vvEPA
United States
Environmental Protection
Agency
Research and Development
Health Effects Research
Laboratory
Research Triangle Park NC 2771
t K< (.00 1 80-018
M,r,' 19HO
Pilot Study—Uses of
Medicare Morbidity
Data in Health
Effects Research
VT> \ "O V
K A rx i
•o . s.
r-P 600/1
80-018
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U S Environmental
Protection Agency, have been grouped into nine series These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are
1 Environmental Health Effects Research
2 Environmental Protection Technology
3 Ecological Research
4 Environmental Monitoring
5 Socioeconomic Environmental Studies
6 Scientific and Technical Assessment Reports (STAR)
7 Interagency Energy-Environment Research and Development
8 "Special" Reports
9 Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL HEALTH EFFECTS RE-
SEARCH series This series describes projects and studies relating to the toler-
ances of man for unhealthful substances or conditions This work is generally
assessed from a medical viewpoint, including physiological or psychological
studies In addition to toxicology and other medical specialities, study areas in-
clude biomedical instrumentation and health research techniques utilizing ani-
mals — but always with intended application to human health measures.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/1-80-018
March 1980
PILOT STUDY—USES OF MEDICARE
MORBIDITY DATA IN HEALTH EFFECTS RESEARCH
by
Irwin J. Shifter and Edgar A. Parsons
System Science, Inc.
P. 0. Box 2345
Chapel Hill, North Carolina 27514
Contract No. 68-02-2782
Project Officer: Wilson B. Riggan
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
HEALTH EFFECTS RESEARCH LABORATORY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
LIBRARY
UU S.
EDTSOTI, M.J. 0881'
-------
DISCLAIMER
This report has been reviewed by the Health Effects Research Laboratory,
U.S. Environmental Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the views and policies
of the U.S. Environmental Protection Agency, nor does mention of trade names
or commercial products constitute endorsement of recommendation for use.
-------
FOREWORD
The many benefits of our modern, developing, industrial society are
accompanied by certain hazards. Careful assessment of the relative risk
of existing and new man-made environmental hazards is necessary for the
establishment of sound regulatory policy. These regulations serve to en-
hance the quality of our environment in order to promote the public health
and welfare and the productive capacity of our Nation's population.
The Health Effects Research Laboratory, Research Triangle Park, conducts
a coordinated environmental health research in toxicology, epidemiology, and
clinical studies using human volunteer subjects. These studies address
problems in air pollution, non-ionizing radiation, environmental carcinogenesis
and the toxicology of pesticides as well as other chemical pollutants. The
Laboratory participates in the development and revision of air quality cri-
teria documents on pollutants for which national ambient air quality standards
exist or are proposed, provides the data for registration of new pesticides or
proposed suspension of those already in use, conducts research on hazardous
and toxic materials, and is primarily responsible for providing the health
basis for non-ionizing radiation standards. Direct support to the regulatory
function of the Agency is provided in the form of expert testimony and prep-
aration of affidavits as well as expert advice to the Administrator to assure
the adequacy of health care and surveillance of persons having suffered immi-
nent and substantial endangerment of their health.
Collection of epidemiological health data is a costly and time consuming
effort. For these reasons, we are always looking for sources of health data
which may be used in our environmental health studies. This report describes
a pilot investigation of the practicality of using Social Security Administra-
tion (SSA) Medicare 20 percent sample of short-stay hospital discharge data to
supplement mortality data in cancer and other environmentally related health
studies.
F. G. Hueter, Ph.D.
Director
Health Effects Research Laboratory
-------
PREFACE
Little is known of the relationship between the complex of chemical and
physical factors comprising the "real life" human environment and the effects
of one or more of these factors on human health. Much of what is known is
derived from laboratory animal observations and accidents or unusual occur-
rences involving humans. While helpful, the non-human data and the atypical
events are generally inadequate from the perspective of relevance to human
environmental exposure parameters. Serious technical questions can also be
posed concerning the instrumentation measuring the composition of the air, and
whether fixed/mobile monitoring data can be re-assembled to characterize
accurately the air actually breathed by humans of different ages and living
patterns.
A supplementary approach to determining factors harmful to human health
in the "real life" environment is to stress geographical areas and/or selected
populations characterized by unusually high/low levels of indicator diseases,
such as respiratory diseases and cancer. The basic objective of EPA being
human health protection, this approach offers the advantages of identifying
areas of greatest need, and of enhancing responsiveness to reduce harmful envi-
ronmental pollutants. An additional advantage is deferment/avoidance of
costly abatement and control measures in areas of lesser need (or no need)
without jeopardizing health.
This approach requires health status details rarely available because of
cost and confidentiality barriers. Considerable EPA reliance has been,
therefore, placed on the low-cost non-confidential mortality statistics
collected and stored on computer tape by the National Center for Health
Statistics for purposes that predate EPA. These tapes exclude much more in-
formation of interest to EPA which is collected in hard copy and presents
considerable cost and clearance problems. Serious attention has not hereto-
iv
-------
fore been directed by EPA to utilization of other established local-state-
Federal health information acquisition systems featuring non-mortality data.
In accordance with its unsolicited proposal that provided the basis for
this project, System Sciences, Inc. staff continued explorations with respect
to environmental health applications of the "Medicare morbidity" files. These
files were created by the Social Security Administration and are being updated
annually for non-environmental purposes. SSA cooperation was outstanding.
Staff of the Program Statistics Division, Office of Research and Statistics,
and Mr. Charles Fisher in particular, immediately perceived the environmental
health relevance of certain data elements. Tapes were prepared and transferred
to EPA that met confidentialy constraints with extracts of most of the data
pertinent to health conditions and the environment. The tapes and data proved
satisfactory for pilot investigation purposes.
The underlying goal of this project is to demonstrate the applicability
to EPA of the partial Medicare files that have been obtained. Assuming appli-
cability, additional SSA information may be provided subject always to confi-
dentiality precautionary requirements. Properly used in conjunction with
emissions, monitoring data, demography, mortality and other "computerized" EPA
resources, the Medicare files have been demonstrated to offer unprecedented
opportunities for contributing to knowledge of relationships between the envi-
ronment and human health.
-------
ABSTRACT
This project is a pilot investigation of the practicability of utilizing
Social Security Administration (SSA) Medicare morbidity data to supplement
mortality data in cancer and other environmentally-related studies. The Medi-
care files are a 20 percent random selection, aggregating 5,000,000 of the
25,000,000 insurees; for this study non-confidential data on 1.2 million hos-
pitalizations for 815,000 persons diagnosed as having a neoplasm, respiratory,
or digestive disease during 1971, 1972, and 1973 were included. The data are
kept current by SSA for their analysis purposes.
Project outputs demonstrate the availability to EPA of a major new data
source for health effects and program formulation analyses. This project pre-
pared the Medicare data for use directly with emissions, ambient air, demo-
graphy, mortality, and other data banks at EPA's National Computer Center.
Outputs include tables and computer-prepared multi-colored maps illustrating
relationships between any selected industry of interest and respiratory or
other diseases of the co-located population.
The Medicare codes show hospitalization date, diagnosis, age, race, sex
and county and state of residence. When used to measure the Pittsburgh 1973
air pollution episode, the files disclosed a significant increase in respira-
tory disease hospitalizations immediately prior to, and during the episode.
With an annual average of 16 hospitalizations to each death, the Medicare
files increase substantially the opportunity for identifying possible rela-
tionships between particular environments and the co-located populations.
Migration aspects can be included as appropriate.
The Medicare files are the only known source that incorporates all cancer
cases systematically, regardless of whether the cancer is fatal. Of new 1979
cancer cases (765,000), about 52 percent (395,000) result in cancer death.
Cancer mortality files exclude, by definition, the non-fatal neoplasms not
vi
-------
meeting the "underlying cause of death" definition as well as deaths from
accidents, and greatly understate cancer prevalence. Project work included
development of a procedure using morbidity and mortality data together for
maintaining, by county, a "running inventory" of increases/decreases in cancer
"survivors," with recommendations for comparative analyses with emissions and
ambient data.
The Medicare files facilitate analysis of possible relationships between
emissions of a specific industry and the disease(s) rates for the co-located
population. In demonstrating the use of these files in response to a special
EPA request, it was found that average Medicare hospitalization rates for each
r
of six respiratory disease classifications are lower for the nation's 97 steel-
producting counties than for the non-steel-producing counties. Similarly, the
41 steel-producing counties of Ohio and Pennsylvania also have lower respira-
tory disease hospitalizations than the non-steel counties in these two states.
This finding was unexpected. Without further investigation, it should not be
concluded from this demonstration that steel-producing industry emissions are
less harmful than emissions of other industries, or that similar conclusions
are justified. A definitive analysis was not within the scope of this
project.
Numerous recommendations are made for applying the Medicare files to
additional analyses of environment-to-health relationships. Prominent among
these are cancer hospitalization trends, by county, augmented with cancer
mortality trends, and available emissions/monitoring measures to identify
areas of cancer increase/decrease possibly related to environmental influ-
ences .
Millions of health status files have been demonstrated to be available
for EPA use, and essentially without cost for data collection or computer
encoding.
vii
-------
CONTENTS
Foreword ill
Preface iv
Abstract • . . vi
Figures x
Tables xi
Abbreviations xiv
I. Introduction 1
II. Conclusions 3
III. Recommendations n
A. Applications 12
B. Supplementary Files-Extensions 18
C. Improvements 20
IV. Data Resources, Methods, and Findings 21
A. Data Resources 21
B. Methods and Findings 27
V. Review and Discussion 101
A. Berks County—Morbidity/Mortality
Comparison by Year 102
B. Daily Morbidity/Mortality During
Pittsburgh Episode of 1973 104
C. Morbidity in Steel-Producing Areas 106
D. Migration of Medicare Claimants 108
E. Findings and Conclusions 109
ix
-------
FIGURES
Number Page
IV-1 U.S. counties with low migration for white males of
all ages . 39
IV-2 U.S. counties with low migration of white males over
age of 44 40
IV-3 High in-migration counties of interest in Florida ... 45
IV-4 High in-migration county of interest in New Jersey ... 46
IV-5 Counties of interest in Pennsylvania 47
IV-6A Levels of employment in two Florida counties for
the years 1953, 1959, 1965, 1970 and 1974 52
IV-6B Levels of employment in five Pennsylvania counties
for the years 1953, 1959, 1965, 1970 and 1974 53
IV-7 U.S. counties in which are located manufacturing
establishments with SIC Code 3312 86
IV-8 Steel-producing counties categorized by Medicare
respiratory disease hospital patients over 64 years
of age, 1971-1973 97
IV-9 Ohio and Pennsylvania counties classified by steel
industry suspended particulate emissions and Medicare
claimant respiratory disease hospital rates 98
-------
* TABLES
Number Page
HI-1 Type of Sampling, Pollutant, and Time Period of Data for
Birmingham, Alabama 16
IV-1 Record Layout for Data Tapes from Medicare Hospital
Discharge Survey 23
IV-2 Counties with Highest White Male Mortality Rates for
all Neoplasms 25
IV-3 County Business Patterns—Florida (Pinellas County) ... 26
IV-4 Sample Point Source Listings from National Emissions
Data System for Particulate and Sulfur Oxides
Emissions (in Short Tons, 1977) 28
IV-5 Sample Display of Medicare Data Assembled Longitudinally
to Show Hospitalization Causes, 1971-1973 30
1V-6 Beneficiary Identification Codes Social Security
Administration Type of Claim 31
IV-7 Beneficiary Identification Code Railroad Board
Employee Type of Claim 32
IV-8 Patient Non-Medical Characteristics from Consolidated
Medicare Records for 1971-1973 33
IV-9 Disease Prevalence from Consolidated Medicare Records
for 1971-1973 34
IV-10 Disease Category Combinations from Consolidated Medicare
Records for 1971-1973 35
IV-11 U.S. Counties with Total Population Greater than 10,000
in Which the Ratio of In-Migration to Out-Migration
for White Males Who Were 45 Years of Age or Older in
1970 is Greater Than 3.00 41
IV-12 U.S. Counties with Total Population Greater than 10,000
in Which the Ratio of Out-Migration to In-Migration
for White Males Who Were 45 Years of Age or Older in
1970 is Greater Than 2.00 43
IV-13 County Population Characteristics 48
IV-14 Medicare Data Characteristics 49
IV-15 County Urban/Industrial Characteristics 50
xi
-------
TABLES (CONTINUED)
Number Page
IV-16 Death Counts by Neoplasm Type and Age Group for Berks
County, PA and Years 1971, 1972, and 1973 58
IV-17 Death Counts by Neoplasm Type for Persons Aged 65 or
Over in Berks County, PA 59
IV-18 Use of Medicare Data to Estimate Cancer Prevalence ... 60
IV-19 Allegheny County, PA, Residents Who Died in the County,
August 1 Through September 22, 1971-1973, with Subtotals
for Underlying Death Cause of Respiratory or
Cardiovascular Disease 62
IV-20 Daily Admissions by Disease Class in Allegheny County,
PA, During the Period August 13 Through September 20,
197] 64
IV-21 Daily Admissions by Disease Class in Allegheny County,
PA, During the Period August 13 Through September 20,
1972 65
IV-22 Daily Admissions by Disease Class in Allegheny County,
PA, During the Period August 13 Through September 20,
1973 66
IV-23 Daily Counts of Allegheny County First Admissions During
August 13-September 10, 1971, 1972, 1973 by Respiratory
Disease Category 67
IV-24 Numbers of Persons Admitted with Indicated Diagnoses
Each 13-Day Period: August 13-25, August 26-Sept. 7,
and Sept. 8-Sept. 20, 1971-1973 68
IV-25 Daily Admissions for Pneumonia and Bronchitis, Emphysema,
or Asthma in the Pittsburgh SMSA During the 39-Day
Period Including the 1973 Air Pollution Episode and
Comparable Period in 1971 and 1972 71
IV-26 Illustrative Distributions of Time Spans from First
Admission for Cancer to Last Admission, 1971-1973 ... 74
IV-27 Average Months and Rank from First Medicare Cancer to
Death of Various Causes 76
IV-28 Calculation of Median Survival Times for Cancer Patients
Age 65 and Older, and Comparison of Relative NCI and
Medicare Rankings 78
IV-29 Frequency Distribution of Numbers of (A) Counties of
Residence, (B) States of Residence, and (C) Counties
of Residence within Indicated State During 1971-1973
for Medicare Claimants Residing in Indicated State
and County 80
xii
-------
TABLES (CONTINUED)
Number Page
IV-30 List of 31 States, 97 Counties, Two Independent Cities
of the United States in Which are Located Manufacturing
Establishments with SIC Code 3312 85
IV-31 Data Entry Form 87
IV-32 Total Number of Medicare Claimants (with Corresponding
Morbidity Rates) in 97 Steel-Producing Counties and
All U.S. Counties for Hospitalizations in 1971-1973
with Diagnoses in Six Respiratory Disease Categories . . 89
IV-33 Respiratory Disease Categories for Which Medicare
Claimant Rates in "Steel-Producing" Counties are
Greater than Respective State Rates 91
xiii
-------
LIST OF ABBREVIATIONS
CO: Carbon monoxide.
EPA: Environmental Protection Agency.
HERL: Health Effects Research Laboratory.
HEW: Department of Health, Education, and Welfare.
ICDA: International Classification of Diseases, Abstracted.
NEDS: National Emissions Data System.
NO : Oxides of nitrogen.
X
OAQPS: Office of Air Quality Planning and Standards.
RAMS: Regional Air Monitoring System.
RAPS: Regional Air Pollution Study.
RSP: Respirable suspended particulates.
SAROAD: Storage and Retrieval of Aerometric Data System.
SMSA: Standard Metropolitan Statistical Area.
S02: Sulfur dioxide.
SSA: Social Security Administration.
SSI: System Sciences, Inc.
TSP: Total suspended particulates.
xiv
-------
SECTION I
INTRODUCTION
This document describes the work and findings of a pilot study investi-
gating the potential utility for epidemiologic research of data from the
Social Security Administration's Medicare short-stay hospital discharge sample
survey. The work was performed by System Sciences, Inc. under the direction
of the Health Effects Research Laboratory of the Environmental Protection
Agency in fulfillment of EPA Contract No. 68-02-2782. Certain printout repro-
ductions, tables, maps, graphs, and data source descriptive materials are
included to meet contractual report requirements for such explanatory
materials related to the various project work steps.
A prime consideration was the development of systematic procedures for
relating morbidity data from this new resource to county level mortality data,
industrial data, environmental data, and demographic data. Many of the pro-
cedures utilized the UN1VAC 1110 at EPA's National Computer Center, Research
Triangle Park, N.C. The procedures facilitate analyses of a single county,
a large number of counties, Standard Metropolitan Statistical Areas, states,
and national trends.
For pilot study purposes, particularly constraints of economy, timeli-
ness, highly confidential files, and the practical risks of unforeseeable
obstacles in applying Medicare medical histories for unprecedented purposes,
only three categories of hospital diagnoses were investigated. Any data that
might jeopardize confidentiality were deleted; only the minimum facts essential
to the pilot study were extracted from the Medicare tapes. The categories
selected are of special interest to the Environmental Protection Agency—all
respiratory diseases, neoplasms, and diseases of the digestive system. The
Medicare tapes for these three disease categories yielded about 1.2 million
-------
hospitalizations, for some 815,000 individuals. Disease definitions and codes
as provided by the Social Security Administration are in conformance with
Eighth Edition, International Classification of Diseases, Abstracted (ICDA).
The study was performed in the following steps. First, the computer
tapes with selected data fields of the Medicare files were verified and re-
formatted for computer-assisted analyses to fulfill study requirements. Then,
counts were made to determine the demographic and other characteristics of the
sample file. Next, population, migration, environmental, and industrial data
were accessed and studied to select areas for further study. Employment his-
tories were obtained for the selected counties to portray industrial devel-
opment patterns in detail.
Three in-depth investigations were performed to demonstrate morbidity/
mortality comparisons: (1) annual morbidity data for one county were examined
to determine the feasibility of obtaining, for environmental health indicator
purposes, rates of death due to selected diseases, and concurrently the accu-
mulating "inventory" of disease experience, which may or may not eventually
cause death in the surviving population of an area; (2) daily admissions
during a period centered on an air pollution episode were studied; and (3)
cancer patient survival times from the Medicare data were compared with
National Cancer Institute times as a partial test of Medicare data reliability.
Another investigation was directed to the use of Medicare patient state/
county residence at times of treatment to assess the impact of migration on
epidemiological studies.
Finally, to demonstrate the utility of the Medicare morbidity data base
for environmental health research, an in-depth analysis was performed com-
paring counties with steel-production facilities to other counties and to the
nation as a whole. For counties in Ohio and Pennsylvania, the comparison
utilized suspended particulate emissions as the measure of the relative pollu-
tion significance of steel industry sources to all "point" sources in each
county. Respiratory diseases were selected as a related measure of health
effects.
-------
SECTION II
CONCLUSIONS
This chapter summarizes findings and conclusions derived from the pilot
study investigating the potential utility for epidemiological research of the
Medicare hospitalization files. Priority and emphasis was given to the
feasibility of applications on behalf of responsibilities of the Environmental
Protection Agency (EPA), particularly airborne pollution of industrial origin.
Enough has been learned in this pilot investigation to conclude that it
is feasible to apply the data bases described herein to make major contribu-
tions to environmental health knowledge by identifying relationships (if any)
between specific industries, emissions, air quality, and selected diseases
indicative of human health status.
The different data bases can also be utilized to determine trends in
environmental health benefits (if any) from air pollution abatement programs
of the recent past—for example, since 1970.
Findings from the preliminary investigation comparing steel- and non-steel-
producing counties by rates of Medicare respiratory disease hospitalizations
suggest the strong possibility of erroneous and misleading assumptions con-
cerning co-locational relationships between certain industries, emissions, and
the available human health evidence.
Little is now known of human health benefits from environmental programs,
although Governmental actions are essentially justified on grounds of pro-
tecting or improving the public health. Government and private industrial
costs can be estimated with some reliability. Benefits are actually measured
in terms of chemical content decreases, with little or no means of estimating
-------
the human health benefits from the decrease. With the Medicare data, compari-
sons of carefully selected areas, with and without improvements in air quality,
by rates and causes of selected hospitalization diagnoses, will shed light on
the prime feature of EPA programs, and about which least is known—the human
health benefits. The next chapter deals with recommendations for further work
in this area.
The following points are believed especially germane to responsibilities
of Health Effects Research Laboratory, Office of Research and Development, EPA.
1. Medicare file contents are credible and reliable descriptors.
2. Privacy and confidentiality constraints can be respected without serious
detriment to the full value of the file.
3. The 20 percent sample selection process is random, with selectees aggre-
gating some 5,000,000 persons of the 25,000,000 eligible, and provides
coordinated local, state, and national representativeness.
4. Because there are over 16 times more hospitalizations than deaths, the
files multiply environmental health perceptiveness, and provide unique
environmental health data not available from mortality files. For
examples, the Medicare files provide the only county-state-national
coordinated source for the following:
a. Non-fatal diseases of environmental health interests, and
b. Potentially fatal cases of environmental health interest where death
intervenes from causes of lower environmental health interest (acci-
dents, strokes to lung cancer patients, etc.).
5. Time-phased histories from first Medicare hospitalization to death,
regardless of changes in patient residence, are routinely computer-
encoded. This capability is particularly relevant to cancer or other
long latency diseases.
-------
6. The hospitalization and mortality files have distinct advantages, and
should be used in mutual support, not exclusion, of one another.
7. The files may be used directly to evaluate the environmental health
status of the country generally, or to compare one area with another with
the precision of any codifiable hospitalization diagnosis, by
a. absolute numbers of selected hospitalization diagnoses
b. rates per X population of selected hospitalization diagnoses
c. trends in absolute numbers
d. trends in rates per X population
e. increasing/decreasing surviving population with selected hospitali-
zation diagnoses
f. increasing/decreasing rates of surviving X population with selected
hospitalization diagnoses.
8. The Medicare hospitalization files can be used to quantify health effects
of unusual temporary environmental conditions, such as an air pollution
episode. (Hospitalization admission dates are routinely computer encoded.)
9. The Medicare hospitalization files offer a unique capability for relating
admissions/rates of selected diagnoses to the chemical composition of the
air. This capability is believed particularly valuable for specific geo-
political areas with the most complete and reliable data, such as St. Louis.
10. Numbers and rates of hospitalizations from environmentally-related
diseases can be used to assess the co-locational relationship to indus-
tries of major environmental pollution-concern. The presence/absence of
significant co-locational health relationship has agency-wide regulatory
and administrative implications, and should be expanded to include assess-
ments of emissions, ambient data, pollutants from additional point and
area sources, and other factors as appropriate.
11. The Medicare hospitalization files are in tape format, readily trans-
ferrable to EPA1s UNIVAC 1110 at Research Triangle Park, and complexities
-------
of county coding differences, longitudinal assemblage of hospitalizations
from annual tapes, and compatibility with EPA SAROAD, NEDS and other
major data bases have been demonstrated to be relatively minor. Special
note is made of the ease and low cost with which multi-color computer-
prepared maps may be made for the United States with detail to the county
level. These maps are helpful in presenting not only the "big picture"
but also in facilitating the analysis of possible interrelationships
between industry, emissions, and selected diseases in the actual en-
vironment .
12. The full value to EPA of the Medicare files cannot yet be assessed, as
many possibilities have not been examined. For example, the classi-
fication system enables differentiation of hospitalization rates by sex
and race between patients who are "primary beneficiaries" or employees,
and those who are Medicare enrollees because of their relationship to the
primary beneficiary. Conceptually, one might thus categorize certain
diseases as environmental or occupational.
The general conclusions of this pilot study are presented below, together
with an abbreviated description of the basis for the conclusions. For de-
tails, the reader is referred to the text, particularly Section IV, Data
Resources, Methods, and Findings.
1. Credibility—The Medicare hospitalization file extracts made available to
EPA and SSI offer credible and reliable health indicators of a population
of particular environmental interest—the elderly, primarily age 65 and
over, aggregating over 25 million, and with 20 percent of these consti-
tuting the Medicare file made available.
Basis—Each of ten neoplasms, selected for definition uniformity with
National Cancer Institute records, were checked for "survival times"
(time to death from confirmed diagnosis), and death percentages attrib-
uted to the neoplasm. There was close correspondence for each of the
ten neoplasm types.
-------
2. National and Local Representativeness—The Medicare sample is large, and
the Medicare population is distributed throughout the country in such
manner that the Medicare tapes provide the potential for continuous
monitoring of this particularly sensitive population through selected,
environmentally-related diseases. This has many applications to EPA
responsibilities for evaluation of pollution sources and the health of
co-located populations. No purpose is perceived that would justify in-
creasing the 20 percent sample, except for special, localized, intensive
ad hoc investigations.
Basis—Each hospitalization is coded by discharge diagnosis, date of
admission, county/state of residence, age, race, sex, and the 5,000,000
persons in the sample reside in every county in the United States. The
20 percent selection process is random.
3. Longitudinal Health Analyses—The Medicare files can be assembled to
show, longitudinally, the Medicare hospitalization history from first
admission to death for each Medicare sample selectee, regardless of
residence changes. Assuming continued cooperation with HEW, EPA thus has
a systematic procedure for obtaining knowledge of changes in health
conditions over time, and non-fatal disease occurrences, by county of
residence at time of hospitalization.
Basis—See example printout presented in text.
4. Accumulative Disease Capabilities Facilitates Environmental Effects
Comparisons—The Medicare files can be assembled to compute the popu-
lation surviving from one or more hospitalizations of selected
environmentally-related diseases. Accumulations of patients by disease
may be by county, counties, or with special permission, by Zip Codes.
There are ten times more Zip Codes than counties. This capability is
unique. EPA applications extend from determining relationship (if any)
between pollution sources and selected diseases, to determining pri-
orities for abatement and control.
-------
Basis—See example calculations for Berks County, Pennsylvania.
5. Change in Rates and Trends of the Surviving Population with Specific
Hospitalization Causes Offers a Powerful EPA Health Analysis Tool with
Particular Applications to Cancer and the Environment—Increases/decreases
of the Medicare population surviving selected environmentally-related
disease hospitalization can be assembled in such a manner to show "small
area" trends, and accumulative status. Conceptually, the increases/
decreases may provide an index of success/failure for EPA programs/
projects justified on environmental health improvement grounds.
Basis—See Berks County, Pennsylvania, example of net annual increase of
surviving cancer cases, by type, and cumulative prevalence. (Any defined
and encoded disease(s) may be selected.)
6. Medicare Hospitalization Files Appear Ideally Suited to Quantifying
Health Effects of Unusual Environmental Conditions—Medicare hospitali-
zations offer EPA significant potential for measuring promptly the
health consequences of adverse environmental circumstances, such as an
air pollution episode. Conceptually, the ambient and hospitalization
measurements could be concurrent. Formal/legal emergency episode
determinations, and countermeasures by Federal, State, and local environ-
mental and public health authorities would be greatly assisted by an
experienced-based air quality-health linkage system.
