studies, and a comparison could not be made. Thus, it was generally
very difficult to say anything about the effect, if any, pollutant
emission factors had on discrepant estimated annual cancer incidences
per million population.
In several instances, the studies referred to more recently
developed emission factors that were used (i.e., the 5-City study) or
not used because it was beyond the scope of the study (i.e., the 35-
County study). Such qualitative statements were used to some extent in
2-29
-------
selecting some cancer risk estimates as better than others. In summary,
except for motor vehicles, trying to identify differences in emission
factors as a source of discrepancy was not very successful.
Modeled vs. Ambient-Measured Concentrations. Cancer risk estimates can
also vary depending on whether they are derived from modeled concentra-
tions or from ambient-measured concentrations. Both methods of
obtaining ambient concentrations from which cancer risks can be
estimated have their own inherent set of limitations (see Table 2-11).
It was beyond the scope of this project to analyze the various limita-
tions of the two techniques for estimating ambient concentrations. For
example, this study did not try to determine whether the most appropriate
models were used in the studies or to try to "correct" the cancer •
estimates to a single model. Similarly, it was beyond the scope of this
project to try to determine whether the proper sampling technique was
used to obtain the ambient samples or whether the sampling point
locations were likely to obtain representative samples.
The study did, however, attempt to use several guidelines or
"thought processes" in evaluating and comparing cancer risks obtained
from modeled concentrations and from ambient-measured concentrations.
These were:
• Unless otherwise noted in a study, all models were assumed to
be appropriate and their results were given equal weight.
• Where modeled and ambient-measured concentrations were used
and risk estimates made, an attempt was made to identify
potential causes for discrepancies based upon known emission
sources. For modeled estimates, this meant trying to identify
emission sources included in the inventory and emission sources
that were excluded. For ambient-measured concentrations, this
meant trying to determine if the locations from which the data
were obtained contained known point sources that might
influence or bias the data.
2-30
-------
TABLE 2-11 ,
SELECTED LIMITATIONS OF MODELED AND AMBIENT-MEASURED CONCENTRATIONS
FOR ESTIMATING CANCER RISK
Modeled Concentration Limitations
1. Many of the dispersion models assume flat terrain and average meteorological conditions.
Rough terrain in the area surrounding a source, such as a valley, can result in concentrations
that are up to one to two orders of magnitude higher or lower than concentrations predicted in
gently rolling terrain.
2. Dispersion modeling often extends to only 20 kilometers from the source. This technique can
lead to understating risk if extending dispersion increases significantly the number of people
•exposed.
3. Dispersion modeling estimates are rarely based on site-specific meteorology. Often, data from
hundreds of kilometers away must be used.
4. Dispersion models do not consider increases in concentrations that could result from re-
entrainment of toxic particles from streets, rooftops, etc. In addition, models do not
account for background concentrations, secondary formation of pollutants, and emissions from
other sources not explicitly included in the analyses.
5. Emission estimates are generated from data and assumptions that could be in error. For
example, although some of the studies (e.g., the 35-County study) incorporate plant-specific
emission estimates whenever possible, the pollutant releases fo.r other sources are frequently
estimated by applying speciation factors against the volatile organic compound (VOC) and total
suspsended .particulate (TSP) data in the National Emission Data System (NEDS). Unfortunately,
some of the information in NEDS is of questionable consistency and quality for the purposes of
quantitative risk assessment.
Ambient-Measured Concentration Limitations
1. A basic limitation is the extrapolation of measurements from a limited number of sites to a
much larger geographic area in order to estimate population exposure. This affects both
estimating exposure within a city from a limited number of sites to estimate average exposure
within the city and estimating nationwide cancer risk from a limited number of geographic
areas.
2. Ambient-measured data collected over long periods of time (e.g., at least one year) are
frequently unavailable, which limits the ability to make statements as to long-term exposures
upon which cancer risk estimates based.
3. All ambient-measured data are subject to errors in sampling and analytical methods.
4. Ambient data may underestimate "true" maximum individual risk (MIR) concentrations because
sampling is limited to a small number of fixed monitoring sites.
2-31
-------
• In the absence of evidence to the contrary, cancer incidence
estimates based on modeled concentrations and those based on
ambient-measured concentrations cannot be summed to obtain a
total risk estimate (i.e., they are not mutually exclusive).
For only one pollutant, 1,3-butadiene, were these two estimates
summed. This was done because the locations for the ambient
data were not identified as having known point sources of 1,3-
butadiene. Therefore, it was felt that a better estimate would
be obtained by assuming the ambient data reflected background
and area-type emissions of 1,3-butadiene (which would include
motor vehicle emissions) to which the cancer risk from the
modeled point sources could be added.
• The South Coast study noted several pollutants for which large
discrepancies between modeled and ambient-measured concentra-
tions occurred and offered potential reasons for such. Other
studies also noted where they believed one methodology may be
underestimating risk. In each case, the studies identified the
modeled estimate as possibly underestimating risk. The reason
most frequently cited for this underestimation was an incom-
plete inventory of emission sources. Modeling biases can also
lead to the underestimation of risk. These discrepancies and
their reasons are noted in the pollutant-by-pollutant analysis
section found in Appendix B. These reasons were considered in
evaluating which risk estimates were "better" than others.
For formaldehyde and carbon tetrachloride, the ambient-measured
concentrations and.derived cancer estimates were selected and
evaluated. For formaldehyde, this was done because it is well
established that formaldehyde is formed in the atmosphere
(secondary formation). Ambient-measured data can account for
this atmospheric-formed formaldehyde, whereas models do not.
In the case of carbon tetrachloride, it is also well known that
carbon tetrachloride remains in the atmosphere long after it
has been emitted. Thus again, ambient-measured data can
account for this "retention" of carbon tetrachloride more
readily than models.
• For comparing between ambient-measured data, the geographic
coverage of the study was considered. It was assumed that risk
estimates based on ambient-measured data from more geographic
locations were better estimates from which to estimate nation-
wide risk than were estimates from single geographic locations.
This led to selecting the Ambient Air Quality study results as
the best estimates of nationwide risk from those estimates
based on ambient-measured data. In fact, most if not all of
the smaller geographic ambient data fell within the range of
data used in the Ambient Air Quality study.
Source Category Definition and Coverage. One of the basic steps in
reducing the data was to determine the various source categories (e.g.,
motor vehicles, electroplating, municipal landfills) covered by the
2-32
-------
studies, and then to assign the risk estimates for each pollutant to
that source category. This is necessary to avoid double-counting. In
this study, specific source categories were used to aggregate and
compare the risk data. For most of the source categories and studies,
assigning risk estimates to the appropriate source category was
relatively easy, as most of the specific source categories were
developed on the basis of the source categories reported in the studies.
In certain instances, however, it was difficult to determine whether or
not a source category in one study was the same as in another. For
example, the source categories "heating," "combustion," "residential,
heating," "coal and oil combustion," and "oil combustion," all appeared
in one or more studies. In this instance, it was very difficult to
determine whether or not the same types of emission sources were being
covered.
Another aspect to source category definition was whether or not
the studies included all of or just some (and which ones) of the types
of emission sources in a particular source category. For example, some
motor vehicle pollutants are exhaust and evaporative emissions as well
as tire wear emissions. Some studies reported only the risk from
/
exhaust and evaporative emissions, while one study included those from
tire wear. The ability of determining the specific types of sources
covered by each study for each source category met with varying success,
because the information needed to ensure an accurate accounting was not
always reported in the available material. In certain cases, we were
able to obtain information beyond that which was published. Thus,
assumptions as to which source categories are mutually exclusive or not
and whether the same set of emission sources are covered in a particular
source category remain, in certain instances, highly uncertain.
2-33
-------
Finally, plant location information from the NESHAP/ATERIS data
base and various EPA documents was used to determine whether specific
plants were located in counties covered by the 35-County study, in the
five cities covered in the 5-City study, in the four IEMP study cities,
in the South Coast geographic area, and in the Southeast Chicago study
area. The relationship of plant locations to the geographic study area
of these other studies was used to assess the potential relationship of
the risk data (whether they were mutually exclusive and could be added,
or whether they were duplicative). Evidence of a match was assumed to
infer a likelihood of double-counting if the two risk estimates were
added. If no plant location match was found, it was assumed to infer a
likelihood of mutual exclusi'veness.
Geographic Scope of the Study. As the primary purpose of this study is
to evaluate nationwide risk, modeled risk estimates from studies that
already have a nationwide scope were generally preferred as better
estimates of nationwide risk than those nationwide risks that could be
extrapolated from the studies with smaller geographic scopes. This is a
somewhat difficult "preference" assumption to make. The smaller
localized studies frequently are based on much more detailed and site-
specific data than are the nationwide studies. Thus, those studies may
do a somewhat better job at estimating likely levels of risk. At the
same time, because they take into account site-specific data, they are
likely to be less representative of conditions nationwide and thus can
not be simply extrapolated nationwide. As this study is in itself a
broad scoping type of study, the broader scoping nature of the
nationwide studies are more consistent with the goals of this study.
Therefore, based upon these considerations, the results of the
2-34
-------
nationwide studies received preference in developing point estimates of
cancer risk.
As noted earlier in this chapter, risk estimates for a few source
categories were available only from the smaller, localized studies.
Nationwide risks were extrapolated from these studies in some instances
(e.g., petroleum refineries). In other instances, so little was known
about the emission source that the cancer risk from only that study was
used in estimating the total nationwide risk. Such sources are part of
one of the general source categories (e.g., unspecified sources,
miscellaneous).
As noted earlier, for ambient-measured risk estimates, those from
the Ambient Air Quality study were generally assumed preferable to those
extrapolated from the smaller, localized studies because of its broader.
geographic scope.
Miscellaneous Specific Considerations. As the studies and various risk
estimation methodologies were reviewed, several additional factors were
considered in evaluating the data.
. The 35-County study noted that the counties studied were
selected, in part, because of the presence of known emission
sources of the pollutants being considered. Thus, the
estimates of annual cancer incidence per million population
calculated for the 35-County study may be higher than the
nationwide population-weighted average. Applying, the 35-
County study's rates directly to the total U.S. population
could result in an overestimation of cancer risk.
• Several methodologies exist for estimating risk from PIC. Each
methodology has its own inherent limitations, and no methodo-
logy has been shown to be better than another. The current
trend in estimating risk has been toward using individual
source category emission factors and developing unit risk
factors that are based upon the mixture of components emitted
.from the source category. For purposes of this study, the
modeled estimates of risk from PIC were selected from those
estimates using this type of risk estimation methodology;
2-35
-------
Limitations and Uncertainties
Limitations
Consistent with the purposes of this study, the analyses in this
report consider cancer risk only from air toxics. Noncancer effects of
air toxics are not included. As noted in Chapter 1, other studies are
being undertaken to examine other health effects, such as subtle
biochemical, physiological, or pathological effects to gross effects,
including death.
The only pathway considered in this report is inhalation.
Potential health risks from ingestion of air pollutants that ultimately
reach humans through the diet or that are directly ingested are not
examined. Neither are the potential environmental effects of direct
deposition and urban runoff of air pollutants to surface water
addressed.
Estimates of cancer risk are based on concentrations of air toxics
found in the ambient air. It was not the purpose of this study to
estimate cancer risk based on exposure to indoor concentrations of air
toxics. As noted in Chapter 1, a separate program has been initiated to
quantify the risk from indoor exposure to air toxics.
Although quantitative risk estimates are reported in this study,
it is important to remember that the reports and studies used do not
cover either all known or potential air toxics or all sources of air
toxics which contribute to outdoor exposure. As noted earlier in this
chapter, the cancer risk estimates in the reports and studies reviewed
cover 90 compounds in approximately 65 source categories. These
compounds represent only a fraction of the total number of compounds
present in the ambient air. Based on a review of studies directed at
2-36
-------
identifying compounds in the ambient air, more than 2,800 compounds have
been identified as existing in the atmosphere,9 some of which may be
toxic at ambient levels. One major factor preventing analysis of more
pollutants is the lack of measurement techniques to obtain ambient
measurements for a number of pollutants. A second major factor is the
lack of data on cancer risk associated with ambient concentrations of
other pollutants. Only about 10 percent (approximately 300) of the
2,800 plus atmospheric pollutants have been tested for mutagenicity or
carcinogenicity. Of these, 97 have tested positive in whole animal
bioassays. The mutagenicity or carcinogenicity of the other 2,500
atmospheric compounds is unknown. The impact on cancer incidence from
these other atmospheric compounds is currently impossible to estimate.
Despite the fact that more than 2,800 chemicals have been
identified in ambient air, a large number of unknown compounds are still
likely to exist. Indeed, atmospheric chemists studying the reactions of
most common urban pollutants are often able to account for only about
one-half of the carbon in their studies. The impact of the unidentified
organic products on cancer incidence is unknown. However, the compounds
for which risk information is available were selected based on evidence
that led to their being suspected carcinogens. Thus, it is possible
that the cancer risk associated with the 90 or so compounds for which
cancer risk data have been obtained represents a much larger proportion
of the total risk than might be suggested by a simple comparison of the
90 compounds to the total number of atmospheric compounds.
9 Graedel, T.E., D.T. Hawkins, L.D. Claxton. Handbook of Atmospheric
Compounds: Sources, Occurrence, and Bioassav. HERL-051a. (1985:
Academic Press, New York).
2-37
-------
The quantitative risks reported in this study are based, in part,
on unit risk factors that are either upper-bound estimates or maximum
likelihood estimates of the carcinogenicity potential of the air toxic
pollutants. In either case, the quantitative estimates based on these
unit risk factors are conservative in that actual cancer cases for these
pollutants may be higher, but are more likely to be lower than the
estimates presented in this study. Thus, the aggregate cancer risk,
which is based on the summation of individual pollutant's cancer risks,
represents a likely overestimate for those pollutants considered.
The amount and quality of information concerning pollutants and
their risk from specific source categories vary considerably. For
example, information on the types of pollutants emitted from motor
vehicles is fairly well established. In addition, emission factors for
most motor vehicle pollutants have been estimated much more closely than
for other source categories because, in part, of the relative ease with
which motor vehicles can be tested. On the other hand, the types of
pollutants from source categories, such as TSDFs and Superfund sites,
are much more likely to vary because the materials that give rise to the
pollutants vary from one site to another. Also, the emission levels of
pollutants from such source categories are much more difficult to
establish because the test methodologies are not as easy to apply as
those for motor vehicles. It should be noted that there is considerable
uncertainty associated with the estimates of risk attributed to
individual pollutants emitted from TSDFs. It is possible that dioxin,
the TSDF pollutant for which the largest risk is estimated, may be
emitted in much smaller quantities, if at all, from TSDFs. Finally,
estimates of risk for some source categories may suffer simply from a
lack of a complete accounting of pollutants.
2-38
-------
Reliable quantitative estimates remain unavailable for many
potentially important source categories (e.g., Superfund sites) and, in
some instances, sources may have been missed in the source assessment.
In addition, quantitative estimates of risk from pollutants formed or
transformed in the atmosphere (secondary formation) remain unquantified
for almost all pollutants. The most important secondary pollutant for
which cancer risks have been quantified to date is formaldehyde.
Formaldehyde is both emitted directly into the atmosphere and formed in
the atmosphere, and atmospheric formation of formaldehyde has the
potential to produce many times the amount directly emitted from most
sources.. The gas-phase transformation products of a variety of common
urban pollutants and air toxics have been shown to be potentially
hazardous. The normal atmospheric reactions of these pollutants produce
a variety of oxygenated and nitrogenated products, such as glyoxal and
peroxyacetylnitrate (PAN), and a variety of unidentified species, which
have been shown to be mutagenic. The total mutagenicity of the
transformation products is often many times greater than the
mutagenicity of the original pollutants. The fact that a gas-phase
product is mutagenic in a bacterial test system suggests, but does not
establish, that a human health risk may arise from exposure to such
products. It is not currently possible to quantify the risk from
exposure to the unidentified, potent gas-phase mutagens produced in
these photochemical reactions. Nevertheless, the evidence to date
clearly suggests that the transformation of ubiquitous, often innocuous,
urban pollutants may add a significant additional risk component to any
assessment of urban exposure and risk.