Basis—The 1973 Allegheny County August 26 - September 7 air pollution
episode, and the days immediately prior thereto, were noteworthy because
of a significant increase in Medicare respiratory disease hospitaliza-
tions. Similarly, more mortalities from respiratory diseases and car-
diovascular diseases were recorded for people of all ages during the
episode than prior to or following the episode. See tables in text.
7. Changes in residence of Medicare hospitalization claimants are unusual.
-------
Basis—Among Medicare hospitalization claimants, 1971 through 1973, in
two Florida and five Pennsylvania counties, there were insignificant
changes in state of residence, and less than five percent residence relo-
cations among counties. Essentially all of the relocations were within
Florida.
8. The presence of a steel-producing facility does not necessarily cause
higher Medicare respiratory disease rates.
Basis—Average Medicare hospitalization rates for each of six respiratory
disease classifications are lower for the 97 U.S. steel-producing coun-
ties than for the non-steel-producing counties. Respiratory disease
rates for white males are significantly higher than for white females.
These facts are based on computer-assisted analysis of approximately
280,000 Medicare respiratory disease patients—see text and tables.
9. The relative quantities of TSP from steel-producing sources in a county
do not necessarily cause high Medicare respiratory disease rates.
Basis—The 20 Ohio and Pennsylvania counties with the lowest percentage
of steel industry TSP emissions, and the 21 counties with the highest
percentage of TSP emissions, showed an average for each of six categories
of Medicare respiratory disease hospitalizations that was lower than the
average rate for the 104 counties in these two states that did not con-
tain steel-producing facilities.
10. Analyses of relationships between carcinogenic hazardous industries,
mortalities, and Medicare neoplasm hospitalizations must include migra-
tion and other factors to avoid misleading conclusions. It is also
necessary to include more than three years of Medicare hospitalization
data to exploit fully the Medicare file source potential for contributing
to knowledge of environmental sources of cancer.
-------
Basis—The Florida counties had high rates of elderly in-migration,
essentially no carcinogenic hazardous industries, and compared with the
Pennsylvania counties studied, higher lung cancer rates, and all cancer
rates. The Pennsylvania counties had a comparatively stable population,
and some chemical, petroleum, rubber, and primary steel industry employ-
ment. Non-Florida exposures to environmental pollution are believed to
account for much of the higher Florida cancer rates.
For the counties studied, no significant carcinogenic industry source was
identified as introduced within a time span that would affect the 1971-1973
hospitalizations. To detect the impact of a newly-introduced carcinogenic
source, it is necessary to have appropriate baseline experience. Three years
of data are insufficient for both baseline and impact measurement of diseases
with long latency periods. For hospitalizations caused or exacerbated by
acute temporary changes in the environment, such as an emergency episode, the
three years of data enable many unanswered questions to be resolved.
10
-------
SECTION III
RECOMMENDATIONS
The Medicare hospitalizations file has been demonstrated in this pilot
investigation to have extensive potential for determining relationships
between the environment and human health. The contents of this constantly
growing file offer the only single source documenting consistently the extent
of fatal and non-fatal diseases related to the environment. From a health
policy perspective, the contents offer the possibility of determining the
human health improvements resulting from the billions of dollars expended, and
planned for expenditure. From a technical perspective, the Medicare file
contents may also provide guidance as to the human health benefits, if any,
resulting from different control and abatement technologies as applied to
improve air quality.
The purpose of this section is to present recommendations responsive to
the project's scope of work requirement for follow-up steps appropriate to the
findings and conclusions of the pilot investigation. The recommendations are
to include, but are not limited to, use of the Medicare hospitalization data
for determining carcinogenic health effects as early as feasible. Recommen-
dations are presented under three broad categories, as follows:
(1) Applications—Apply the Medicare hospitalizations file to analyses/
products of current HERL and EPA interest;
(2) Supplementary files, new data, and pilot investigations—Extend
and/or expand the Medicare pilot investigation files to obtain
knowledge of different relationships between the non-industrial or
background environment and human health, between different cate-
gories of urbanized-industrial areas and the health of the colocated
11
-------
population, and explore the feasibility of alternative techniques
for utilizing supplementary data files collected for non-EPA pur-
poses on behalf of HERL and EPA responsibilities; and
(3) Imp rovements—Verify, re-format, and otherwise improve the reli-
ability, accuracy, and utility of the 1971-1973 computerized Medi-
care files presently in hand, and others on order.
The recommendations listed for follow-up to this pilot investigation do
not purport to be complete. Several are closely related to one another.
Collectively, the recommendations present extensive and intensive steps now
feasible to be taken by EPA in bridging the knowledge gap between evidence of
human health and the measureable components of the air being regulated, and
controlled by EPA.
In preparing these recommendations, it has been assumed that readers have
some technical familiarity with the contents and limitations of EPA infor-
mation-acquisition systems. An understanding of certain recommendations will
be helped by a review of the following chapter, which describes the data
resources utilized in this pilot investigation. Other recommendations are
based on the demonstrated credibility and reliability of the Medicare hos-
pitalizations data, as presented briefly in the preceding section. These
additional recommendations are directed to applications not attempted or not
considered seriously in the absence of the pilot investigation findings.
A. APPLICATIONS
1. Investigate Alternative Methodologies for Identifying Areas of
Increasing Cancer Prevalence and Recommend a Methodology for Use
by EPA
Integrate cancer mortality and Medicare cancer hospitalization files
for selected areas, and feature trend analyses in the investigation
and comparison. (Note: The widely publicized NCI "Cancer Atlas" is
based on accumulated deaths over 20 years, and does not distinguish
12
-------
between counties with increasing/decreasing deaths or death rates.)
Identify counties with increasing trend rates believed especially
significant, compare with ambient/emissions data, and discuss pos-
sible causes. Emphasis is to be placed on a method for identifying
areas with increasing cancer, using ongoing local-state-Federal
information-acquisition procedures.
2. Utilize Medicare Morbidity Data to Measure the Human Health Effects
of Air Pollution Episodes
Using the methodology developed for analysis of the Pittsburgh air
pollution episode, determine the relationships between (a) other
episodes and/or selected significant temporary changes in the en-
vironment and (b) increases in hospitalizations for selected dis-
eases. The Pittsburgh methodology is to be expanded to include
emissions and ambient monitoring where data permit. Where feasible,
include daily mortality by cause.
3. Determine the Existence of a Relationship Between Improving Air
Quality and Improving Environmental Health
Select approximately ten counties with air quality that has improved
significantly over the last decade, as measured by one or more
pollutants to be selected in consultation with EPA. Select an equal
number of counties, matched insofar as feasible, in all respects
except for improved air quality. Select an equal number of counties
on the basis of deteriorating air quality, as measured by specific
pollutants. Determine hospitalizations and rates of environmentally
related diseases for each set of counties. Analyze and discuss
findings.
4. Determine the Existence of a Relationship Between Deteriorating
Air Quality and Environmental Health
13
-------
This application focuses on a relationship hypothesized as opposite
to the preceding investigation. The work steps are essentially
identical in concept.
5. Relationships Between Medicare Hospitalizations, Mortality and
Specific Pollutants (TSP, S02>
(RSP), etcQ for Selected Areas
Specific Pollutants (TSP, SO^, Respirable Suspended Particulates
Compare/contrast selected Medicare hospitalization diagnoses by
county, with emissions/monitoring data for the same areas. Include
provision for coordination with National Air Pollution Index moni-
toring sources/information. Investigate possibilities of deter-
mining threshold existence by comparisons of areas of poor air
quality (exceeding standards for specific pollutants) and areas of
good air quality.
a. Analyze the extensive pollutant monitoring information obtained
in the multi-million dollar St. Louis project with respiratory
disease hospitalizations for each county covered by the St.
Louis area RAMS network. Perform a time series analysis using
these data and the fine and coarse particulate fractions for
elements of interest (sulfur, lead, total mass, etc.). These
size fractions are readily available from the automatic dichot-
omous sampler data with which SSI is already working. Other
readily available factors which can be taken into consideration
include hourly measurements of temperature, wind speed and
direction, precipitation, etc., from the RAPS data base.
b. Explore relationships between human health thresholds, and one
or more pollutants. In consultation with the EPA Project
Officer, prepare supplementary tabulations for unusual health
status occurrences to determine the existence of a minimum
level of a particular pollutant or minimum mixes of certain
pollutants. Also, for selected areas/times with high hos-
pitalization rates, investigate further the hypothesis that
increased morbidity rates precede elevated mortality rates.
14
-------
(1) For one selected high steel-production county (Pittsburgh,
Allegheny; Birmingham, Jefferson) chart and analyze the
correspondence between daily TSP/SO-/RSP levels and hos-
pital admissions rates for Medicare respiratory disease
admissions for the same year(s). Particular attention
will be paid to selected periods to be defined in con-
sultation with EPA as pollution episodes. Outputs: Time-
series plot, Julian day, and brief analysis.
(2) For Jefferson County (Birmingham), extract daily Medicare
hospital admissions for respiratory diseases, for the
years 1971 through 1973. For the same time period, ex-
tract TSP, RSP, and/or other data in size specifications
and sampling types to be determined in consultation with
EPA. See Table III-l for type of sampling, and particle
sizes. Note data date availabilities. Outputs: Julian
day time series plots, and brief analyses comparing air
quality with Medicare hospitalizations.
(3) Compare/contrast Birmingham respiratory disease admissions
with other selected areas (not necessarily characterized
by steel production concentrations) for which ambient
sampling data are available. See Table III-l for types of
sampling, pollutant, areas and time periods for which the
data sampling was conducted. Detailed specifications to
be developed in consultation with EPA project officer.
Outputs: Time series plots, tabular summaries and statis-
tical and interpretative analyses.
6. Relationships Between Medicare Hospitalizations, Mortality, and
Air Quality
a. Mortality and Medicare Morbidity Comparisons—Select disease
categories believed most related to environmental conditions,
and rank the counties in the United States by absolute numbers
15
-------
Table III-l
TYPE OF SAMPLING, POLLUTANT,
AND TIME PERIOD OF DATA FOR BIRMINGHAM, ALABAMA
Type of
Sampling
Hi Vol TSP
0.01 - 50 ym
Cassett
0.01 - 26 ym
HASL Cyclone
AEC Curve
0% @ >10 ym
50% @ 3.5 ym
100% <§ < 2 ym
Multistage
Andersen Ranges
0.01 - .93 ym
.93 - 1.75 ym
1.75-2.4 ym
2.4 - 5.5 ym
5.5 - 50 ym
CHAMP RSP
Andersen
0.01 - 3.5 ym
3.5 - 26 ym
Type of
Analysis
TSP, N0~, SO^
RSP
RSP
RSP
RSP, RNO~ RSO^T
NRSP
Birmingham,
Alabama
10/69-9/76
6/71-9/76
6/71-9/76
4/72-2/73
12/75-9/76
16
-------
of Medicare patients, total hospitalizations, and rates per
resident population over 64 years of age. Compare with mor-
tality data. The disease categories selected are recommended
to be identical with one or more of the 56 categories for which
SSI has already calculated age-adjusted mortality rates, by
county for the approximately 10 million deaths occurring in
1968-1972.
b. Prepare an Environmental Health and Pollution Atlas of Basic
Facts—
(1) Apply the multi-color county-based computer-mapping pro-
gram to prepare a series of U.S. maps identifying envi-
ronmental disease hospitalization prevalence (respiratory
diseases, lung cancer, all neoplasms, etc.) by quintiles
(or other classification), the presence of environmentally
significant industries, fuel consumption by type/quantity,
and other factors. Coordinate the "atlas" content with
the maps and annual updating procedures of OAQPS in such
manner to facilitate routine updating of environmental
health status, trends, and changes.
(2) Apply the computer-mapping procedure to include ambient
measures of air quality. Discuss findings, with special
attention to counties with highest hospitalizations and/or
highest level of pollutants.
(3) Analyze demographic characteristics of high/low counties
to rationalize differences, and "map" findings.
7. Explore Relationships Between Sources and Types and Hospitalizations/
Mortality Health Indicators
Select industries of pollutant significance, or critical source
types within one or more industries and compare the co-locational
17
-------
significance of the emissions from these sources in terms of hos-
pitalizations and mortalities. Compare findings with areas com-
parable except for the absence of the industry or critical source
type.
B. SUPPLEMENTARY FILES - EXTENSIONS
1. New Data
a. Obtain Medicare hospitalization files on all diseases, and more
current years (the present tapes made available by HEW for
1971-1973 have only neoplasms, respiratory diseases, and
diseases of the digestive system).
b. Obtain the non-confidential Social Security Administration
computerized file showing the continuous work history, annually
since 1957, by employer and county, of a one percent sample
(about 1.5 million total workers).
2. Linkages Between Industrial/Environmental Exposures and Hospitalization/
Death Diagnoses
Develop linkages, to the extent possible, between the continuous
work history file and the Medicare morbidity file. It is feasible
technically and at insignificant costs for both Medicare patients
and the continuous work histories to be identified by the same
algorithm to yield a unique, non-confidential, non-Social Security
identification number. The possibilities are now being explored
in HEW; complications are the novelty of the concept, and assuring
compliance with bureaucratic precautions for safeguarding
confidentiality.
Investigate the feasibility of utilizing the continuous work history
file to improve knowledge of environmental exposures by industry and
18
-------
county, and relate industrial years of employment exposures to
causes of hospitalization/death under Medicare.
a. A pilot investigation may build upon the project just finished
1
by selecting known Florida counties with high in-migration
Medicare hospitalizatlon/death cases, and determining the
working-life industry/locational history of individuals
experiencing selected diseases such as all cancers or specific
types of neoplasms.
b. Concurrently, another pilot investigation could select one or
more carcinogenic hazardous industries, and for each industry
select employees of one or more pre-determined age groups as of
1957. The age groups should be such as to expect Medicare
hospitalizations and/or death to have occurred to a significant
number of selectees since 1957, and beginning about 1970.
Conceptually, the two investigations are as follows:
(a)
(b)
For Medicare hospitalizations/
deaths from selected causes
of Florida Medicare residents,
For Employees of Carcinogenic
Hazardous Industries of Appro-
priate Ages in 1957,
Determine Industrial and
locational (State/County) work
history.
Determine Medicare hospitali-
zation/death causes, and migra-
tion location (State/County).
3. Death/Disability Rates by Industry/Employment - Indicators of
Environmental Pollution
Using the continuous work history file, investigate death/disability
rates as indicators of environmental health problems and effects
among employees in industries known to be sources of environmental
19
-------
pollution. Include the identification of the ten (10) counties with
highest and lowest death/disability rates. Relate to air quality
for the counties identified. Compare rates of selected Medicare
hospitalization diagnoses for the high/low counties with (a) air
quality and (b) death/disability rates.
C. IMPROVEMENTS TO MEDICARE FILES IN HAND
1. Verify the Medicare hospitalizations by comparisons with morbidity
files of recognized health research authorities—for example,
supplement the Medicare neoplasm hospitalizations-National Cancer
Institute comparison with a similar comparison of respiratory
diseases with National Heart and Lung Institute files, if possible.
2. Update, and make consistent the county codes identifying Standard
Metropolitan Statistical Areas to achieve compatibility with NEDS
and SAROAD.
3. Identify the advantages, and difficulties, of obtaining complete
consistency between Air Quality Maintenance Areas and Medicare
hospitalizations occurring to residents of these areas.
4. Develop a transfer code enabling interchangeability of the Julian
date of Medicare hospitalization with the more commonly used EPA
system of month, day, year.
5. Sort tape by State, County, Claim Number, Admission Date to give
chronological history of claimants within a state and county.
6. Sort tape by SMSA, State, County, Claim Number, Admission Date to
give same chronological history as above except by SMSA.
7. Run maps similar to previous maps showing morbidity rates and other
data in terms of SMSA's.
20
-------
SECTION IV
DATA RESOURCES, METHODS, AND FINDINGS
This section is presented in two parts. The first part is comparatively
brief, and describes the major resources consulted and the types of data
assembled. The second part describes the tasks performed in this pilot
study, together with observations and commentary concerning the usefulness of
task products.
A. DATA RESOURCES
A major objective of this pilot study was to determine the environmental
health value of certain data banks, developed for other purposes, by demon-
strations with EPA's computer configuration and its associated data bases.
"New" data was not to be collected. Instead, emphasis was placed on exploit-
ing existing information sources, preferably computer-encoded.
The following description covers the major data files used in this pro-
ject. Included are example printouts. Incidental sources of information are
included as appropriate in the text discussing a specific demonstration or
application.
1. Medicare Files
The approximately 25 million Medicare beneficiaries who are provided
medical treatment under Social Security represent the largest population group
for which morbidity data are collected regularly and compiled systematically
nation-wide by a single program. This group includes essentially 100 percent
of the population 65 years of age and older. Collectively, this group experi-
ences about 5.5 million hospital admissions annually. They account for about
half the hospital cancer cases in the United States.
21
-------
System Sciences, Inc. (SSI) obtained tape reels from the Social Security
Administration containing data from the 1971, 1972, and 1973 Medicare short-
stay hospital sample discharge survey. Table IV-1 shows the record layout for
each of the tapes. The State and County codes of field numbers 4 and 5 per-
tain to the place of residence of the claim beneficiary. Records for which
the discharge diagnosis is "neoplasm" (ICDA 140-239), "disease of the res-
piratory system" (ICDA 460-519), or "disease of the digestive system" (ICDA
520-577) are included in these files; records with other codes are available
from the Social Security Administration, but were not deemed necessary for
purposes of the pilot study.
There are approximately 400,000 admission/discharge records for each of
the three years. This represents a 20 percent sample of qualifying claims and
includes multiple records for some individuals. The Social Security numbers
are scrambled in accordance with a single, unique algorithm to permit a longi-
tudinal matching of individual claim histories, by means other than use of the
Social Security number. This provision enables assignment to each Medicare
beneficiary of a unique alphanumeric identifier, but with the individual's
privacy and identity fully protected. The Medicare sample selectee procedure
is "permanent"; that is, once in the sample, the selectee's Medicare history
is routinely up-dated until death through the claims/insurance/payment pro-
cess. Also, the longitudinal file starts with the first Medicare claim for
each individual.
2. POPATRISK Data Base
&
A data base, called "POPATRISK," was developed by SSI to provide the EPA
with a user-oriented data base containing recent county-based information for
all counties in the contiguous United States, including population demographics,
population mobility, climatology, emissions, air quality, and age-adjusted
death rates.
Population at Risk to Various Air Pollution Exposures: Data Base "POPATRISK";
Contract No. 68-02-2269, for Health Effects Research Laboratory, EPA/Research
Triangle Park, North Carolina. EPA-600/1-78-051, June 1978.
22
-------
>
H
,0
td
H
rrj
rH
0)
fa
O
co
4J
ct
CU
4J
c
O
o
CO T)
C C
o w
•H
1 i
CO 60
O CU
tM FQ
0)
ft
•O H
rH
CU >-3
•H
fa 0)
N
•H
C/)
Q)
B
td
S3
rU
rH
cu
•H
fa
p
CU §
rH O
cu eg a
•H E Ai
ed cu C
a fe a
1 1 1
H CN en
H rH
rH
O
•H
M
CU
B
3
a
td
ft
rH
td
rH
^_l
CU
n
c
a
•H
rH CU
CJ CO
. .
H CN
cu cu
o o
goo
O O rS r^i
M C 4J 4J
cu d 3 3
a 3 o o
o o
T3 T3
7.1 s s
QJ M CU CU
4J CU 4-1 4-1
•H .C cd ed
,C 4J 4-1 4J
^O W CO
II
rH
cd
1
o
CM
rH
co
3
4-1
cd
4-1
W
cu
60
J_J
cd
f"{
o
CO
•r-l
p
.
^
cu
P!
0
a
1
o
r— I
a
4J
cu
M
o
B
1
rH
co
CN
rH
rH
Cd CO
C cu
o w
•H 0
4J C
•H 60
T3 cd
-d -H
< p
.
co
-
T
^b
vO
XT
tJ
o%
en
CN
1
0
-3"
rH
CO
cu
tl
cS
*~^
p
M
CN
en
o>
CM
"*
cu co
60 0)
V-l CO
ed O
X! C
O 60
co cd
•H -H
p p
.
CT.
1
§
cu
1 TJ
0
CTi O
CT\
CT\ <3}
•" S
<; to
S 4-1
CO -H
60
a -H
•H -rj
1
4J en
o
1 3
4-1
AS 0
£3 CtJ
cd
rH M
« o
m
en
en
en
en
1-1
cd
•H
o
•H
n_i
cu
1
1
-------
"POPATRISK" contains approximately 27.5 million characters and is in
SYSTEM 2000, Version 2.80 format, facilitating access with minimal user com-
puter training. Population demographics are as of the 1970 Census; population
mobility is described spanning the years 1965 to 1970 for six sex-race cate-
gories in seven age groupings for both "in" and "out" migrants; climatology
information contains county summaries of temperature, precipitation and hours
of sunshine; county point and area source emission estimates are provided for
five criteria pollutants—TSP, SO-, NO-, CO, and ozone—based on the NEDS-USER
file; air quality information is based on 1974 data contained in SAROAD; age-
adjusted death rates were computed for the combined years 1969, 1970, and 1971
for four sex-race categories in 50 groupings of ICDA categories (eighth revision),
3. "Industrial Correlates" Project Statistics
Under contract with the EPA Office of Toxic Substances, SSI systemati-
cally related mortality rates of specified disease categories to certain
aspects of the environment with the primary focus being upon the relationship
between community proximity to industrial effluents/emissions and the various
mortality rates. One of the products of that research effort was a ranking of
the white male and white female age-adjusted death rates (per 100,000; ages
34-75) in all the counties of the United States. A "hard copy" listing of
the top 500 counties in each disease category was also prepared. An example
ranking of the top U.S. counties by white male mortality rates for disease
category 04 (all neoplasms—ICDA 140-239) is displayed in Table IV-2.
4. County Business Patterns
A primary data resource for documenting industrial development in the
study area was County Business Patterns, published by the U.S. Department of
Commerce and based on Social Security Administration reports of first-quarter
earnings. Industry sectors are classified by 2-, 3-, and 4-digit Standard
Industrial Classification codes. A sample page of data for Pinellas County,
Florida, 1975, is shown in Table IV-3.
24
-------
Ta1,'.- IV-2
COUNTIES WITH HIGHEST V.'HITh MALE MOKfALITY RATLS FOR ALL NEOPLASMS (ICDA 140-239)
(Ages 34-75, Races per 100,000)
RANK
RATP
ST-cnu NAM?
001.
002.
003.
004.
005.
006.
007.'
008.
009.
010.
Oil.
012.
013.
014.
015.
016.
017.
013.
019.
020.
021.
022.
023.
024.
025.
026.
027.
023.
029.
030.
031.
032.
033.
034.
035.
036.
037.
033.
039.
040.
041.
04.?.
043.
044.
045.
046.
047.
043.
049.
0^0.
1310
852
7°0
770
767
721
697
654
653
652
622
622
61C
608
500
574
564
561
555
549
548
546
534
533
530
523
520
520
517
509
5C9
507
506
499
496
405
493
492
Vo
4? 6
405
401
430
471
470
478
47fr
476
47^
47S
55-070
06-00^
00-079
30-069
3fl-OQ7
12-129
7 5-0 19
13-103
13-049
12-035
16-025
26-079
33-085
13-061
13-253
48-311
46-075
32-0?7
12-041
51-043
01-1 13
22-OT7
5^-065
45-065
28-161
51-127
51-137
03-053
13-02^
51-131
M-131
24-037
28-103
13-007
17-003
30-037
24-107
17-153
32-01 5
20-021
13-237
48-31 3
51-053
23-033
40-115
22-101
13-171
39-121
2 «-0! 1
21-1 °3
MEN?MriNFF, wis
ALPJNF, C\L
"lNr?AL, CTL
PFT'* fLFIiMt '•'T'-i
3 I LI INGS, N D
WAKIJLLA, FLA
M A N'T U C < t-' T , M A S
LONG , GA
CHA"LTO\, GA
C| Ar, LFP , FLA
C A !•' A S , IT A
KALK ASK 4, MIC
SI O'lX, M D
CLAY, G\
S 6 HI NDLF, GA
•'C "IULLF.N, TFX
JpMP5t C T
P c D 5 H r \G T NCV
GILCH?IST, FLA
r.LA°KF, VA
r>USS?LL, iLA
ST. Br.R^AKD, LA
MPRCA^^ V) V
MCCH";MICK, S C
YAL^P'ISHA, MIS
\'FV> Kr\'TT V\
DRj5"Gr, v,i
MTSDAL.", C3L
TUY1N, GA
SUP' Y, V^
•NfiRT HA'lf T"1^, V A
ST. ^49 YS , vr)
NCXUqf-F, 'MS
T4KPR, <~. 4
il F/A'inF 7 , ILL
GOLD EM V A! 1 CY, M PN
"ALTI'.'n^c CITY, '10
PUL-ISKI, [LI.
^_|«^^;'1CO) N|^V
rwFC"-iKrTt KAN
0 LI T M A M , r, ;
MAD I SPV, TFX
n i \i'..< I "'ii = , VA
°1E SOT 0, '-US
IT TAW A , MK'L
ST. M A Q Y , I. \
L A N I I'l, G A
•VOPI.F, OH
ypLrVAR, ^ jc
orc^Y, KY
25
-------
112
Table IV-3. SAMPLE PAGE
COUNTY BUSINESS PATTERNS — FLORIDA
Table 2. Counties—Employees, Payroll, and Establishments, by Industry: 1975-
-Continued
IK
M4
X41
M4J
344]
1444
3*4*
3411
347
3471
W»
34W
3S
353
3i3$
344
3542
344*
3S4S/
355
3SSI
IS*
3SM
35»
1541
X
XI
3«I7
36S
34J6
3671
3732
3*41
3»
I»
»4
3*49
39*
39S1
399
3993
41
411
41(1
41 It
412
413
42
4J1
44
44C
44M
47
4/J2
48
403
419
4«J
491
—
FINCLLAS— Commuen
Fub'icxcd sfwciu'ai n.ei«i
Uet.il door* >a*n nnd Irirn
fabricated p'ais work (boiler mops)
Pre(jt>(tcjleo meial buiijirxjs
Screw machine prcducts bo Hi etc
Set** machine prnductt
Uetai scmcr* dec
PI itinfl ltd pQliiiHing
Mi*c 'jbf>c«lcd mcinl product*
FMirieared meial cruducts ncc
Uachineiy *»c«pi e'cctntat
Coniiruclion and rrUtitd tnachinory
Conveyor* and convcymi} equipment
U.ichme loo's rrietai loirnmg *vpcs
S&i-riai dtci too'* jigs 4 iixturcs
MjCtiifO tool «icc<***ori««
M'tc machinery riccpt c'cct'ica'
CiecitK and electronic •qu-pTienr
cSHE^=.__,,
CLtlromc -o^.to,!
Sh-p building 3nd >ennq
Sorg.C*' IrtO -ncaic^l ^n«iiiur-cnt»
OoMhJln,,c ,oo«,
pS-"'^"~"™«'-"">»«''«
Mrtcellaneou* n-unuljcture*
Snjn* *nfl JOvo'tiiinq display*
Truckm9 and •artjm.j-i.n-j
Truck'ng local arid long dnianc*
Water (.ansporrnnon ;«rr,ce»
Arrangemeni cl i» jnjoorutiort
Numbar of
Match 1 2
10&0
1S2
(C)
MB
(C)
1S9
203
2 DOS
(Cl
46)
S44
to
310
(0)
272
40)
(Cl
(6)
443
448
8197
(1)
(t)
(Cl
) 41$
IE)
(fl
479
It!