2-39
-------
Uncertainties
There are many uncertainties associated with the methodologies
that are available for making the risk estimate used in the reports and
studies that formed the data base for this report. The following
uncertainties are among the more important ones to keep in mind. The
list is not all inclusive, and additional uncertainties are identified
throughout the report.
• Cancer incidences presented in this report are based on the
assumption that emission levels and ambient levels for each
pollutant either "averages out" over a 70-year period to equal
the concentrations used in the calculations of annual incidence
or remain constant for that period of time. In reality,
emissions and air quality will vary from year to year. Because
the amount and direction of variation is unknown, it is unclear
how much this assumption affects the results.
• All of the analyses assume exposure to air toxics occurs where
people reside. This assumption does not consider the
possibility that people may move throughout the urban area and
change their homes several times during their lives. In
addition, few plants may operate or be expected to emit at the
same level for 70 years, though the areas in which they are
located may remain industrial. Thus, future exposures may be
either worse or better than the old environment. Because
exposures are simulated over a 70-year period, it is unclear
how much this assumption affects the results.
• All of the risks assume continuous outdoor exposure. This
assumption ignores the fact that people spend the majority of
their time indoors, and thus are exposed to indoor atmospheres,
which can be significantly different from the outdoor
atmosphere. Indoor concentrations of certain pollutants (e.g.,
radon, tobacco smoke, formaldehyde, and other VOCs) are
commonly several times higher than outdoor concentrations.
Estimated cancer risk to such indoor pollutant concentrations
suggest that cancer risks based solely on outdoor exposure may
be understated for such pollutants. On the other hand, the
extent to which certain pollutants (e.g., trace metals)
penetrate indoors is large unknown. If emissions of a
pollutant do not penetrate completely indoors and if there are
no indoor sources of that pollutant, then cancer risks based
solely on outdoor exposure will have been overstated.
• All risks are assumed to be additive. This assumption can lead
to substantial errors in risk estimates if synergistic or
antagonistic interactions occur. Although dose additivity has
been shown to predict the acute toxicities of many mixtures of
2-40
-------
.similar and dissimilar compounds, some marked exceptions have
been noted. Consequently, additivity assumptions may
substantially overestimate risk in some cases and underestimate
it in others. The available data on mixtures are insufficient
for estimating the magnitude of these errors. Based on current
information, additivity assumptions are plausible for component
compounds that induce similar types of effects at the same
sites of action.
Unit risk factors used in this study have been generated, in
most instances, using EPA approaches or models. Most of the
resulting unit risk factors are generally regarded either as
plausible, upper-bound estimates or as maximum likelihood
estimates. The linearized multistage procedure used to derive
these factors leads to a plausible upper limit to the risk that
is consistent with some proposed mechanisms of carcinogenesis.
Such estimates, however, do not necessarily give a realistic
prediction of the risk. The true value of the risk is unknown,
and may be as low as zero.
Cancer unit risk values are subject to much uncertainty and in
many cases are preliminary estimates. The risk estimates in
the reports are based on layers of assumptions concerning the
health effects of chemicals, the degree of human exposure, and
the way these substances interact inside the human body. For
example, the weight of evidence of carcinogenicity for the
compounds identified in this report varies greatly, from very
limited to very substantial. Further, the extent of evaluation
and health review performed varies considerably among
compounds. As additional scientific information is acquired,
these values could change significantly, as they have in the
past, and thus the magnitudes and relative importance of
particular pollutants can change.
In developing its unit risk factors, EPA uses a nonthreshold,
multistage model, which is linear at low doses, to extrapolate
from high-dose response data to the low doses typically caused
by exposure to ambient air. In other words, carcinogenic sub-
stances are assumed to cause some risk at any exposure level.
If the true dose-response relationship at low doses is less
than linear, then the unit risk estimates err on the high side.
Many of the individual pollutants have specific uncertainties
that affect their potential contribution to cancer risk.
Chapter 3 and Appendix B identify these uncertainties.
Chromium, formaldehyde, and PIC are three of the major
contributors, based on this study, to cancer risk. Each have
specific uncertainties that may significantly affect the
estimate of cancer risk attributed to them. These
uncertainties are highlighted below.
In the case of chromium, only the hexavalent form has been
proven to be carcinogenic. The percentage of total chromium
that is hexavalent is known to vary considerably depending on
2-41
-------
the source. For example, hexavalent chromium is less than 1
percent of total chromium emissions from coal and oil burning
combustion, while it is nearly 100 percent of total chromium
emissions from cooling towers and electroplating.
Nevertheless, considerable uncertainty remains as to the
exposure to hexavalent chromium versus total chromium
emissions.
In the case of formaldehyde, a number of unresolved issues
concerning its carcinogenesis have been the subject of a
considerable amount of scientific debate. At the center of
this debate are questions concerning such issues as the
mechanism of action of formaldehyde at the molecular level, the
shape of the dose-response curve, the importance of irritation
and the role of the mucus blanket, and the significance of
endogenous formaldehyde. The EPA has determined that the
95-percent upper confidence limit on risk for formaldehyde,
based on data from a 24-month animal study conducted by the
Chemical Industry Institute of Toxicology (CUT), is the
appropriate statistical estimate to use in assessing human
risk. This is consistent with the EPA Guidelines for
Carcinogen Risk Assessment, which state that in the absence of
compelling biological information on the mechanism of action,
the linearized multistage procedure should be used to derive an
upper bound estimate of risk. The EPA does not recommend the
use of maximum likelihood estimates of cancer risk based on
animal data; such estimates are highly unstable (i.e., small
changes in the data may cause orders-of-magnitude fluctuations
in the risk estimates). The EPA is currently evaluating new
scientific data on formaldehyde and will publish an update to
the 1987 assessment at some time in the future.
There has also been disagreement over whether to consider the
incidence of both malignant and benign tumors in rats or
whether only the malignant tumors are significant. The unit
risk factor based on total tumors is approximately 14 times
higher than the unit risk factor based on malignant tumors
only. The current consensus is that only the malignant tumors
should be used to assess the human cancer risk from
formaldehyde. There appears to be little evidence that benign
tumors progress to any of the malignant tumors seen in the CUT
study. The unit risk factor based on malignant tumors only is
used in this report to estimate cancer incidence from exposure
to formaldehyde.
In the case of PIC, there are several sources of uncertainty.
There are a number of methodologies available to estimate risk
from PIC. Some of these methodologies use BaP as a surrogate
for both PIC emissions and unit risk value. Others use PIC-
specific emission factors and unit risk factors or comparative
potency factors. The estimate of cancer incidence is'seen to
vary by a factor of 200 depending on which methodology is used.
While no one methodology has been shown to be a better
methodology for estimating risk from PIC, this study uses the
2-42
-------
methodology that relies on PIC-specific emission factors and
unit risk factors or comparative potency factors.
The unit risk factors for estimating risk from PIC are of
particular concern. As noted previously in this chapter, many
of the unit risk factors used in this study have been approved
by EPA. The most important exception to the use of EPA
approved unit risk factors are the group of compounds known as
PIC. There is no current EPA unit risk factor for this group,
although unit risk numbers are available for some of the
compounds (e.g., BaP) that compose PIC. The 1985 Six-Month
Study used a unit risk factor of 4.2X10"1 Ug/nr)~ for PIC.
This unit risk factor was derived in a highly unusual manner,
and represents an initial attempt at quantifying the potential
risk from PIC. Any estimate based upon this unit risk factor-
is highly tentative.
Other unit risk factors for estimating the cancer risk from PIC
have become available more recently. These unit risk factors
represent estimates of risk from PIC mixtures emitted from
specific source categories (e.g., motor vehicles, hazardous
waste incinerators). Some of the more recent unit risk factors
were estimated using what is known as the comparative potency
approach.10 Even though the more recent factors are also
uncertain and have not received the same level of scrutiny by
EPA as for other unit risk factors, it was felt that they were
an improvement over the PIC unit risk factor used in the 1985
Six-Month Study. Thus, the risk estimates from PIC presented
in this report reflect the use, where possible, of the more
recently developed unit risk factors for specific PIC mixtures.
Another source of uncertainty associated with PIC is the
selection of the appropriate unit risk factor for diesel
particulates, which are included with this group of compounds.
Unit risk factors identified for diesel particulates range from
2xlO~5 to IxlO"4 Ug/m3)"1. The EPA has not yet determined a
single best estimate of the unit risk factor for these
particulates. Thus, the estimate of risk from all sources of
PIC includes the range of risk created by the range in the unit
risk factor for diesel particulates.
Major uncertainties exist for many other chemicals addressed in
this report. For example, there is considerable debate in the
scientific community concerning the mechanism of carcinogenic
action and the estimation of cancer potency for dioxin.
Another unresolved issue concerns the relevance to man of the
kidney pathology observed in rats following exposure to
gasoline vapors. A detailed discussion of the uncertainties
associated with risk estimates for these and other chemicals is
outside of the scope of this report.
10 For a brief discussion on the comparative potency approach, see
page B-110 of Appendix B.
2-43
-------
In summary, the portion of the entire cancer risk represented by
pollutants and source categories not covered in this study is unknown.
It is expected that the pollutants and source categories covered are
among the most likely major contributors to cancer risk based on our
current state of knowledge regarding carcinogenicity of pollutants and
sources that emit those pollutants. As new information is obtained,
other pollutants and sources may be found to be as important, or even
more important, contributors to cancer risk.
As a result of the limitations and uncertainties identified above,
the numerical estimates presented in this report should be viewed only
as a rough indication of the potential for cancer risk caused by a
limited group of pollutants found in the ambient air. Many of the risks
cited in this report are almost certainly inaccurate in an absolute
sense. The best use of the risk estimates is in describing the broad
nature of cancer risk posed by these air toxics and by making relative
comparisons of risks across pollutants and sources.
2-44
-------
CHAPTER 3.0
THE MAGNITUDE AND NATURE OF THE CANCER RISK
In thfs study, the magnitude and nature of the air toxics problem
were evaluated based upon the results of a diverse collection of reports
and studies. These reports and studies cover many pollutants and source
categories, They also cover varying geographic areas, ranging from
city-specific studies to nationwide studies. The methodology used to
estimate the magnitude and nature of the cancer risk nationwide from
this diverse collection of reports and studies was described in the
previous chapter.
In this chapter, the overall magnitude of the cancer risk is
.presented first. The magnitude of the cancer risk is presented in terms
of annual cancer cases and lifetime individual risk. The nature of the
cancer risk problem is then described in terms of individual pollutants,
source categories, and geographic variability. Finally, the results of
this study are compared with those of the 1985 Six-Month Study.
It is important to understand that these estimates reflect the use
of either an upper bound or a maximum likelihood estimate of unit risk;
that is, for the pollutants examined, the actual cancer risk may be
/
higher but is more likely to be lower. As discussed in Chapter 2, this
occurs because of the manner in which EPA calculates the unit risk
factors for toxic pollutants.
3-1
-------
Magnitude of the Cancer Risk Problem
The magnitude of the cancer risk is presented first in terms of
total nationwide cancer cases per year and then in terms of lifetime
individual risk. Both measures of the magnitude of cancer risk play an
important role in the understanding of the problem and in the
development of air toxic control strategies and regulations. Detailed
analyses for those pollutants that were initially identified as
potentially resulting in ten or more cancer cases per year nationwide
are found in Appendix B.
Annual Cancer Cases
The estimates of nationwide annual cancer cases for 26 specific
pollutants are presented in Table 3-1. The remaining pollutants are
grouped together under "Miscellaneous." Annual cancer incidence was
calculated by dividing the estimated lifetime incidence levels by 70
years.1
Both range and point estimates of nationwide annual cancer cases
are presented in Table 3-1. These estimates were derived, in most
instances, from annual cancer incidence estimates based on both modeled
and ambient-measured concentrations. The estimates under the column
"Range" reflect a narrowing of the total range of nationwide annual
cancer incidence that can be calculated from the various studies. As
seen in Table 3-1, the range of estimates is about two-fold in size,
being approximately 1,400 to 2,900 cancer cases per year.
1 The unit risk factors used in this study represent the chance of
contracting cancer from a lifetime (70 years) exposure to a given
concentration of that pollutant. It was assumed that the resulting
lifetime incidence levels could be divided by 70 to represent annual
incidence levels.
3-2
-------
TABLE 3-1
SUMMARY OF ESTIMATED NATIONWIDE ANNUAL CANCER CASES BY POLLUTANT
ESTIMATED ANNUAL CANCER CASES6
POLLUTANT
EPA
CLASSIFICATION6
RANGE
Totals
1,366-2,909
POINT0
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
Acrylonitrile
Arsenic
Asbestos
Benzene
1,3-Butadiene
Cadmium
Carbon tetrachloride
Chloroform
Chromium (hexavalent)
Coke Oven Emissions
Dioxin
Ethyl ene di bromide
Ethylene dichloride
Ethyl ene oxide
Formaldehyde
Gasoline vapors
Hexachl orobutadi ene
Hydrazine
Methyl ene chloride
Perchl oroethyl ene
PIC9
Radionuclides1
Radon1
Tr i chl oroethyl ene
Vinyl chloride
Vinylidene chloride
MiscellaneousJ
Bl
A
A
A
B2
Bl
B2
B2
A
A
B2
B2
B2
B1-B2
Bl
B2
C
B2
B2
B2
A
A
B2
A
C
~
13-14
8-68
82-126
143-181
244-266
6-16
31-47
29-115
113-283
7-11
2-125
25-68
16-45
5-6
124-240
19-76
9
6
3-6
6-13
438-1120
1-3
2
5-13
13-25
0.5-10
15
13
68
88
181
266
10
41
115 „
147-265d
7
2-1256
68
45
6
124 .
19-76f
9
6
5
438-1120h
3
2
7
25
10
15
1,726-2,706
NOTE: Values in this figure are not absolute predictions of cancer occurrence
and are intended to be used in a relative sense only. The dose-
response relationships and exposure assumptions have a conservative
bias, but omissions due to uncharacterized pollutants (either directly
emitted or secondarily formed) and emission sources, the long-range
transport of pollutants, and the lack of knowledge of total risk from
multi-pollutant exposures will offset this bias to an unknown extent.
a These estimates are based on unit risk factors that may overstate the actual
risk. The unit risk factors for arsenic, benzene, cadmium, and hexavalent
chromium are maximum^likelihood estimates. The unit risk factor for
3-3
-------
FOOTNOTES TO TABLE 3-1 (continued)
asbestos is a "best" estimate, very similar to the value that would be
obtained if the maximum likelihood estimate was calculated. Unit risk
factors for PIC have no classification. All other unit risk factors are
upper-bound estimates.
b For a discussion of how EPA evaluates suspect carcinogens and more
information on these classifications, refer to "Guidelines for Carcinogen
Risk Assessment" (51 Federal Register 33992). The EPA classifications used
in this report are:
A « proven human carcinogen
B - probable human carcinogen (Bl indicates limited evidence from human
studies and sufficient evidence from animal studies; 82 indicates
sufficient evidence from animal studies but inadequate evidence from
human studies)
C = possible human carcinogen
0 If a range is shown, it was considered unreasonable to select a point
estimate.
d Range primarily reflects uncertainty with the exposure to hexavalent
chromium from cooling towers. Some uncertainty to actual exposure to
hexavalent chromium from all sources exists because the percent of total
chromium that is hexavalent is still being evaluated for most sources.
e Range reflects great uncertainty associated with exposure to dioxin from
treatment, storage, and disposal facilities, and from municipal waste
combustors. Other uncertainties associated with dioxin estimates include
sampling method and extrapolation from tetrachlorinated dibenzodioxin (TCDD)
to the other dioxin subspecies.
f Range reflects different assumptions as to which portion of gasoline vapors
is carcinogenic. The upper end of the range assumes all gasoline vapor is
carcinogenic; the lower end assumes only the C6 and higher fraction of the
gasoline vapor is carcinogenic.