(81
407
106
1C]
146
i?9
A 728
669
298
(C)
(C)
-------
5. NEDS Emissions Inventory
One project task required identification of counties with steel-production
facilities and determination of the fraction of point source emissions attrib-
utable to these sources. Retrievals were made from EPA's National Emissions
Data System for information about individual SIC 3312 point-source emissions
of Total Suspended Particulate and Sulfur Oxides. An illustration of the
data obtained is shown in Table IV-4.
B. METHODS AND FINDINGS
This second and major part of the Data Resources, Methods and Findings
section describes the tasks performed in this pilot study and principal
results or findings, as well as appropriate observations and comparisons
useful in assessing the utility of the task products. A discussion of initial
data processing is presented, as well as the general characteristics of the
sample population, and a description of the methodology for choosing study
areas and documenting their industrial development. The section concludes
with descriptions of three investigations undertaken in this project; (1) a
comparison of the availability and uses of morbidity and mortality data, (2)
evaluation of the significance of Medicare claimant migration, and (3) a
demonstration of morbidity/mortality analyses in steel industry counties.
The latter demonstration features computer-prepared multi-color United
States maps displaying, by county, rates of Medicare hospitalizations for
respiratory disease together with the relative importance of the steel
industry as a source of TSP emissions. This inexpensive technique not only
facilitates analyses, but is extremely effective in communicating the com-
plexities of environmental health interactions.
1. Initial Processing of the Medicare Data Files
The tape reels containing data from the 1971, 1972, and 1973 Medicare
short-stay hospital sample discharge survey were received from the Social
Security Administration by the end of the first month of the contract perfor-
mance period. There was a single tape for each of the three years. The
program documentation accompanying the files indicated 373,877, 408,204,
27
-------
1 ,
1 <
o
f
i
i
1
H- <
ee
o.
i
5P i
H r-» i
co c> :
f •-!
CO
, - ' K
HZ 1 *
So i •
S H :
1
CO g ; [
O O '
l-l I
cn co J
HZ °
S ^ ! a
, u
d 2 • ' a
So u
o M ;
H CO L
< '£ c
"T z «
> 33
^ 3-^1
O E* a i
H *-* 1
03 •/! ><;
to o o
H Z
H — 1 '
J 3 I
CO
CJ Q
B! Z
0 f
EH i-
£5 , J
SB , ;
p* H
E-i .
32* c
Oj CU *
< C4 -
"2 i
| 3
, «
1 <•
i
1 L
! L
I *
1.
,
! 4
!
( •
j c
1 «
i '
3 i
1
1 !
g
i
... 1
r~ '. \
J •* M <0
- o
K • •• I
• • • h™ P*
M O Z o
t. cc. K. o
L O O 1
0 a. cj
• ui O
L Of. '• 1
: u. ! ** ""
i o OH
: '• t
u a: . ft cj
S Ul XI
> > n
U. C*4
U_ »-»
ui r'**
">
u
0
I/)
•
4 i 1. 1
o cj
* <-> a
c <
f Ul — 1
L ^ •<
< CO
S 0 "- vi
r •• is z _j
J -O cj r *-
»i "-• Q. 01
• - UJ
c cj — H-
« 0 1
> z ui ae
J F- —1 -»
L. !•* O CM
J _J CC *—
o >-
O « Q. S >-
1 Z O O CD
ft -a £r cj
» ••* h- Ui
i o uj o i-
i 2 >. 2 <
k H-< £
* • 1 »« >-<
D tr> 'O a un
5 -I O O Ul
1
(VI
1
1
» 1
1
1
1
^
> ir>
UJ «-f fO c
^ S
< ..
> .. i- r
H-I D Z (
ct (r H- ,
a. o o
cj a. c
a. c:
T U
1/1 O O
cc •
m Qi rt t
2 < (
3! Ul *
O >• II
U (
u, «
U '
1
c
i
i
CM I
o c
" CJ .
a: «
< Ul
a. ^
i t
o o >- .
3- •• CE Z .
3- -O cj *- '
f*> ""^ A.
V CM I i- i
ee cj o f
< 0 .
IS £ Ul .
-J »- -J
Ul H-r O <
ui _i ce •
k- O !—
cn < a. z: •
z o o c
l/l < tt CJ
I-1 )—
3 O ui O >
Z i. 2
O IT -0 0
C •< O O I
,
D L
3 I-
1 •
>» :
3 II
3 C
e
3 .
i e
i
1 u
C
J i
j ;
n :
<
•^
^
i
i
i
j
n
u r
j
2 f
I C
J •
a C
Q
f\ C
_J :
•*
£
u :
— c
« *
C t
^
•v. I
- L
h
— t
C
jj
— ^
4
C
-, C
/I
i/
J
o
J f-«
V
I
> • •
- o
Y, X.
U 0
• Ul
L OL
r u
n o
K.
U CE
2 <
Z UJ
3 >-
J
O
- O
c
< U,'
L ^
•) J
^ c
r •• c
T 0 V
>* oa '.
£ c
f
J *•
V f
4->
- C
/I < £
- 2 <
/l < i
^ ^
3 O L
• i— «
^ \n
? "
-o
K1 c
C
• •
1- f
Z C
t-i C
0
a. c
0 i
•
K i
c
L
n
u. c
u. •
u '
f
c.
u
L
C
c
3 >- I.
S 2 .
J S' >
- a :
~ *-H |
J 3 >•
O •
r ui s
^
- -J •
-, 0 c
J CC >
D *—
J. 7-. >
3 O t
£ CJ
- t
ij O h
S. S
C 0 fc
.7) O L
<
U
f
0 U
3 I-
1 <
•• ;
D *-
5 Q
a
^
3 •
a
* -j
a
j ;
n 2
c
Nj
M
H
r\
1
J
^
1
J O
J
r »-
t a
j «
* a
0
n c
j -
-» ^
c
ij >
Q
4 <
E L.
-
\ L,
* L
I-
"• V
a
w
u
^ ••
«
E «
-< C
/I C
u c
1
1
> ; O
1
1
i '
! i
i ! f i
1
! ;
!w i *
o r-
j r* fO CO UJ
0 H.
r •• I <
» • * H" f*» ^b
< D 20 _
c ce •— o cz
. o o l a.
o a. CM i
» Ul O •*
L ce 10.
: u. z
1 O OH L/>
: • cc
j a: M' o ui
: < u z
: ui i/i -*
,» 1. Cj)
u.
* cj a >-^
: < ct
i ui ; -J — a-
N"> ^~^ CL Oi
. CM Z — U J.
c cj o >- a:
r O « «,
3 ^ Ul E O
i i 1 ~ j
^ t-< o r\j u.
J _) CE — u.
o >— >-
1 . -y>
. Z 0 0 CC I
>) « C£ CJ vO
•- I- Ul
3 O Ul O I— 3
•7 5^ J? ^
. i- E ..
„ .,«_•« ^
j IP o o i/i o
3 — o o ui a
i
i
'
! i
t j
t !
I I
I
I ;
> 00
•* '• m gg
• • ' h* ^
0.20
Of t- O
0 0 I
cj a. CM
U) O
ce ' i
u
o on
. •
or o o
< CJ
UJ ' ^>
>• i n
U CM
U. •"
w fn
' PO
u
! cj
i/l
_
Ul
O • 0
Ul ' _l
< ' CD
J
O 1— JO
• • LO 2 _J
-o U s: —
fO ^-* CL »i:
CJ X P-. Ul
CJ D f—
O <
2 ui x:
>—_)-»
^-i O CV*
_J CC —
O r-
1 Q. 2 >
Z O O CE
— i 21
.. >. -,
* * r*- o ^«-
u*> *C O v/l
-> O 0 Ul
1
28
-------
and 414,434 records for the respective years. These numbers were indepen-
dently verified as identical with the 20 percent sample of records discussed
in this report.
A computer account was established at the EPA National Computation
Center in Research Triangle Park, and the Medicare data tapes were registered
there. The data were copied onto system tapes, and the original reels were
released to SSI. An initial reading of the tapes assured the readability and
appropriate formatting of the various data-field items.
To allow statistical computations of sample characteristics on a per-
patient, as well as a per-record, basis, and to develop a chronology for
individual claimants regarding diagnosis, discharge status, beneficiary resi-
dence, etc., the three years of Medicare records were merged and sorted by
claim number. A sample of the longitudinally merged and sorted records,
grouped by claim numbers, is shown in Table IV-5. Each line in the table is
one record, representing a single admission and discharge. The claim number
consists of a twelve-digit number obtained by scrambling the Social Security
number under which a claim was filed, followed by a one-or-two-place suffix
identifying the beneficiary involved. Beneficiary identification codes are
listed in Tables IV-6 and IV-7. Thus a unique claim number represents a
specific individual. The records in Table IV-5 are further ordered by date of
admission to provide a chronological history for each individual, for the
years 1971, 1972, and 1973.
2. Sample Characteristics
Following creation of the longitudinal file, counts were taken of the
items shown in Tables IV-8, IV-9, and IV-10. Table IV-8 indicates that the
consolidated data file contains about 1.2 million admission/discharge records
(representing a sum of records for the three years) filed under 815,000
unique claim numbers. Although there may be instances of individuals filing
for benefits under more than one Social Security number, one would expect the
proportion of such cases to be negligible and that 4,075,000 (five times
815,000) is a good approximation to the number of individuals in the United
States treated as in-patients under Medicare provisions. There were 1,196,000
29
-------
Jl
' 1
CO
CT*
rH
rH
r-.
CTi
rH
^
CO
W
CO
CJ
53
O
M
H
t*3
M
^c
H
M
P-i
CO
o
^
O
PC
eo
O
EH
S_)
|J
^r
55
M
5
EH
M
O
o
^
p
. "j
pq
5
w
CO
CO
^j
H
P
M
2
§
M
w
yj
^4
hJ
Cu
p
W
j_J
P_I
JS
^
Cfl
•
m
I
£>
H
W
r4
H
0*
60
^
CO
CO
at co
60 -H
M CO
CO O
X! C
CJ 60
CO cfl
tH iH
P P
T3
rH CO
Cfl -H
G CO
O O
•H C
4J 60
•H Cfl
*UJ -rl
-o P
0)U
60 CO
rl- 3
Cfl 4-1
JS cfl
CJ 4-1
CO Cfl
•H
P
01
4-1 G
cfl O
P -H
CO
C co
Cfl -H
•H B
rH •§
3 <3
^i
4J
G
O
o
0)
4-1
CO
4J
Cfl
^rt
01
3
Cfl
X
fll
W
eo
M
0)
XI
3
53
B
•H
Cfl
rH
O
CO CO
O O
X X
CO CO
CTN CT*
- r-
o o o o
^rf b^ b^ i
CO CO CO CM
CTN o as vo
-•a1 *^ *sj~ in
O O rH rH
O O O O
S3- CM rH •<}•
00 vO si" vO
rH CO O CO
rH rH CM CM
r~^ r-^ r^ r*-»
O O O O
in in m in
rH i-H rH rH
m m in in
0 O 0 O
CM CM CM CM
*""3 r"5 *-"} *"*3
co co co co
CT* O"* CT* 0s
vO vO vO VO
CM CM CM CM
O O O O
m m m in
CM CM CM CM
r*- r-^ ^^ t^
o o o o
o o o o
o o o o
o o o o
m
r-
o
X
rH
O>
rH
i-H
CO
oo
rH
CO
r-»
o
m
rH
m
o
, t
^1
CM
h-j
CO
0s
vO
CM
O
in
CM
r^
0
o
o
o
m m
vO vO
O O
vO vO
CM C*l
m m
i-H rH
CM CM
VD vO
rH rH
rH rH
0 0
m oo
o o
CO CO
CO CO
r** r*-»
0 0
rH rH
CO CO
in m
o o
^ ^
O"\ OS
00 00
px. f^.
rH rH
-a- -a-
r^ r*^
O O
o o
o o
o o
m in m
vO vO vO
O O O
vO vO vO
CM CM CM
in in m
CM CM CM
O"* CT* CT\
rH rH rH
in in m
O I-H rH
000
r-. c-« o
m in cr>
i-H rH rH
CO CO co
r- f~- r^
o o o
oo oo oo
sj- *<}" CT\
0 O
r*^ f-«*
ON O\
LO LH
*d~ **d~
i"1** r^
0 O
o o
o o
0 0
Ol
C
o
C
cfl
frj
4-1
01
J»j
o
*"O E
cfl II
Q) rH
TJ
II "
rH 0)
C
r, O
0) G
> II
•H 0
rH
cfl
O CO
•H
Cfl
.. o
co G
3 60
4J cfl
CO -H
4-1 P
Cfl
rH
0) CO
60 G
rl 0
X* 4-1
CJ -H
CO *X}
iH TJ
0 <
CJ TJ
rt
3
o
1
3
II
c3 «v
3 M
O CD
G X!
y 4-1
3 °
II CO
CO
fV
•* .y
0) U
tO i— 1
g o
0) II
m CM
II
CM •>
Ol
•• 4J
O) -H
rH X!
B if
II rH
rH
• •
.. 0)
X o
Ol cfl
co c6
cO X>
G C
'h 4-1 rH CO O 4J C
r cfl oiOiXiiHco-
^-'01 4J CD CO rH 60
TJ m cfl o o ••> G
>, B G oi cx^-v 3
W G4J-rl 60rH QCMrH
W -H CO X CO CX, }-i
•H J-IO-HB4JI-.T3
uia-H M'O cfl oicri d
•H Q> UH (X XBrHcfl
•4-1 T3 CX rH Ol ^-'
Oi C CO CO cfl CO CO
C 0) -H C 4J CO 3
XiXl C-HCOO3CJ
O Cfl 4J rH 01 C
<| -H • -H 4-) rH O
COX3cfl.-.'OO>CeX)-i
Cfl U 0) <^N T3 _C Ol J3
>>, CO Cfl . -H B ^
l-i r-~ CM CD oi o
CO CT> C VO • • 0) 4J
EirH cflm/-NCD >-i co B
-H 4J ^-^ X d >. CD
r-lC-H rHOCflCDCfl
CL. -H i-H C O\ -H rH
O O , CX
CflTj CXrH^CO G V-l O
rl O O -H Cfl O 01
CD-HJ-lCJCflB 4JC
cfl X! 4J -H T3 •> CO
4-IOIO)4-lCflrH}-l4-l
01 B X! iH C^ -H G
r-l cfl 4->j20)CTvp.cfl
XI Cfl CJ 01 H CD C
•rl TJ <4-< C t-l 01 60
60C4J O OX! C t-t -rt
•rl CO O M 4-1 iH rH
r-l G CO XI **-< CO
0) •. -H 4J CD O S
CM CO 4-1 O CO G
COr^-H-H-HCOOE •>
-rlCT> rHG>-H-HCOrH
rH X 3 0 CO cfl .
r-l OO}-lCOCOrHCM
CO C iH -H X! -H -H CX vo
3-HXI4J ox: B OrH
*T^( J3 M *T3 fl)
•r") CN Q) 0 "O n3 C el)
> « > O C CD
•rl •* O -H M CO O 4-1 3
"tl rH O T3 «4H 5dCfl
C r^ c « 4-> co a
•H O> O) CO XI G C
rH rO cO 4-1 O Xw 60 «
4.1 -H ^ Cfl -H 4J -H Ji
COCCD OICO-HrH4J
W -rl 01 G 13 CO ^ CO CO
•H HO -rl B 01
«4-< <3" -H C B « TJ
MH CO -H TJ Ol >-,
CU-vOCO C04JS-IC
X! CO -rl Tj -H CO TH
4«> rj ^^ B Ol 4-1 XJ T3
O4-I134JCO ^ CTJ
•••HCcflcOJ-i OOI
COC03 G-H «cj4-i
0) CO OX!-r|l4H/-NOI CO
60-HO4JB J-iCOG
cfl B X J-i co oi -H
AiTj CO-H OliHXl CO B
G cfl -H co 4-i ,-G B cfl j»i
-r4 X 30)
Ol -H O O Ol
TJ XH 60 rG *H M-l E CO ^
C3 4-1 C CD -rl CD S-l
COlHCfl.->CQO)COCOO>
cij O X B cfl U cj cfl
4-1 ~ CO T3 B rH
Cj CU 4-1 CTi Cfl C CO
•X3rHO.
H-4 E 4-1 0) = Cfl
Ol -H B Ol = O
4J -H 4-1 0) M CO
•rl CJ -rl -H Cfl 0) Cfl
rJ E X 0) r-l
OG3T3-H<4-ICO!-IO)
H-Hpcfl coO-Hcflj3
^H|
0)
4-J
cO
rO
c^
O
iH
CO
CO
iH
B
T3
cd
Ja
#«
CO
•H
CO
o
C
60
cfl
•H
'O
fx,
Jin
«\
K^
4-1
C
3
O
CJ
£x^
^v^
*rj
(U
CX
3
o
i-l
60
t
01
rl
O)
XI
^"»
CO
B
cfl
4J
Cfl
TiJ
0)
JS
H
•
0)
4-1
CO .
4-1 CO
co oi
CO
C r^»
•H rH
5 G
-H n3
*j
0)
« 4J
1-4 CO
ai 4-i
"i rH
3 -H
C 0
cfl
•H
cfl O
rH 4J
CJ
CO
S^ 0)
XI -H
rl
CO O
•H 60
01
>i 4-1
CO cfl
rH a
CX
co s-i
•H Ol
4-1
CO O
•H
£ S
30
-------
TABLE IV-6
BENEFICIARY IDENTIFICATION CODES
SOCIAL SECURITY ADMINISTRATION TYPE OF CLAIM
A Wage Earner (Old-age or Disability)
B Aged Wife
B3 Second Claimant
B8 Third Claimant
Bl Husband
B4 Second Claimant
B2 Young Wife (entitled child in her care)
B5 Second Claimant
B7 Third Claimant
B6 Divorced Wife
B9 Second Claimant
Cl Children (youngest child)
C2 (next to youngest child, etc.) (CIO would appear as CA, Cll as
CB, C12 as CC, etc.)
D Aged Widow
D2 Second Claimant
D8 Third Claimant
D4 Remarried Widow (60 years of age or older)
Dl Widower
D3 Second Claimant
D5 Remarried Widower (62 years of age or older)
D6 Surviving Divorced Wife
D7 Second Claimant
E Mother (Widow)
E2 Second Claimant
E7 Third Claimant
El Surviving Divorced Mother
E3 Second Claimant
Fl Father
F2 Mother
F3 Stepfather
F4 Stepmother
F5 Adopting Father
F6 Adopting Mother
F7 Second Alleged Father
F8 Second Alleged Mother
Jl Primary (less than 3QC's and entitled to HIB)
J2 (3 or more QC's and entitled to HIB)
J3 (less than 3 QC's and not entitled to HIB)
J4 (3 or more QC's and not entitled to HIB)
Kl Wife's (less than 3QC's and entitled to HIB)
K2 (3 or more QC's and entitled to HIB)
K3 (less than 3QC's and not entitled to HIB)
K4 (3 or more QC's and not entitled to HIB)
M Uninsured (Not entitled to HIB, but qualified for SMIB)
Ml Uninsured (Qualifies for HIB but requests only SMIB)
T HIB Entitlement (deemed insured)
31
-------
TABLE IV-7
BENEFICIARY IDENTIFICATION CODE
RAILROAD BOARD EMPLOYEE TYPE OF CLAIM
10 Railroad employee annuitant
11 Survivor joint annuitant (election to reduced annuity to guarantee
widow payment)
13, 43, or 83 * Widow with child in her care or child alone of a deceased
railroad employee
14 or 84 *Spouse of a railroad employee
15 or 85 *Parent of a railroad employee
16, 46, or 86 *Widow or widower of a railroad employee
*1 in subscript denotes RR annuitant involved
8 in subscript denotes RR pensioner involved
32
-------
TABLE IV-8
PATIENT NON-MEDICAL CHARACTERISTICS
FROM CONSOLIDATED MEDICARE RECORDS FOR 1971-1973
Item Count
Admission/Discharge records 1,196,515 —
Unique claim numbers 815,000 100.00
White males 359,017 44.05
White females 386,674 47.44
Non-white males 25,843 3.17
Non-white females 24,589 3.02
Unknown race and/or sex 18,877 2.32
Discharged Alive 710,323 87.16
(100.00)
1 admission 527,231 (74.22)
2 admissions 121,322 (17.08)
3 admissions 36,965 ( 5.20)
4 admissions 13,388 ( 1.88)
5 admissions 5,591 ( 0.79)
6 or more admissions 5,826 ( 0.82)
Discharged Dead 104,677 12.84
(100.00)
1 admission 58,612 (55.99)
2 admissions 25,385 (24.25)
3 admissions , 11,352 . (10.84)
4 admissions 4,986 ( 4.76)
5 admissions 2,176 (2.08)
6 or more admissions 2,166 ( 2.07)
Additional diagnosis (coded 1 - yes) 737 5^3 61.65
Age 65 or younger in 1971 102,400 12.56
(100.00)
i 52,757 (51.52)
Females 49,643 (48.48)
Railroad Retirement beneficiaries 26,596 3.26
33
-------
TABLE IV-9
DISEASE PREVALENCE FROM
CONSOLIDATED MEDICARE RECORDS FOR 1971-1973
Item Count
Discharge Diagnosis (Neoplasms) *
ICDA 140-145, buccal cavity 4,439 . 0.54
ICDA 146-149, pharynx 1,479 0.18
ICDA 151, stomach 7,290 0.89
ICDA 153, large intestine 25,843 3.17
(exc. rectum)
ICDA 154, rectum/rectosigmoid junction 10,334 1.27
ICDA 157, pancreas 6,177 0.76
ICDA 150, 152, 155, 156, 158, 159,
other digestive 6,352 0.78
ICDA 162, trachea/bronchus/lung 23,409 2.87
ICDA 160, 161, 163, other respiratory 4,133 0.51
ICDA 170-174, bone/connective tissue/
skin/breast 38,318 4.70
ICDA 185, prostate 24,652 3.02
ICDA 188, bladder 13,085 1.61
ICDA 180-184, 186, 187, 189,
other genitourinary 18,145 2.23
ICDA 191, brain 960 0.12
ICDA 190, 192-199, other and unspecified 25,202 3.09
ICDA 204-207, leukemia 6,450 0.79
ICDA 200-203, 208, 209, other lymphatic
and hematopoietic tissue 10,905 1.33
ICDA 210-228, benign 29,668 3.64
ICDA 230-239, unspecified nature 19,634 2.41
Discharge Diagnosis (Respiratory Diseases)
ICDA 460-474, acute infection/influenza 57,615 7.07
ICDA 480-486, pneumonia 100,347 12.31
ICDA 490-493, bronchitis/emphysema/asthma 62,465 7.66
ICDA 500-519, other respiratory 66,584 8.17
Discharge Diagnosis (Digestive Diseases)
ICDA 530-537, esophagus/stomach/duodenum 85,119 10.44
ICDA 570-577, liver/gallbladder/pancreas 95,468 11.71
ICDA 520-529, 540-569, other 233,110 28.60
977,183 119.90*
A unique claim number will be counted no more than once within a
disease classification category, but a claim history (with more
than one admission/discharge) may contain more than one category
of discharge diagnosis.
34
-------
TABLE IV-10
DISEASE CATEGORY COMBINATIONS FROM
CONSOLIDATED MEDICARE RECORDS FOR 1971-1973
Item Count
Unique claim numbers 815,000 100.00
Unique claim numbers for which
complete admission/discharge
hisotry shows:
Neoplasm diagnoses only 200,902 24.65
Respiratory disgnoses only 208,527 25.59
Digestive diagnoses only 330,022 40.49
Neoplasm and respiratory diagnoses 15,441 1.89
Respiratory and digestive diagnoses 30,486 3.74
Neoplasm and digestive diagnoses 26,746 3.28
Neoplasm, respiratory, and
digestive diagnoses 2,876 0.35
35
-------
admissions, about 1.5 admissions per claimant. This compares with an average
of 1.7 admissions per patient during a two-year follow-up period found by the
Third National Cancer Survey.
More than 91 percent of the claimants were white, with white males
accounting for about 44 percent of the sample (or 359,017 individuals).
Close to 13 percent of the sampled individuals admitted to hospitals for
treatment were discharged dead; there were approximately 28 percent more
admissions for those eventually discharged dead during the sample years than
for those discharged alive following each recorded hospital admission.
Nearly 62 percent of the records showed a diagnosis in addition to
the ICDA-coded discharge diagnosis. As the presence of other condition(s)
is indicated, but not its disease code, this data field cannot now contribute
to environmental health effects analyses until coded. It constitutes an
additional epidemiological resource in an area where comprehensive and uni-
formly defined statistics are generally inadequate for environmental analyses.
There were 102,400 claimants (12.56 percent of the sample) who were 65
years of age in 1971 or later and were probably making their first claims
under Medicare. A field on the unabstracted files maintained by the Social
Security Administration is labeled "Formerly Disabled" and could provide
some further insights on this point.
The claim number also identifies railroad employees and the spouses
or surviving beneficiaries covered by health insurance provided under the
Railroad Retirement Act (RRA). Although this industry is not of primary
research interest to this project, it was thought that the suffixes to claim
numbers in the Medicare data identifying RRA claims offer an excellent oppor-
tunity to compare the morbidity/mortality differentials of working husbands
and their wives on a nationwide basis for a single industry.
Two types of analysis were performed using the four-digit ICDA-coded
discharge diagnosis. In one procedure the neoplasms, respiratory and diges-
tive disorders were partitioned into 19, four and three categories, respec-
tively, and the numbers of claimants whose hospitalization showed at least
36
-------
one admission for a category were registered. This provided a disease preva-
lence indicator, and the sample distribution is presented in Table IV-9. The
cancer diagnosis code groupings were chosen to incorporate cancer types with
high incidence and relatively low five-year survival rates. The lower-
survival-rate (or high death rate) consideration was of importance since only
three years of Medicare morbidity data were available with which to facilitate
morbidity/mortality comparisons.
The second procedure characterizes the complete hospitalization history
associated with each claim number by whether neoplasm, respiratory, or diges-
tive disorders have occurred exclusively or in combination with another type
of disorder. For example, a claimant with a history of admissions for cancer
may be classified as neoplasm diagnoses only, neoplams and respiratory diag-
noses, neoplasm and digestive diagnoses, or neoplasm, respiratory, and diges-
tive diagnoses. Results are shown in Table IV-10.
Of the 815,000 Medicare claimants in the hospitalization sample, 245,965
had an admission for which the discharge diagnosis was "neoplasm" (ICDA 140-
239); 257,330 had an admission for respiratory disease; and 390,130 were
admitted for a digestive disease. Since this Medicare file includes records
for only three stated disease groupings, it is possible that admission/discharge
histories with a neoplasm and a non-abstracted disease classification, such as
cardiovascular disorder, are counted in the "neoplasm diagnoses only" category.
It is interesting to note that 90.73 percent of the individuals in the sample
did not have diagnoses in more than one of the abstracted categories (neo-
plasm, respiratory, and digestive).
3. Choosing the Study Areas for Selected Morbidity/Mortality Comparisons
To select counties of particular interest in this study of correspon-
dences between area morbidity and mortality statistics, retrievals from the
POPATRISK data base were made, and other data resources were examined. Infor-
mation concerning population distribution, migration, air quality, and indus-
trial employment were among the factors considered. For example, POPATRISK
contains Census migration survey data covering the years 1965-1970, from which
counties with stable populations and those with significant changes in the
37
-------
number of residents over that time period can be determined. It was con-
sidered particularly important to differentiate between "retirement" counties
characterized by large fractions of older persons who had lived/worked else-
where during their earlier years, and "non-retirement" counties with a younger
and more stable population.
In consultation with EPA, attention was focused on U.S. counties having
1970 Census population greater than 10,000. Within this population specifi-
cation, 31 counties with low migration (both in- and out-) for white males of
all ages, and 45 counties with comparably low in- and out-migration for white
males over the age of 44 were identified. There are 15 counties which appear
on both lists. The maps in Figs. IV-1 and IV-2 show the locations of the
counties meeting these criteria.
Tables IV-11 and IV-12 indicate the counties with high migration, either
in- or out-, among white males over 44, and with large differences between
rates of in-migration and out-migration. More specifically, Table IV-11 lists
counties with ratio of in/out migration greater than 3.0. For the same
population category, Table IV-12 shows the counties with an out/in migration
ratio greater than 2.0.
Because the older age groups are of primary interest, and because changes
in morbidity and mortality statistics may be more easily attributable to pop-
ulation increases due to in-migration than to losses of out-migrants, only the
75 counties of Fig. IV-2 and Table IV-11 were considered further. For those
75 counties, absolute numbers of white males over the age of 64 and the per-
centage of that group which changed county of residence between 1965 and 1970
were obtained from POPATRISK.
The POPATRISK file also provided emissions and monitoring measures for
TSP and S0?. The emissions values are NEDS estimates expressed in tons per
year; the air quality values are the arithmetic means, for each county, of the
annual geometric means of data from individual population-based monitoring
sites. In cases where monitoring data were not available, a regression equa-
tion using NEDS data for TSP and oxides of nitrogen (NO ) was used to estimate
X
a county air quality index with regard to TSP.
38
-------
o
V
•H
V-l
W
CD
CJ
CU
o
M
l-i
a) co
r* .^
(U CU
PQ PQ
CM rH
C CU rH
Cfl CO 4-1 CU 1-1
d 4J £ CO C
' O CU 3
•H 43 -H
CO M d
rH Cfl 3
PQ U -j
3.2
O C N
cfl 3
C
cfl
CO
o cu
(U 4-)
so o
CO O
P-l CO
CU rH
13 rH
d cu
CO &
X T)
Q) rH
rH CO
<; cj
T3
K
o
rJ CO
0) CU
4-1 O
3 4J
Pi CO
CO
>, (U
M ^
4J 3 -H
C JC3
•H W
O
O ,0
o
O 0)
- t-l
O CO
CtJ
O
rH
•U I
10
4J tH O
CO 4J
0) r-l
noc
W) <4-l O
tH
c a -u
o o cd
•H -rl rH
4J 4J 3 ^-N
co crj a, cu
rH M O M
3 e»o a cu
(X -H X!
O B !>>
a i w c
•u c 5
rH 3 3 O
n) O O J3
W O CO
o t>
w a o a)
cfl r-~ CD
J! O^ O
•HO
(U
14-1
O
CO C
(U H
•H 60 "4-1 4J
4-> 1-1 O CO
C S cu
3 I -u £
O C C
CJ -H CU CU
CJ C3
• CU rJ O
CO rH CU d
~> 8 ^ "•
C*i CO
CD
C 4J
CO C
42 3
4J O
• o co
60 -H CO rH
•H J3
rH -H
CJ
o
o
4J
•H
d
CO
•H
o
A
3
CO
Q)
cfl
J-l
o
d
o
CO
CD
o
d
CU
CO
39
-------
cj
£
M
CS
•H
3
c
CO
M
fw
g
^y^
CO
0) iH
•H £
M CU
M >J
>-.
4J
•rl
CJ
M
CU
4-1
CO
cu
o
(-1
o
£5
cd
•H
CO M M
Ai iH 43
1-1 co e
CU rH CO
M pq cj
^^
cu
4J CD
C -rl
CU M
CJ W
is
tO 4-1
£ W
tO cfl
»s o
0 C
tfl Cfl
f-3 (J
rH
rH
pn
T", r'-'.xh^l' '"- " ;?_.^-, r""":V>>r -ViUr^c^ Jr -*:,"'••—'VHtafii'
^SUv;^
' t""^ I J ' *•
"~. .,~il I
Hrff
—^ ; i— — -~ -, • , t H
> 43 »
CU I O
CO
d)
o
M
a)
o 7^
M cO
P,
CO *J
rl Cfl
00 rH
•H 3
0 CU
I O
•U Cu
3
O >.
-a c
c 3
to o
o
c
o o
•H r^
4J CTi
tO rH
M
00 CU
•H X
e 4-1
C >*H
•H O
CU
3
43
Q
CO
4-1
CU CU CU
Ai C O
O CU K
O Q) G
VJ rl O
cj O S
u
> G
cu o
43 -rl
4-1
CU
C
O
CU
4-> rH CJ
cd a cu M
•H S 00 3
T3 3 G O
CO CO Cfl UH
3J J2 £ ^
<3 <3 W i—1
4-1 4-1
CO CO
cu
G C
C O
O -rl
43 rH
CU -rH
M M
CU CU
H >
40
-------
U 0
M pa
§ o
CO •
Z W co
o||
O W H
- H
o M erf
rH EC W
5 H
^D ooiAcocrirHOr-^voovor^ioiooocMco CN o
O> CM i— 1 vD«^-COOvDr— CO^^OOI — rHOO-J-CM rH i^« uo co CM vo OS
CM rH rHCMCNCMrHrHrHrHrHrHiHrHrHCOiHCM
& von r^r^iocsir^c^Ocor-.cocT\-sj-covOrHu-) vo co
lO CO<)- CMCOrH-d-OOlO\DcOP^<- £j CU3M crJrH3-rHcdrHcd 4J-HO OrHCOCJ COM CO
MjcS XCUO Octf4JrH(-iOO'd!U>'dS-iOrHCOC)-iO dU §
erf o erf cdi — i J r^^CTHOD'HdcucUto^dcottJtrJ'HttJ W o t—t W)
-------
W
0 0
S S o
yt o
to
S! W en
oi|
0^ W H
cT M oS
rH ffi W
3i H
<§ OS W
O H to O
•w
<
M
H o «n
Z M sf
S H
O ^3 fT1
o w 2
• w ts
co 05
f-J **
0 3
*J ®
Sfl
*
± *^
in n)
^f V4
00
4J
o
*
pj
o
io CO
•^ ^
S S
1
f3
M
•K
c
J
rH g CO
CO M
*J Q) 60
H i5 S
i 1
H
4J
a
8
vo O CO iO ON O O iO r**
iH CT> OvO >^-iH lOrH iO
CO CO CO "^ CO CO CO *^ CO
IO O iO CD vO <^ 00 O^ cO
ON ON r^ ?^*H r^>» co 10 *—\ co
* • • • •• •• •
rHO cncN OCM CMCM CM
"
I^% rH *^" *>O IO -^" CNI 00 CM
rHin min VOCM oo co
VOCO O^CM CM*VO ONO^ 00
rH
vO r*^ ^D r*** ^D oo ^^ ON ON
mm vo^a- est-H CT>OO CM
rHrH rHrH rHrH rHrH rH
_,. _^^ ,— . .A Qk Q^ _., ,— .Q
l^'l-I VOCM c^oC p^cM CO
COrH CMCO rHCM CM-*^" CM
a
o
•H
4J
(0
rH
3
O
PH
CO
3
CO
o
J^
4J
M "S §
PH f« C ^ CO O
PLJ W MO CVlr-HU
f-H CO ^t W CO O O ^d CU
coo crfj^ co S co £5 i— i j-i u-i
COCOCU 0)cOZC()rH«ri-dOOC)W B^
C/D 03 ^ (X fl) £3 fl) r~l ^s C3 £* pd CO
M (UcdU (Tj O W t-C *H W d)O*~^ s *
42
-------
so
0 O
M EC
§ OT §
55 W CM
M J
CD K
CD tjj £H
o" H Pi
rH EC W
iS H
3 Pi [iJ
EC O PH
H PM O
Pi 53 to
won
H M
CM •< H O
1 fvj pq O">
> 0 0 rH
H H
53 g 53
W O 1 M
tj H £25
3 H H Pi
3J «! W
H (J O P
t» H tJ
PM O
O 53
PJ O Pi
M O
rJ H
-<
H o m
S H
O
cd g cd
4-> p
O 0) 60
H 4J -r)
•H g
C? 1
I2 4J
3
o
*
Cl
vO
ir>rH O O\ O CT>CO rH CMO vO
• • • • • •• • •• •
in-* oo co m om oo -jco CM
rH
r-~r~ tTi oo O rHr^ ^D Oco cr\
COCO CO °^ vO vOOO CO O OO VD
CMrH rH rH rH COrH CM rHO O
cooo cy> sj- -d- OCM rH cor-- co
• • • • • •• • •• •
cor-~ o -3- rH I-H^O oo cMr^ oo
CMCM CO CM CM «*CO CO CMrH rH
CT* O r^ rH r^* CT\ CM co CM in -^f
• • • • • •• • •• •
-a- o CM oo o CMOO oo i-ir-- oo
HrHrH rHCMrHrHrH
<
g
o
K Pi
en cu >H 53 c ~i53cdC3t>,i-ia) pd >i t-i OJCJO-C
O OCdhJ QJ53 >^U t>i53 bO[?PQ QJPi OJ4Jpi CJ
W Ot-liJr-IW OJM cdO'Hcdpd^CO M-iHrH 3
OUOHpq
4-1
td
ex
o
PH
CO
CO
g
43
-------
Noting that 30 of the 75 counties listed had a white-male-over-the-age-of-
64 population greater than 2,000 and that only in Florida, New Jersey and
Pennsylvania did a majority of the listed counties satisfy this older popula-
tion criterion, it was decided, in consultation with the EPA Project Director,
that closer examination of Medicare statistics in ten Pennsylvania counties,
14 Florida counties, and one New Jersey county would be appropriate for
further narrowing the field of candidate counties for use in the morbidity-
mortality analysis to follow. The maps of Figs. IV-3, IV-4, and IV-5 show the
25 counties which were considered.
Enumerations of patient characteristics in the selected counties were
obtained from the Medicare admission discharge records. In addition, popu-
lation and industrial data were compiled for each of these counties. Findings
are given in Tables IV-13, IV-14, and IV-15.
At a meeting with the EPA Project Director, the merits of the county
choice alternatives were discussed and selection of the study areas (two
Florida counties and a region in eastern Pennsylvania comprised of five con-
tiguous counties) was made. Broward and Pinellas counties in Florida have the
largest populations among the "high migration" counties; they offer an excel-
lent opportunity to examine the potential of the Medicare data in estimating
migrants' contribution to disease prevalence and specifying other character-
istics of this relatively mobile population.
Berks, Lackawanna, Lancaster, Luzerne, and Schuylkill counties in Penn-
sylvania were also selected for analysis. They are among the most "stable"
(low migration) counties nationwide, and, by forming a contiguous region,
offer an opportunity to contrast patterns and causes of hospital admissions
among the Medicare-eligible sample population over a fairly constrained and
demographically homogeneous area.
4. Industrial Development Patterns in the Selected Counties
Following designation of the study areas, documentation of industrial
development in the seven counties (two in Florida, five in Pennsylvania) was
44
-------
Florid,i
JACKSONVILLE
•••UK
. _ Jo. r /
TALLAHASSEE f /
FORT LAUDERDAU-HOH.Yy»oqO. Vv. \ \/
\,\\ «okuj.\ \\ I »6i(l.iUec«l«t|'t*
\\\\ A \V:\.V\V\ .i-Jcx.r
LEGEND
© Place* ol 100.000 or more inhabitants
• Places of 60.000 to 100.000 inhabitants
O Places of 25,000 to 50000 inhabitants outvdc SM$A's
i Standard Metfopo'-lan
Statistical Areas (SM$A s|
Fig. IV-3. Counties of interest in Florida are identified by diagonal lines;
all are "nigh in-migration" counties (wm 45+ in-migration to
out-migration ratio > 3.00).
OJS. DEPARTMENT OF COMMERCE &3CMI md Economic Sutiftia Adminiltfailon BUREAU OF THE CENSUS
45
-------
New Jersey
PATERSON CLIFTON-PASSAIC
^l**"
.,/' ^
UNKX an
JERSEY CITY
:irr
LEGEND
® Places of 100,000 or more inhabitants
• Places of 50.000 to 100,000 inhabitants
O Central cities of SMSA's with fewer than 50.000 inhabitants
O Places of 25,000 to 50,000 inhabitants outside SMSA's
Standard Metropolitan
Statistical Areas (SMSA's)
Fig. IV-4. County of interest in New Jersey is identified by diagonal
lines; Ocean County is a "high in-migration" county.
US. DEPARTMENT OF COMMERCE Social and Economic Swittia AdmlnlitratkMi BUREAU OF THE CEN*U*
46
-------
£ 2
o "2
o i2
O O
a
o
CJ
to
o
I—I
4J
3
O!
C
•H
c
o
•r-l
d bo
•H -H
G E
txl I
> -ui
•H 3
>. O
W
c -a
c c
a) e)
PL,
l
a a
•r-l -rH
•U +
w m
•u
a
CO
1)
US.DCPARTMENTOf COMMERCE Socul and Economic Sutiitiu Adminmr«lion BUREAU OF THE CENSUS
47
-------
CO
CJ
H
w
M
Crf
M
H
c_)
^
S
rH O
1
> &
H O
» J
W H
-I <
PQ ( i
<« £3
2
r:
o
u
01 C VD
rH O
JO iH CU 4J
14H S 4-> 60 3
O cfl
3 P-. 0
C
«* o
(1) VD iH
J-B| ^J
CO CU CO
IS 60 M
e
O M
_____
CU B
rH O
Cfl *H
<4-l ^ 4J
O co
Q) rH
B-« 4J 3
•H Q.
iS °
^j.
eu co Vj vo
JJ CU CU
•rH rH ^ CU
j£ CO CD 60
5 a * *\ »V
CO rH CO rH ST
OOCTiCTirH in ^ r— r~ CO vo tr» CM r^ CT» in CM
CTirHrHCMO vOCTirHCTvvO rHvOvT-r^-rH rHr^-rHCOrH mmvOvOCM
OCMrHrHi-H COin
ocriomsj- o\rHcoOrHcr.in oornooOrH
CNCOCOrHcM rHCMCOCNCN rHCOCOCOrH rHrH rH rHrH
i — !CT»r^OvO CT»COCT>f — inr^-r^ r^vocMoo^t rHcocMrHcN r^voovovo oococOrHiTi
omcT.*~a- OCOCOI^CM vOi-Hvocoo
Or^o>ooc^ ininr^.oom oomcMOoo mvomvocTi CO(Jh4J3fnIil *$ cfl-, PH^ CUP^PncO -,
>>-))-lrHXl >iH cfl4JCU gOCUCflcfl CU^d-Hra4J CUAiOCU3
OCfl4JrHoO T3CUCMO rHCOCucU rHVJCOeC -HCJCNrC
M^CtHO'H pSCUcOcflCfl Cflcfl-HcflO H^rHCflCU MCflcfl3O
paouaw MI-JS^O p^p^p-iwo
-------
00
CO
n
H
CO
M
PS
w
2 H
V °
Cq
rH 3
W
W rj
t-H
§ 3
H <;
Q
W
3
CJ
1 1
R
H
S
CD
cd o)
S3 >-« 4->
O CD o td
w
4J
K""1 r-3
01 rl Q) 0)
rH O -rl rH
S td td g"
M PH Cd
0) -rl CO
^J Of t^
•H CO 01 0
.C fv
c e
cd cd
Q) g CO
td cd co
2 rH I^
O CTi
-l r-H
•H f-H
0) c
Selected
Counties
|-~ CO rH rH 0>
oo cJ** \D cy\ co
rH rH rH CM CN
-* rH m VO rH
r^* rH rH rH
rH
in CD co o*\ *3"
oo ^o <}• r^^ CM
CO CM CN CO CO
rH r- rH in VD
co in co in *^*
CT*
00 CTi C^ in CN
rH •— 1 CN ~* »J
<}• CN rH rH rH
f>
CN
Broward FL
Charlotte FL
Citrus FL
Collier FL
Highlands FL
rH o m o o
co in vo r~ tTi
rH rH rH rH rH
oo CM r-. o> oo
CM CN
*^ in t^ r^» •^
00 CM CO rH O
CO CO CO *st" CO
rH f^ -J CO CM
vO co ^D r^ oo
m in oo CM co
rH -^ >Cf rH rH
Indian River FL
Lee FL
Manatee FL
Martin FL
Osceola FL
VO O ON rH OO
ON \o r^ co CN
rH rH rH CM CM
VD ~-J r~- CM rH
oo CM vo in vo
rH
VO 1"- CM in rH
m \o m r>- r-
CO CO CO CO CO
oo O in ON r^
co in co in vo
cr» o o o>
CM O vo CT\ rH
CM <)• vO VO 1^
n rt
rH CM
Palm Beach FL
Pasco FL
Pinellas FL
Sarasota FL
Ocean NJ
m oo o^ CM o
1^ i-H rH vO O
rH i-H rH rH rH
vo m r>« oo <•
r- CM I-H
rH
rH m co o> in
m r^ CM vo CM
CO CO CM CM CO
00 CM O> rH O
O rH in rH rH CM
00 in CM -* rH
ft
CN
Allegheny PA
Berks PA
Blair PA
Cambria PA
Centre PA
-vj- m o co co
in rH rH r-» OO
rH rH rH rH
r^ o\ co oo CM
rH CM CO CM
rH O» rH r-- CTl
O rH CTi \O in
CO CO CO CM CM
O in cr» o o
CO VO O CM CN
i-H rH CN CN i-H
CM r~-
-------
CJ
M
H
CO
p_l
CU
13
o
s
m 5
T* "
)
H 3
W
[V] _^
,j» tr^
?9 w
S g
£3
M
•"•^
Z
1
><
£-1
^5
0
u
c
o
rH -H
14H Cfl 4-J
O 4J CO
O rH
fr-e H 3
p.
o
PM
0)
60 C!
cfl HO)
1 1 lii l_j
a) O co 3
> cu .u
< >-i cu a
cu r*» cd
rH & O M-l
Cfl B rH 3
3 •§ & C
*i
nn (-I*
4-J W
*H *"^. ^**«.
^ CO
fll
IX*
&*5 O^ >%
1 O
O rH
CM O,
W
CO
W5 *J
C §
°1l
(H 4J CO
CU O -rl
J3 Cd rH
B ^W rO
3 3 to
53 C u
cd co
S w
- — ~ —
C
o
•H s
m 4-> nj
Ofri o
tw »^— '
rH rJ
&*9 3 *->
CU
O
PM
Tj CO
CU CU
4J *H
CJ 4J
a) c
rH 3
0) o
to o
mr^-inooo cnco<3~<)-oo <3-oomcMoo cMOoomm o oo in r^ m
rJ < r-3r-Jj3fnri, PM< cSMPMrH
4-l(ii T3 &$ £n cOrJCOcO'-J C<3<3 PM CD -H
T34J MC cu cfl cuf^icd-usz; cupMp-icd ~.
3 J-l J-l rH ,G -H cd-UCU BCJCUcdcO CUJ
-------
undertaken. The number of employees in each of several distinct industrial
activities was used as the measure of industrialization. Employment data from
County Business Patterns for the years 1953, 1959, 1965, 1970, and 1974 were
obtained for each of the seven counties at the two-digit level of Standard
Industrial Classification (SIC) code, except for industry codes identified by
EPA's Implementation Planning Program as relating to significant sources of
TSP or SO . For the latter industry groups, data were obtained for the most
detailed code available.
A perspective of chronological changes in total employment and the main
types of employment in the selected counties is presented in Fig. IV-6A, for
the Florida counties, and Fig. IV-6B, for the counties in Pennsylvania. The
categories of primary metal industries, and chemical, petroleum, and rubber
displayed in the bar charts of Fig. IV-6 were chosen to facilitate comparison
with the example study of steel-industry counties, to be described later.
The directors of the statistical divisions of the Pennsylvania and
Florida Departments of Commerce also supplied data maintained by their respec-
tive offices concerning the chronological development of industrial employment
within the study counties. The Bureau of Statistics, Research, and Planning
of the Pennsylvania Department of Commerce forwarded Industrial Census Reports
for the five counties of interest for each of the years 1970-1976. In addi-
tion to historical trends of employment, production, and wages, there is a
complete four-digit SIC categorization of establishments, with a distinction
between "production (and related) workers" and "all others." These reports
were first printed in this format in 1961. Prior year data (back to 1950)
were included in Productive Industries Reports. The Division of Employment
Security of the Florida Department of Commerce sent information on historical
employment data (including sample reports and descriptions of data limitation)
for Broward and Pinellas counties. Most data were provided in a two- or
three-digit SIC code. However, the data from County Business Patterns,
Bureau of Census, were more generally useful for consistency, uniformity, and
the demonstration purposes of this project.
51
-------
Number of
Employees
243.IK
180K
160K
140K
120K
100K
8 OK
60K
40K
'74
'70
'6!
'59
20K
'74
'70
'59
'53
BROWARD
PINNELLAS
Legend
•<—Total Employment
Manufacturing
Employment
Chemical, Petroleum,
—• and Rubber Industries
Employment
Primary Metal
Industries Employment
Fig. IV-6A. Levels of Employment in Two Florida Counties
for the Years 1953, 1959, 1965, 1970 and 1974
52
-------
00
C
vH
I
cd
4-1
3
C
C
§
>,
O
rH
a
a
„ 0)
B -H
3 r(
01 4-1
rH CO
O 3
M T3
4J rj
Q) M
PU
@
ex
rH S
CO W
4-1
0) W
rH .JO 0) -H
SrO S >~l M
_ 3 ?>-. u 4-1
•H Ed O CO CO
0 rH E 3
0) T3 CX -H *O
x c a M c
O CO W PL, M
1 1
TA
m
CO
in
U 1
vD
ro
tn
?K
ferf
33
c_>
CO
M
co
QJ
•H
4J
C
3
O
0
CO
•H
rH 13
>s C
CO CO
C
C O
Q) t-v
o
in
vD
OS
m
in
§
»
i>
g
%
^
^
^
i|
LANCASTER
"ck
ro
in
cj>
4J CTi
C m
0) CTi
ex in
e cr>
W rH
<*-! M
O M
CO
co ai
rH fH
0)
> QJ
0) .C
1-4 *J
PQ
VO
I
LM CO
O 0)
0)
M ,
Q) O
m
Oi
in
rt
60
•H
W
o
o
o
oo
O
CN
53
-------
A particularly interesting publication received was "Occupational Employ-
ment in Manufacturing Industries in Florida," in which estimates of total and
proportional numbers of employees within occupational subdivisions of the
three-digit SIC's 201-399 were derived from a 1974 survey of 5,550 firms in
the state. This information was obtained for the first time in 1974. The
survey will be repeated at three-year intervals, and holds considerable
promise as an analytical tool for epidemiological studies of Health Effects
Research Laboratory, EPA, in linking industries with occupations.
5. Morbidity/Mortality Comparisons
A fundamental question being addressed by this study is whether the
morbidity data collected as part of the Medicare program adequately reflects—
or, indeed, extends—knowledge of disease incidence and prevalence that can
otherwise be obtained from mortality statistics. This section presents three
illustrations of the manner in which the Medicare records can be retrieved for
purposes of health effects investigations; each example focuses on a different
geographical area and a different time interval.
A brief review of mortality information is helpful for an understanding
of the complementary role of morbidity information. The National Center for
Health Statistics receives copies of death certificates from vital statistics
offices in each of the 50 states and U.S. territories, then codes much of the
non-confidential information for computer storage, research purposes, and the
preparation of summary statistics. Mortality rates computed from these data
have, to this time, been based almost exclusively on the single code for the
"underlying cause of death." Items on the death certificate form pertaining
to other causal conditions manifest at the time of death are not coded for
computer-assisted analyses.
The significance of use of the "underlying cause" alone can be appre-
ciated by reference to a recent study of the 1969 death certifications for the
United States. In that year, malignant neoplasms of some kind were "men-
tioned" over 565,000 times as contributory, associated and/or the underlying
cause of death on the certificates of some 368,000 individuals. Of these,
54
-------
323,000 died with some malignant neoplasm designated as the underlying cause;
customary epidemiological analyses would be limited to these 323,000 cases.
Within the neoplasm total were 15,989 "mentions" of secondary malignant neo-
plasms of the lung, with 766 of them designated as the underlying cause of
death. Use of the underlying cause alone would ignore over 15,000 cases, and
significantly understates the prevalence of secondary malignant neoplasms of
the lung. Lung diseases are particularly important to EPA and its concern
with human health effects.
Two ways of obtaining more complete information concerning the numbers of
people affected by certain diseases are (1) use of all medical descriptors
entered on the death cTLificition foi - .;nd (2) use of medical histories
and/or hospitalization records. The Medicare file is an example of this
second alternative. These two ways are not mutually exclusive. They are
complementary. With complex interactions of the real world, neither the
mortality nor the morbidity records can be expected to meet fully the needs of
the environmental health analyst.
There are advantages and limitations to each of the two types of records.
The death certificate, especially when augmented by an autopsy, is generally
complete and accurate; but is restricted to observations concerning the death.
Medical histories, as represented by the Medicare hospitalization sample,
often provide medical information well before the time of death, including
possible chronological development of disease conditions. Medical histories
also give some indication of geographical location which may be related to
pollution levels, or time-specific events detrimental to health. Unfortu-
nately, there is no source of medical history data that meets all environ-
mental health analysis needs. The Medicare files utilized are more accessible
ami extensive than other medical-history sources, but nevertheless the files
pertain only to the older members of the population and are a 20 percent
sample. At present, the Medicare data has only one disease code with each
hospital admission/discharge—a limitation identical with mortality data.
Nonetheless, the Medicare data obtained from the retrievals described
below do give an important perspective on the problem of identifying the
55
-------
number of individuals within each county recognized each year as having
certain diseases.
Morbidity trends are obviously more timely than mortality trends, and
offer the singular advantage of possible interventions before death results.
As a further practical consideration, there are about 16 times more hospital
admissions than deaths.
It is noted that the POPATRISK data base was not used as the source of
mortality data for mortality/morbidity comparisons because POPATRISK contains
age-adjusted, rather than age-specific, rates and would not be comparable to
the simple counts obtained from the Medicare record files. For convenience,
the mortality data used were provided directly by the State of Pennsylvania.
a. Berks County—By Year—
The first morbidity/mortality comparison example is for one county,
Berks, in Pennsylvania. A single county was chosen because of the wealth of
demographic, environmental pollution, economic and general statistical infor-
mation that is collected and coded for computer-assisted analyses by standard
county identifiers. With these identifiers and codes, widely-used computer
procedures can be directed to inter-county comparisons, as well as assembling
the data for comparisons by metropolitan areas and states.