9 PIC (products of incomplete combustion) is a group of pollutants that have
not been very well defined and for which EPA has not developed a
classification. It is composed of some pollutants, such as BaP, for which
EPA has developed a classification. BaP is a B2 pollutant (probable human
carcinogen).
h Range reflects the use of two unit risk factors for diesel particulates.
1 From sources emitting significant amounts of radionuclides (and radon) to
outdoor air. Does not include exposure to indoor concentrations of radon
due to radon in soil gases entering homes through foundations and cellars.
* Includes individual pollutants primarily from the TSDF study and the
Sewage Sludge Incinerator study.
3-4
-------
The estimates under the column entitled "Point" reflect an attempt
to derive a single estimate of nationwide annual cancer incidence. For
most pollutants, a reasonable point estimate could be selected. Point
estimates are reported in this column when either only one study
reported that pollutant (and as a single point estimate) or the analysis
for that pollutant (see Appendix B for discussion) suggested that a
single point estimate was a better indicator of risk than a range. For
four pollutants, as discussed below, this could not be done. For these
four pollutants, a narrower range was estimated. As seen in Table 3-1,
the "point" estimates narrow the overall range slightly, to between
approximately 1,700 and 2,700 cancer cases per year nationwide. Between
25 and 40 percent of this range is attributable to the cancer risk
estimated for products of incomplete combustion (PIC). As noted in
Chapter 2, there are many uncertainties associated with estimates of
cancer risk from PIC. .
As noted above, a point estimate was not reasonable for four
pollutants. These four pollutants are PIC, dioxin, gasoline vapors, and
hexavalent chromium. In the case of PIC, the large range is created
primarily by the uncertainty of the unit risk factor associated with
diesel particulates, which are included in the estimates of risk for
PIC. A single unit risk factor has not been identified by EPA's Office
of Research and Development for diesel particulates. Instead, a range
of unit risk factors, from 1.0 x 10"9 to 2.0 x 10"5 (^g/m3)"1, has been
identified by EPA's Office of Mobile Sources. This' range was used to
estimate the cancer risk from diesel particulates and is reflected in
the total estimated cancer risk from PIC.
In addition, uncertainty lies in the methodologies available for
estimating risk (as discussed in Chapter 2 and Appendix B) for PIC. One
3-5
-------
methodology uses benzo(a)pyrene (BaP) concentrations and BaP unit risk
factors to estimate risk from PIC. This methodology assumes BaP is the
only carcinogenic component of PIC. More recently developed
methodologies use source-specific PIC emission factors and unit risk
values to estimate risk. None of the methodologies have undergone a
high degree of scrutiny at this time. Thus, although this report uses
the more recently developed methodologies to estimate cancer risk from
PIC, sufficient uncertainty remains concerning all methodologies
associated with PIC risk estimation such that a range might still have
been selected as the reasonable best estimate for PIC, even if a single
unit risk factor could be identified for diesel particulates.
The range for dioxin is the result of difficulties with the
sampling methodologies used to estimate emissions of dioxin and with the
methodology used to extrapolate risk from tetrachlorinated dibenzodioxin
(TCDD) to the other dioxin subspecies. In addition, much of the range
is the result of the uncertainty associated with the risk of dioxin from
treatment, storage, and disposal facilities (TSDFs) for hazardous waste.
Although the TSDF study allows the calculation of a single point
estimate (of 91 cancer cases per year), the underlying emissions data
are very uncertain. Actual cancer cases attributable to dioxin
emissions from TSDFs could be considerably less. Finally, early data on
municipal waste combustors (MWCs) showed a wide range of estimated
annual cancer cases (approximately 2 to 20). Recent revisions to the
MWC study suggest that the estimated risk attributable to dioxin may be
one-half this estimate. For these reasons, no attempt was made to
develop a point estimate for dioxin.
For gasoline vapors, the range in estimated risk reflects the
uncertainty over quantifying the emissions that are associated with the
3-6
-------
cancer-causing portion of gasoline vapors. It has been suggested that
only a portion (i.e., only those C6 and higher components), rather than
all, of total gas vapors are carcinogenic. At this time, it is
uncertain as to which provides a better estimate of the emissions of
concern.
For chromium, the range reflects uncertainty over the ratio of
hexavalent chromium to total chromium emissions for various chromium
emission sources. Several studies (e.g., the 5-City study) attempt to
consider available information on the estimated ratios of hexavalent
chromium to total chromium for specific sources. For cancer risk
estimates based on ambient-measured concentrations of chromium,
estimating cancer risk is complicated by the fact that the sources that
contribute to the ambient measured chromium concentrations are not
identified. Thus, estimating what fraction of total measured chromium
may be hexavalent is even more difficult and uncertain. This degree of
uncertainty makes any single estimate untenable, and therefore a range
has been retained at this time.
Lifetime Individual Risk
In addition to annual incidence, the magnitude of cancer risk from
air toxics can be described in terms of an individual's lifetime risk.
The lifetime individual risk is a measure of the probability that an
individual will develop cancer as a result of exposure to the ambient
concentration of an air pollutant over a lifetime (i.e., a 70-year
period).2 The ambient concentration used to calculate lifetime
2 Lifetime individual risk is calculated as follows:
Lifetime individual risk = (exposure concentration) x (unit risk factor)
3-7
-------
individual risk may be measured or model-predicted. Lifetime individual
risk at a particular location is calculated by multiplying the unit risk
factor by the estimated long-term average exposure at that location.
Where the average ambient concentration is representative of an entire
geographic locale (e.g., a city), the term "areawide" or "urban-wide"
lifetime individual risk can be used.
Frequently, the lifetime individual risk is reported as "maximum
individual risk" (MIR). Maximum individual risk refers to an estimate
of the maximum level of lifetime individual risk to which a person could
be exposed. The MIR is calculated at the specific location near an
emission source where the highest long-term average concentration is
predicted. It is best characterized, especially when developed as part
of preliminary risk assessments, as a rough measure of the potential
maximum individual lifetime cancer risk associated with exposure to the
maximum modeled long-term concentration. The MIR is not an appropriate
measure of the risk level affecting the entire population residing near
a particular facility, but rather only to the individuals residing at
the specific point of estimated maximum exposure.
The highest predicted modeled concentration may or may not always
occur at a point where an individual actually lives. When the highest
predicted modeled concentration is found to occur in an inhabited area,
the term "maximum exposed individual" (MEI) may be used to refer to the
maximum individual risk to which an individual is exposed.
Highly spatially-resolved models are recommended for calculating
MIRs. These models, such as EPA's HEM-SHED, calculate individual risks
close in (<1 kilometer) for all potential receptor locations around
specific point sources. Some model-based studies, however, use a larger
3-8
-------
spatial setting of grid cells, and thus may not identify the highest
modeled concentrations. Similarly, measured concentrations are unlikely
to identify the highest concentrations because of the too few monitoring
sites generally used in most studies. Thus, in monitoring-based studies
and model-based studies, it is often more appropriate to refer to the
"maximum individual risks" reported as either "highest observed
individual risks" or "highest grid-cell individual risks," respectively.
For purposes of this study, the term "maximum individual risk"
(MIR) is used to refer to the highest lifetime individual risk reported
in the various studies and reports. MIRs were estimated for individual
sources (e.g., waste oil combustors, POTWs), individual pollutants
(e.g., arsenic, benzene), and locations (e.g., traffic intersection,
\
geographic locale). MIRs for individual sources reflect the aggregate
risk associated with multiple pollutants emitted from that source. MIRs
for individual pollutants reflect the risk for that pollutant either
from an individual plant within a particular source category (e.g.,
waste oil combustors) or from sources across multiple source categories
(e.g., the Ambient Air Quality study). MIRs for locations reflect the
aggregate risk associated with multiple pollutants and sources.
Table 3-2 summarizes the maximum individual risks reported in the
various studies for individual pollutants and facilities. Almost all of
the studies reported maximum individual risks of at least 1.0 x 10~4.
Many studies showed maximum individual risks of 1.0 x 10"3 or higher.
Where appropriate and where possible, footnotes are used to further
clarify the types of lifetime individual risks and the procedures used
to calculate them.
3-9
-------
TABLE 3-2
SUMMARY OF MAXIMUM INDIVIDUAL RISKS OF CANCER AS
REPORTED IN THE VARIOUS STUDIES
STUDY
Waste Oil
Combustion
Hazardous
Waste
Combustors
TSOF
Sewage Sludge
Incinerators
Municipal Waste
Combustors
POTW
Coal and Oil
Combustion
Drinking Water
Aerators
Asbestos*1
South Coast
Southeast
Chicago
INDIVIDUAL SOURCE/
POLLUTANT/LOCAT I OH
Individual source3
Arsenic
Cadmium
Individual source
Individual source0
Individual source
Individual source6
Individual source
Individual source9
Arsenic
Beryllium
Cadmium
Hexavalent chromium
POM
Formaldehyde
Individual source
Fabricating
Hilling
Renovation
removal
dispose
Demolition
removal
disposal
Benzene
Hexavalent chromium
Grid cell (populated)1
MAXIMUM
LIFETIME
INDIVIDUAL
RISK
1.8X10"4
1.6x10"*
2.1X1Q"5
<1x10"7 to 1x10"4
2x1 O"2
5x10"2
10~5 to 10~3
4.5x10"2
7x10";!
4x10";?
2x10";!
1x10";
8x10"'
5x10"°
1x10"'
2x10"8 to 2x10"5
2x1 0'3.
3x10"S
6x10"£
3x10"s
4x10"!j
7x10"3
10"* to 10"!?
10"4 to 10"-5
9x10"4
REFERENCES
1
2
3
4.
S
6
7
8
9
10
11
IEMP-BaltimoreJ
lEHP-Santa Clara
Benzene
Chloroform
Hexavalent chromium
Five others
Traffic intersection
Benzene
Ethylene oxide
1.0x10
1.1x10
<3.6x10
<6.8x10
3x10
2x10
2x10
-4
-4
-4
-5
-4
-4
-4
12
13
3-10
-------
TABLE 3-2 (concluded)
SUMMARY OF MAXIMUM INDIVIDUAL RISKS OF CANCER AS
REPORTED IN'THE VARIOUS STUDIES
STUDY
lEMP-Kanawha Valley1
lEHP-Philadelphia
Ambient
Air
Quality"1
NESHAP/ATERIS
Data Basen
POLLUTANT/SOURCE
CATEGORY
Institute
Chemical mfg.
PIC
Arsenic
Cadmium
Hexavalent chromium
Chloroform
Benzene -
1,3 butadiene
Carbon tetrachloride
Ethylene dibromide
Ethylene dichloride
Formaldehyde
Hethylene chloride
Styrene
Perch loroethylene •
Tried loroethylene
Vinyl chloride
Vinylidene chloride
Acetaldehyde
Acrylonitrile
Arsenic
Benzene
Beryllium
Butadiene
Cadmium
Carbon tetrachloride
Chloroform
Hexavalent chromium
Coke oven emissions
Ethylene dichloride
Epichlorohydrin
Ethylene oxide
Hexach lorobenzene
Formaldehyde
Methylene chloride
Perch loroethylene
p-dich lorobenzene
Styrene
Tri ch I oroethy I ene
Vinyl chloride
Vinylidene chloride
MAXIMUM
LIFETIME
INDIVIDUAL
RISK
8x10"3
2.2x10"4
8.4x10"-*
3.9x10"*
3.3x10"*
3.7x10",
6.4x10"*
1.7x10"*
1.3x10"*
5.2x10";!
7.9x10"^
1.1x10,
3.0x10"*
9.6x10"*
.8x10"°
.2x10";!
.2x10";!
.Ox10"j!
.7x10"5
5.0x10^*
3.8x10",
1,2x10"*
6.0x10";:
1.9x10";
3.2x10-,!
1.2x10-*
5.7x10-^
2.0x10-^
1.8x10-!
3.4x10-^
1.1x1
-------
FOOTNOTES TO TABLE 3-2
a The MIRs for the individual source and the pollutants are assumed representative of the entire
population of waste oil combustors.
k Range covers individual HIRs for each modeled facility in the source category.
c MEI to "highest annual average ambient concentration around a TSDF."
^ Based on 10th percentile of all sewage sludge incinerator test data for a non-specified facility.
e For existing facilities. Range associated with MIRs at different types of municipal waste
combustors.
f MIR is associated with one specific POTW. Other POTWs have lower MIRs.
9 HIRs were calculated for three types of boilers (industrial, commercial, utility) and two types
of firing (oil-fired and coal-fired) for each type of boiler. This MIR is_associated with an
oil-fired, commercial boiler. The range of MIRs estimated was from 2 x 10" (oil-fired, utility
boiler) to 7 x 10~5 (oil-fired, commercial boiler). The MIRs for the individual pollutants are
associated with oil-fired commercial boilers except for POM (coal-fired commercial boiler) and
radionuclides (coal-fired industrial and utility boilers). For additional information, see
Appendix C, page C-10.
^ MIR not absolute maximum, but reasonable estimate of highest risk expected.
* An MIR of 5 x 10"^ was estimated for a grid cell, but census data indicated that no one was
living in that grid cell.
J" Based on highest average value reported for the pollutant at any of the monitoring sites.
k MIR for the traffic intersection is associated with risks from four pollutants. The MIR for
benzene is based on maximum concentration at a traffic intersection. The MIR for ethylene oxide
is based on maximum concentration at a hospital.
1 Site of MIR is near a specific facility in Institute, WV, and is based on exposure to six
pollutants.
m Based on highest arithmetic mean concentration observed.
n The lifetime individual risks from the ATERIS database are highly uncertain. The ATERIS contains
data from all stages of air toxics analyses, from the very preliminary to the more detailed.
3-12
-------
REFERENCES TO TABLE 3-2
1. Peters, W., Duggan, G., and R. Fegley. Waste Oil Combustion Cancer Risk Assessment. Technical
Staff Paper. October 1987. page 3.
2. U.S. EPA, Office of Solid Waste. Regulatory Impact Analysis for Hazardous Waste Boilers and
Industrial Furnaces. Draft. Exhibits 7-6, 7-9, 7-12, and 7-14.
3. U.S. EPA, OAQPS. Hazardous Waste TSDF - Background Information for Proposed RCRA Air Emission
Standards. Volume 1 - Chapters. Preliminary Draft EIS. March 1988. p. 6-10.
4. U.S. Environmental Protection Agency. Standards for the Disposal of Sewage Sludge. Proposed
Rule. February 6, 1989. 54 FR 5783.
5. U.S. EPA, Office of Solid Waste and Emergency Response. Municipal Waste Combustion Study:
Report to Congress. EPA-530-SW-87-021a. June 1987. p. 86.
6. Memorandum. R.B. Lucas, U.S. EPA, Chemicals and Petroleum Branch, to J. Padgett, U.S. EPA,
OAQPS. New Study on the Air Toxics Problem in the United States - POTW Emissions. July 29,
1988. 3 pages.
7. Peters, W.D., U.S. EPA, Pollutant Assessment Branch. Coal and Oil Combustion. July 25, 1988.
6 pages.
8. Memorandum. W.D. Peters, U.S. EPA, PAB, and S.W. Clark, U.S. EPA, STB to R.G. Kellam, Program
Analysis and Technology Section, and A.H. Perler, Science and Technology Branch. Risks
Associated With Air Emissions from Aeration of Drinking Water. November 10, 1985.
9. U.S. EPA, ESED. National Emission Standards for Asbestos - Background Information for Proposed
Standards. Draft. March 5, 1987.
10. South Coast study, p. vi-6.
11. Southeast Chicago study, p. 43.
12. lEMP-Baltimore study. Tables V-8 and V-14.
13. lEMP-Santa Clara study, pp. 3-82 and 3-112.
14. lEMP-Kanawha Valley study, p. 4-94.
15. lEMP-Philadelphia study, p. VI-22.
16. Ambient Air Quality study. Tables 9 and 10.
17. ATERIS Database printouts. 1989.
18. Table E-1. Major Lifetime Risk and Cancer Incidence for the Four Major AN Source Categories.
(Personal communication from I la Cote, USEPA, to Ken Meardon, PES).