The Pennsylvania Department of Health compiled mortality data for Berks,
Lackawanna, Lancaster, Luzerne, and Schuylkill counties according to disease
categories specified by System Sciences, Inc., identical to the 19 cancer
diagnosis subdivisions presented in Table IV-9. There are five-year age
groups through age 74, also 75-84 and 85+. Statistics in all seven counties
cover the individual years 1971-1975, as requested. For pilot analysis
purposes, data were aggregated, without differentiating by sex or race of
the deceased. The mortality statistics thus obtained provide an aggregate
basis for comparison with the Medicare-derived statistics.
The retrievals from the 1971-1973 Medicare files counted hospitalizations
for residents of the seven selected study counties, including Berks. Counts
56
-------
by county, year, disease category and age were taken of the following: the
number of hospitalizations in which the claimant was discharged dead and the
number of new cases of cancer each year partitioned by the year of death of
the claimant or the fact that the death of the claimant is not indicated.
Morbidity and mortality data for Berks County only are presented and discussed
at this time.
The cancer mortality statistics provided by the Pennsylvania Department
of Health for Berks County for the years 1971, 1972, and 1973 are shown in
Table IV-16 under the four age groups 65-69, 70-74, 75-84, and over 84 years
with 19 subdivisions of the neoplasm disease classification. The 92 deaths
over the three-year period (30, 24, and 38 deaths in the respective years)
give a small number of cases in each of the age-year-disease categories;
little can be determined from these data with regard to mortality trends.
Table IV-17 consolidates the age categories for the mortality statistics
from both the Medicare file and the Pennsylvania vital statistics office; the
counts are by year and cancer type for Berks County residents over the age of
64. While there is an apparent upward trend from 1971 to 1973 in the number
of hospital deaths reported for cancer of the trachea/bronchus/lung (ICDA 162)
and the miscellaneous classification (190, 192-199), there is undoubtedly a
high degree of variability in the sample data, and there is not a comparable
indication of trend from the statistics of the Pennsylvania Department of
Health. It can be seen in Table IV-17 that the deaths reported by Medicare
for Berks County are not a uniform proportion of the number of deaths as
reported by the Commonwealth of Pennsylvania.
The changes in the number of cases in the specified disease categories
during each year are presented in Table IV-18. A table for each of the years
1971, 1972, li)73 shows the net change in cases for each disease category; that
is, the number of new cases first appearing in the years, minus the cases
removed by death during the same year.
Table IV-18 also shows the net change over the entire three-year period
covered by the consolidated Medicare files available for this project, and
57
-------
vO
00
CO
cu
60
00
1
m
***
co
0)
60
1
O
^
CO
cu
60
ON
1
m
xO
co
cu
60
CO
rH
CM
ON
rH
ON
rH
CO
„!
T""1
CN
ON
rH
rH
rH
CO
r-
ON
rH
CM
O\
rH
rH
ON
rH
CO
f-
ON
rH
CN
ON
rH
^
r-^
ON
O O O
0 O 0
O O O
O O 0
O O O
\ O O
O O rH
O O rH
0 O O
O 0 O
0 O 0
O O 0
lasms
.45, buccal cavity
i.49, pharynx
stomach
X r-i «— i
P- r I »
Oil "
0 xO rH
®
&l rH rH rH
rH rH O
O O O
O rH O
rH O O
o o o
CN O O
rH O O
rH O O
CN O O
o o o
O O O
rH rH O
£j
0
•rH
4-1
large intestine
(exc. rectum)
rectum/rectosigmoid June
pancreas
co -
-------
w
4J ro
C r--
cu o>
B rH
•O 4J
-O CU M
0)4-1 CO
4-1 rl CX
cfl O CU
o cx jz n
•H Q) 4J
t3 Crf CO CO CN
C CU -rl r^
M CU P C 0>
CX CO rH
J3 >,UH >
<3 4-1 H O rH
PL, -H >sj3
Is 6 cu w 4->
" 10 CO C rH
>H rl CO 3 C CO
H CU rH Cd CU CU
§,0 CXO P-c D3 -H
BO r--
O 3 CU CO O 4-1 O
<-J !S JS cd 4J O -H
CO
t>
*
w
pa
2!
M
& 13 C co
W CU O r-~
> ,C C -rl CTv
O 4J CO -r( 4-1 rH
•rH Cd Cd CO
Pi! & 4J N
O CU 43 -H
tJ CX O rH
m co >-,^-/ co
vO CU H 4-1
Q CO -H
Q S -H CX
Cd TJ 03 CO 03
CJ CU CO O O CN
rJ CO Cd a CU
erf cu u j: rH
O 43 -H 0 B CX rH
PL, B TJ co O B r>.
3 C -H M cfl cr\
W 13 M Q I4H CO rH
Pu
E
r-<
S
to
3
p-i
o
w
a
>-'
«
C/l
H
(5
O
CJ
*~H
33
^
*-N
TT"]
Q
QJ
TJ
0
CJ
«n
>% OU"i — M-H, Ocd
•I-J 6 rH M iJ CO C
•H CU CO QJ3l-iCj4-l rHvH
> d-H •>>X<1)GJCO ±4
crj vH -— x CO vO-HcJJddcl ~3
O uEo m44d4JdOJ r^-o
X C034J rHCOQOOM C04J
rHd 04-JCJ CJU O J3 rH-ri
crj ?-, 4J o cu *• co 43 «-^--^ g
oi-i Gcj>-icoiri-H^,coaJc:a) ~cu
O C3X-H J-i^-cd'-OTJ cdvo C'rH U MO CO
3.CCJ EOrH 0) rH O ^i ca CJ CO
eO CX, CO O . 3 ^ j^j; JDMiJ'OrH!-)
2 gto'jj-ioxcuo" co-rj cu
= -"OMXCldCM^aJr-l » OcrJ " £i
« ino>i-irjo- m u~i m u-v \.o o r^ co co 03
r-HrHrHrH rH rH rH rHrHrH rHr-ir-H
CN CN CO in O
-* rH CM
O> O rH O" O CO
cn rH ^H
O CT\ rH CO O O
o Is-. CN in o o
O 0- w ca
a_J rn t
I— f UJ r*
3 -rH 3
»l J i , 1
>-4 *^ 4~A
TJ OJ rrt
C ^ cj c
CO 4J -H
O 4-i -a-
t* m -i
r^ U/ ^J
0 --H .,4
tf~; f^j (7\ Q |,
tv »r^ O CU »p4
O S CN 0 C U
-U to CJ
A JV M M .„! X-*
'*-l< ^ J I~T ^TX,
^ u co S c w
C^ a O CD CJ r*
C r-H rH rrj gv M
^ cr\ r^-^ oo d CO C7\
43 rH O O ffj CN m
CM CM CN CN
" ^r i i it
rH O vr 0 O Q
0 CA O O rH CO
•H r-l CM CM CN CM
ro
-------
TABLE IV-18
USE OF MEDICARE DATA TO ESTIMATE CANCER PREVALENCE
(Berks County 1971, 1972, 1973)
Hospital Diagnosis
ICD Type/Group
140-145X, buccal cavity
146-149X, pharynx
151-151X, stomach
153-153X, large intestine (exc. rectum)
154-154X, rectum/rectosigmoid junction
157-157X, pancreas
150, 152, 155, 156, 158, 159, other digestive
162-162X, trachea/bronchus/lung
160, 161, 163X, other respiratory
170-174X, bone /connective tissue/skin/breast
185-185X, prostate
188-188X, bladder
180-184, 186, 187, 189X, other genitourinary
191-191X, brain
190, 192-199X, other and unspecified
204-207X, leukemia
200-203, 208, 209, other lymphatic and
hematopoietic tissue
210-228X, benign
230-239X, unspecified nature
Net Annual
increase in
Surviving Cases*
1971 1972 1973
3
1
4
13
5
2
1
6
6
20
16
9
7
0
7
0
5
21
16
0
2
1
19
5
2
1
6
1
18
8
8
5
0
10
3
2
13
13
0
0
4
16
9
0
2
5
0
16
8
7
9
1
11
2
1
15
5
.Cumulative
Prevalence/
Survivors*
1973
3
3
9
48
19
4
4
17
7
54
32
24
21
1
28
5
8
49
34
*Excludes those known to have died, each year, regardless of cause of
death. For example, a 1971 lung neoplasm diagnosis (ICD 162), who dies
in 1971, 1972, or 1973 of heart disease or any other cause is not classed
as a survivor in the year in which death occurs.
60
-------
represents all disease cases for Berks County Medicare claimants who were
alive through 1973. Table IV-18 demonstrates the potential use of Medicare
data for estimating the increasing and/or decreasing population burden of
diseases of special interest to EPA on a county-by-county basis. The annual
data illustrate the availability of equally important trend data. Trends,
together with increasing/decreasing population with certain diagnoses, in and
out of hospitals, present a capability believed to have important applications
in studies of environmentally-related morbidity. The morbidity data resource
may also prove invaluable as an indicator of the health improvements resulting
from pollution abatement and control requirements of EPA. To have suffici-
ently large numbers for statistical analyses, however, it may be necessary to
aggregate age and/or disease categories.
b. Morbidity/Mortality During the 1973 Allegheny County Air Pollution
Episode—
The relevance of the Medicare data with respect to air pollution episodes
was demonstrated by an application centered on a specific episode—Allegheny
County in 1973. This episode was selected because of previous day-by-day
mortality studies by EPA, and the readily available mortality data. It will
be recalled that the Medicare files show day of admission as well as disease
cause. Retrieval and analytical steps can thus be keyed directly to the days
during or after an episode and to any area of interest in the United States
consisting of one or more counties.
The EPA study of health effects related to an air pollution episode in
Allegheny County, Pennsylvania, from August 26 through September 7, 1973.
Findings revealed that the total number of county-resident deaths (all causes)
for each of the 13 days from August 26 through September 7 exceeded the
corresponding 1971-1973 daily mortality averages.
Table IV-19 displays the daily mortality totals for Allegheny County from
August 1 through September 22 for the years 1971, 1972, and 1973. Deaths with
either respiratory or cardiovascular disease as the underlying cause are also
presented in this same table. These latter data show a greater variability,
61
-------
TABLE IV-19
ALLEGHENY COUNTY, PA. RESIDENTS WHO DIED IN THE COIINTY, AUGUST 1 THROUGH SEPTEMBER 22,
1071-1973, WITH SUBTOTALS FOR UNDERLYING DEATH CAUSE OF RESPIRATORY OR CARDIOVASCULAR
DISEASES
D tTE
710001
710SU 2
710 80 3
-- 710804
710805
710806
7 10 80 7
71060 8
710809
- 710810
710811
- 710812
710813
710315
7 10 81 6
710817
710813
710819
710820
710821
710822
710823
710324
71032?
--710826
710827
710828
7 ID 82 3
710830
710831
— 710901
710332
7 1C 90 3
710904
710905
710305
--710907
71Q9C8
710909
710910
710911
710312
-710313
710914
7 10 31 5
7 10 91 6
7 ID 91 7
710918
710919
710920
710921
710922
NO. DE»D IN
TOT*L
38
42
34-
44
35
46
"55 "
40
-38 -
39
35
SI
•— •"*
42
36
45
42
56
39
51
32-
34
44
4 1
36-
34
- 32
34
35
42
35-
31
32
49
56
50
- -itl
52
42
39
44
33
35-
26
37
40
52
35
- 32
51
40
45
RESP
2
0
1
3
1
1
2
_ _ _«i
0
—-T
5
3
— -2
4
1
1
1
c
0
- - 1
T
1-
0
-- 6
1
0
5
1971
C.V.-
17
•7 T
26
21
23
16
24
19
1£
2S
16-
21
-»e _
16
- 25
19
25 -
32
28
17
21
19
17
20 -
16
14
18
2S
20
2 -18-
0
1
0
2
2
3
i
1
1
- -2
T
- — • — - i. -
1
- - 3
7
1
1
1
3
1
2
IE
20
29
38
26
23
29
22
21
- 22
19
IS
16
15
18
30
17
- 16
32
31
29
DUTE
- - —
720801
720803
720804
720805
72030G
720807
-720808
720809
720810
720011
720812
720813
720815
72081S
720317
720318
720319
-720320
720621
720322
720823
720824
720825
-72032S
720327
720323
72082S
72C330
T 20831
-720901
720 SO?
720903
720904
720905
7?0906
7203H7
720<303
720909
720910
720911
720912
- 720913
720914
720315
720315
720317
720918
7?0913
720320
72CS21
720322
NO. DC 40 IN
TOUL
47
«i -»
37
31
35
37
44
38
- -31
23
-- 31
42
30
- 40
35
- - 50
38
52
34
40
45
54
37
37
44
33
44
41
26
44
29
20
42
33
- • 37
28
40
33
45
37
50
34
38
44
43
50
40
47
43
39
RESP
2
2
2
1
2
0
1
- Q -
2
1
2
3
2-
1
5
3
1
1
2
2
1
3—
5
1 -•
2
2
1
1 -
5
0
0
1 -
2
— _^
1
2
1
0
1
0-
1
2
4
1
3
2
O.
3
0
1972
C.V.
19
17
16
17
18
27
22
— 18
14
-.-15
23
16
26
19
- - 30
21
25
24
22
32
34
— 20
17
•— 26
17
24
25
— 11
27
11
11
- 23
20
• - 24
11
19
1 7
26
20
— 26
17
23
24
24
33
21
24
26
24
BITE
NO. DE»D
IN
TOTAC RESP
7 30 30 1
7 3C SO 3
730804
730805
730805
730807
730809
7 30 81 0
730811
730312
7 30 81 3
-7 -»n D-> it
730315
730815
7 30 81 7
-730818
730819
-730820-
7 30 82 1
7 30 82 2
730823
7 30 32 4
730825
- 7 30 325-
7 30 82 7
730828
7 30 82 9
7 30 S3 0
7 30 83 1
- 730201
7 30 90 2
730903
7 30 90 4
730905
730306
7 31 90 7
730903
730 30 3
730910
7 30 31 1
730912
- 730913
7 3C 31 4
7 30 91 5
730916
730917
7 30 91 8
730919
730320
730921
7 30 92 2
51
3 j
38
36
44
- 44
41
44
36 - -
43
--33
45
->f
50
42
41
31 -
34
7 a
55
47
33
40
37
- - 52 -
E S0
p 59
I "
S 51
0 53
-B 58 - -
E 57
44
50
56
43
45
37
38
44
- 43 ---
37
2 H
43
49
41
26
50
27 - -
43
56
4 7
G
n
0
i
4
i-
2
1
1
0
Q
2
3
•7
2
0
0
1
2
0
1
2
0
1
2
2
3
6
7
2
3
5
7
3
3
6
5
6
Z
0
0
0
1
2
2
0
4
1
4
2
1
2
3
1973
C.V.
24
-22
22
20
19
24
24
—-26
26
16
20
15
24
T *?
25
- 23
23
In
17
- 25
30
25
20
?4
17
- 29
28
2C
33
26
33
32
29
23
33
31
23
IT
1C
22
25
26
19
13
21
27
21
IS
25
16
25
30
21
62
-------
as might be expected; but the generally higher number of deaths during the
episode days of 1973 are apparent in these two disease categories, as well.
To determine whether morbidity measures parallel the peaking effect noted
in the mortality data for the episode period, retrievals were accomplished
from the 1971-1973 Medicare files which give a variety of details about hos-
pitalizations during the time in question in Allegheny County, as well as each
of the other counties of the Pittsburgh Standard Metropolitan Statistical Area
(Beaver, Washington, and Westmoreland counties).
One retrieval provided data on the number of admissions for each day of
the 39-day period from August 13 through September 20—the episode days are
the middle 13—for the following respiratory disease categories: acute
respiratory infections, except influenza (ICDA 460-466), influenza (470-474),
pneumonia (480-486), bronchitis, emphysema, and asthma (490-493), other dis-
ease of the respiratory system (500-519), and all respiratory disease (460-519)
Tables IV-20, IV-21, and IV-22 summarize the first retrieval of morbidity
for Allegheny County. For the pneumonia and bronchitis/emphysema/asthma
disease categories (ICDA 480-486 and 490-493, respectively), the counts of
admissions are higher during the episode period than during any of the other
(eight) 13-day intervals for which data is displayed. In those retrievals
showing the classification for all respiratory diseases (ICDA 460-519), the
numbers pertaining to the episode, likewise, appear to be significantly larger.
A second retrieval indicated the number of first admissions for each
respiratory disease category during the same 39 days of August and September
(i.e., only the first hospitalization of a claimant history in a disease
grouping was recorded). This retrieval is reproduced in part in Table IV-23.
The findings are consistent. During the episode, total admissions for res-
piratory diseases were higher, and the total of first admissions alone was
higher than the eight other 13-day periods.
For a third retrieval, the 39 days were divided into three 13-day inter-
vals and the number of people admitted for respiratory disease during each of
63
-------
TABLE IV-20
DAILY ADMISSIONS BY DISEASE CLASS IN ALLEGHENY COUNTY, PA, DURING THE PERIOD
AUGUST 13 THROUGH SEPTEMBER 20, 1971
Date
460-466x
1
Aug 13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Sept 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
1
0
470-474x
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
ICDA Group
480-486x3 490-493x4
0
0
0
1
0
1
0
1
0
1
1
0
1
0
0
0
0
0
0
0
0
0
0
0
1
0
2
1
0
0
0
0
0
0
0
0
1
1
1
0
0
1
0
0
0
1
0
0
2
1
0
1
0
1
0
2
0
0
1
1
0
0
0
0
0
0
0
1
2
0
0
2
0
0
0
0
0
0
500-519x"
0
1
1
1
1
0
1
0
1
0
0
0
0
1
0
0
0
0
0
1
0
0
0
2
1
1
1
1
1
0
0
0
0
0
0
0
0
1
0
460-519x
0
1
2
2
1
1
2
1
1
4
2
0
2
1
1
0
2
0
0
3
1
0
0
2
2
1
3
2
2
2
0
0
2
0
0
1
1
3
1
»Acute respiratory infections, except influenza
-Influenza
.Pneumonia
-Bronchitis, emphysema, and asthma
,0ther diseases of the respiratory system
All diseases of the respiratory system
64
-------
TABLE IV-21
DAILY ADMISSIONS BY DISEASE CLASS IN ALLEGHENY COUNTY, PA, DURING THE PERIOD
AUGUST 13 THROUGH SEPTEMBER 20, 1972
Date
460-466x
Aug 13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Sept 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
0
0
0
0
0
1
0
0
0
0
0
0
0
1
1
0
0
0
0
0
0
0
0
1
0
0
1
0
0
0
0
0
0
0
1
0
0
0
0
470-474x
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
ICDA Group
480-486x3 490-493xA
500-519x 460-519x
1
0
0
1
0
1
0
0
0
1
0
2
1
0
0
0
0
0
1
1
0
0
0
1
0
2
0
0
1
0
0
1
2
0
0
0
0
0
1
0
0
0
0
0
1
0
0
0
1
0
0
0
0
0
1
2
1
0
0
0
0
0
0
2
0
1
0
1
2
1
0
1
1
0
0
0
0
0
0
0
1
1
0
0
0
0
0
1
0
0
1
0
1
0
2
0
2
1
0
0
0
0
0
1
1
0
0
2
0
0
1
2
1
0
1
0
0
1
0
1
2
0
3
0
0
0
3
0
2
2
1
2
1
4
1
3
2
0
0
0
2
2
3
3
0
2
4
1
1
4
3
2
0
1
0
1
1
^Acute respiratory infections, except influenza
-Influenza
.Pneumonia
-Bronchitis, emphysema, and asthma
,Other diseases of the respiratory system
All diseases of the respiratory system
65
-------
TABLE IV-22
DAILY ADMISSIONS BY DISEASE CLASS IN ALLEGHENY COUNTY, PA, DURING THE PERIOD
AUGUST 13 THROUGH SEPTEMBER 20, 1973
Date
460-466x
470-474x
ICDA Group
480-486x3 490-493x4
500-519x 460-519x
Aug 13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Sept 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
0
0
0
1
0
0
0
3
0
0
1
0
1
0
0
0
0
0
0
0
1
0
0
1
0
0
0
0
0
1
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
1
0
0
3
0
0
0
0
0
0
1
1
0
2
0
0
4
0
0
1
0
1
2
2
1
1
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
1
2
0
0
0
0
0
0
0
0
3
1
0
1
1
2
0
0
0
0
0
0
0
1
0
1
0
0
1
0
0
0
0
0
1
2
0
1
2
0
3
1
0
2
0
2
1
2
2
1
0
1
0
0
1
1
0
1
0
1
0
0
1
0
1
0
2
0
1
3
0
1
1
1
3
3
2
2
4
5
1
1
3
2
2
3
2
2
8
1
1
3
1
4
4
2
2
1
1
0
2
2
0
2
2
3
1
1
3
0
-Acute respiratory infections, except influenza
-Influenza
.Pneumonia
-Bronchitis, emphysema, and asthma
.Other diseases of the respiratory system
All diseases of the respiratory system
66
-------
I
a
3 o
_ o
5 u a- a!
O^ >
rtp » g
- u
£ £ o
££ ^o
»-l W
OS ^ CD i
rt U C !
r^ X 1
CT\ H CD i
S ^ 0
Qi I
* o o
2ta
w < a- u c ,
AS 2 2- "i
2 2 c ^ n '
H £ " r" a
= u •" ° = '
iJ W W^ — s
U X i_) ~" '
3 H o
< C <
0< u S C,
3S ^ cc c.
B! z x OJ a •
3 O
-•=•-; -a ^ CD
•-: S H § £, •' =
o " 3 *-j £!)t sc ;^-
M o 2 c -v. c:
CT! < i-J 1,
C/l «£ U
M 33 J "' ^ ~
X O O C ^ ^
< PJ w ^ ^r r: , ,
H .e •« s -
y> - ^ ^ —
Oi W 2ti i— i — 1
KH . 0 ^ ' U | C
tj-t M Pu fl
_ ^ - " -J - ' C
H* X § ^ - = . >
2 eel HH uj
3 O cd -< _, ^
O O W u -v, CL , -
O W Ov nj
H 4J — r-, f-
>« <1 S_, -n — — L
£35 r< c c
tJ a *j
3 ui i o — o ' ;
U I« 0 c ^ C3 l c
ui <; n I
*-J uj u —. !
j tn a £ =o 3 i ..
Q £-> !
fe '^ "; 1 C
O \" O ^ r~^ ~^ '
cii Z C — C
2 e ^ 3 c -. -
r- ' t— -i C • — ^
z <: 3 j — ^ ; c
^D £S d
O M O ,-. ' r
CJ c- W CN ^ 1
CO CO ( ^" ^- ' C.
>< tj <^ t—
-J a: uj ^ -3 ~. , L
•j tn -^ o ! c
- -< L.
o 03 a ^i
-^ ~" , D
•^ § « ^ i
<^ 2- J 1
l a) < i
> « 1
M
fc1 x j x
^ •• „ 1 O 3
-1 M ?, 0 ' r~
3 g d 3° • 3
H z t: i
£ a ! r
^ J3 . 1*.
<| » I a
Cl
" !
a o —i ' — (NJ
cocao
c — c ' — IM
a o a i a J
o — a j — M
o o o ; o a
o o o • — —
a o o | a a
a — o i — (^
o o a ! cj a
D CD O , (V fM
O O O ! O CD
t
a o o i a o
o c a I o o
i
o a o a c
o a o ! a o
3 a — , o ~
c c o ' c c
-« a — — ""i
3 o a i o a
3 o o a CD
:D a ~D , -3 o
o a c ' o CD
0 O 0 0 0
I
3 O IM i 3 --J
r a o CD o
r; a c ' a c
3 O O ; O O
3 a — o -.
"3 C O , C C f-i
r^
— ^ _ M ON
3 O C C CD "
3 — — , O !N.
-> CD O ( T CD
3 C C 1 CD Q
3 c o | o a
D _, _ , o c\i
- C CD , ~ CD
|
D - rs, 1 cj 3-
D a o j o o
3 a o — — i
-I l- _j ! iJ Cj
D — CD i C —
D a a ! o o
3 a — i , ~ ixi
3 i-; c , o a
— ^ ' CD —
5^5,30
J O 3 , — —
D: o _i o
^ -* O '^ ^
3 o cj , cj a
3 -O — , — ~J
O C ! C C.
1
3 3 0 | 3 3
3 J D ' -3 a
: ~- =1 - -
i
X X X X
• o -*i j* >
J. "N »^ — ^
3- 3- ! _1 -1
.T CD , 0 3
aj i> c: *c
» 3" , Jl »
1
O O *•
o a c
c; o <
O CD C
1
- 0 fc
D O <
CD O f
a a c
o o c
CD a f
- o 1
o o c
oat
a c c
1
O D C
o a F
o o t
D C t
1
o c *
a a j
o a *
o o c
a cc i
o o ^
CD 0 F
O CD 1
C C E
O O E
— a :
c c c
.- D :
CD 0 C
CD O •
o a :
i
c a r
c, o t
0 0 t
a o
a c 1
o a t
a o i
o o
o a i
a o i
CJ 0
C 0 !
J 3
0 0
C- CD i
CO"
C 0 t
o 3 :
c. c
n CT (
o o
a 3 J
o o i
X X
J 3-
-0 *• J
3 T
O 3 ;
>£ r*- 3
» »
-, — O r O O (M —
3000 o
1
3 O a 0 3
3 0 O 0 CD
[
3 CD 0 0 0
3 cj a p a
D o o a -
3 c CD c a
-, a — INI aj O
3 a a a p o
« O (V Kl "-" O
30 3 a £ a
: M
-i „ 0 — 0
3 CD o a a
1
3 f\J M 3* 3
r c c p c
~- c — o
3 CD O CD CD
3 C — fM CD
3 CD O CD .— I O
L 0 CD — °M C.