19. U.S. EPA. National Emission Standards for Hazardous Air Pollutants; Benzene Emissions from
Haleic Anhydride Plants. Ethvlbenzene/Styrene Plants. Benzene Storage Vessels. Benzene
Equipment Leaks, and Coke By-Product Recovery Plants. Proposed rule and notice of public
hearing. July 28, 1938. 53 FR 28496.
20. U.S. EPA. Coke Oven Emissions from Wet-Coal Charged By-Product Coke Oven Batteries --
Background Information for Proposed Standards. EPA-450/3-88-028a. April 1987. p. E-30.
21. Memorandum. L.J. Zaragoza, Pollutant Assessment Branch. Hexachlorobenzene Exposure and Risk
Assessment. December 1.1, 1984. Docket No. A-84-39, Item II-B-1.
3-13
-------
The NESHAP/ATERIS data base showed maximum individual risks of 1.0
x 10"3 or higher for 12 pollutants in at least one location.3 These
risk estimates are associated with individual sources within specified
source categories. The EPA re-estimated the maximum individual risk for
182 of the 205 facilities that were identified as having maximum
individual risks of 1 x 10"3 or higher in the ATERIS data base. This
was done by collecting current information on emissions from these
facilities under Section 114 of the Clean Air Act. Overall, the new
analysis showed that estimates of maximum individual risk increased for
12 facilities and decreased for 170. One reason for the generally lower
risk estimates is that some of the companies have taken steps since the
previous emissions data were collected to reduce emissions through
process changes and control devices. The new risk estimates generally
suggest that the maximum individual risk estimates in the ATERIS data
base are too high. Nevertheless, the estimates for some of the 182
facilities analyzed continue to be of serious concern.
On a source category basis, some of the source-specific studies
identified a single maximum individual risk value and others reported a
range of MIR values. In some instances, only the highest maximum
individual risk associated with a specific facility was reported (e.g.,
POTWs and TSDFs). Other facilities in such source categories would have
lower MIRs than those shown in Table 3-2. In other instances, a single
MIR was reported that could be expected at a typical, but unspecified
facility in a source category (e.g., waste oil combustors). Where a
3 The ATERIS contains data from all stages of air toxics analyses,
from the very preliminary to the more detailed. It is not considered an
authoritative source for verified estimates of risk attributable to
individual point sources, and should not be relied upon as credible
estimates of individual source cancer risks. Therefore, the estimates of
MIR in the ATERIS data base are subject to significant uncertainty.
3-14
-------
range of MIRs are shown, the values cover either all of the MIRs for
each of the facilities modeled (e.g., hazardous waste combustors,
drinking water aerators) or the highest MIR expected at typical
facilities within the source category (e.g., municipal waste
combustors).
Maximum individual risks associated with individual sources within
a source category can vary. Table 3-3 illustrates this variation using
the distribution of MIR values associated with hazardous waste
combustors. Under current conditions (referred to as "baseline" in the
table), the majority of these hazardous waste combustors have MIR values
of less than 10"7; however, several may have MIR values of 10"4,
depending on the type of waste being burned. After compliance with
proposed regulations, the highest MIR value decreases to 10"5, and there
is a reduction in the number of hazardous waste combustors associated
with each level of MIR risk. A second illustration is provided in Table
3-4, which shows the distribution of MIRs associated with coke ovens.
Another way to examine MIR is to look at the number of people
exposed to various MIRs. This is illustrated in Table 3-5, which shows
the distribution of people exposed to MIRs for drinking water aerators.
Areawide lifetime individual risks are shown in Table 3-6. These
risks are in the 10"5 to 10~4 range. Compared to the maximum individual
risks for the corresponding cities shown in Table 3-2, the areawide
risks are approximately one order of magnitude lower.
Lifetime individual risks were calculated in the Ambient Air
Quality study at monitoring sites in four cities (see Table 3-7).
Lifetime individual risks in the Ambient Air Quality study were defined
as "the sum of lifetime individual risks for metals, BaP, VOC, and PIC
at a monitoring site within a city where a battery of air toxic
3-15
-------
TABLE 3-3
DISTRIBUTION OF MAXIMUM LIFETIME INDIVIDUAL CANCER RISKS TO THE MOST EXPOSED
INDIVIDUAL FROM HAZARDOUS WASTE COMBUSTORS - BOILERS AND FURNACES
MAXIMUM
INDIVIDUAL
RISK
>1 X 10"4
1 X 10~4
1 X 10"5
1 X 10"6
1 X 10"7
<1 X 10~7
Total
>1 X 10~4
1 X 10'4
1 X 10~5
1 X 10"6
1 X 10~7
<1 X 10~7
Total
TYPE OF
WASTE
Base Case
High Risk
CONTROL DEVICE PERFORMANCE
Base Case"
Baseline
0
0
10
61
103
778
952
0
19
100
167
198
468
952
After Compliance
0
0
6
48
56
650
795C
0
0
73
52
35
595
755C
Pessimistic"
Baseline
0
0
10
65
101
777
952
0
21
102
167
207
456
953
After Compliance
0
0
6
48
72
634
759C
0
0
73
58
36
585
752C
NOTE: Numbers in table indicate the number of hazardous waste combustors associated with each maximum individual
risk level.
SOURCE: U.S. EPA, Office of Solid Waste. Regulatory Impact Analysis for Hazardous Waste Boilers and Industrial
Furnaces. Exhibits 7-6, 7-9, 7-12, and 7-14.
8 "Base case" assumes "typical" removal efficiencies for control devices.
"Pessimistic" assumes removal efficiencies of control devices for toxic metals and hydrogen chloride are several
percentages points lower than in the base case in most cases. For organic compounds the difference is several
fractions of a percent in most instances.
c Difference in total device due to some devices that discontinue burning due to the regulations.
3-16
-------
TABLE 3-4
MAXIMUM LIFETIME INDIVIDUAL CANCER RISKS FROM COKE OVEN EMISSIONS
MAXIMUM LIFETIME
INDIVIDUAL RISK
> 10"2
10'3 to 10'2
10'4 to 1(T3
NUMBER OF
COKE OVENS
13
25
5
SOURCE: Appendix E. Coke Oven Emissions Risk Assessment for Wet-Coal
Charged Coke Oven Batteries.
3-17
-------
TABLE 3-5
DISTRIBUTION OF MAXIMUM INDIVIDUAL CANCER RISK
AT 22 DRINKING WATER AERATORS
MAXIMUM
INDIVIDUAL RISK
1.9xlO'5
1.3X10'5
9.5X10'6
4.6xlO'6
2.9X10'6
l.SxlO'6
1.4X10"6
l.lxlO'6
l.OxlO"6
1(T7
lO'8
NUMBER OF
FACILITIES
1
1
1
1
1
1
1
1
1
11
2
NUMBER OF PEOPLE
EXPOSED
439
7
4
28
33
30
2
1
11
208
13
SOURCE: Memorandum. W.D. Peters, US EPA, Pollutant Assessment Branch,
and S.W. Clark, US EPA, Science and Technology Branch, to R.G.
Kellam, US EPA, Pollutant Assessment Branch, and A.M. Perler,
US EPA, Science and Technology Branch. Risks Associated with
Air Emissions from Aeration of Drinking Water. November 13,
1985. Table 5.
3-18
-------
TABLE 3-6
AREAWIDE LIFETIME INDIVIDUAL CANCER RISKS FOR SELECTED CITIES
CITY/LOCALE
Philadelphia6
Santa Clarac
Southeast Chicago
City Ae
City Be
City Ce
City De
City Ee
Kanawha Valley*
Baltimore9
South Coast*1
AREAWIDE
LIFETIME
INDIVIDUAL RISK
4.0x10~5, 1.2x10"4
4x10"5
2.2x10"4
1.4x10"4
4.3x10~4
2.0x10"4
7.0x10"4
2.7x10"4
5.0x10~4, 1.2x10~3
1.3x10"4, 3.3x10"4
1.2x10~4, 2.1x10"4
NUMBER OF
POLLUTANTS
7
--
30
20
20
20
20
20
18
9
12
MAJOR POLLUTANT
CONTRIBUTORS TO
INDIVIDUAL RISK8
Benzene, carbon tet.
1,2 dichloropropane
Carbon tetrachloride
Coke oven emissions,
Cr+6
Formaldehyde,
PIC, 1,3-butadiene
1,3-butadiene, PIC,
Cr+6, formaldehyde
1,3-butadiene,
formaldehyde
PIC, 1-3-butadiene,
Cr+6
PIC, formaldehyde,
Cr+6
Ethylene oxide,
1,3-butadiene
Benzene, Cr+6
Benzene, Cr+6
Note: In some instances, the areawide lifetime individual risk was calculated by dividing total
lifetime cancer cases by exposed population. Where possible and as appropriate, these
estimates were adjusted based on unit risk factors used in this study.
a Cr+6 = hexavalent chromium
Carbon tet. = carbon tetrachloride
k IEMP Philadelphia study, p. V-27, Lower estimated based on modeled data; higher estimate, on
monitored data.
c IEMP Santa Clara study, p. 3-80.
Southeast Chicago study, p. 38.
e Five City study, p. 53.
* IEMP Kanawha Valley study, pp. 4-116 and 4-117. Higher estimate based on box model
concentrations; the lower, on Gaussian model analysis. These estimates are for the
entire Kanawha Valley study area.
9 IEMP Baltimore study. Tables V-7 and V-13. Range created by range of estimated risk
for hexavalent chromium and cadmium.
h South Coast study, p. VI-11. Lower estimate based on modeled data; higher estimate on
ambient measured data.
3-19
_
-------
TABLE 3-7
SUMMARY OF LIFETIME INDIVIDUAL CANCER RISKS
FOR SELECTED CITIES
CITY
Los Angeles
Baton Rouge
Boston
Chicago
LIFETIME
INDIVIDUAL
RISK3
6.6 x 10'4
3.8 x 10'4
3.0 x 10"4
3.2 x 10'3
NUMBER OF
POLLUTANTS"
17
16
11
14
MAJOR POLLUTANTS
CONTRIBUTING TO
LIFETIME INDIVIDUAL
RISK
Formaldehyde, PIC
Ethylene Dichloride,
PIC
PIC, Chromium
(hexavalent)
PIC, Formaldehyde
SOURCE: Ambient Air Quality Study, Table 8.
a These risks are the sum of the lifetime individual risks for a number
of pollutants using the estimated annual average concentration at a
monitoring site within each of the four cities.
b Includes nickel, but no cancer incidence was attributed to nickel.
3-20
-------
pollutants is being monitored." As seen in Table 3-7, lifetime
individual risks on the order of 10"4 and higher were found. The
magnitude of lifetime individual risks is affected by the number of
pollutants as well as the particular pollutants included. The number of
pollutants monitored ranged from 11 in Boston up to 17 in Los Angeles.
None of the cities had data on 1,3-butadiene, a pollutant found in this
study to be one of the major contributors to risk. In addition,
formaldehyde data, another major contributor to risk, were unavailable
for Baton Rouge and Boston.
Nature of the Cancer Risk Problem
The nature of the cancer risk problem is examined by looking at
the relative contributions of individual pollutants and source
categories to total estimated nationwide annual cancer incidence. In
addition, the geographic variability of the cancer risk is examined by
comparing reported ambient concentrations of selected pollutants,
estimated annual cancer incidences, and estimated lifetime individual
risks.
Individual Pollutants
Table 3-8 presents the percent contribution of individual
pollutants to the total estimated cancer cases. The percent
contributions were calculated using the point estimates presented in
Table 3-1. Where a range is indicated in Table 3-1, the midpoint was
used to estimate the pollutant's potential relative contribution.
Figure 3-1 illustrates the results presented in Table 3-8.
Based on the estimates in Table 3-1, five pollutants -- PIC, 1,3-
butadiene, chromium, benzene, and formaldehyde -- account for
approximately 70 percent of the total estimated annual cancer cases.
The reader is reminded that there is considerable uncertainty associated
3-21
-------
TABLE 3-8
RELATIVE CONTRIBUTION OF INDIVIDUAL POLLUTANTS TO
TOTAL ESTIMATED CANCER CASES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
POLLUTANT
PIC
1,3-Butadiene
Chromium (hexavalent)
Benzene
Formaldehyde
Chloroform
Asbestos
Arsenic
Ethyl ene di bromide
Dioxin
Gasoline vapors
Ethyl ene dichloride
Carbon tetrachloride
Vinyl chloride
Acrylonitrile
Cadmium
Vinyl idene chloride
Hexachl orobutadi ene
Trichloroethylene
Coke Oven Emissions
Perch! oroethyl ene
Hydrazine
Ethyl ene oxide
Methyl ene chloride
Radionuclidesa
Radon3
Miscellaneous
Totals
PERCENT
CONTRIBUTION
35.2
12.0
9.3
8.2
5.6
5.2
4.0
3.1
3.1
2.9
2.1
2.0
1.9
1.1
0.6
0.5
0.5
0.4
0.3
0.3
0.3
0.3
0.3
0.2
0.1
0.1
0.7
100.0
NOTE 1: Values in this figure are not absolute predictions of cancer occurrence and
are intended to be used in a relative sense only. The dose-response
relationships and exposure assumptions have a conservative bias, but
omissions due to uncharacterized pollutants
-------
o
co
o
CM
o
T—
o
0
CL
3*.
"p
NOTE: Values I" this figure are not absolute predictions of cancer occurrence
and are Intended to be used in a relative sense only. The dose-response
relationships and exposure assumptions have a conservative bias, but
omissions due to uncharacterized pollutants (either directly emitted or
secondarily formed) and emission sources, the long-range transport
of pollutants, and the lack of knowledge of total risk from multi-
pollutant exposures will offset this bias to an unknown extent
| 1 ,3-Butadiene
mium, hexavalent
.
u
; §
i
= pC".?V
< • {•'.'•
'/ ' ' • J • • '
Hi
Others
Acrylonitrile
Cadmium
\\\\\§ . Benzene
Formaldejiyde
i
'.'.' i .'.'.
i'.'.' s.'.'.
s •' !"•
* " JiL
£
S
~>, '.'.'
b
-
Vinylidene chloride
Hexachlorobutadiene
Trichloroethylene
•i
CD
1
= '
I
R"5!
yi
Coke oven emissions
Perchloroethylene
Hydrazine
CD
C
IT
? ••
« . .
c
1
111
Ethylene oxide
Methlyene chloride
Radon
Radionuclides
56 other pollutants
Gasoline vapors
Ethylene dichloride
Carbon tetrachloride
Vinyl chloride
S . . i
: . . . t
ititiip
CO
03
O
k.' '.
0 10 20 30 4(
PERCENT CONTRIBUTION (%)
CD
•a
T3
CD
•S •£
.i §
•4—' "tli
CO _5
m o
"CO D-
"o >•
i O
Q S3
C CD
g >-
"3 CD
.g Q_
4±; CO
^ CD
8 %
"^
.> CD
I i
CD CO
DC O
CO
CD
D)
LJL
iNvinnod nvnaiAiQNi
3-23
-------
with the absolute risk estimates for many of the pollutants examined in
this study. The relative contribution of any one pollutant is subject
to change due to the uncertainties associated with the risk estimates.
In addition, the relative contribution may change as new information on
as yet unquantified air toxics is obtained.
Source Categories
As noted earlier, over 60 source categories were identified in the
studies and reports used. Table 3-9 summarizes the estimated annual
cancer cases associated with 40 specified-types of source categories,
one aggregate category, and two unspecified categories. The relative
contributions of the individual source categories are illustrated in
Figure 3-2. It is important to remember that not all source categories
that emit air toxics were covered in the studies and reports used for
this study. Thus, the relative contributions presented in this study
can only reflect the source categories that were covered. As new
information is developed, these relative contributions could change,
perhaps significantly.
The estimates presented in Table 3-9 reflect the range estimates
for modeled estimates only. (By their nature, ambient-measured data do
not distinguish between sources.) Although secondary formaldehyde is
not a modeled source category per se, two studies (the 5-City study and
the Southeast Chicago study) attributed,the difference between ambient-
measured concentrations and the modeled concentrations to the secondary
formation of formaldehyde. It is these results that are included in
Table 3-9. Based on the range estimate for the modeled estimates,
between 1,430 and 2,538 cancer cases per year are estimated.
Individual source categories have frequently been grouped in two
ways: (1) mobile vs. stationary and (2) point vs. area sources. Mobile
3-24
-------
TABLE 3-9
SUMMARY OF ESTIMATED CANCER CASES BASED ON MODELED AMBIENT CONCENTRATIONS, BY
SOURCE CATEGORY
ANNUAL
SOURCE CATEGORY CANCER CASES
1.
2.
3.
4.
5.
6.
7.
8f
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
Motor Vehicles
Secondary
Formaldehyde
Electroplating
TSDFs
Woodsmoke
Asbestos, Demolition
Unspecified (point)
Cooling Towers
Gasoline Marketing
Solvent Use/Oegreasing
Unspecified (area)
PVC/EDC/Vinyl Chloride
Iron and Steel
Sewage Sludge Incinerators
Municipal Waste Combustors
Petroleum Refineries
1,3-Butadiene Production
Styrene'butadiene Rubber
Production
Coal and Oil Combustion
POTWs
Smelters
Commercial Sterilization/
Hospitals
Pesticide -Production/Usage
D rye lean ing
Pulp and Paper Manufac-
turing
Chlorinated Drinking Water
Ethylene Dibromide
Production
Polybutadiene Production
Ethylene Oxide Production
Ethylene Dichloride
Production
Waste Oil Burning
Asbestos Manufacturing
Asbestos Renovation
Glass Manufacturing
Hazardous Waste Combustors
Paint Stripping
Pharmaceutical Manufac-
turing
769-1,461
106-154
120
49-140
89
81
27-92
0.01-111
24-75
22-36
21
19
17-18
13 •
2-22
8-14
10
10
8-10
6
3-4
3-4
3.4
3
2.1
1.7
1.5
1,2
1.2
0.8
0.6
0.5
0.4
0.4
0.3
0.22
0.2-0.4
(percent of
total)
(54-58)
_„,.
(7.4-6.1)
(8.4-4.7)
(3.4-5.5)
(6.2-3.5)
(5.6-3.2)
(1.9-3.6)
(0.0-4.4)
(1.7-3.0)
(1.5-1.4)
(1,5-0.8)
(1.3-0.7)
(1.2-0.7)
(0.9-0.5)
(0.1-0.9)
(0.6-0.6)
(0.7-0.4).
(0.7-0.4)
(0.6-0.4)
(0.4-0.2)
(0.2-0.1)
(0.2-0.2)
(0.2-0.1)
(0.2-0.1)
,
(0.1-0.08)
(0.1-0.08)
(0.1-0.06)
(0.08-0.05)
(0.08-0.05)
(0.08-0.05)
(0.04-0.02)
(0.04-0.02)
(0.03-0.02)
(0.03-0.02)
(0.02-0.01)
(0.02-0.01)
(0.01-0.02)
PRINCIPAL POLLUTANTS
PIC, 1,3-butadiene
Formaldehyde
Hexavalent Chromium
Dioxin
PIC
Asbestos
Arsenic, formaldehyde
Hexavalent Chromium
Gasoline Vapors, Benzene
Perch loroethylene, Methylene
Chloride
Carbon tetrachloride
Vinyl chloride
Coke Oven Emissions, Benzene,
PIC
Cadmium, Vinyl Chloride
Dioxin
Gasoline Vapors, Formaldehyde
1,3-butadiene
1,3-butadiene
Arsenic
Vinyl chloride
Formaldehyde
Ethylene Oxide
Benzene
Perch I oroethy I ene
Chloroform
Chloroform
Ethylene Dibromide
1,3-butadiene .
Ethylene Oxide
Ethylene Dichloride,
Arsenic
Asbestos
Asbestos
Arsenic
Hexavalent Chromium
Methylene chloride
_
Chloroform
3-25
-------
TABLE 3-9
SUMMARY OF ESTIMATED CANCER CASES BASED ON MODELED AMBIENT CONCENTRATIONS, BY
SOURCE CATEGORY (concluded)
ANNUAL (percent of
SOURCE CATEGORY CANCER CASES total)
38. Benzene Fugitives
39. Nitrite Elastomer Produc-
tion
40. ABS/SAN Production
41. Asbestos Fabrication
42. Benzene Storage
43. Other
Total
NOTE: Values in this
0.2 (0.01-0.01)
0.16 (<0.01)
0.13 (<0.01)
0.13 (<0.01)
0.1 (<0.01)
6-13 (0.4-0.5)
1,430-2,538
figure are not absolute pr
PRINCIPAL POLLUTANTS
Benzene
Acrylonitrile
Acrylom'trile
Asbestos
Benzene
Hexavalent Chromium, radon
'edictions of cancer
occurrence and are intended to be used in a relative sense only. The
dose-response relationships and exposure assumptions have a
conservative bias, but omissions due to uncharacterized pollutants
(either directly emitted or secondarily formed) and emission sources,
the long-range transport of pollutants, and the lack of knowledge of
total risk from multi-pollutant exposures will offset this bias to an
unknown extent.
Estimated incidences is approximately equally divided between point and area
sources.
3-26
-------
CO
o
a «
.9 CD
O >
s? ®
.2 Q_
r^ w
O Q)
O CD
C/D O
11
£= o
c; '.£5
O Cd
O Z
CD -O
> CD
CD
CO
CD
•5
D)
U_
3-27
-------
sources are represented by the "motor vehicle" source category in
Table 3-9. All other source categories, except secondary formaldehyde,
make up stationary sources. It is sometimes difficult to distinguish
between point and area sources. The primary distinguishing feature is
the number of sources in a source category. "Area" sources are
generally considered to have too many individual sources to develop
source-specific data from which to estimate risk for each source. Point
sources, in contrast, are "few" enough in number to be located
individually to allow source-specific data to be developed, from which
risks can be estimated for each individual source. To estimate risks
from sources, EPA has developed model algorithms for area sources and
for point sources.
For purposes of this study, source categories have been designated
as either an "area" source or a "point" source depending on whether the
cancer risks were estimated using an area source model algorithm or a
point source model algorithm. Using this basis, the following
individual source categories are considered area sources:
• motor vehicles
• woodsmoke
• asbestos, demolition and renovation
• gasoline marketing (service stations only)
• coal and oil combustion (residential only)
• solvent use/degreasing
• drycleaning
• pesticide usage
• chlorinated drinking water
• paint stripping
3-28
-------
All other sources (except secondary formaldehyde) are considered to be
point sources in this study.
The estimated cancer incidence-from secondary formation of
formaldehyde for selected source categories is shown in Table 3-10.
This apportionment was based on the relative percent contributions
calculated for specific source categories in the 5-City study. The
cancer incidence from secondary formaldehyde was distributed among the
major classifications (i.e., mobile vs. stationary and area vs. point),
as discussed below. Approximately one-half of the cancer incidence from
the "other" source category was from point sources and one-half from
area sources. The specific source categories in Table 3-9, however, do
not include the estimated cancer incidence from secondary formaldehyde
shown in Table 3-10. Instead, a separate "source category" for
secondary formaldehyde is shown.
Examining mobile versus stationary sources, approximately 58
percent of the estimated total annual incidence is estimated to occur
from motor vehicles (including cancer risk from secondary formaldehyde).
Stationary sources account for approximately 42 percent of the total.
Of the major stationary sources, two of the top six -- electroplating
and cooling towers — are related to hexavalent chromium, which accounts
for the entire estimated risk from these two source categories. The
second largest stationary source category is TSDF, in which dioxin is
estimated in the study on TSDFs to contribute 65 percent of the total
estimated 140 annual cancer cases.4 In general, while many stationary
source categories emit a number of different pollutants, the majority of
risk is attributable to a select few in each source category.
4 Because of the great uncertainty associated with the estimate of
dioxin emissions from TSDFs, this estimate could be substantially lower.
3-29
-------
TABLE 3-10
DISTRIBUTION OF ESTIMATED CANCER CASES FROM SECONDARY
FORMALDEHYDE FORMATION AMONG SOURCE CATEGORIES8
SOURCE
CATEGORY
ANNUAL CANCER
CASES IN U.S.
PERCENT
CONTRIBUTION
Area Sources
Motor Vehicles
Solvent Use
Gasoline Marketing
Area Source Subtotals
45
38
10
93
34.8%
28.9%
8.3%
71.9%
Point Sources
Petroleum refining
Chemical Manufacturing
Point Sources Subtotal
8
5
13
5.9%
3.7%
9.6%
24
18.4%
Total Secondary Formaldehyde
130
100%
8 Distribution of secondary formaldehyde based on data from the 5-City
study.
3-30
-------
Table 3-11 presents the results in Table 3-9 on an area versus
point source basis. Area sources are found to contribute approximately
75 percent off the total number of annual cancer cases (including those
from secondary formaldehyde) with point sources contributing
approximately 25 percent of the total. Of the area sources, the major
source is mobile sources, contributing 78 percent of the total annual
incidence attributed to area sources (including the estimated 45 annual
cancer cases attributed to mobile sources from secondary formaldehyde,
as shown in Table 3-10). For point sources, the largest category is
electroplating, which accounts for almost 25 percent of the total point
source-related annual incidence. Although the estimates in Table 3-11
add up to "100 percent of the risk," the reader is reminded that this
study does not include risk estimates from all known sources. The
relative contributions of the types of sources, therefore, could change
as additional data on other sources are obtained.
Geographic Variation
As has been stated, the primary purpose of this study is the
estimation of nationwide cancer risks. The various studies used to meet
this goal illustrate the variation in exposure to different pollutants
and in the resulting cancer risk that exists between geographic areas on
a county-to-county and city-to-city basis as well as on an intra-city or
intra-region basis. Table 3-12 presents ambient-measured concentration
data for several selected pollutants and cities. As can be seen in this
table, the variation in ambient concentrations depends on the pollutant
considered. For example, for the selected cities, ambient benzene
concentrations differ by less than a factor of 2. Two of the pollutants
vary by factors of approximately 4 to 5. For the other two selected
pollutants, ambient concentrations vary by factors of approximately 12
3-31
-------
TABLE 3-11
CONTRIBUTION OF AREA VS. POINT SOURCES
TO NATIONWIDE ANNUAL CANCER CASES
SOURCE TYPE/
INDIVIDUAL SOURCE
CATEGORY
ANNUAL
CANCER
CASES3
PERCENT CONTRIBUTION TO..
Nationwide Source Type
Total Total
Area Sources
Mobile Vehicles
Woodsmoke
Asbestos, demolition
Gasoline Marketing
Solvent Use/Degreasing
Unspecified/Other
Commercial Sterilization/
Hospital
Pesticide Usage
Drycleaning
Chlorinated Drinking Water
Coal and Oil Combustion
(residential onl>)
Asbestos, renovation
Paint Stripping
Secondary Formaldehyde
1,115
89
81
46
29
21
3.5
3
3
2
2
0.4
0.3
93
56.2
4.5
4.1
2.3
1.5
1.1
0.2
0.2
0.2
0.1
0.1
0.02
0.01
4.7
75.0
6.0
5.4
3.1
1.9
1.4
0.2
0.2
0.2
0.1
0.1
0.03
0.02
6.2
Subtotal Area Sources
Point Sources
1,487
75.0
100
Electroplating
TSDFs
Unspecified
Cooling Towers
Chemical Users and Producers
Iron and Steel
Coal and Oil Combustion
(non- residential)
Sewage Sludge Incinerators
Municipal Waste Combustors
Petroleum Refineries
Miscellaneous
POTWs
Manufacturing
Gasoline Marketing
Secondary Formaldehyde
ubtotal Point Sources
120
94
59
56
43
17
8
13
12
11
11
6
6
3
38
497
6.0
4.8
3.0
2.8
2.2
0.9
0.4
0.7
0.6
0.6
0.5
0.3
0.3
0.2
1.9
25.0
24.1
19.0
11.9
11.2
8.7
3.5
1.5
2.6
2.4
2.2
2.2
1.1
1.3
0.6
7.5
100
TOTAL - All sources
1,984
100
Based on raid-point of estimate from Table 3-9.
3-32
-------
TABLE 3-12
COMPARISON OF MEASURED AMBIENT CONCENTRATIONS OF
SELECTED POLLUTANTS IN SELECTED CITIES
POLLUTANT
Benzene
Chloroform
Ethylene dibromide
Methylene chloride
Perch loro«ithylene
A
8.2
1.0
0.06
6.1
1.5
B
7.9
6.2
0.05
3.7
3.6
c-
8.4
4.6
0.2
2.1
5.9
CITY
D
10.8
4.6
0.04
5.2
4.5
E
9.8
17.4
0.07
7.6
2.4
F
13.10
16.6
-
-
5.8
G
8.8
1.2
-
-
0.5
NOTE: All numbers are in terms of
SOURCE: Ambient Air Quality Study, data worksheets.
Key; A = Bakersfield, CA
B = Newark, NJ
C = Philadelphia, PA
D = Elizabeth, NJ
-E = Camden, NJ
F = Baltimore, HO
G = Baton Rouge, LA
for perch!oroethylene and 17 for chloroform. The degree of variation
presented in Table 3-12 depends upon the pollutants and cities selected.
Nevertheless, the point is still the same regardless of which pollutants
or cities are selected -- ambient concentrations vary between cities.
Ambient concentrations can also vary within a city or within a
specified geographic locale (e.g., the South Coast Air Basin, the
Kanawha Valley). Table 3-13 presents ambient-measured concentrations
for selected pollutants at different locations in Baltimore, the South
Coast Air Basin, and the Kanawha Valley. In general, the variation in
ambient concentrations within each area is approximately the same as the
variation in Table 3-12 for comparable pollutants.
As might be expected, the variations in ambient concentrations for
pollutants can lead to variations in the number of cancer cases between
geographic areas and in the estimates of cancer cases per year per
million population. Results from eight studies are presented in Table
3-14. As seen in this table, annual cancer cases varied from a low of
0.03 per year to a high of 128 per year. This reflects a combination of
3-33
-------
5? g
5 M
I £
a
-
8.,-
+-» -M
.8
•8
-§
«- -i- c
3 4-»
O 0)
S JB
_, O •—
O -t-> E
3-34
-------
TABLE 3-14
VARIATION IN ANNUAL CANCER CASES AND
CANCER RATES DUE TO EXPOSURE TO OUTDOOR
AIR TOXICS BY GEOGRAPHIC LOCALES
GEOGRAPHIC
LOCALE
Baltimore
Kanawha Valley
Philadelphia
Santa Clara
City A
City B
City C
City D
City E
Southeast
Chicago
South Coast
County A
B
C
D
E
F
G
H
I
J
K
L
M
N
0
P
Q
R
ESTIMATED ANNUAL
CANpER CASES
2.95-7.15
1.77
0.42
1.85
3.1
67.7
1.1
14.1
6.7
1.26
19-33
128.3
38.4
24.1
22.8
14.6
12.4
7.8
7.6
6.6
5.5
4.8
3-1
1.6
1.1
0.6
0.2
0.1
0,03
CANCER CASES PER YEAR
PER MILLION POPULATION
1.9-4.7
17.8
0.26
1.4
2.1
6.2
2.8
10.0
3,9
3.2
1.7-3.0
17.2
20.4
12.5
38.3
11.3
14.6
9.2
9.7
8.7
3.5
19.2
2.8
1.9
3.0
4.2
1.2
0.6
1.6
b STUDY
lEMP-Baltimore
lEMP-Kanawha Valley
lEMP-Philadelphia
lEHP-Santa Clara
5 City
5 City
5 City
5 City
5 City
Southeast Chicago
South Coast
35 -County3
Selected counties.
" Derived by dividing estimated annual cancer cases by the population in
the geographic locale.
3-35
_
-------
different exposure levels and the size population exposed to those
levels. Cancer rates varied from a low of 0.26 to a high of 38.3 cancer
cases per year per million population. These are equivalent to areawide
lifetime individual risks of approximately 2 x 10"5 to 3 x 10"3 for the
exposed populations. In general, a lower absolute number of cancer
cases corresponded to a lower cancer rate. A notable exception is
County K from the 35-County study, where a "modest" number of estimated
annual cancer cases (4.8) had one of the highest cancer rates (19.2
cancer cases per year per million population).