3 a o p ^'J_°
.. a — «g -i
3 a o p a
\j a o r\j c
_, 0 0 P 0
j 0 0 0 —
2 U O CD 0
— ~ fl CJ
3 o a p u
1
D O CD C J
j tj a L- o
3 O O C/ M
3000 a
3 O D a O
3 C 0 tn CD
M _* O »O O
3 O a U 3
3 a a ID a
D O C CD C
—
D 0 O D 0
D 3 — i- 0
D C C CD O
^ CD CD CD C~'
D 0 0 p 0
, 0 0 i. 3
3 0 U p =
j
< X X X X |
D 1 > J- vC
u !> — — -3 '
r 3 Ji jn 3 j
3 o o b CD ;
L. CF C A •£ i
r 3- 01 » r \
\ \
o • a o
a a o
c. c. o
a a i o
o — a
a ' o o
0 — 0
o o o
0 ! tM O
O 1 C3 C
o ' rvj a
a , o a
O — 04
O ' C 0
0 0 —
0,0 a
a' - -
ceo
a a o
o o o
00 —
a a o
CD 3- to
o a o
o o a
CD c a
oca
CD 0 0
CD IV O
O C C
C- CD C
O O 0
c — a
CD O O
CD — O
c, , a 3
a o o
CD a o
c: o c
CD ' a o
CJ O 'NJ
o t- o
a o —
a o a
o o CD
c. u o
a a —
000
O M O
O C CD
O CD O
000
O O O
C C, C
O *-. C3
000
0 ' — 3
, 0 Oj
XXX
J •£ "O
^ J ^ 1
j" a- -T '
o a o
r* a> c? ,
3" J * '
)
0 0
0 0
~ n
O CD
5 c
- N
O < 0
O (M
O O
— 3
0 0
— J
0 0
a —
a o
a i*>
c- o
a o
a -
O 0
— CO
a 3
-M NJ
c c
r* -1
o i_
C 3-
O 0
CV (V
D 0
— ' .%•
Q J
O "i
c c_
N •*>
*^ . i
c c
„ __^
0 0
-3 _
-j CJ
>• ' X
0 J-
-1 .n
0 0
O .3
kfl 3*
67
-------
TABLE IV-24
NUMBERS OF PERSONS ADMITTED WITH INDICATED DIAGNOSES EACH 13-DAY PERIOD:
AUGUST 13-25, AUGUST 26-SEPT. 7, AND SEPT. 8-SEPT. 20, 1971-1973
*
Class
140-239
460-466
471-474
480-486
490-493
500-519
520-577
Aug.
13-25
23
01
00
06
06
06
41
1971
Aug 26-
Sept. 7
44
00
01
01
05
06
44
Sept.
8-20
48
02
00
06
05
04
53
Aug.
13-25
36
01
00
07
02
04
48
1972
Aug. 26-
Sept. 7
42
03
00
05
06
07
44
Sept.
8-20
39
02
00
05
07
08
43
Aug.
13-25
44
06
00
07
04
12
47
1973
Aug. 26-
Sept. 7
(Episode)
44
02
00
13
08
12
36
Seot
8-20
54
03
01
02
03
OQ
40
ICDA Classes
140-239 Neoplasms
460-466 Acute respiratory infections, except influenza
470-474 influenza
480-486 Pneumonia
490-493 Bronchitis, emphysema, and asthma
500-519 Other respiratory diseases
520-577 Diseases of the digestive system
68
-------
the 13-day time periods were counted. Table IV-24 shows these counts. During
the episode period, 23 persons were hospitalized for respiratory diseases.
This was the largest number, by far. The next largest number of respiratory
hospitalizations was 17, the 13-day period just preceding the episode. Medi-
care hospitalizations for respiratory diseases appear related to the quality
of air that characterizes an episode.
Finally, for those individuals who were admitted during the 13-day
episode interval with a respiratory disease, counts were taken of those who
died following a final hospital admission before the end of the same calendar
year. That portion of the calendar year was divided into episode period, the
remainder of the month 01 September, a^d the months of October, November, and
December.
During each year 1971 and 1972 there was no Medicare respiratory disease
death during the interval corresponding to the 1973 episode, while there were
four such deaths during the actual episode in 1973. The only other deaths
indicated by this retrieval were one for respiratory disease in November 1971,
and one for digestive disease during December 1971. There were no other
Medicare deaths.
An interesting example of the connection between morbidity and mortality
is found in the four individuals (4) admitted to a hospital for treatment
during the 1973 episode who died of a respiratory disease before discharge.
Only three claimants admitted for a respiratory illness during the same time
span in either 1971 or 1972 died of respiratory disease following hospital
admission in the remaining three months of the respective calendar years.
However, the small numbers do not justify conclusions.
Admissions for neoplasm and digestive disorders do not show a peaking of
admissions during the pollution episode, such as that found for the respira-
tory disease categories. This finding was expected. It was performed as one
of several tests of the credibility of the Medicare morbidity data. These two
sets of admissions dat re not reproduced or discussed further.
69
-------
Expanding attention from Allegheny County to the larger geographical
area, Table IV-25 presents admissions statistics from all four counties of the
Pittsburgh SMSA for pneumonia and bronchitis/emphysema/asthma in each of the
years, 1971, 1972, and 1973. While the counts from Beaver, Washington, and
Westmoreland counties do not, in the absence of data from Allegheny County,
exhibit any pronounced peaking during the episode days of 1973, the daily
total for residents for the SMSA (Allegheny included) does not obscure the
elevated levels of admissions for the respiratory ailments during that time
interval. The total number of admissions for pneumonia during the episode is
about twice the average of the other eight intervals; for bronchitis/emphysema/
asthma it is also about twice the average for the other eight intervals.
An indeterminate number of the higher quantity of admissions may be
attributed to the unusually warm Allegheny County temperatures that coincided
with the 13-day episode. Episode days numbers 1, 12, and 13 had maximum
temperatures of less than 90° Fahrenheit; days 2 through 11 experienced 90°
and above. By contrast, in only two days of the 104 days comprising the other
eight intervals were there maximum temperatures of 90° or above. Data for
northeastern and midwestern U.S. cities indicate higher mortalities of all
kinds occur as temperatures exceed 80°, and it is reasonable to include the
coincidental higher temperature as a factor contributing to the increased
respiratory disease hospitalizations noted in Table IV-25.
Summarizing this section, various Medicare morbidity measures did reflect
a peaking effect during the Pittsburgh air pollution episode in the same way
that mortality data did. As indicated in several tabulations, total admis-
sions for pneumonia, bronchitis/emphysema/asthma, and all respiratory disease
increased during the episode, as did counts of first admissions for these
respiratory diseases. There is also some indication of a "build-up" in res-
piratory disease hospitalizations coincident with the time period immediately
preceding the air pollution episode. With a 20 percent sample, the three
categories of respiratory disease measures are essentially the same. In a
larger sample, small differences would be expected, but each should reflect
similar health effects.
70
-------
TABLE IV-25
DAILY ADMISSIONS FOR PNEUMONIA AND BRONCHITIS, EMPHYSEMA, OR ASTHMA IN
THE PITTSBURGH SMSA DURING THE 39-DAY PERIOD INCLUDING THE 1973 AIR POLLU-
TION EPISODE AND COMPARABLE PERIOD IN 1971 AND 1972
1971
Pneumonia
1972 1973
Bronchitis, Emphysema, or Asthma
1971 1972 1973
Aug. 13
14
15
16
17
18
19
20
21
22
23
24
25
1
0
0
2
1
2
1
1
0
1
2
0
1
1
1
0
1
0
1
0
0
0
1
0
3
1
1
1
0
0
3
0
0
0
0
0
1
1
2
TOTAL
12
Aug.
Sept.
TOTAL
Sept.
26
27
28
29
30
31
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
0
0
0
0
1
0
0
0
0
0
0
2
0
3
2
2
0
1
1
0
0
0
0
0
1
1
1
0
0
1
0
0
1
1
1
0
0
2
0
2
8
0
0
1
0
0
1
2
0
0
0
2
1
1
0
3
0
0
4
0
3
1
0
1
2
2
_2
18J
2
1
0
3
0
0
0
0
0
0
1
2
1
> Episode
0
0
1
1
0
0
1
0
0
3
2
1
_JL
10
0
1
1
2
0
0
2
0
0
0
1
0
1
1
0
0
0
0
1
0
0
1
1
1
0
0
1
0
1
2
1
0
1
1
0
0
0
0
0
0
0
0
0
0
2
0
1
2
0
0
0
0
0
1
0
2
0
2
1
1
0
0
0
0
ID
7
0
0
1
2
1
0
0
0
0
0
0
2
0
fi
1
1
0
2
3
2
1
1
1
2
0
0
_0
14 ,
0
1
0
1
1
0
1
1
0
1
1
0
0
> Episode
TOTAL
10
71
-------
The Pittsburgh data indicate that increased respiratory disease hos-
pitalizations can be expected to occur during an air pollution episode. On
the other hand, analyses of Medicare hospitalizations for digestive and
neoplasm diseases showed no relationships to air pollution episode timing.
These two findings indicate the Medicare files have sufficient reliability as
indicators of the effects of the environment on human health to justify
further investigation.
c. "Survival" Times as an Indication of Medicare Data Reliability—
The Medicare hospital discharge survey file is not the only source of
national morbidity statistics, although none can match Medicare's extensiveness
and detail. Bi-weekly reports on respiratory and other diseases are published
by the Center for Disease Control, and cancer incidence data are compiled by
the National Cancer Institute (NCI) from information supplied by tumor regis-
tries in nine selected states across the United States. Both are widely used.
The NCI registry attempts to be particularly thorough for the areas in which
it obtains data.
It is feasible to assess the representativeness and utility of the Medi-
care data for a variety of epidemiological research purposes by comparing
findings from these other sources with data on the same subject contained in
the 20 percent 1971-1973 Medicare file sample. Two comparisons were made.
First, average "survival" times from first cancer diagnosis in a claimant's
hospitalization history to the time of death are compared with findings of the
NCI Surveillance, Epidemiology and End Results Reportings (SEER) program.
Second, the Medicare and NCI data were ranked in terms of deaths related to
total patients with a specific neoplasm.
It is noted that NCI uses a somewhat different coding procedure and its
data include cases that pre-date the Eighth ICDA coding used in the Medicare
data. To reduce possibility of errors, ten NCI cancer types were selected
that are believed most directly comparable, and least likely to be involved
with coding complexities. NCI classifies the population of Medicare age under
two age classes; 65 to 74 years, and over 75 years of age. These differences
are taken into consideration, as discussed below.
72
-------
A comparative "survival time" methodology identical with that presented
herein may be applied with respiratory diseases in a context of determining
the intervals between first admission and death for parts of the country that
differ in environmental exposures. The methodology may also have applications
in evaluating the comparative severity of air pollution episodes in different
parts of the country, and/or the severity of a particular episode in reducing
the time-to-death interval below "normal" for that area.
Time-to—death—A retrieval from the Medicare sample data file was de-
signed to provide information on the numbers of individuals with a cancer
diagnosis who die and the approximate time span between the first admission
(during 1971-1973) for a diagnosed cancer and the last admission, following
which the claimant is discharged dead due to either a cancer or non-cancer
cause.
This retrieval is illustrated in part by Table IV-26, which reproduces
most of the detail of the first page of the printout. As shown in the examples
in Table IV-26, a distinction is made between first admissions for cancer
which are the same as the specified ICDA-code cause of death and those admis-
sions for which the discharge diagnoses are in a different cancer category
than that ultimately causing death. A third line of data shows the distribu-
tion of times from the first admission for the cancer to which death is attrib-
uted to the time of the claimant's last admission (that is, cases in which an
admission for a different cancer category occurred earlier in the claimant's
hospitalizations history). The fourth line of data shows cancer-diagnosed
claimants dying of non-cancer causes.
These details are significant for epidemiological studies, and some are
particularly relevant to possible relationships between the environment and
condition of human health. For example, environmental factors have been
variously evaluated as causing 60 to 90 percent of cancer, and this Medicare
morbidity/mortality file offers the only source for including that proportion
of cancer cases which terminate in death from non-cancer causes. It is also
noted from the four examples in Table IV-26, that considerable numbers of
73
-------
1
H
ul
O <3^
H -I
I
a: *-<
W r~
O cj\
c/l
g§
"S
z w
^g
H S
U3
Crf O
M td
si
os ^
ft. O
00
tn 3
§8
PH
cog
w »:
X to
S|
«a
o a
w M
H O
3 Pj
05 <
a: U
H tf:
« M
i-J Q
O •-•
U
>
3
— » X
0 "3
2
i
o-
t*
.1 a
Z 1
t «-
r
< (M
a. r-
V) I
. 0
X
1-1
*•
o
1
1 r-
ijg
ul
O.S
O C
*•• «
O C
1
(
J2 '
0 C
0 (
f— f
r- t
0 C
O I
> (j
IS:
3 «-f
- O C
3 O C
-i .
5
1
3
3
3 OO ,
1 :
:
>j o«- ;
3 O C
3 O t.
M M .
M O C
3 o e
3 0 C
3 C
J :
I
4
3 C
3 | t
3 i
1
»
f
• in N.
O (M OO
o c
0 <
>-i ce '.
_J ul I
OC VJ
< z i
Ul < 9>
o f_
a «-
ui a: o
h» ui O
»- a. o
M I- O
X 0 '
a
ac
!
ac
3 OC
3 OC
3
t
± (
'
i
v> <
ul
0 C
U I
ui :
ac •
0.1
• L
< 1
U
3 1 f
1C , I
U '
LA 0 O C
r- o oc
:l f
c O ^3 r
3 O
3 O
3 O
3 O
3 0
3 O
>J K)
cK r- r- O O
1 O O C
tv o o c
a:
•« ae
J ul
Z w ; l
< Z | I
w o ; i
j a . >— »
C i
C ' t
J
1 1
ac z
ul C
3 1
x z •
fc—
Ul OC Q II
o 1,
i Z t
i 4t .
u •
u c
J Z l
E ul <.
c s .
&
U
J
z
u or a i
— ui a p t
-SO' I
•4 1- O
3 O
3 0
•*
M
1
r-
X 00
Z t
O ro
!"
< rj
a. i-
v> 1
O
111 T-
S
1—
t O*
1
N.
Z
!•
4t
UJ
n a
u
a oc
j
ac
UJ
n ae
^ Ul
a: u
< z
uj « r
11 UJ U t
J U
a r
u z ui ce <
1C «
). u
1
co , a: z
a <; v
u O
< I Z
ac !•
ui ac o "
W Ul
z u .
< Z i
M <
ce
t~ ul t
f« X
1-1 ^>
X 0
a
^
si
o..
O f
o c
0 C
0 C
DOC
DOC
3
3
3
3
3 P- -O
IM r*\ O "••
O r
OC
n
V
n
3
=>
3
3
oc
-Of.
J 0 C
3
3
a: '
i
i
V) I
ul
a t
ul t
U 1
UJ
oc <
0. t
a: ;
UJ <
••
•C
U
a
X
u
.j
s
<
•J
1
e
3
rv u
I •<
o- c
1C
z t «•
O 1*1 r
ir
a. ««- y
u> I a
O c
ul «• C
C
*"
c
r
,J
5 O C
3 r- C
.0.
^ CD *
• r- »•
• (SI C
30C
3 O 0
* f~ r
y r- c
3 0 C
> O
• CH
3 O
J O
a «r
- (M
3 O
3 0
CO N. •* T-
• O >o ry ^
T
• IM C
3 O
o o o o
oc
ul
*•* OS
-1 UJ
ac w
< Z
ui < «-
W OO
0 ' 0
ui ae t-
t— UJO
I- X 0
i-i H- O
X O
o
X 3E «
hB
ui oe o "
E UJ
U W 1
E Z
' ' X
! Z I
a w 31 < a u 31 « < a : t
M4
J
VI
•4 ** f
/I W> *
^
0
V) Vt I/I t/t
m c s: t :
z o ~<
O i < •
VI Kl (•• *•
a a t
•4 I V> Ul Ul I
a (M i x :
0 H- < .
• O O to i
3 Oil
Z 0 Z Z
U 1 -4 c
E ce a -
K 0 f
/I u. j • c
i i o c
E Z i Z C
O O O 0 !
| -N !•* •
V> VI 1
•4 *>H '
/> V)
I/I I/I (/I V! <
D uj u
O X I
\J H- 4
301.
3 1
3 z :
u c
4 Z I
E Ul f
C X
J 31
u ul
C C '
C < t
1 VI k
1
E 2.
c
u
•J
E
«
X
3
A.
1
E
a
e
b
u
^
i,
fc
L
C
C
c
u
2
t H
4_)
3
O
ff
H 4-1
a «
ij — t
33 -J
»-> 3
US rH
3 U
^ C
j
«
^ .£
•H j b
a 1 r- J
•O c
«
o c
O ' O
z
0 0 0 O
i *»4 *
'• (/I C
* >-« t
t (/) 1
•<
n
w tfl vi w
C vt: C E E Z VI XKCX >vt
3
z o a i
3 a i z
O £
«•«< o ««c<< a «c «
> a <
c < •
a
<
z
o
•>">• VI O»— **t-*H»! VI fTlh-H*H-H» V>
to O V) V) V) bO (/» INJ v> 4rt fcO VI V> 1
C W1
a 0 C
< O O <
C X :
3 O <
C
3
O at ac ac at
«J t*> k*
.j
< UJ ul UJ Ul
X X X X
M
H* ^
IA
•»
*r
< o
a -J
w *•-
M a
u
^ »H »
• >• ^
4
•
t
E (M
a Occ
t o o o
O OC JC
J
— v> i/» w> (rt to
X O X
CX a <
DO < (
- c a
DOC
c c
3 O <
c
3
o
•c
a ac , o cs ac uc ac
! _l
< UjUJlUUl! < UIU
X X X £
M «H
**• h* 1
0>
fl^
1
< o
a "4T
LJ f"
" 2
I
k
1
i
< <
a i
_1
U ul LU <
r ac r c
'^ H
*•
A
p*
1
n
p»
9
J
•4
• tw
•4
w
1
1
1
a :
i Ul
c w ;
C Z L
.• < :
OC
u u
c z
U 3> <
U ul u
c c a
• < «
3 v> t,
I 1
E Z 3
u
L OC
C 0
1 u.
1
E Z
o o o o
-
V
1
c
* PH p
0 V> V
^ -^
/> VI
T st
S x a"
a a c
C X
> a
< < < <
(M *
O <•-
IO
Kl 11
a c
- >- H
C C 1
3 0 C
o ac ae t
u
U uj U
XXI
t ' •
j
fO
1 lf\
T~
1
< t **^
a ' m
w ^*
m a
u
•I -4 "
• *• h
• f
1 VI
C X
3 O
t ae
J Ul
C X
• Iw
3J O
T3
<" 5
u c-
y en
-O
0)
3) S
-. jj
^
—
-n H
~Z. .
"• 5
3 —
— -^
-T O
•"•j -j
•~" Ji
T *
Jl '-C
-j en
'J
1)
C "3
12 *-<
-
*c *"*
T D
*" -^
— „
15 w
C
s 3 _>
"• "° ^
3 ^ i
•j 5
J) T3
— ' D
S — 1 CJ3
13 •" J-i
5 ^ 2
C CJ
X 3 -B
U ^ T1
— t cn
l-i 7) 1)
0 Z X
•*-( (C Tj
a u
s c
•Q OJ O
a _S -7*
'j " ^
S U S
•-H _^ Tj
H H ctJ
74
-------
cancer hospitalizations are for cancer types different from the cancer type
that eventually terminates in a cancer death.
This Medicare information retrieval provided the basis for Table IV-27
which shows the average time span in months for the ten selected NCI-Medicare
disease categories from first Medicare cancer hospitalization record to death.
Also shown is the relative rank in parentheses of the particular cause of
death. It is noted that the Medicare tapes available were for only three
years, 1971-1973, and some cancers (prostate, bladder, colon, and rectum) have
comparatively long median survival times. Patients may therefore have experi-
enced Medicare hospitalizations prior to 1971, but for purposes of this
retrieval it was assumed that the earliest cancer admission listed on the tape
was their first cancer hospitalization admission. The net result of this
assumption is to understate the average time span from first admission to
death. For cancer types in which death occurs relatively quickly, the under-
statement is not significant. However, for those cancers with average or
median survival times of several years, an average survival time calculated
from tapes showing 36 months at most would necessarily underestimate the
actual "survival" duration. Nevertheless, if this qualification is kept in
mind, and supplemented with ranking of the time intervals from "first" admis-
sion to death, the findings should provide a basis for evaluating consistency
of Medicare files with NCI files for the same cancer types.
Table IV-27 presents, by cancer disease category, the average number of
months from actual or assumed first Medicare cancer admission to death.
Separate averages were calculated where more than one cancer type was in-
volved. For those cases where death is known to have occurred, although for
a non-cancer cause, the average interval to death was also calculated.
Rankings are shown within parentheses for each of the four broad "first-
admission-to-death" classes.
The Medicare rankings as calculated and shown in Table IV-27 are almost
identical with the NCI rankings of median survival times for these same
diseases. There is also a close consistency between the median survival times
for each cancer type, as calculated for NCI's experience, and the average
75
-------
TABLE IV-27
AVERAGE MONTHS AND RANK FROM FIRST MEDICARE CANCER
TO DEATH OF VARIOUS CAUSES
Disease Category
Pharynx
146-149*
Stomach
151*
Large Intestine
(Exc. rectum)
153*
Rectum/Recto-
sigmoid Junction
154*
Pancreas
157*
Trachea/Bronchus/
Lung
162*
Prostate
185*
Bladder
188*
Brain
191*
Leukemia
204-207*
(a)
Months Rank
7.22 (5)
6.52 (2)
8.67 (7)
8.36 (6)
5.08 (1)
7.14 (4)
8.76 (8)
10.11 (10)
6.64 (3)
10.06 (9)
(b)
Months Rank
5.83 (6)
5.78 (5)
/
7.47 (7)
8.19 (8)
5.62 (3)
5.14 (2)
9.34 (10)
9.23 (9)
4.00 (1)
5.74 (4)
(c)
Months Rank
4.86 (6)
4.18 (3,
5.19 (7)
7.12 (9)
2.66 (1)
4.32 (5)
6.18 (8)
7.96 (10)
3.64 (2)
4.24 (4)
W)
Months Rank
10.93 (9)
7.18 (4)
8.88 (6)
9.66 (7)
6.42 (1)
6.58 (2)
9.85 (8)
11.37 (10)
8.00 (5)
7.15 (3)
*Causes of Death Selected for Comparability with NCI "End Result In Cancer."
(a) First cancer admission is other than cancer cause of death
(b) First cancer admission is same as cause of death
(c) Time from first admission for same cancer as cause of death when first
mention is for other cancer
(d) Non-Cancer Death
Categories of first admission for cancer given in left margin
76
-------
survival times derived from the Medicare tapes for these same types of cancer.
Although the Medicare and NCI ranks are essentially the same where the first
cancer diagnosis/admission is also the cause of death, the average months to
death calculated from the Medicare tapes are significantly less than the NCI
median number of months for four longer survival types of cancer—prostate,
bladder, colon, and rectum. As indicated above, the 1971-1973 Medicare tape
limits cause this difference to be expected.
NCI/Medicare rankings of "survival" times—Table IV-28 presents the
median survival times for the ten cancer categories as calculated from NCI
files for NCI patients 65 years of age and over. The median survival times
shovn in column fe) '--IP b.->. r jatod dire Lly to column (b) of Table IV-27, as
both are concerned with the survival time from diagnosis/admission of the
cancer type that eventually caused death. The times are comparable with the
limitations of the Medicare dat.o taken into account, as discussed above.
Table IV-28 also presents relative rankings, in parentheses, from NCI and
Medicare patient descriptors files, by cancer type. The NCI rankings are of
median survival times; the Medicare rankings are the highest to lowest per-
centages of deaths among those admitted with indicated cancer in Medicare
records. These percentages of death-to-admissions varied from 47.5 percent
(ICD 157, pancreas) to 13.0 percent (ICD 188, bladder). The percentages were
calculated on the assumption that the higher the percentage of deaths for
particular cancer types, the lower the survival time from hospitalization
diagnosis to death. As Table IV-28 shows, this assumption is almost perfectly
correlated with the NCI rankings; of median survival times.
In summary, there is consistency, and a close relationship between
Medicare Morbidity/mortality data and comparable tabulations of the National
Cancer Institute for those „itegorie? of cancer defined in identical or near-
identical terms. Medicare provides a nationwide sample, and NCI offers more
thorough coverage in relatively restricted areas. Neither Medicare nor NCI
purport to have complete files. Available data are not definitive as to which
is more "correct." Medicare morbidity/mortality files provide data not avail-
able through NCI that are relevant to EPA1s environmental health mission, and
77
-------
TABLE IV-28
CALCULATION OF MEDIAN SURVIVAL TIMES FOR CANCER PATIENTS AGE 65 AND OLDER, AND
COMPARISON OF RELATIVE NCI AND MEDICARE RANKINGS
NCI Neoplasm
Subclasses
Number of Patients
Assumed 65-74 over 75
ICDA // (a) (b)
Medical
Survival Time (yrs) Survival Relative Rank
65-74 over 75 Time (yrs) NCI Medicare
(c) (d) (e) (f) (g)
Pharynx 146-149
Stomach 151
Colon 153
Rectum 154
Pancreas 157
Lung and Bronchus 162
Prostate 185
Bladder 188
Brain 191
Leukemia 204-207
801 436
3,490 2,722
6,324 5,215
3,760 2,878
1,768 1,292
6,783 2,304
5,382 5,694
3,474 2,916
1,990*
459 955
1.0 0.9
0.4 0.3
2.2 1.0
1.9 1.0
0.2 0.2
0.4 0.2
3.6 2.0
3.0 1.4
0.3+
0.2 0.2
1.0
0.4
1.7
1.5
0.2
0.3
2.8
2.3
0.3
0.2
(6)
(5)
(8)
(7)
(1)
(4)
(10)
(9)
(3)
(1)
(7)
(4)
(6)
(8)
(D
(2)
(9)
(10)
(5)
(3)
(a) Number of patients age 65-74 diagnosed with disease, 1955-1964.
JL
(b) Number of patients age 75 and older diagnosed with disease, 1955-1964.
(c) Median survival time (years) for patients age 65-74 diagnosed with
disease, 1955-1964.*
(d) Median survival time (years) for patients age 75 and older diagnosed
with disease, 1955-1964.*
(e) Median survival time (years) for patients age 65 and older diagnosed
with disease, computed from columns (a)-(d).
(f ) Relative rank of median survival times in column (e).
(g) Relative rank of percentages of deaths among those admitted with indicated
cancer In Medicare records.
+ Reported only for patients age 65 and older.
*From data compiled by End Results Section, Biometry Branch,
National Cancer Institute, End Results in Cancer, Report No. 4, DREW'Publication
No. NIH 73-272.
78
-------
it is believed plausible, prudent, and appropriate for EPA to evaluate Medi-
care tapes generally as possessing at least the same credibility and authori-
tativeness as NCI tape sources.
6. Medicare Claimant Migration
Linkages between environmental pollution and adverse human health effects
are difficult to determine because of the long time lag between exposure and
appearance of some effects, and due to movement of individuals in and out
of the exposure area during that time interval. Comprehensive, detailed
migration data needed to address such moves explicitly do not now exist in
adequate detail, although stc-ps ^r<"- underway to remedy this deficiency in
significant respects. The Medicare hospital discharge file, showing county of
residence at the time of each admission, as well as age, ICDA code, and other
factors, may be of use in determining migration patterns of a population
believed likely to exhibit the delayed effects of environmental exposures.
To provide some indication of the migration analysis value of the file,
retrievals were made from the merged Medicare files (which had been sorted by
claim identification code) to determine relocations of residents who, at the
time of at least one Medicare admission for treatment during 1971-1973, were
resident in one of the seven study counties in Florida and Pennsylvania.
Counts were made of the number of counties and number of states listed as
place of residence for each claimant over the three-year period. In addition,
the number of counties within Florida or Pennsylvania given as place of resi-
dence were shown. Enumerations were performed separately for males and females,
with a further subdivision for white or non-white race. Tables were prepared
by county of residence, as well as by race-sex category. Table IV-29 sum-
marizes these statistics.