Variation in lifetime individual risk between and within cities
can also be examined. Areawide lifetime individual risks for selected
cities were presented earlier in Tables 3-6 and 3-7. The lifetime
individual risks among the cities/locales shown in these two tables,
however, were essentially on the same order of magnitude with one
another (approximately 10"4). The areas shown in Tables 3-6 and 3-7 are
urban areas, with the exception of the Kanawha Valley, which is
classified primarily as rural. The Kanawha Valley, on the other hand,
is a fairly highly industrialized area. Because of the large number of
factors that differed in deriving these risk estimates, it is difficult,
if not impossible, to say why such a relatively narrow range is
observed. The small range may point to a relatively consistent areawide
lifetime individual risk regardless of the urban area or industrialized
area in which one lives.
Table 3-15 presents maximum individual risks associated with
various cities or specific geographic locales. The maximum individual
risks in Table 3-15, however, are not necessarily directly comparable to
each other, because they vary in manner in which they were estimated and
3-36
-------
TABLE 3-15
VARIATION IN MAXIMUM LIFETIME INDIVIDUAL RISK, BY LOCATION
MAXIMUM LIFETIME
LOCATION INDIVIDUAL
RISKS
South Coastb
Southeast Chicago0
Baltimore
Philadelphiad
Santa Clara
Kanawha Valley6
Belle
Charleston
Institute
Nitro
lo-3
1 x 10"3
r x io"4
7 x IO'6 to
3 x IO"4
3 x IO"4
3 x 10'3
6 x 10'3
8 x 10'3
8 x 10'6
COMMENT REFERENCE
MEIa to hexavalent
chromium
MEI, additive,
30 pollutants
MEI to benzene and
to chloroform
MEI, additive,
9 pollutants
MEI to emissions at
a traffic intersection
additive,
3 pollutants
additive,
4 pollutants
additive,
6 pollutants
additive,
9 pollutants
1
Z
3
4
5
6
NOTE: Values have been adjusted to reflect unit risk factors used in this study wherever possible.
a MEI = maximum exposed individual
Maximum exposed individual risk based on model-predicted exposures in the South Coast study
rather than the monitored exposures in the.study.
c 8 x 10' of additive lifetime risk was attributed to five pollutants from steel mills. Southeast
Chicago study, p. 43.
MEI risks were calculated for eight locations in the city. The range reflects the high and low
MEI risks estimated.
e The individual risks calculated in the IEMP-Kanawha Valley Study were for neighborhood sites with
suspected highest exposures from point source pollutants.
3-37
-------
REFERENCES TO TABLE 3-15
1. South Coast study, p. VI-6.
2. Southeast Chicago study, p. x.
3. IEMP - Baltimore study, Table V-8. Based on exposure at the site of
maximum average concentration.
4. IEMP - Philadelphia study, p. VI-49.
5. IEMP - Santa Clara study, p. 3-82,
6. IEMP - Kanawha Valley study, pp. 4-73, 4-84, 4^95, and 4-99.
3-38
-------
the number of pollutants included in the risk estimate. For the
locations shown, maximum individual risks are in the neighborhood of
10"3 and 10~4. The results from the lEMP-Philadelphia study illustrate
that the MEI level can vary within subareas of a city by a factor of 40
(7 x 10'6 to 3 x 10'4), while the lEMP-Kanawha Valley study's results
show a 1,000-fold difference in maximum individual risk for one of the
locales in the Kanawha Valley versus the other three locales (10~6 vs.
10"3). Table 3-16 shows additional details on the variation in maximum
individual risk within the Kanawha Valley.
The pollutants that are the most important contributors to annual
incidence and the source categories that emit those pollutants and to
the individual risk in a geographic area can also vary from one area to
another. Table 3-17 illustrates some of the annual incidence variation
for five pollutants and the source categories that emit two of these
pollutants across six selected cities. For example, among the
individual pollutants, benzene is estimated to contribute a relatively
consistent percentage, between approximately 5 and'10 percent of total
cancer cases for the six cities. In contrast, 1,3-butadiene is seen in
Table 3-17 to contribute a much wider range, between 6 and 48, percent
of total cancer cases across the five cities. Among source categories,
road vehicles are consistently a major contributor,to annual incidence
attributed to benzene in each city, contributing between 45 and 81
percent of the total cancer cases attributed to benzene, and are the
most important source category of benzene-related incidence in five of
the six cities. In contrast, the relative contribution of "iron and
steel" to benzene incidence varies dramatically between cities, ranging
from 0 percent in four of the six cities to 25 percent in City D and
over 50 percent in Southeast Chicago. Along the same lines, 100 percent
3-39
-------
TABLE 3-16
ESTIMATES OF MAXIMUM LIFETIME INDIVIDUAL CANCER RISKS IN
NEIGHBORHOODS SURROUNDING FACILITIES IN THE KANAWHA VALLEY
TYPE OF
INDIVIDUAL RISK
Highest
(number of people
exposed)
Range of Maximum
Risks in
Remaining
Neighborhoods
Average Risk
Population in
locale
(number of people)
Belle
2.8x10"3
(600)
7x1 0"5. to
4x1 O'4
2.2x10"4
15,530
LOCALE
Charleston/
South Charleston Institute Nitro
6x10"3 8x10"3 8x10"6
(2,700) (1,300) (1,453)
2x10~4 to 3x10"3 2x10"4 to 2x10"3 9x10"7 to 6x10"6
2.8x10"4 1.U10"3 3.2x10"6
51,750 22,390 9,990
NOTE 1: Values have been adjusted to reflect the unit risk factors used in this study.
NOTE 2: These risk estimates are based on pollutants from point sources; risk from area
source pollutants are not included.
SOURCE: lEMP-Kanawha Valley Study, pp. 4-73, 4-77, 4-84, 4-88, 4-95, 4-97, 4-102, and 4-106.
3-40
-------
TABLE 3-17
CITY-TO-CITY VARIATION IN RELATIVE CONTRIBUTION OF SELECTED
POLLUTANTS TO TOTAL ANNUAL CANCER INCIDENCE
ITEM
Pollutant
(Source Category)
Benzene
(Road vehicles
(Gas marketing
(Iron & Steel/Steel
Mills
1,3-Butadiene
(Road vehicles
(Chemical Mfg.
Formaldehyde
Chromium
Methylene chloride
CITY3
A
8.8%
73%
7%
0
19.3%
100%
0
34.5%
5.2%
2.3%
B C
10.1% 4.8%
81% 67%
3.6% 0
0 0
23.9% 48.4%
100% 17%
0 83%
23.0% 24%
21.3% 16%
3.9% 1.6%
D
9.6%
56%
1%
24%
16%
100%
0
6.9%
14%
2.2%
E
7.0%
63%
3%
0
13%
100%
0
18.9%
16%
3%
SOUTHEAST
CHICAGO
4.8%
45%)
2.3%)
52%)
6.4%
100%)
0)
17%
16%
0.2%
5-City study. Derived from data worksheets.
Southeast Chicago study, p. 33. Relative contributions have been adjusted
based on unit risk factors listed in Table 2-6 in this report.
3-41
-------
of the estimated 1,3-butadiene related cancer cases are due to emissions
from road vehicles in five of the six cities. For City C, however, over
80 percent of the 1,3-butadiene related cancer cases are attributed to
chemical manufacturing and less than 20 percent to road vehicles.
Table 3-18 illustrates the variation in maximum individual risk
for individual pollutants. The data are taken from the lEMP-Baltimore
study, and show the exposed population to each pollutant as well. The
range in "maximum" individual risks is from 10"6 to 10"4. In Table 3-19,
the individual pollutant contributors to the highest estimated
individual risk grid cell in the Southeast Chicago area are presented.
As seen in Table 3-19, coke oven emissions contribute over 77 percent of
the total individual risk.
Table 3-20 illustrates the areawide lifetime individual risk
associated with individual pollutants based upon data from the IEMP-
Philadelphia study. Both monitored and modeled results are presented.
The range of areawide lifetime individual risks for individual
pollutants is from 10"6 to 10"5.
Variation in lifetime individual risk across source categories is
illustrated in Tables 3-21 through 24. Tables 3-21 and 22 report
maximum individual risks for two cities. Specific sources show maximum
individual risks in the range of 10"7 to 10"4.
Tables 3-23 and 24 show areawide lifetime individual risks for
specific sources. Table 3-23 shows area and point sources in Santa
Clara, while Table 3-24 shows area and point sources for the Kanawha
Valley. In the Santa Clara study, area and point sources are found to
be the major contributor to total areawide lifetime individual risk. On
the other hand, point sources are found in the Kanawha Valley to be the
major contributor to total areawide lifetime individual risks. Both
3-42
-------
TABLE 3-18
MAXIMUM LIFETIME INDIVIDUAL CANCER RISKS IN BALTIMORE
BY INDIVIDUAL POLLUTANT8
MAXIMUM LIFETIME
POLLUTANT INDIVIDUAL RISK
._.
Benzene
Tri chl oroethyl ene
Perch! oroethylene
Ethyl ene di chloride
Chl orof orm
Carbon tetrachloride
1,2-dichloropropane
Chromium (hexavalent)c
Cadmium6
l.OxlO'4
6.7xlCT6
5.4xlO'6
6.8xlO"5
l.lxlO'4
Z.lxlO'5
3.6xlO'5
0 to 3.6xlO'4
0 to 3.6xlO'6
EXPOSED
POPULATION15
i
48,771
48,771
14,270
12,880
23,997
23,997
16,848
490,690
118,411
NOTE: Values have been adjusted to reflect the unit risk factors used
in this study.
SOURCE: ' lEMP-Baltimore Study, Tables V-8 and V-14..
a Except for cadmium, individual risks were calculated using the maximum
observed ambient concentration measured across all monitoring sites.
Measured cadmium concentrations were below detection limits. For screening
purposes, the Baltimore study calculated risks assuming a range in ambient
concentrations from zero to the upper end of the detection limit (about
0.002
b The exposed population is in the grid cell at the monitoring site of maximum
concentration.
c Range indicates possible ambient levels of hexavalent chromium, from 0
percent to 100 percent.
d Range indicates possible ambient levels from 0.00 ng/mz to the upper end of
the detection limit (about 0.002 /t
3-43
-------
TABLE 3-19
RELATIVE CONTRIBUTION OF INDIVIDUAL POLLUTANTS TO
MAXIMUM LIFETIME INDIVIDUAL RISK OF CANCER
IN THE SOUTHEAST CHICAGO AREA
POLLUTANT
MAXIMUM
INDIVIDUAL RISK
% CONTRIBUTION
Coke Oven Emissions3
Benzene
Chromium
Formaldehyde
POM
Arsenic
Cadmium
Carbon tetrachloride
7xlO'4
6xlO'5
5xlO'5
4xlO'5
3xlO"5
2xlO'5
IxlO'5
IxlO'5
77%
7
6
4
3
2
1
1
Others
-------
TABLE 3-20
AREAWIDE LIFETIME INDIVIDUAL RISKS OF CANCER:
MONITORED VS. MODELED AMBIENT AIR CONCENTRATIONS
IN PHILADELPHIA
POLLUTANT
Chloroform
Ethylene dichloride
Carbon tetrachloride
Benzene
Trichloroethylene
1 , 2-di chl oropropane
Perch! oroethyl ene
CUMULATIVE
AREAWIDE LIFETIME
MONITORED
6.9xlO'6
L.OxlO'5
2.7xlO"5
S.OxlO"5
2.7xlO"6
2.2xlO'5
2.8xlO'6
1.2xlO"4
INDIVIDUAL RISK
MODELED
4.6X10"6
l.OxlO'5
r.5xl(T6
1.9xlO'5
1.7xlO'6
9.0xlO'6
2.1xlO'6
4.8xlO"5
NOTE: Values have been adjusted to reflect the unit risk factors
used in this study.
SOURCE: lEMP-Philadelphia, p. V-27.
3-45
-------
TABLE 3-21
ESTIMATES OF MULTI-POLLUTANT LIFETIME
CANCER RISKS TO THE MOST EXPOSED INDIVIDUAL
TO VARIOUS SOURCES IN PHILADELPHIA
MEI
LOCATION
Northeast Water Control Plant
Refinery B
Chemical Manufacturer
Plastic Cabinet Mfr.
Pharmaceutical Mfr.
Garment Mfr.
Refinery A
Industrial Dry Cleaner
MAXIMUM
INDIVIDUAL
RISK
6.2 x 10'5
1.4 x 10'5
2.3 x 10"4
8.2 x 10"7
3.2 x 10'4
1.7 x 10'5
3.1 x 10'5
2.8 x 10"5
COMMENT
8 pollutants
3 pollutants
3 pollutants
1 pollutant
3 pollutants
1 pollutant
3 pollutants
1 pollutant
NOTE: Where possible and as needed, the values have been adjusted to
reflect the unit risk factors used in this study.
SOURCE: lEMP-Philadelphia Study, p. VI-49.
3-46
-------
TABLE 3-22
ESTIMATED CANCER RISK TO MAXIMUM EXPOSED
INDIVIDUALS TO ORGANIC GASES IN SANTA CLARA
FOR SELECTED SOURCES
SOURCE
TYPE
Traffic Intersections
Hospitals
Pharmaceutical Manufacturer
Computer Equipment Mfr.
Industrial Facility
Fuel Pipeline
Drycleaners
Sewage Treatment Plants
Gasoline Station Pump
Groundwater Aeration
MAXIMUM
INDIVIDUAL RISK
3 x 10'4
2 x 10"4
1 x 10'4
4 x 10'5
3 x 10'5
2 x 10'5
1 x 1CT5
5 x 10'6
4 x 1(T6
2 x 1(T7
SOURCE: lEMP-Santa Clara study, p. 3-82.
3-47
-------
TABLE 3-23
AREAWIDE LIFETIME INDIVIDUAL RISK OF CANCER
FROM LIFETIME EXPOSURE TO ORGANIC GASES
IN SANTA CLARA
SOURCE
CATEGORY
Burning of Waste Material
Combustion of Fuels
Degreasers
Drycleaners
Fuels Distribution .
Industrial Solvents Coating
Mobile Sources
Off-Highway Mobile Sources
Other Chem./Indust.
Other Organics Evaporation
Pesticides Usage
Area Source Total
25 Point Sources Total
Carbon Tetrachloride
AREAWIDE
INDIVIDUAL
RISK
4 x 10'8
1 x 10'6
8 x 10'7
8 x 10'7
1 x 10"6
3 x 1(T6
1 x 1(T5
3 x 10'7
2 x 10'7
4 x 10'7
8 x 1CT7
2 x 10"5
6 x 10"6
1 x 10'5
TOTAL
4 x 10
-5
SOURCE: lEMP-Santa Clara study, p. 3-80.
3-48
-------
TABLE 3-24
ESTIMATES OF AREAWIDE LIFETIME INDIVIDUAL RISKS OF CANCER
ACROSS AREA AND POINT SOURCES IN THE KANAWHA VALLEY
SOURCE TYPE
Area
Gasoline Marketing
Heating
Road Vehicles
Solvent Use
Waste Oil Burning
Area Subtotal
Point
TOTAL
Population in
Locale
LOCALE
Belle
l.lxlO"6
l.SxlO'5
S.lxlO'5
7.7xlO"6
6.8xlO'7
5.3xlO'5
2.2xlO~4
S.OxlO'4
15,530
Charleston/
South Charleston
1.4xlO"6
1.3xlO"5
5.1xlO'5
l.SxlO'5
1.2xlO'6
S.OxlO'5
2.8xlO"4
3.6xlO'4
51,750
Institute
7.9xlO"7
l.SxlO'5
2.6xlO'5
6.3xlO"6
6.5xlO'7
4.7xlO"5
l.lxlO'3
l.lxlO'3
22,390
Nitro
3.4xlO'7
4.2xlO'6
9.8xlO'6
l.Sxl'O'6
2.4xlO'7
1.6xlO"5
S.OxlO'6
1.9xlO"5
9,990
SOURCE: lEMP-Kanawha Valley Study. Tables 32, 40, 45, 52, and 54. Values
for area sources could not be adjusted using the unit risk factors
in this study. However, the net effect is expected to be small.