Clearly, a vast majority of the claimants in this three-year data sample
show a single county of residence. Only the non-white residents of Pinellas
County, Florida, have a migration rate greater than five percent, and that is
based on a very small number of cases. The rates in the Florida counties are
79
-------
TABLE IV-29
FREQUENCY DISTRIBUTION OF NUMBERS OF (A) COUNTIES OF RESIDENCE, (B) STATES
OF RESIDENCE, AND (C) COUNTIES OF RESIDENCE WITHIN INDICATED STATE DURING
1971-1973 FOR MEDICARE CLAIMANTS RESIDING IN INDICATED STATE AND COUNTY
Broward County, Florida
White Nonwhite
A. Number 1
of 2
Counties 3
B. Number 1
of 2
States 3
C. Number 1
of 2
Counties 3
within Florida
A. Number 1
of 2
Counties 3
B. Number 1
of 2
States 3
C. Number 1
of 2
Counties 3
within Florida
A. Number 1
of 2
Counties 3
B. Number 1
of 2
States 3
C. Number 1
of 2
Counties 3
in Pennsylvania
Males
2,326
118
1
2,350
94
1
2,421
24
0
Females
2,043
67
2
2,066
44
2
2,089
23
0
Males
Females
63
2
0
63
2
0
65
0
0
46
1
0
47
0
0
46
1
0
Males
2,606
90
2
2,632
75
1
2,691
17
0
Pinellas County, Florida
White Nonwhite
Females
2,443
59
3
2,452
52
1
2,494
11
0
Males
28
2
0
28
2
0
30
0
0
Females
35
2
0
35
2
0
37
0
0
Males
Berks County, Pennsylvania
Whit£ Nonwhite
Females
569
2
0
570
1
0
570
1
0
545
9
0
548
6
0
551
3
0
Males
6
0
0
6
0
0
6
0
0
Females
5
0
0
5
0
0
5
0
0
80
-------
TABLE IV-29 (continued)
Lackawanna County, Pennsylvania
White Nonwhite
A. Number 1
of 2
Counties 3
B. Number 1
of 2
States 3
C. Number 1
of 2
Counties 3
in Pennsylvania
A. Number 1
of 2
Counties 3
B.
A.
B.
C.
Number 1
of 2
States 3
Number 1
of 2
Counties 3
in Pennyslvania
Number 1
of 2
Counties 3
Number 1
of 2
States 3
Number 1
of 2
Counties 3
in Pennsylvania
Males
520
5
0
523
2
0
522
3
0
Females
513
0
0
513
0
0
513
0
0
Lancaster County,
White
Males
540
10
0
545
5
0
545
5
0
Males
830
13
0
831
12
0
842
1
0
Females
544
5
0
556
3
0
557
2
0
Luzerne County,
White
Females
811
14
0
815
10
0
821
4
0
Males
3
0
0
3
0
0
3
0
0
Pennsylvania
Females
0
0
0
0
0
0
0
0
0
Nonwhite
Males
4
0
0
4
0
0
4
0
0
Pennsylvania
Females
6
0
0
6
0
0
6
0
0
Nonwhite
Males
1
0
0
1
0
0
1
0
0
Females
0
0
0
0
0
0
0
0
0
81
-------
TABLE IV-29 (continued)
Schuylkill County, Pennsylvania
White Nonwhite
Males
A. Number 1
of 2
Counties 3
B. Number 1
of 2
States 3
C. Number 1
of 2
Counties 3
in Pennsylvania
459
5
0
461
3
0
462
2
0
Females
432
1
0
432
1
0
433
0
0
Males
0
0
0
0
0
0
0
0
0
Females
1
0
0
1
0
0
1
0
0
82
-------
generally greater than those in the Pennsylvania counties. No hospitalization
history gives more than three residence addresses. The number of individuals
admitted to hospitals in more than one study county is inconsequential.
The examination of Medicare claimant migration did not indicate as
widespread a relocation pattern as one might expect from the retirement age
population, especially in the high in-migration counties of Florida. It may
be that most individuals who change residences do so shortly after the retire-
ment date, while still in relatively good health. The data do indicate that
once hospitalization occurs, few individuals change their residence.
The Medicare records , ?., > u efore, not a good source of data either on
migration patterns, or on the contribution of previously incurred environ-
mental/occupational exposures to health statistics of various geographic
regions. A more complete representation could be found by examining the
claimant's employment history by industry and county in conjunction with the
Medicare data. Steps have been initiated to obtain this industry/county data
for the working population prior to their retirement years, and for linking
county/industry of employment with county or counties of retirement.
7. Demonstration—Mo"'bidit3r/Mortality Analyses in Steel Industry Counties
Considerable interest was expressed within EPA for determining whether
the techniques and information being used in this project could be applied to
an investigation of a single industry Dn a county level of detail. The steel
industry was of particular concern because of litigation initiated by steel
Industry repr.-jeutacives concerning EPA enforcement of a standard for Total
Suspended Partirulate. Under the direction of the EPA Project Officer, the
morbidity <,xpeiience of tiie Medicare population in steel-producing counties
ras featured i, the example application. It is emphasized that the work on
iliis task is designed only to demonstrate the versatility of the data bases
for EPA purposes and not to establish the validity of arguments on either side
of the issue under contention.
83
-------
a. Identification of Steel Industry Counties and Data Collection—
The work involved identification of the major steel-producing counties of
the United States, and comparisons of demographics, pollutant emissions, air
quality, morbidity and mortality rates with U.S. totals and statistics from
steel-producing and non-steel-producing counties. A list of locations of
steel-producing establishments meeting the requirements for Standard Indus-
trial Classification 3312 was provided by EPA. This classification includes
blast furnaces (including coke ovens), steel works, and rolling mills. A
compilation of the counties with establishments classified under SIC 3312 is
given in Table IV-30; there were 97 such counties in 31 states. Figure IV-7
shows the geographic distribution of these counties (except Honolulu, Hawaii).
The form used for entry of all relevant data elements is shown in Table
IV-31. The first of its six sections is administrative, containing the state
and county names, the EPA/SAROAD number (the identifier used by the SSI-
developed POPATRISK data base), and the Social Security Administration state-
county number used with its Medicare hospitalization survey files.
The second section contains Bureau of Census county demographic infor-
mation retrieved from POPATRISK: total county population in 1970, county
population divided by county land area, estimated percent of population
living within town or city boundaries, percent of population that is white,
percent female, percent over the age of 64 years, absolute number of white
males and white females over the age 64, and proportion of white males and
white females over 64 years of age in 1970 who resided (a) outside the county
in 1965 and within the county in 1970, or (b) inside the county in 1965 and
outside the county in 1970.
The third section contains totals for emissions (in tons per year) of TSP
and S02, the summation of emissions by individual point sources within the
3
county, and the arithmetic means (in yg/m ) of the site-specific geometric
means of ambient TSP and S02 measurements taken over one-hour or twenty-four-
hour sampling intervals in 1973 at county monitoring sites designated for
population-oriented, source-oriented, and background surveillance, respec-
tively. Emissions values are taken from EPA/NEDS files, monitoring data are
from SAROAD.
84
-------
TABLE IV-30
LIST OF 31 STATES, 97 COUNTIES, TWO INDEPENDENT CITIES OF THE
UNITED STATES IN WHICH ARE LOCATED MANUFACTURING ESTABLISHMENTS
WITH SIC CODE 3312--BLAST FURNACES (INCLUDING COKE OVENS),
STEEL WORKS, AND ROLLING MILLS
(Source: NEDS)
Alabama
Etowah
Jefferson
California
Alameda
Los Angeles
San Bernadino
Colorado
Pueblo
Connecticut
Fairfield
Delaware
New Castle
Florida
Hillsborough
Martin
Georgia
Fulton
Hawaii
Honolulu
Illinois
Cook
Madison
Whiteside
Indiana
Allen
Henry
Howard
Lake
Marion
Porter
Vipo
Kentu -ky
Boyd
Campbell
Daviess
Louisiana
Tangipahos Parish
Maryland
Baltimore City
Baltimore
Michigan
Macomb
Wayne
Minnesota
Ramsey
St. Louis
Missouri
Ft, Louis City
SI. Louis
New Jersey
Middlesex
New York
Albany
Chautauqua
Erie
North Carolina
Mecklenberg
OhiiD
Ashtabula
Butler
Cuyahoga
Jefferson
Lake
Lawrence
Lorain
Lucas
Mahoning
Richland
Scioto
Stark
Trumbull
Uut .dngton
Oklahoma
Tulsa
Oregon
Multnomah
Yamhill
Pennsylvania
Allegheny
Beaver
Berks
Bucks
Butler
Cambria
Chester
Pennsylvania (cont.)
Dauphin
Erie
Lawrence
Lebanon
Mercer
Mifflin
Montgomery
Northampton
Northumberland
Philadelphia
Venango
Warren
Washington
Westmoreland
South Carolina
Darlington
Georgetown
Richland
Tennesse
Hamilton
Rnox
Roane
Texas
Austin
Chambers
El Paso
Gray
Guadalupe
Harris
Morris
Tarrant
Utah
Utah
Virginia
Buchanan
Washington
King
West Virginia
Brooke
Cabell
Hancock
Marion
Wisconsin
Milwaukee
85
-------
4J
•H
M)
cs
W -H
4J ^
C! rH
01 O
CO T3
•H C
iH CO
,0
tO «
•U CO
co .M
a) M
o
w> >
c
•H f-l
t-l 0)
3 a) •— «
w *j d
O M g
tO O
. W
C CO
n3 C 4-1
8 > §
•O O C
0) «
•U CU -H
CO ^ -H
O O (0
O ^
3 -U
J3 -H C
003
•H G O
J3 -H O
C CO iH
•H CO 3
O ,H
CO tfl O
CO B
M
•
00
86
-------
TABLE IV-31
DATA ENTRY FORM
Serial //
/ Section 1
State
/
Population Ppn.
/ /
Males Females
Over 64 Over 64
/Mi.
%
65+
County
/ /
2 % Urban % White
/ ,
Male % Female
IN-MIGR 65+ IN-MIGR
SAROAD #
/ /
% Female
/
% Male
65+ OUT-MIGR
SSA #
/
% Over 64
/
% Female
65+ OUT-MIGR
Section 2
Emissions: TSP
so2
Monitoring: TSP
(Site Means) SO
County Total
County Total
Population
Population
Point Sources
Point Sources
S'jiicce Background
Source Background
TSP SO Sec-
Mi2 Mp" tl10n ,-
% 3312 TSP
% 3312 SO
X
1
\.'n Top
500)
TSP Emissions
SO Emissions
Mean TSP Predict
Monitoring
White
Section 4
Mortality Rates from
All Neoplasms (ICDA 140-239)
Cancer of Buccal Cavity /Pharynx (140-149)
Cancer of Digestive Organs/Peritoneum (150-159)
Cancer of Respiratory System (160-163)
Cancer of Bladder (188)
Diseases of the Respiratory System (460-519)
Acute Inter stitial/Bronchopneumonia (484,485)
Bronchitis, Emphysema, Asthma (490-493)
Male
Female
".-. • I care
Morbidity:
Section 5
All Respiratory Diseases (ICDA 460-519)
Acute Infection, Except Influenza (460-466)
Influenza (470-474)
Pneumonia (480-486)
Bronchitis/Emphysema/Asthma (490-493)
Other Respiratory Diseases (500-519)
White Claimants / Rates per 1000
Male Female / Male Female
Mortality
Rates:
Male
Female
Section 6
Influenza and Pneumonia (470-486)
Bronchitis (490-492)
Asthma (493)
Other Bronchopulmonic Disease (460-466,
500-519)
Cancer of Buccal Cavity/Pharynx (140-149).
Cancer of Digestive Organs/Peritoneum (150-159)
Cancer of Respiratory System (160-163)
Cancer of Urinary Organs (188,189)
87
-------
Indication is given in the fourth section of each steel-producing county's
rank, if applicable, in the top 500 counties of the United States with regard
to TSP emissions, SCL emissions, mean predicted for TSP monitors (calculated
by an algorithm for counties in which TSP and NO emissions fall within a
X
specified range), the mean of county population-oriented TSP monitoring sites,
and white male/white female mortality rates for all neoplasms, certain sub-
groupings of malignant neoplasms, and three respiratory disease categories.
The emissions and ambient ranks were derived from the POPATRISK data base.
The mortality ranks were done under an earlier SSI-EPA contract, "Investiga-
tion into the Industrial Correlates of Environment Related Mortality."
Section five presents a compilation by SSI of data concerning white males
and white females from the Medicare sample. Statistics were obtained of the
number of admissions/discharges and of the number of unique claim numbers with
diagnoses of six respiratory disease classifications. The sixth division of
the data entry form displays mortality rates, averaged over the years 1969 to
1971 (POPATRISK data base) for white males and white females of all age groups
in four respiratory disease and four cancer categories.
b. Comparison of Morbidity in Steel-Producing Counties with State
and U.S. Totals—
Total numbers of white male and white female Medicare claimants requiring
hospitalization in 1971-1973 for each of the six respiratory illness classi-
fications for the 97 "steel" counties were obtained and divided by the respec-
tive total white male and female populations over the age of 64 years. Rates
per 1,000 for each of the six disease categories (male and female) were com-
puted and compared with total U.S. ratios and those of the complementary set
of U.S. counties (i.e., without data from the 97 "steel" counties). As seen
in Table IV-32, for each of the six disease groups for each sex, the steel
county morbidity rates are lower than the corresponding total U.S. rates.
These findings should not be construed as evidence that steel industry
emissions are not harmful. Individuals more susceptible to respiratory dis-
eases caused or aggravated by a co-located steel plant may have died before
88
-------
to
to o
H H CO
^2 ^c i*3
pj N M
H (4
B 3 8
H H W
O H H
S S5 6
2 o
O S3 W
S3 to
Prf <
O O W
S3 fn co
M M
Q CO £3
55 PJ]
O M 51-1
to ps o
§ 8 pi
8M
• P-i
CO CO
CM si • w
co p -3 g
> e j 2
H e d M
>> Ja o
H §
S3
O
O CO vf
0) VO
W CO OJ
W 2 ofi
H <
to oj
4-* J-l
•H OJ
rj ^>
^E CJ
tn rH ON CO rH in
rH vO CM vD CT> O
rH O CO rH vO CM
ON vO CM vO CTN'CO
VO CM rH CM
00 CM O
H
^^
vO CJ
VD O
»y "H
*~+ t W
ON O cd
rH VO s~* rH
to *^~ co ^**^ 3
I VivX 0\ o\ cX
^^ «^" rH O
vo n) i m P*
•»
• •a- -sr >» »-i
C* o o o o. w
o «H r-« oo e «
4J 4J *& •& W 1-1
Ctl O ^^ ^x *^ *rH
|4 01 CO DM
vl *M Ctt CO *H CO
ex c M -H o oj
OJ M C C tH Oi
OJ 0) O X
PCS ft) 3 g 0 M
*J rH 3 d CU
rH 3 «*-! QJ O JC
^ssss
to
cu
rH
1 +
O vO
O CU
O 4J
rH -H
to 3
H
K CO
CU
rH
& +
OJ vD
4J
•rH
^
CO
4» -*
rH VO
cd
g eu
p4
H X O
o JS
H
. .
CO CO <±
• OJ vD
rt QJ
^
CU
•H CU
rj ^
^3? Cj
t^ co oo 3" CO CO
rH
^••^
vD
VO
^^
^°** \
O\ O
rH VO s~^
•n -* co <-^
1 ^-^ ON ON
O St rH
vo cd i m
- co w X
• \ )-<
fs C 1 I X O
w O CD o a. 4J
o "H r-» oo 6 d
0} O **-* *^^ **^ *H
t4 cu co ex.
^ «*H ed cd tH to
ex c N TH u cu
w M ci d TH ccj
o> cu o x
ec! cu 3 6 o M
*J rH 3 CJ CU
•-H 3 m QJ O X
r-J 0 pj C »-* 4J
<: <: M PL, « o
vO
vO
^T
oo
VD
•k
O
rH
m
vO
m
^>
vO
•.
r*. •
C
O
Tt
CO
rH
a.
o
.
o
•rl
4J
cd
0
•H
«4H
CO
CO
cd
rH
CJ
CU
CO
t
rH Cd
CO
to
o a
U-l C1
s
CO -»H
4~> a)
a •-*
3 cj
o
O rH
CO 3
cu -o
•a -H
X d
W M
* +-
89
-------
attaining Medicare age/eligibility, or may have relocated to a less hazardous
environment. Available resources did not permit introduction of these and
other factors affecting conclusions on possible environmental health
relationships.
Steel county morbidity rates were then compared with the corresponding
rates of their respective states. It is noted that the state rate included
data from its steel-producing counties. Table IV-33 indicates the counties
with respiratory morbidity rates higher than state rates. It is also noted
that for only male disease categories 1 and 4 (i.e., all respiratory disease
and pneumonia) was there a majority (actually 51.5 percent) of steel counties
with rates greater than their corresponding state rates.
c. Investigation of Relationship Between Morbidity and Pollution Levels—
To determine whether a relationship could be found between manifested
health effects and proportion of TSP and SO point source emissions attribu-
X
table to steel production facilities, a retrieval from NEDS keyed to sources
with SIC 3312 was accomplished. The emissions values of the steel-producing
sources were totaled by county and divided by respective county point source
totals. This percentage was then entered on the data entry form, along with
computations of TSP and SO estimated total emissions per square mile of
X
county geographic area. The percentage provided an estimate of the relative
amount of the total emissions that could be attributed to the steel producing
plants in each county.
Data for the 41 steel-producing and the 114 non-steel counties of Ohio
and Pennsylvania (the states with the largest number of steel-producing
counties) were examined to discern possible patterns of relationships among
variables for which values have been obtained on the data entry forms. The
NEDS retrieval of emissions values from individual SIC 3312 point sources
indicated that a number of counties with significant steel production opera-
tions were omitted from the earlier listing provided by EPA (for which the
steel/non-steel county analysis was requested). The counties added for this
two-state analysis were Belmont, Columbiana, Marion, Muskingum, and Tuscarawas
in Ohio and Lancaster, Pennsylvania.
90
-------
TABLE IV-33
RESPIRATORY DISEASE CATEGORIES FOR WHICH MEDICARE CLAIMANT RATES
IN "STEEL-PRODUCING" COUNTIES ARE GREATER THAN RESPECTIVE STATE RATES
STATE COUNTY/CITY
Alabama Etowah
Jefferson
California Alaraeda
' Los Angeles
San Bernadino
Colorado Pueblo
Conne ct i cu t Fair fie Id
Delaware New Castle
Florida Hillsborough
Martin
Georgia Fulton
Hawaii Honolulu
Illinois Cook
Madison
Whiteside
Indiana Allen
Henry
Howard
Lake
Marion
Porter
Vigo
Kentucky Boyd
Campbell
Daviess
Louisiana Tangipahos Parish
1
X
X
X
X
X
X
X
X
X
X
X
X
X
Maryland Baltimore City"*" 1
Baltimore I x
Michigan Macomb
Wayne
Minnesota Ramsey
St. Louis
Missouri St. Louis City"1"
St. Louis
X
X
WHITE MALES
(
2
X
X
X
X
X
X
X
X
X
X
X
SAT]
*4
X
X
X
X
X
X
IGOl
4
X
X
X
X
X
X
X
X
X
X
X
X
X
IY
c
X
X
X
X
X
X
X
X
X
X
X
6*
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
1
X
X
X
X
X
X
X
X
X
X
x
X
WHITE FEMALES
<
2
x
x
X
X
X
X
X
X
:ATI
3
x
X
X
X
X
X
X
X
X
X
X
E:GOI
4
X
X
X
X
X
X
X
X
X
X
X
IY
r
X
X
X
X
X
X
X
X
x
X
X
6*
X
X
X
X
X
X
X
X
X
X
x
x
X
91
-------
• RESPIRATORY DISEASE CATEGORIES FOR WHICH MEDICARE CLAIMANT RATES
IN "STEEL-PRODUCING" COUNTIES ARE GREATER THAN RESPECTIVE STATE RATES
STATE COUNTY/CITY
New Jersey Middlesex
New York Albany
Chautauqua
Erie
North Carolina Mecklenburg
Ohio Ash tabula
Belmont"*"
Butler
Columbiana+
Cuyahoga
Jefferson
Lake
Lawrence
Lorain
Lucas
Mahoning
Marion+
Muskingum*
Richland
Scioto
Stark
Trumbull
Tuscarawas+
Washington
Oklahoma Tulsa
Oregon Multnomah
Yamhill
Pennsylvania Allegheny
Beaver
Berks
Bucks
Butler
Cambria
Chester
Dauphin
Erie
i
Lancaster^
Lawrence
Lebanon
Mercer
1
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
WHITE MALES
C
2
X
X
X
X
X
X
X
X .
X
X
X
'
X
X
X
X
X
X
X
X
:ATI
3
X
X
X
X
X
X
X
X
X
X
X
:GOB
4
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
.Y
5
X
X
X
X
X
X
X
X.
X
X
X
X
X
X
X
X
X
X
X
X
i
6
X
X
X
X
X
X
X
X
X
X
X
1
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
WHITE FEMALES
C
2
X
X
X
X
X
X
X
X
X
X
X
X
X
X
:ATE
3
X
X
X
X
X
X
X
X
X
X
x
x
X
X
X
X
GOF
4
X
X
X
X
X
X
X
X
X
X
X
X
X
x
X
X
X
.Y
5
X
X
X
X
X
X
X
X
X
X
X
X
X
X
x
x
X
£
6
X
X
X
X
X
X
x
X
X
x
x
x
x
X
92
-------
TABLE IV-33 (continued)
RESPIRATORY DISEASE CATEGORIES FOR WHICH MEDICARE CLAIMANT RATES
IK "STEEL-PRODUCING" COUNTIES ARE GREATER THAN RESPECTIVE STATE RATES
STATE COUNTY/ CITY
Pennsylvania Mifflin
(Continued) Montgomery
Northampton
Northumberland
Philadelphia
Venango
Warren
Washington
Westmoreland
South Carolina Darlington
Georgetown
Richland
Tennessee Hamilton
Knox
Roane
Texas Austin
Chambers
El Paso
Gray
Guadalupe
Harris
Morris
Tar rant
Utah Utah
Virginia Buchanan
Washington King
West Virginia Brooke
Cabell
Hancock
Marion
Wisconsin Milwaukee
1
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
WHITE MALES
(
2
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
;ATI
3
X
X
X
X
X
X
X
X
X
X
X
X
X
:GOE
4
X
X
X
X
X
X
,
X
X
X
X
X
X
X
X
X
X
IY
5
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
6*
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
1
X
X
X
X
X
X
X
X
X
X
X
X
X
WHITE FEMALES
C
2
X
X
X
X
X
X
X
X
X
X
X
X
:ATE
3
X
X
X
X
X
X
X
X
X
X
:GOF
4
X
X
X
X
X
X
X
X
X
X
X
X
IY
5
X
X
X
X
X
X
X
X
X
X
X
X
X
X
6*
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Category 1: All Respiratory Diseases (ICDA 460-519)
2: Acute Infection, Except Influenza (460-466)
3: Influenza (470-474)
4: Pneumonia (480-486)
5: Bronchitis/Emphysema/Asthma (490-493)
6: Other Respiratory Diseases (500-519)
Data not available at this f.ime.
93
-------
Cluster diagrams were prepared indicating observed relationships between
(1) the Medicare hospitalization rate among white males for all respiratory
diseases (ICDA 460-519) and monitored ambient TSP concentrations, (2) the
Medicare hospitalization rate among white females for bronchitis/emphysema/
asthma (ICDA 490-493) and TSP emissions, and (3) the Medicare hospitalization
rate among white males for "other respiratory/bronchopulmonic disease" (ICDA
460-466, 500-519) and white-male age-adjusted mortality rate for the same
disease classification.
The first two sets of cluster diagrams showed no significant associations
for either the Ohio or Pennsylvania counties. The third set of cluster
diagrams (Medicare hospitalization rates and age-adjusted mortality rates) for
"other respiratory/bronchopulmonic disease" for white males showed the most
pronounced correspondence. Correlation coefficients were computed separately
for Ohio and Pennsylvania steel counties and were above 0.60 for Pennsylvania
non-steel and Ohio steel counties. It was 0.21 for the Ohio non-steel counties.
d. Geographic Distribution of Morbidity—
The male and female morbidity rates (i.e., number of Medicare claimants
hospitalized in 1971-1973) in each of six respiratory disease categories for
the 155 Ohio and Pennsylvania counties were listed in rank order, and the
respective ranks were entered onto the individual county data entry form.
Ranks were also entered on "working maps" of Ohio and Pennsylvania showing the
morbidity rates for ICDA categories 460-466 (acute infection, except influenza)
and 490-493 (bronchitis/emphysema/asthma).
There was no consistent steel industry locational significance to the
county morbidity rankings displayed on the maps. However, shortage of funds
and time precluded meteorological modeling of any kind. These same constraints
made it impossible to include other industries in the analysis.
In summary, no consistent relationship between steel production and
morbidity rates in the same counties has been shown by this limited pilot
investigation. The absence of such a relationship is believed to be extremely
important to evaluations associating adverse human health effects, such as
94
-------
respiratory diseases, to airborne emissions from the steel industry alone or
to a significant degree. Similarly, the absence of a strong steel industry-
morbidity relationship demonstrates the need for refined industry-morbidity-
mortality research to assist in guiding abatement and control research
priorities on the one hand, and establishing regulation and environmental pro-
tection priorities on the other.
8. Computer-Prepared Maps Comparing Pollution/Morbidity Factors
To demonstrate the relevance to health effects analyses of geographical
relationships derivable from the Medicare morbidity tapes, several different
procedure^ were explored. These procedures feature combining statistics with
maps of the United States that show data and geography to the county level of
detail. Some maps were prepared entirely by computer, utilizing techniques,
programs, and equipment readily available to Health Effects Research Labora-
tory. Others were prepared by "hand," such as the preceding Ohio and Pennsyl-
vania maps whic h identify respiratory disease rates of steel counties by
encircling the data.
For the computer-prepared maps, EPA Research Triangle Park equipment was
used exclusively. The maps can display data from Medicare morbidity files and
many other sources, relatively inexpensively, in multi-color format, and with-
out significant delays or additional specialized programming expenditures.
Display of statistical data in map format is helpful analytically in
revealing efficiently the presence, and absence, of relationships that are
difficult to detect from tabulations alone or without detailed familiarity
with geopolitical spatial relationships. For health effects analyses, it is
important . > no-e th • t computer techniques enable designation of one or two
variables s or man;, variables. The variables are not limited to the Medicare
n.crbidity tapes, but can be selected from EPA and many other data files.
Two maps were prepared, utilizing EPA facilities, to assist in the
analysis of possible relationships between respiratory diseases (as quantified
by the Medicare morbidity tapes) and steel-producing facilities. Respiratory
95
-------
diseases were selected as indicative of adverse health effects caused by
industrial airborne emissions of concern to EPA.