Values for point sources have been adjusted using the unit risk
factors in this study.
3-49
-------
studies show similar risks from area sources (10~5 range). The
difference in relative contributions is due to the presence or absence
of point sources. The Kanawha Valley is a relatively heavily
industrialized area, with significant point sources, whereas the Santa
Clara area is much less industrialized. Thus, the relative
contributions of point and area sources to total areawide lifetime
individual risks is consistent with the character of the two study
areas.
The areawide lifetime individual risk data from area sources in
the Kanawha Valley Study show relatively consistent percentage
contribution among the same source category between locales. Among
area-type sources, "mobile sources" as a source category is found to be
the largest contributor to areawide lifetime risks in both the Santa
Clara and Kanawha Valley studies.
Comparison with the Results from the 1985 Six-Month Study
The results of the present study are compared with the results of
the 1985 Six-Month Study. This is done in two ways. First, a
comparison of estimated nationwide cancer cases is made to examine the
magnitude of the problem. Second, a comparison of the nature of the
problem is presented by examining the pollutants and the source
categories that appear to be the greatest contributors to risk.
Magnitude of the Problem
Tables 3-25 and 3-26 compare the cancer rates (i.e., annual cancer
cases per million population) arid annual cancer cases, respectively,
estimated for three studies presented in the 1985 Six-Month Study with
the point (or range) estimates of this study. As seen in these two
tables, the present study's low end estimated total cancer cases per
year per million population and the nationwide number of annual cancer
3-50
-------
TABLE 3-25
COMPARISON OF ANNUAL CANCER CASES PER MILLION POPULATION
WITH 1985 SIX-MONTH STUDY
POLLUTANT
Arsenic
Benzene
1,3-butadiene
Cadmium
Carbon tetrachloride
Chloroform
Chromium (hexavalent)
Dioxin
Ethylene Oxide
Ethylene dibromide
Ethylene dichloride
Formaldehyde
Gasoline vapors
Perch loroethylene
Trichloroethylene
Vinyl chloride
Vinyl idene chloride
Other
NESHAP
0,02
0.14
<0.001
0.04
0.06
<0.01
1.43
--
0.21
0.12
<0.01
0.01
N/A
0.01
0.04
0.05
<0.01
0.11
1985 SIX-MONTH
35 County
0.02
0.39
<0.001
0.02
0.004
0.002
0.29
--
N/A
0.02
0.03
0.21
0.15
0.14
0.15
0.17
N/A
0.35
STUDY
Ambient Air
Quality
0.26
1.02
--
0.06
0.19
0.07
1.05
--
N/A
N/A
0.05
0.83
N/A
0.10
0.08
--
0.27
0.01
THIS STUDY
0.28
0.75
1.11
0.04
0.17
0.48
0.61-1.1
0.008-0.52
0.03
0.28
0.19
0.52
0.08-0.32
0.03
0.03
0.10
0.04
0.13
Risk Estimates from Other EPA Efforts
a
Asbestos
Gasoline Marketing
PIC
Radionuclides
0.50
0.20
2.65
0.07
0.50
--
2.60
0.07
0.50
0.20
2.68
0.07
0.37
•• --
1.83-4.67
0.02
TOTAL
5.6
4.9
7.4
7.2 - 11.3
NOTE: Values in this figure are not absolute predictions of cancer occurrence
and are intended to be used in a relative sense only. The dose-response
relationships and exposure assumptions have a conservative bias, but
omissions due to uncharacterized pollutants (either directly emitted
or secondarily formed) and emission sources, the long-range transport
of pollutants, and the lack of knowledge of total risk from multi-
pollutant exposures will offset this bias to an unknown extent.
a Except for PIC in the 35-County study, these estimates of cancer incidences
were not part of the individual results of the NESHAP, 35-County, and
Ambient Air Quality studies. The 1985 Six-Month Study included these
estimates for these pollutants to provide for a more complete accounting of
information available to the 1985 Six-Month Study.
Includes radon.
3-51
-------
TABLE 3-26
COMPARISON OF ANNUAL CANCER CASES WITH
1985 SIX-MONTH STUDY
1985 SIX-MONTH STUDY
THIS STUDY
POLLUTANT
Acrylonitrile
Arsenic
Benzene
1,3 butadiene
Cadmium
Carbon tetrachloride
Chloroform
Chromium (hexavalent)
Coke Oven Emissions
Dioxin
Ethylene Oxide
Ethylene dibromide
Ethylene dichloride
Formaldehyde
Gasoline vapors
Hethylene chloride
Perch loroethylene
Trichloroethylene
Vinyl Chloride
Vinyl idene chloride
Other
NESHAP
0.42
4.7
32.3"
0.01
8.5
14
0.27
330
8.6
--
47.8
26.7
0.9
' 1.6
--
1
2.9
9.7
11.7
0.04
2.9
35 County
4.2
1.1
18.5
0.01
1.1
0.2
0.1
13.4
2.4
--
--
1.0
1.5
10
6.8
--
6.7
6.8
8.2
--
0
Ambient Air
Qua I i ty
..
60
234
--
14.6
43
17
242
--
--
--
--
11
191.3
--
7.4
22
18
--
62
1
13
68
181
266
10
41
115
147-265
7
2-125
6
68
45
124
19-76
5
6
7
25
10
30
Subtotals
504
207
Risk Estimates from Other EPA Efforts
TOTAL
1291
234
1539
1,195-1,493
Asbestos
Gasoline Marketing
PIC
Radionuc I ides/Radon
115
46
610
16
23.7
--
125.1
3.3
115
46
615.4
16
88
(see gas vapor)
438-1120
5
1716
1,726-2,706
NOTE: Values in this figure are not absolute predictions of cancer occurrence
and are intended to be used in a relative sense only. The dose-response
relationships and exposure assumptions have a conservative bias, but
omissions due to uncharacterized pollutants (either directly emitted or
secondarily formed) and emission sources, the long-range transport of
pollutants, and the lack of knowledge of total risk from multi-pollutant
exposures will offset this bias to an unknown extent.
a Except for PIC in 35-County study, these estimates of concern incidences were
not part of the individual results of the NESHAP, 35-County, and Ambient Air
Quality studies. The 1985 Six-Month study included these estimates for these
pollutants to provide for a more complete accounting of information available
to the 1985 Six-Month study.
3-52
-------
cases are essentially the same as the original Ambient Air Quality study
in 1985 Six-Month Study, which was larger than the other two studies in
the 1985 Six-Month Study. The upper end of the estimated cancer rate
and annual cancer cases of the present study are approximately 1.5 to 2
times larger than the NESHAP or Ambient Air Quality studies in the 1985
Six-Month Study. Using a 1986 U.S. population of 240 million, the
present study estimates up to approximately 500 to 900 more cancer cases
per year nationwide than either the NESHAP or Ambient Air Quality study
results in the 1985 Six-Month Study.
There are several factors that account or may account for this
apparent increase in estimated risk. One factor is that this study
includes more pollutants for which risks have been estimated than were
included in the 1985 Six-Month Study/ Most of these pollutants are the
result of the Sewage Sludge Incinerator, Hazardous Waste Combustion, and
TSDF studies being available for inclusion.. On an individual pollutant
basis, the potentially most important addition from TSDFs is dioxin, for
which up to 92 annual cancer cases were estimated based on data in the
TSDF study. As shown in Table 3-26 by the large range in risk (2 to 125
annual cancer cases), there is substantial uncertainty associated with
the risk estimate for dioxin in this study.
A second factor that accounts for an increase in the estimated
cancer risk is the changes, some of .which are significant, that have
occurred to unit risk factors. Table 3-27 compares the unit risk
factors used in the 1985 Six-Month Study with those used in this study
for those pollutants for which the unit risk factor has changed. As
seen in this table, the unit risk factors have changed in a few
instances by relatively small amounts (±25 percent). In some instances,
3-53
-------
TABLE 3-27
COMPARISON OF UNIT RISK FACTORS
POLLUTANT
Acryl amide
Benzene
BaP
Beryl 1 i urn
1,3-Butadiene
Cadmium
Chloroform
Epichlorohydrin
Ethyl ene di bromide
Ethyl ene oxide
Formaldehyde
Gasoline vapors
Methyl chloride
Methyl ene chloride
Nickel (subsulfide)
Perchl oroethyl ene
Propylene oxide
Styrene
Tri chl oroethyl ene
Vinyl chloride
Vinyl idene chloride
MAY 1985
1.7xlO'5
6.9xlO'6
3.3xlO'3
4.0xlO"4
4.6X10'7
2.3xlO"3
l.OxlO'5
2.2xlO'7
S.lxlO'4
3.6xlO"4
6.1xlO"6
7.5xlO'7
1.4xlO'7
l.SxlO'7
3.3xlO'4
1.7xlO"6
1.2xlO"4
2.9xlO'7
4.1xlO"6
2.6xlO"6
4.2xlO'5
JUNE 1988
l.lxlO'3
8.3xlO"6
1.7xlO"3
2.4xlO'3
2.8xlO"4
l.SxlO'3
2.3xlO'5
1.2xlO'6
2.2xlO"4
-A
1.0x10 4
l.SxlO"5
6.6xlO"7
3.6xlO'6
4.7xlO'7
4.8xlO'4
5.8xlO'7 -
3.7xlO'6
5.7xlO"7
1.7xlO'6
4.1xlO'6
5.0x10'^
% CHANGE
+ 6400
+ 20
48
+ 500
+ 60770
22
+ 130
+ 445
57
72
+ 113
12
+ 2,470
+ 161
+ 45
66
97
+ 97
59
+ 58
+ 19
3-54
-------
the change has been large (over 100 percent) and, in the case of 1,3-
butadiene, the change has been over a 60,000 percent increase.
A third factor that accounts for the apparent increase in
estimated cancer risk is a more complete accounting of sources that
contribute to cancer risk. As noted above, a potentially significant
»
new source category is TSDFs. Another important source category is
electroplating.
The more extensive body of information available to this study has
helped provide for a more complete accounting of source categories and
pollutants. The apparent increase in estimated cancer risk, therefore,
should not necessarily be viewed as a problem that has become worse.
Rather, the estimates in this study, which are based on new and more
complete information, simply suggest that the problem may be larger than
previously thought.
Nature of the Problem
The nature of the air toxics problem can be described in several
ways: which pollutants contribute the most to the cancer risk; which
sources contribute the most to cancer risk; and how does cancer risk
vary from one geographic region to another. Since this study found
geographic variations to be of a very similar nature as those reported
in the 1985 Six-Month Study, only the first two aspects of the problem
will be compared.
Individual Pollutants. For the most part, the same pollutants
found to contribute the largest percentages to total annual cancer risk
in the 1985 Six-Month Study are also found to be among the larger
contributors in the present study. These compounds include hexavalent
chromium, PIC, asbestos, benzene, carbon tetrachloride, ethylene .
dibromide, arsenic, and vinyl chloride.
3-55
-------
As seen earlier in Table 3-25, eight of the individual cancer
rates (i.e., annual cancer cases per year per million population)
calculated for pollutants under the present study fall within the range
of cancer rates created by the three studies in the 1985 Six-Month
Study. The cancer rates for two pollutants (gasoline vapor and PIC)
bound their respective estimates in the 1985 Six-Month Study. This
indicates that the magnitudes of incidence for these ten pollutants have
been estimated to be approximately the same. For formaldehyde, the
cancer rate calculated in this study is lower than that in the Ambient
Air Quality study found in the 1985 Six-Month study, but higher than the
other two analyses in the 1985 Six-Month Study. The decrease most
likely reflects better data and measurement techniques available than
were used in the original Ambient Air Quality study.
For the other 10 pollutants identified in Table 3-25, the cancer
rates calculated in this study fall outside the range created in the
1985 Six-Month Study. Of these pollutants, four -- ethylene oxide,
trichloroethylene, asbestos, and radionuclides -- show a decrease in the
estimated cancer rate. For ethylene oxide and trichloroethylene, most
of this decrease can probably be attributed to the change in the unit
risk factor. For asbestos, the change reflects better emission factors.
For radionuclides, a new risk analysis was conducted using updated
information on the number of facilities, radionuclide emissions to the
air, and control technologies. The net effect of the updated
information was a decrease in the estimated risk for radionuclide
exposure.
Six pollutants (including dioxin, which was not included in the
1985 Six-Month Study) show an increase in estimated risk. For arsenic,
a modest increase in annual cancer cases is estimated (from 60 to 68 per
3-56
-------
year), which is apparently due to higher measured ambient
concentrations.
For ethylene dibromide (EDB) the estimated cancer rate has
increased in spite of a decrease in the unit risk factor. This has
likely occurred due to the absence of a risk estimate for EDB in the
1985 Six-Month Study from the Ambient Air Quality study. As noted in
the present study, modeled estimates appear to underestimate actual
ambient concentrations. The present study based the risk estimate on
ambient-measured concentrations. Thus, the net effect is an increase in
the estimated cancer rate for EDB, with an increase in estimated cancer
cases from 27 to 68 per year nationwide.
For both ethylene dichloride and chloroform, the updated Ambient
Air Quality study's estimates of cancer risk were selected for the risk
estimate. The increase in the estimated cancer cases from chloroform
can be attributed in part to an increase in its unit risk factor. For
both pollutants, the increase may be simply attributable to a more
recent and larger data set that shows higher ambient concentrations than
before.
The most dramatic increase is associated with 1,3-butadiene. This
has occurred for two reasons. One reason is the increase in the unit
risk factor, from 4.6 x 10"7 to 2.8 x 10"4, an increase of over 600
times. The second reason is that ambient-measured concentrations of
1,3-butadiene were not a part of the 1985 Six-Month Study and the major
source of ambient 1,3-butadiene -- motor vehicles -- were not included
in the other two studies in the 1985 Six-Month Study. These two factors
combined to increase the estimated nationwide cancer risk due to 1,3-
butadiene from 0.01 cancer cases per year to almost 270 per year.
3-57
-------
Source Categories. In the 1985 Six-Month Study, area and point
sources were found each to account for approximately one-half of the
aggregate incidence in both the NESHAP and the 35-County studies. When
PIC was included (by using BaP as a surrogate), areas sources were found
to be dominant, -accounting for over 75 percent of the incidence in both
the NESHAP and the 35-County studies. This result was noted as being
consistent with the fact that PIC was estimated to account for a large
portion of aggregate incidence, and that nearly all BaP emissions
appear to come from area sources (principally motor vehicles and fuel
combustors in small heating units).
Earlier in this chapter, Table 3-11 summarized the estimated
contributions of individual source categories to total cancer risk by
area vs. point source. Area sources were estimated to contribute
approximately 75 percent of the total nationwide annual incidence and
point sources, approximately 25 percent. The two studies, thus, show
essentially identical estimates of the relative contribution of area vs.
point sources in spite-of some significantly important pollutants and
source categories included in the current study that were not included
in the 1985 Six-Month Study.
Table 3-28 presents the results of the Southeast Chicago study in
terms of area vs. point and mobile vs. stationary sources. In that
study, point sources are estimated to contribute approximately 48
percent of the total estimated annual incidence in the Southeast Chicago
area, and area sources approximately 30 percent. (Approximately 20
percent was attributed to background pollutants, the sources of which
were not identified.) These relative contributions of area vs. point
sources are very different from the nationwide split estimated. The
3-58
-------
TABLE 3-28
CONTRIBUTION OF SOURCES TO ESTIMATED ANNUAL
CANCER CASES AND AREAWIDE LIFETIME INDIVIDUAL RISKS
IN SOUTHEAST CHICAGO
SOURCE TYPE/
CATEGORY
Point
Steel Mills
Chrome Platers
Other Industrial Sources
Sewage Treatment Plants
Total Point
Area
Home Heating
Consumer Sources
Mobile Sources
Waste Handling
Total Area
Background Pollutants
Mobile
Stationary
Background
ANNUAL
CANCER CASES3
0.41
0.185
0.016
0.001
0.612
0.127
0.05
0.22
0.001
0.398
0.26
0.22
0.79
0.26
AREAWIDE
LIFETIME
INDIVIDUAL RISK
7.3 x 10'5
3.3 x 10"5
2,9 x 10'6
1.8 x 10'7
1.1 x 10"4
2.3 x 10'5
8.9 x 10'6
3,9 x 10'5
1.8 x 10'7
7.1 x 10'5
4.6 x 10'5
3,9 x 10'5
1.4 x 10'4
4.6 x 10'5
PERCENT OF
TOTAL
32
15
1
0.1
48
10
4
17
0.1
31
20
18
62
20
a Southeast Chicago study, p. 33. Values were adjusted to the unit risk
factors used in this study.
b Calculated by multiplying annual cancer cases by 70 and dividing by
population of study area (i.e,, 393,000).
3-59
-------
larger share attributed to point sources in the Southeast Chicago area
is likely due to locally high risk from steel mills.
Mobile sources are also seen to be a relatively lower contributor
to total annual incidence versus stationary sources in. the Southeast
Chicago area. This again is likely due to the locally-high risk from
steel mills. In addition, risk from heating appears to be higher than
the nationwide estimate, thus further reducing the percent of total
annual incidence attributed to mobile sources.
The total areawide lifetime individual risks for the Southeast
i
Chicago study are similar to those reported earlier in this chapter for
the Santa Clara study (Table 3-23) and the Kanawha Valley study (Table
3-24). Excluding consideration of "background pollutants," areawide
lifetime individual risks from area sources are again in the 10"5 range,
with mobile sources being the major contributor to areawide lifetime
individual risk followed closely by home heating. Point source
contribution to areawide lifetime individual risk in Southeast Chicago
is higher than the area source contribution, as was seen in the Kanawha
Valley study. This seems consistent with the relative nature of the
study areas (Southeast Chicago has significant point source contribution
from coke ovens.)
3-60
-------
4.0 SUMMARY AND CONCLUSIONS
In this chapter, the results of this study are summarized and
conclusions are drawn with regard to the magnitude and nature of the
cancer problem associated with outdoor exposures to air toxics in the
United States. These results are also compared to those contained in
the 1985 Six-Month Study.
As has been discussed throughout the report, the results of this
study are subject to various limitations and uncertainties. The
numerical estimates presented in this report, therefore, should be
viewed as rough indications of the magnitude of potential cancer risk
caused by a limited group of pollutants found in the ambient air. Many
of the absolute values for individual pollutants are almost certainly
inaccurate. The best use of these estimates is in describing the broad
nature of ;the cancer risk posed by these toxic air pollutants and by
making relative comparisons of risks between pollutants and source
categories.
Magnitude of the Cancer Risk
Annual Cancer Incidence
Based on the pollutants and source categories examined, nationwide
annual cancer incidence is estimated to be between 1,700 and 2,700
cancer cases per year (see Table 3-1). This is equivalent to between
7,2 and 11.3 cancer cases per year per million population (1986
population of 240 million). Approximately one-third of this cancer risk
4-1
-------
was attributed to exposure to products of incomplete combustion (PIC).
Of all the cancer risks estimated in this study, the greatest degree of
uncertainty is mostly likely associated with the cancer risk estimate
for PIC.
A range of annual cancer incidence is reported as the result of
uncertainties associated with primarily four individual pollutants that
also are estimated to be among the largest individual contributors to
cancer risk. These four pollutants are PIC, dioxin, gasoline vapors,
and hexavalent chromium. The uncertainties identified are associated
primarily with: (1) the inability at this time to select a single unit
risk factor from a range of unit risk factors for diesel particulates,
which are included with PIC; (2) the sampling and extrapolation
methodologies for dioxin; (3) the identification of the cancer-causing
portion of gasoline vapors; and (4) the portion of total ambient
chromium that is hexavalent. Although point estimates were made for
most pollutants, the lack of a range does not mean there is no
uncertainty associated with the absolute magnitude of the cancer risk
estimate.
The 1985 Six-Month Study presented three separate analyses that
showed a range of cancer rates from approximately 5 to 7.4 cancer cases
per year per million population. The results of the current study
estimated a cancer rate of between 7.2 and 11.3 cancer cases per year
per million population (see Table 3-25). Using a total 1986 U.S.
population of 240 million, the results of this study show approximately
500 to 900 more cancer cases per year (comparing lower and upper
ranges). This "increase" does not necessarily indicate a growing
problem, but is more likely the result of analysis of more air toxic
pollutants than were considered in the 1985 Six-Month Study and, in some
4-2
-------
instances, a better accounting of sources (e.g., sources that emit 1,3-
butadiene). Even though this study has a broader data base to draw upon
than was available to the 1985 Six-Month Study, it is recognized that
cancer risk estimates are being made for a only a portion of total
ambient air pollutants and for a portion of all sources. In addition,
quantitative risk estimates from pollutants formed ,or transformed in the
atmosphere (secondary formation) remain unquantified for almost all '
pollutants. Evidence to date suggests secondarily-formed pollutants may
pose a significant component of total cancer risk. Based on these
considerations, the actual magnitude of the problem, therefore, can
easily be larger than estimated in this study. On the other hand, the
estimates presented in this study are based on the use of unit risk
factors that are either upper-bound estimates or maximum likelihood
estimates of the carcinogenicity of a pollutant. Quantitative estimates
derived from the use of these unit risk factors, therefore, could
overstate the true risk from a pollutant. The net effect of these and
other uncertainties (e.g., assessing exposures) on total risk is
unknown. It is expected, nevertheless, that the pollutants and source
categories considered herein are among the major contributors to cancer
risk from air toxics based on our current state of knowledge.
Lifetime Individual Risks
Maximum lifetime individual risks of 1 x 10"4 (1 in 10,000) or
greater were reported in almost all of the studies examined for this
report (see Table 3-2). Risk levels this high were reported for such
specific sources as major chemical manufacturers, waste oil
incinerators, hazardous waste incinerators, publicly owned treatment
works (POTWs), steel mills, hospitals, traffic intersections, and
hazardous waste treatment, storage, and disposal facilities (TSDFs).
4-3
-------
Risk levels reported for areawide lifetime individual risks, which
includes risk from point and area sources, were generally around 10
(see Table 3-6).
On an individual pollutant basis, maximum individual risks of 1 x
10"4 or greater were reported for 16 of the pollutants included in the
NESHAP/ATERIS data base (see Table 3-2). Twelve of these pollutants
were estimated to have maximum individual risks of 1.0 x 1()"3 or
greater. These'estimates of risks are related to specific sources.
However, because of the nature of the assessments contained in, the
ATERIS data base, there is a very large degree of uncertainty associated
with some of these estimates for specific sources.
Multi-pollutant lifetime individual risks in four urban areas due
to exposure to 9 to 16 pollutants (at one monitoring site in each urban
area) ranged from 3 x 10"4 to 3 x 10"3 (see Table 3-7). These estimates
were based on ambient-measured data and generally cannot be related to
specific point sources.
While the present study shows the estimate of nationwide cancer
cases to be somewhat larger than was estimated in the 1985 Six-Month
Study, the maximum and areawide lifetime individual risks estimated in
*.
the present study are nearly identical to those estimated in the 1985
Six-Month Study. The broader scope of the present study has resulted in
identifying additional types of sources (e.g., TSDFs, POTWs) that can
contribute to significant maximum individual risks.
Nature of the Cancer Risk
Individual Pollutants
As discussed in Chapter 2, there is considerable uncertainty with
the absolute risk estimates for some of the pollutants examined in this
4-4
-------
study. Nevertheless, the available information indicates seventeen1 of
the approximately 90 pollutants examined may each account for 10 or more
cancer cases per year nationwide. Of these, thirteen may each account
for 40 or more cancer cases per year. These thirteen are: PIC;
1,3-butadiene; hexavalent chromium; formaldehyde; benzene; chloroform;
asbestos; dioxin; arsenic; ethylene dibromide; gasoline vapors; ethylene
dichloride; and carbon tetrachloride.
The seventeen compounds that are estimated to contribute at least
10 excess cancer cases per year nationwide appear to be most frequently
associated with high maximum individual risks. However, other compounds
may be.the most significant contributor to the maximum individual risk
for a particular city. For example, coke oven emissions in the
Southeast Chicago study contributed over 75 percent of the highest
estimated lifetime individual risk. Individual compounds, such as
epichlorohydrin and styrene, that have small aggregate cancer incidences
may also be associated with high maximum individual risks (greater than
1 x 10'4).
For the most part, the individual compounds found to be the more
important contributors to cancer risk in the present study are the same
as those found in the 1985 Six-Month Study. The most significant
difference is the addition of 1,3-butadiene to the list of potentially
important contributors. Dioxin may also be a significant contributor,
but the uncertainties associated with its risk estimates make it
difficult to conclude this at this time. Several pollutants, on the
other hand, appear to be somewhat less of a factor in terms of aggregate
Acrylonitrile, arsenic, asbestos, benzene, 1,3-butadiene, cadmium,
carbon tetrachloride, chloroform, chromium (hexavalent), dioxin, ethylene
dibromide, ethylene dichloride, formaldehyde, gasoline vapors, PIC, vinyl
chloride, and vinylidene chloride.
4-5
-------
cancer risk, but not necessarily in terms of maximum individual risk.
Changes in the pollutants identified in the present study and in the
1985 Six-Month Study as the more important contributors are primarily
due to the broader scope of the present study and to newer estimates of
the unit risk factors for the individual pollutants.
Sources
As in the 1985 Six-Month Study, a wide variety of sources
contribute to aggregate incidence and individual risk (see Table 3-11).
Motor vehicles were found to be the largest contributor to nationwide
annual incidence, contributing almost 58 percent of total incidence
(including estimated risk attributable to the secondary formation of
formaldehyde). The risk associated with the secondary formation of
formaldehyde was estimated to account for 6.5 percent of the total
estimated incidence (130 annual cancer cases). Of these 130 annual
cancer cases, 93 are estimated to be attributable to volatile organic
compound (VOC) emissions from area sources (including 45 from mobile
sources) and 37 from point sources (see Table 3-10). Electroplating
(6%) was the third largest contributor to aggregate incidence as a
result of chromium emissions. The next five major contributors were
TSDFs (5%); woodsmoke (5%); asbestos, demolition (4%); gasoline
marketing (3%); and solvent use/degreasing (3%). Unspecified point
sources (3%) and cooling towers (3%) were the ninth and tenth largest
contributors to total annual incidence.
In general, a significant portion of the cancer risk from specific
sources was usually due to a few pollutants, even where a source emitted
many different pollutants. For example, over 70 pollutants were
included in the analysis on hazardous waste combustors, but two
pollutants (cadmium and hexavalent chromium) were estimated to be
4-6
-------
responsible for almost 90 percent of the estimated cancer cases from
hazardous waste incinerators and three pollutants (cadmium, hexavalent,
chromium, and arsenic) for almost 90 percent of the estimated cancer
cases from hazardous waste boilers and furnaces.
Both mobile and stationary sources were found to contribute
significantly to total nationwide annual incidences. Considering both
direct emissions to the atmosphere and secondary formation of
formaldehyde, mobile sources were estimated to contribute approximately
58 percent and stationary sources approximately 42 percent of total
annual incidence. Area sources were found to contribute approximately
75 percent and point sources approximately 25 percent of the total
cancer incidence (see Table 3-11).
The relative contribution of the aggregate types of sources (i.e.,
point vs. area, mobile vs. stationary) to total annual incidence can
vary significantly for specific geographic areas. For example, the
Southeast Chicago study showed point sources contributing almost 50
percent (vs. the 20 percent noted above) and stationary sources
approximately 60 percent (vs. 42 percent from above) of the total annual
incidence estimated for Southeast Chicago (see Table 3-28). These
differences are most likely due to* the significant contribution to risk
from steel mills in the Southeast Chicago area.
With regard to lifetime individual risk, reported maximum
individual risks usually were associated with specific point sources,
such as industrial facilities or chemical manufacturers. Based on the
information in the lEMP-Santa Clara study, the levels of maximum
individual risk associated with individual area-type sources (e.g.,
gasoline marketing, degreasers, waste oil burning) appear to be lower
than those found for sources typically included in a point source
4-7
-------
category (see Table 3-22). However, the lEMP-Santa Clara study found a
maximum lifetime individual risk of 10"4 for at least one traffic
intersection. On the other hand, not all point sources have high
maximum lifetime individual risks associated with them. In fact, the
majority of point sources in some source categories have maximum
individual risks of 10"6 or less.
As noted earlier, areawide lifetime individual risks were
generally lower than the maximum individual risk values within
comparable geographic locales. The relative contribution of area and
point sources to areawide lifetime individual risks can vary from one
locale to another. For example, the lEMP-Santa Clara showed area
sources contributing approximately 50 percent of the areawide lifetime
individual risk and point sources approximately 15 percent (see Table
3-23). (The remaining 25 percent was from carbon tetrachloride, which
was not allocated in that study to either area or point source.) In
contrast, the Southeast Chicago study shows point sources contributing
approximately 48 percent and area sources approximately 31 percent of
the areawide lifetime individual risk (the remaining 20 percent was from
formaldehyde and carbon tetrachloride, which were not allocated to
either area or point source) (see Table 3-28).
Among area sources, mobile sources were found to be responsible
for between 50 and 60 percent of the areawide lifetime individual risk
(see Tables 3-23, 3-24, and 3-28). Solvent use and heating in the IEMP-
Kanawha Valley study (see Table 3-24) and home heating in the Southeast
Chicago study (see Table 3-28) were identified as having areawide
lifetime individual risks approximately one-half to one-quarter as large
as those associated with mobile sources.
4-8
-------
Geographic Variability
Exposure to individual air toxics varies on a city-to-city basis
as well as on an intra-city basis. For some pollutants, such as
benzene, the variation appears to be relatively small, less than a
factor of two (.see Table 3-13). For other pollutants, the variation is
higher, ranging to a factor of almost 20. For the pollutants compared,
the degree of variation in ambient concentrations for a particular
pollutant apparently can vary by the same degree within a city as
between cities.
The variations in ambient concentrations for individual pollutants
can lead to variations in the number of cancer cases and the cancer rate
(i.e., cancer incidence per year per million population) between
geographic areas. In spite of the differences in risk attributable to
individual pollutants, areawide lifetime individual risks were found to
be generally the same between the geographic locales examined in this
study (see Table 3-6). Particular geographic locales may have
substantially higher areawide lifetime individual risk. If this occurs
in a relatively sparsely populated locale, a low absolute number of
cancer cases would mask a high cancer rate and this higher-than-average
areawide lifetime individual risk. In a similar manner, a relatively
low areawide lifetime individual risk may mask a significant maximum
individual risk that affects a small portion of the local population.
Most of the geographic locales reviewed in this study showed
comparable maximum or highest estimated lifetime individual risk levels
(see Table 3-15). However, this does not mean that the same number of
people are exposed to that level, of risk in each city.
The pollutants and source categories that are the most important
contributors to risk (annual incidence and maximum individual risk) in a
4-9
-------
geographic area will vary from one area to another. For some
pollutants, the variation may be relatively small and the primary source
will be the same between areas. For example, benzene was found to
contribute between approximately 5 and 10 percent of the total annual
incidence in six cities, with between 45 and 80 percent of the benzene-
related cancer incidence attributed to motor vehicles (see Table 3-17).
Other pollutants show a wider range of variation, and the relative
contribution for some pollutants can be dramatically affected by the
presence of major point sources. For example, in five of the six
selected cities, 1,3-butadiene was estimated to contribute between 6 and
24 percent of the total cancer incidence, all attributable to motor
vehicles. In the sixth city, over 48 percent of the total cancer
incidence was attributed to 1,3-butadiene. Of the 1,3-butadiene-
related cancer incidence in this city, over 80 percent was attributed to
chemical manufacturing plants and less than 20 percent to motor
vehicles.
In general, the results and conclusions of the present study are
consistent with those drawn in the 1985 Six-Month Study regarding
geographic variability.
4-10
------- |