The first map, Fig. IV-8, is a cut-out of a computer-prepared U.S. map
identifying counties in Ohio and Pennsylvania and adjacent areas listed in the
NEDS data base as containing one or more SIC 3312 steel-producing establish-
ments. These "steel" counties are then sub-classified graphically, depicting
one of four categories of respiratory disease hospital claimants rates per
1000 white males over 64 years of age. The respiratory disease data are taken
from the Medicare morbidity tapes; the population data from POPATRISK. It
will be recalled that county-by-county comparisons with state average rates
for six different respiratory disease categories are shown in Table IV-33.
The second computer-prepared map, Fig. IV-9, illustrates the use of this
technique in analyzing possible "steel production" and respiratory disease
relationships in greater detail. In this more complex map, the hospital
respiratory disease rates for the counties that do not produce steel are shown
in blue.
This more complex map also differentiates among the steel-producing
counties in terms of the estimated TSP emissions from the steel industry as a
percentage of all point sources in the county. In Ohio and Pennsylvania there
are a total of 21 counties with steel plant estimated TSP emissions aggre-
gating up to 12.7 percent of the county's point source TSP emission. These 21
counties are shown in red.
The 20 remaining steel-producing counties have steel-producing sources
that emit from 12.8 percent to 86 percent of the TSP point source emissions in
each county. These 20 counties are shown in black.
For each of these three categories of counties (no steel production,
counties with steel industry sources aggregating 12.7 percent or less of all
TSP emissions, and counties with steel-producing sources emitting more than
12.7 percent of all TSP emissions), the Medicare respiratory disease hos-
pitalization patient rates are shown by one of four categories of rates per
1,000 white males over 64 years of age.
96
-------
£*,
o
CO
•H
a
CO
CU
ce
CU
en
CO
CU
en
a
Ul
V
a.
ft
c
^
B
^
o
c
o
CO
iH
3
a.
0
a<
0
o
o
-^
OJ
>
o
j-J v|
l-c —I
O OJ
CU •
^H e/3
CU I
01 C
££
O U
3 C
T3 3
O C
!-l O
a
.
co o
o
.c u
o
(U CO
to n
CO 4-i
4) X
en U
I ;r^.
O ^J U
O O 4J
3 -rf
E -i
•t)
TJ CJ
01 CO
N |
CO 00 3
3 " §•
^ •» 9
1-H U O
1-) I—t
CO I
01
cy Dfi
•rf U CO
u cu
O > CO
50 3 01
tU -H
u C ^J
CO O
O -
en co o
OJ —I
•H 3 r
-J C, ^H
COD
3 C. 01
C A-l
•j o a
o i
=n c; c
c ^-< o
— ( Z
O 1-4 =
00
I
vVW
97
-------
ratory
•H
tn
01
ai
ease
mants
cn -H
Q .2
CJ
*>
>s
)w
O
60
01
CO
U
.H
QJ
01
4-1
CX
O
&
0
o
o
^
01
a
c
0
T<
cn
cn
B
U
DM
£_|
U-
0
.
O
Z
-3-
OJ
o
SJ
4J
C
01
CJ
1-1
01
0-
cn
u
it
C
3
O
0
CN
v£
v|
_1
U
H
C/5
1
Z
z
1 ]
0
ft
vD
1
VC
J
U
C/5
1
Z
o
z
0-
CN
C?* O CN *£> O"i O
co in so r^ co in
II v| 1 1 l
r^ O c*1 r~* C
i— Oi vo t— ov
r-~ f*^ r^. r^
CN CN CN CN
<-t t— t l-t 1-*
J J
u [t]
H H r r r =
t/) t/3 _3 k-3 -3 *J
1 t fi3 tiJ hJ tol
s: z; uj w uj w
O O H H H 6-«
z 2: c/i y^ vi v.
uncissa
-H
V | | |
n"i (*•••
\c r^«
S 2 S
CD CO CO
I 1 t
ao co ac
CN CN CN
— I rM M
J 1-J (-Z
u^ to uj
H H H
SSB
v£J sD sD
O
m
\
o
t*
2
CO
t
00
CN
iH
t-H
Cd
u:
u^
fN
00 C C
C O -H
CD E CR
I W >,
CD CC
01 a.
CO E
CO 3 O
u cj
co d T-t
T3 JM CO
3 CJ
01 ac
cn c T-I
o .2
0) *H D.
—( JJ CO
CO O -J
> 1- J=
U-l 4J
O ±>
eti^: -*
01 JJ i-t
cu 4J
c3 u-i
i-l O
&,
K iH
•H CO
•a c
0) 00
o c
cn
cd -•
ia
co o c
E z in
(1)
c
•H
rH
^
O
c
u
o^
CN
O
a.
CO
*ts
(U
_o
g
OJ
CO
c
CD
a,
&
to
G.
£
&
*w
QJ
CO
j_i
p\
r— t
— '
-H
-J
c
•H
*-"
01
OJ
4-1
OJ
E
CO
CO
a
C
T3
QJ
CO
c
U-j
ft
,^
<
>
.Z
w
B
CM
trf
c
o
cu
c
QJ
QJ ^j
4J U
3 CO
C.-H
II
•o
cn <
CL 4J CX
E O E
3 d -H
cn
c -a e
o c co
u co 1-4
oo
cn •• o
F-< 01 *-l
•H 60
C C
• JJit
OT3
01 CO
O .Q CO
^^"
CJ OJ
J-l CO
^t AJ OJ
= c S
E CD
B -a
T3 Q. C
N 3
o o e
Q. 3
OJ o
>, > C->
3
a.
CJ
U-l
O
d
o
•H
4~>
1^4
4J
0)
3
•H
M
•— •
OJ
jr
CO
•H
Q
i-t
3
i^
t
U
d
(D
P
V4
CJ
O
d
o
c
0)
f.
•***
D-
g
a)
to
>-,
CO
(0
JJ
to
p
Lj
•H
.
4J
QJ
C
O
U
J
a
c
CO
o
CO O
CJ
^
F-H CJ
C 0)
o ^
OJ
•"-> C
C-H
01
cn-o
QJ OJ
h-t U
O.3
•o
01 O
CO O.
.
^,
01
4J
o
rH
O.
c
CJ
a
— O
CD *J
11 C
^ B
0 0)
*
C B u-
0) TH O
.
01 0)
>•, 1- >,
^ CC
O -3-
•U *H ^
01 -3
it CU W
i_ Z 01
-H >
X TJ O
J CO c.
c cn c
3 —. B.
0 H
o o< o
o
nj OJ c
c c
CO m IJ
> 3 01
— CJ O.
>, •
cn
cn
a.
"O
o>
98
-------
As a partial analysis of the co-locational significance of steel-producing
facilities to a county's Medicare respiratory disease rates and experience, it
is interesting to compare the counties that have the highest respiratory
disease rates.
Steel-producing counties—Only two of the 20 "black" counties (the
higher percentage groups of steel industry emissions) are also in the highest
category of respiratory ^ibt.aoe ra^as. The other 18 counties are equally
distributed among the three remaining classes of respiratory disease rates.
These data do not show a direct association between steel industry TSP emis-
sions and higher respiratory disease rates for the co-located county
population.
Among the 21 "red" counties (the lower percentage group of steel industry
emissions), only four counties are in the highest category of respiratory
disease rates. An equal number of counties is in the lowest category of
respiratory disease rates. Of the remaining 13 counties, seven are in the
next highest category of respirator" disease rates. No clear conclusion can
be drawn from this group of steel- producing counties and their respiratory
disease experience with the population of Medicare age.
Non-steel-producing counties—The health status of the population in non-
steel-producing counties, if measured in Medicare respiratory disease rates,
is lower than that of the steel-producing counties. This generalization
applies both to the "steel" counties with a low percentage of emissions from
steel producers, and to the counties in the higher category of steel production
emission sources.
Of the "non-steel" .-ouuties, 34 had the highest of the four respiratory
disease rate classes; 21 fell in the next highest category; 29 in the next
lower category, and the remaining 30 counties had the lowest category of
Medicare respiratory disease cases.
Incomplete analysis of the non-steel-producing counties, in terms of
Medicare respiratory disease rates, indicates there are many sources of
99
-------
emissions, types of emissions and other factors, that must be taken into con-
sideration in comparing steel and non-steel counties. Many industries and
natural events contribute to air pollution in specific localities. It is
quite possible that a detailed analysis would disclose hazardous sources
within certain non-steel counties that would explain their higher respiratory
disease rates. Time and budget did not permit an analysis of possible pollut-
ant sources in the non-steel-producing counties, or of other factors to be
considered in a thorough analysis. Besides meteorology, to account for
emissions from upwind sources, other pertinent factors are population re-
locations of the adversely affected, and death rates by age for both steel and
non-steel counties.
The respiratory disease morbidity rates of the steel- and non-steel-
producing counties raise strategic, tactical, and technical questions for
achieving health-improvement or health-protection goals by reducing emissions
on a selected industry basis. Many of these questions may be answered by
analyses of trends in respiratory diseases compared with estimated emissions
and technically reliable monitoring of the ambient air. On the other hand, if
reduced emissions are not being reflected in reduced respiratory disease, then
an entirely new set of questions arises as to the validity of emissions-
reducing programs and associated billions of dollars of expenditures that have
been justified on health protection and improvement grounds.
100
-------
SECTION V
REVIEW AND DISCUSSION
This project featured "non-confidential" portions of the computerized
Medicare hospitalization 20 percent sample file containing neoplasm, respira-
tory, and digestive disease diagnoses for the years 1971, 1972, and 1973.
Extracts in tape foim were rec. vcd from Social Security Administration, the
records were sorted and otherwise reformatted for convenient use on EPA's
Research Triangle Park computer, retrievals of pertinent data were programmed,
and illustrative presentation of the results made. Other data resources
supplied information on county industrial development, demography, air qual-
ity, emissions, and indicators of he'1th status.
The geographic focus of the ctudy varied according to task, with most
attention being given to three areas in Florida and Pennsylvania. For some
aspects of the work, statistics were compiled nationwide. A special analysis
compared counties with and without steel-production facilities in terms of
respiratory diseases, and with county-by-county details in Ohio and Pennsyl-
vania. At several junctures, data from counties with air pollution episode or
other relevant aspects were given consideration and appropriate data incor-
porated.
Since the objective of this work was to demonstrate the potential utility
of the Medicare hospitalization file and not to conduct a definitive analysis
of disease patterns, little time was available with which to compile compre-
hensive environmental statistics for all the counties and to prepare analyses
in depth. Rather, data sets were chosen to feature and demonstrate methodolo-
gies for applying the Medicare morbidity data in the context of interests of
primary concern to Health Effects Research Laboratory.
101
-------
The credibility/reliability of the Medicare files was evaluated by
comparisons with National Cancer Institute "survival times" for patients 65
years of age and over. At the national level of aggregation, and with quali-
fications described in the text, the Medicare record contents are very closely
comparable with NCI record contents. The Medicare data may in fact be more
representative of the national population over age 64 than NCI data, which are
taken from two statewide tumor registries and about a dozen metropolitan areas
aggregating about 10 percent of the total United States population. The
Medicare records are from all U.S. counties; the sample includes five million
of the 25 million covered under the Medicare and Railroad Retirement Act
insurance procedures.
From a health effects perspective, it is believed necessary that national
control and abatement programs of EPA justified to protect human health be
quantifiable systematically in human health terms. No alternative source is
available and superior to that of the Medicare data files and their potential.
Major project activities were discussed in Section IV. Results of
various computer-assisted analyses are presented there, and summarized below.
A. BERKS COUNTY—MORBIDITY/MORTALITY COMPARISON BY YEAR
The retrievals by age, sex, race and time of hospitalization such as
demonstrated for residents of Berks County in Section IV, give a clear indi-
cation of the power of the Medicare data to measure accumulating environ-
mentally related diseases with a comparatively long latency and time-to-death
period (cancer) in a county's resident population. A procedure for estimating
the net change in numbers of cases with selected cancer diagnoses was demon-
strated for each of the years 1971, 1972, 1973. The kind of "running inven-
tory" of individuals with diseases of special environmental interest demonstrated
in Section IV is superior in certain respects to mortality statistics and in all
cases is a unique, valuable indicator of environmental health in local areas.
Mortality descriptors alone have many deficiencies as indicators of the
human health status with respect to the environment. The most prominent of
102
-------
these statistical deficiencies are the long time lag between exposure and
death, and the high probability that death will occur from causes unrelated to
unhealthy environmental conditions. Morbidity descriptors, on the other hand,
are more responsive as human health indicators to changes in the environment—
whether good or ill. The mortality and morbidity files should be used in
mutual support. Air quality improvements can be expected to improve health
while degradation cf tl'e air can be expected to hatve the opposite effect.
By comparing ambient measures, amissions, and morbidity/mortality experiences
of counties, metropolitan areas, and other regions, EPA control and abatement
priorities based on human health considerations may also be quantified sys-
tematically from Mth hos;-italization and "running inventory" of survivors of
selected environmentally related diseases.
Distribution of cases by age at the time of disease identification, or at
any other specified time as is feasible from Medicare records, offers an
outstanding opportunity to note significant increases in disease occurrence
levels which may be caused by the introduction of a hazardous condition in a
particular locality. The Medicare files facilitate detailed comparisons among
localities with respect to time of disease diagnosis, and patient age/sex/
race. If there has been a definitive ind time-specific change in county
environmental quality and a disease onset is expected from prolonged exposure
to the conditions introduced, the procedures demonstrated with the Berks
County Medicare data may be useful in determining the latency period for the
disease and confirming/denying the association with the condition being
evaluated. No comparable timely opportunity is offered by mortality data.
Longitudinally assembled admissions/diagnoses data, in conjunction with
the survival-time statistics illustrated in Section IV can be of great value
in monitoring environmental health in terms of cases being added to or sub-
tracted from the county-resident environmentally-related disease pool. Con-
siderable potential is seen in the richness of morbidity-related information
available in the Medicare files which is not available through mortality data
based only on underlying cause. This additional detail would be of use in
studying health/environment histories of both deceased and surviving individuals.
103
-------
Tabulations of new cases conceptually assembled routinely from Medicare
records, when combined with knowledge of the time-spans and development
sequences between first manifestations of diseases and later death, can serve
to provide EPA with early indicators of (a) factors in the county or region
which may pose increased health risks and (b) the effectiveness of control
measures in removing health-threatening factors.
In considering potential uses of the Medicare files for the types of
analysis demonstrated in Berks County, the limitations of the data must be
kept in mind. Being a sample, the Medicare claimant records may not be rep-
resentative of the true prevalence of the diseases reported, particularly in
areas with small numbers of residents of over relatively few years. The
medical histories are incomplete; most claimants become eligible for Medicare
at age 65, only in-patient hospital treatments are reported, and deaths which
do not occur during a hospital stay are not indicated. The Continuous Work
History Sample (CWHS) of the Social Security Administration offers some remedy
for these shortcomings, as the CWHS reports all deaths in its sample. Time,
work scope and budget did not allow this CWHS computerized file to be utilized.
B. DAILY MORBIDITY/MORTALITY DURING PITTSBURGH EPISODE OF 1973
The daily morbidity statistics compiled for Allegheny County and the
other counties of the Pittsburgh Standard Metropolitan Statistical Area are of
special interest to the EPA1s Health Effects Research Laboratory as supple-
ments to mortality data, and are particularly relevant to EPA's continuing
investigation of "excess deaths" during and immediately following a pollution
episode such as that of 1973. Because hospital admissions/discharges for a
disease do not have the unique conclusiveness which characterize the event of
death (and attributed underlying cause), three different classifications of
morbidity were used in the 1971, 1972, and 1973 retrievals of Medicare hos-
pitalization data. These enumerated (1) all admissions during the episode's
13 days and the 13 days before and after the episode, (2) only the indiv-
idual's first admission for a disease classification during the time period,
and (3) the number of individuals admitted during each 13-day sub-interval.
104
-------
Analysis of this Medicare data discloses that the respiratory disease
hospitalization counts were not above the three-year average each day of the
1973 episode, as was true for the "all-causes" mortality data. From the
perspective of environmental linkages exclusively, there were significantly
more respiratory disease cases occurring during the 13-day episode period than
in the other eight 13-day 1971-1973 reference intervals. The occurrence of
four deaths from respiratory disease fol^wing admission during the episode
was in striking contrast to the one or two hospital deaths recorded for each
of the previous two years but the small numbers do not justify conclusions.
For the Pittsburgh episode, the additional respiratory disease hospitaliza-
tions, immediately prior to the episode and during the episode, appear to
provide a clear environmental health justification for the episode declaration
and accompanying precautionary measur.s. It would be interesting to extend
the Pittsburgh analysis to other areas experiencing episode or near-episode
conditions to determine if they also experienced abnormally high respiratory
disease hospitalizations.
The morbidity data offer two major advantages over the mortality data for
studies of changes in daily rates. First, the morbidity data include non-
fatal disease conditions, and, therefore, are not constrained to the stringent
requirement of death. This means higher daily counts of cases, a greater
variety of cases, a more complete understanding of environmental health ef-
fects, and the practical feasibility of obtaining more definitive information
from the patient. Second, the date of hospitalization on the Medicare record
is generally the same as, or shortly after, the date of environmental inter-
est, while the date of death for mortality cases has a more variable and
uncertain relationship to the date of fatal illness manifestation and to
environmental dates of interest.
The daily morbidity data provided by the Medicare files may be of par-
ticularly great value in the process of pollutant standard-setting and emer-
gency abatement procedm i. The data may be used (as in Pittsburgh) to
demonstrate relationship- between air quality and human hospitalizations for
certain diseases. It nv nlso be possible to quantify emissions/ambient/
disease correspondence, to specify pollutant level(s) which precede(s) a
significant increase in :tal admissions.
105
-------
No other source of morbidity data is known which can so readily provide
the detail available through the Medicare files for any county or aggregates
of counties, including the daily, monthly or other time period of hospital-
ization. If more geographic or disease detail is needed, it is believed quite
feasible to expand the present 20 percent sample to 40 percent, or 80 percent
or higher sample. The additional costs are believed relatively minor, as data
on all Medicare hospitalizations are routinely required for reimbursements, no
new steps are involved, and expanding the sample only involves additional ICD
encoding and related computerization steps.
C. MORBIDITY IN STEEL-PRODUCING AREAS
The comparison of morbidity rates in counties with steel-production
establishments versus morbidity rates in non-steel counties across the United
States was designed to show how the various component data resources can be
brought together for analytical purposes. The steel industry was selected
because of EPA's special interest. Other industries can also be selected, as
well as groups of industries. Similarly, counties and their associated
morbidity/mortality may be classified and analyzed as rural-urban, including
estimated emission categories, ambient categories, summer-winter temperature,
fuels consumption, and so forth.
This steel-producing and non-steel-producing counties comparison, de-
scribed in Section IV, has demonstrated a number of informative ways by which
the Medicare computerized files help determine whether the operations of a
specific industry pose health risks to the surrounding resident population.
Methodological refinements can help the Medicare data to be used even more
effectively for this kind of epidemiological research function.
The fact that the hospitalization rates for respiratory disease were
generally lower in the sHeel-producing counties than elsewhere, and showed no
clear relationship between the relative importance of steel-producing sources
of TSP and respiratory diseases runs counter to expectation. Common sense
suggests that steel industry emissions are not healthful. Further, adverse
health effects from high TSP levels and from exposure to steel production
106
-------
processes have been documented. For example, during the Allegheny County
episode discussed above, respiratory disease hospitalizations increased, in
conformance with expectations.
A number of factors affect the validity of findings based on Medicare
hospital admissions in counties with a particular industry. First, the
Medicare sample, predominantly individuals over 65, has not been demonstrated
to represent the health effects of younger age groups to a polluted environ-
ment. Some diseases claim their victims at an earlier age, and higher pollu-
tion levels may have exacted conclusive effects before the victims reached the
Medicare age. Hospital admissions for Medicare residents of industrialized
and more polluted areas nay also i- .", jwrv than exp'cted because the more
susceptible individuals may have relocated to more healthful environments.
With increasing age, many other disease/accidents present important competing
risks, thus masking the full health imp-let, of environmental pollution in
heavily industrialized areas.
The use of county level of detail tor the study of morbidity and mortality
data poses a problem in examining the contributions of point-specific sources
such as steel mills. The industrial facility should, at the very least, be
determined to be a major emitter of a pollutant within the county. In the
present project, all steel-producing counties were considered and although the
relative importance of steel industry sources was determined, the significance
of the variations was not analyzed. Time and budget constraints also limited
the consideration given to the presence of other types of industrial establish-
ments. With tha F,T'-'>/NEDS data base and the Census/SSA files of four-digit
SIC employment, these latter difficulties can be overcome.
Finally, the scope of this project did not permit examination of some
important variables; more thorough investigation of all possible correspon-
dences would be required in a definitive study. Inclusion of additional
variables (ambient measures, estimated emissions, and trends for steel and
non-steel producing counties) should increase the health-hazard-detection
sensitivity of the methods already described and suggest new pollution abate-
ment strategies based on human health protection. Reduction of emissions
107
-------
across-the-board for a particular industry, such as steel, represents one of
many alternative approaches with advantages and disadvantages to be considered
carefully before implementation.
In summary, the comparatively lower respiratory disease experience of the
steel producing counties suggests the necessity for further study, for if
confirming/verifying results are obtained, serious questions are raised con-
cerning acceptable background levels, abatement and control of the particles
(size, composition) that are harmful and abatement/control policy trade-offs
between health purposes and aesthetic purposes.
D. MIGRATION OF MEDICARE CLAIMANTS
The examination of Medicare claimant migration (Section IV) did not
indicate as widespread a relocation pattern as expected from the retirement
age population, especially in the high in-migration counties of Florida.
Extremely low rates of county/state relocations were observed in the Medicare
sample study areas of Florida and Pennsylvania; once hospitalized, changes in
residency are few. It may be that most individuals who change residences do
so shortly after the retirement date, or while still in relatively good health.
The Medicare records alone are, therefore, not a good source of data either on
migration patterns or on the contribution of migration to health effects of
various geographic regions, but must be used in conjunction with other sources.
A more complete representation of county/state environmental exposures
prior to age 65 can be obtained from longitudinal employment and location
history computerized files now being obtained from Social Security Administra-
tion. These files may be linked with the Medicare morbidity files to show
both industry and location during pre-retirement years and subsequent re-
locations. With this linkage, long latency period diseases may be identified
and analyzed in conjunction with environmental exposure areas.
108
-------
E. FINDINGS AND CONCLUSIONS
The Medicare hospital discharge survey sample is a source of morbidity
data with unique value for disclosing and monitoring relationships between the
environment, and the health of t -\ important segment of the population. It is a
growing data base of medical records, is routinely collected, covers all parts
of the country,, LJ ' aintc Lned in "compctrrized" format, and the files may be
assembled longituuinally to disclose the complete Medicare hospitalization
history of all sample selectees, regardless of disease or place of hospitali-
zation. These files provide daily information on primary causes of each
hospital admission geographically identified by the claimant's county of resi-
dence. The Jex, race ana ago wf Lae cj.ainkiu«. ac che time of admission are
also shown.
The data from the Medicare file ---in be utilized to estimate national,
regional, state, or county totals of ho. pit;1 admissions by discrete types of
diseases. In assessing the ..., edibility of the Medicare data, a close cor-
respondence was shown to exist: Tvith other morbidity/mortality surveys and files.
Procedures, with tables, ^err demonstrated which measure for each year
the addition to and loss from the pool of residents having specified diseases.
The capability for estimating the population that has experienced one or more
hospitalizations, by ICD code, for each of the 3,050 counties, has many environ-
mental hca1th app1ications. It provides an additional perspective to raw
admissions counts, or counts of mortalities by cause. The scope of these
eav lton;r>ental healm applications could not be addressed in this pilot project,
but the procedure demonstrated provides an additional tool for comparing
cr"!_ities and IP, tr> "olitan areas by population burden of environmentally-
1 i.."! di ;e3'to. Comparisons an be sharpened by utilizing ambient and
' ' Lo.-s ieL._ ' , .."ft" .''or the areas b< ~ ng analyzed to differentiate, if pos-
sible, between (a) "normal" or "background" case-load levels of the diseases
of particular interest; (b) additional case-loads that might be attributed to
chronic low-level exposures to pollutants; and (c) higher case-loads caused by
acute or emergency episodic exposures to pollutants. Daily counts were taken
109
-------
which noted the peaking effect of hospital admissions during the pollution
episode in Allegheny County, Pennsylvania, 1973.
A preliminary study of a single industry (steel) was performed, demon-
strating how the files may be used in conjunction with other EPA computerized
files to assess the relationship between selected diseases and the presence
or absence of specific industrial sources of emissions.
The extent of migration exhibited in Medicare claimant hospitalization
histories was investigated with the finding of relatively few changes in
residence after a hospitalization occurs.
The pilot study demonstrations indicate the flexibility of the Medicare
data for a variety of purposes and suggest that, with appropriate supple-
mentary refinements, this resource may be developed to have important uses in
the Health Effects Research Laboratory and elsewhere in EPA. As examples, the
Medicare files offer assistance in emergency episode determination and conclu-
sion decisions, differentiating between industry and non-industry sources of
environmental diseases, in focusing regulation and enforcement resources on
factors most injurious to human health, and other evaluations in EPA that
directly or indirectly are concerned with human health and the environment.
110
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT NO. 2.
EPA-600/1-80-018
4. TITLE AND SUBTITLE
Pilot Study - Uses of Medicare Morbidity Data In
Health Effects Research
7. AUTHOR(S)
Irwin J Shiffer & Edgar A Parsons
9. °ERFGRMING ORGANIZATION MAME AND ADDRESS
System Sciences , Inc.
P.O.Box 2345
Chapel Hill , NC 27514
12. SPONSORING AGENCY NAME AND ADDRESS
Health Effects Research Laboratory
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
March 1980
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
1AA817
11. CONTRACT/GRANT NO.
68-02-2782
13. TYPE OF REPORT AND PERIOD COVERED
U.S. Environmental Protect]in
Research Triangle Park, no <:/7'ii
14. SPONSORING AGENCY CODE
EPA bOO/11
is. SUPPLEMENTARY NOTES
16. ABSTRACT
-"- the practicability of utilizing Social
morbidity data to supplement mortality
This project is a pilot investigation
Security Administration (SSA) Medicare
data in cancer and otner environmentally-belated studies. For this study
non-confidential data on 1.2 million hoLpitalizations for 815,000 persons
diagnosed as having a neoplasm, respiratory, or digestive disease during 1971,
1972, and!973 ware deluded. The data ere kept current by SSA for their
;*naly- ;s purpose
The Medicare files are the only known source that incorporates all cancer cases
systematically, regardless of whether the cancer is fatal. The Files facilitate
analysis of possible relationships between emissions of a specific industry
and the disease(s) rates for the co-located population.
Numerous recommendations are made -for applying the Medicare files
analyses of environment-to-health relationships. Prominent among
to additional
these are
cancer J.ospi te. M^ation trends, by oounty, augmented with cancer mortality tetads
arc available emnis?ions/monitoring measures to identify areas of cancer
increase/decrease possibly related to environmental influences.
Tht3pagei
, UNCLASSIFIED
122. PHICH
EPA Form 2220-1 (Rev. 4-77) »
------- |