EPA-600/1-79-024
August 1979
MODEL FOR MEASURING THE HEALTH IMPACT FROM
CHANGING LEVELS OF AMBIENT AIR POLLUTION:
MORBIDITY STUDY
by
Tsukasa Namekata, Bertram W. Carnow,
Zanet Flournoy-Gill, Eileen B. O'Farrell and Domenic Reda
Occupational & Environmental Medicine Program
School of Public Health
University of Illinois at the Medical Center
Chicago, Illinois 60680
Contract No. 68-02-2492
Project Officer
Dr. Wilson Riggan
Population Studies Division
Health Effects Research Laboratory
Research Triangle Park, North Carolina 27711
HEALTH EFFECTS RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
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 or recommendation for use.
ii
-------
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 estab-
lishment of sound regulatory policy. These regulations serve to enhance 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 program in toxicology, epidemi-
ology, 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 criteria documents on pollutants for which national ambient
air quality standards exist or are proposed, provides the data for registra-
tion of new pesticides or proposed suspension of those already in use, con-
ducts research on hazardous and toxic materials, and is primarily responsible
for providing the health basis for noniodizing radiation standards. Direct
support to the regulatory function of the Agency is provided in the form of
expert testimony and preparation of affidavits as well as expert advice to
the Administrator to assure the adequacy of health care and surveillance of
persons having suffered imminent and substantial endangerment of their
health.
The report examines the association of day-to-day variation in ambient
pollution concentrations and morbidity as measured by emergency room visits
for cardiac and respiratory diseases in two major hospitals in the City of
Chicago. This is the morbidity phase of a research project to quantitate the
association between health effects and ambient pollutant concentrations in a
major metropolitan area.
F. G. Hueter, Ph.D.
Director
Health Effects Research Laboratory
111
-------
ABSTRACT
This study quantitatively examines the relationship between human health
and ambient air concentrations of the major pollutants in the city of Chicago.
The study consists of two major parts; a morbidity component and a mortality
component. This report describes the morbidity analysis in which linear
regression models have been developed to quantitatively estimate the degree
of the air pollution contribution to emergency room visits for cardiac and
respiratory diseases in two major hospitals in the city of Chicago.
Data on emergency room patients were collected on every Tuesday,
Wednesday and Thursday from April 12, 1977 to April 14, 1978. Air measure-
ments used in the study are total suspended particulate, sulfur dioxide and
nitrogen dioxide from 22 sites of the Chicago Air Sampling Network (CASN).
Also used are measurements of sulfur dioxide, nitrogen dioxide, nitric oxide,
carbon monoxide and ozone from the Illinois E.P.A. monitoring site on Polk
Street at the University of Illinois Medical Center, the site closest to the
hospitals. Daily climatological measurements used are from Midway Airport,
as published by the U.S. Department of Commerce.
Based on pollution measurements (TSP, SC>2 and N02) from CASN and the
patient's home address, exposure levels were estimated for each of the 21,000
patients. A daily average exposure level was obtained by dividing the sum
of the individual exposure levels by the number of patients in a specific
disease category. With the pollution measurements from Polk Street, one
reading was used as an alternate of an average exposure index.
A dependent variable in multiple regression analysis is the percentage
of excess visits for each disease group, which is a ratio of the number of
visits to the average number of visits in a specific disease group, expressed
as a percentage. Three types of independent variables are used; air pollutants,
climatological measurements and days of the week as dummy variables. Only
one pollutant is included in each equation, since the number of observations
differs from one pollutant to the next. Based on the significant associations
between the pollutants and the disease groups, holding climatological and
days-of-the-week variables constant, the variation due to the pollutant is
estimated.
iv
-------
According to the results, sulfur dioxide based on patient exposure levels
can account for about 13% of the variation of emergency room visits for acute
bronchial and lower respiratory infections and about 22% for total cardiac
diagnoses. Nitric oxide based on measurements from the closest site to the
hospitals can account for about 7% of the variation of visits for total
respiratory diagnoses, 6% for allergic conditions and upper respiratory
infections, 4% for total cardiac diagnoses and 4% for hypertension and vascular
heart diseases. Total suspended particulate, carbon monoxide and ozone do not
show significant associations with any disease groups.
This report was submitted in fulfillment of Contract No. 68-02-2492
by the University of Illinois School of Public Health, under the sponsorship
of the U.S. Environmental Protection Agency. This report covers a period from
February 15, 1976 to October 15, 1978, and work was completed as of December 31,
1978.
-------
CONTENTS
Disclaimer
Foreword
Abstract
Figures
Tables
Acknowledgement x
1. Introduction 1
2. Conclusions 2
3. Recommendations 3
4. Background Information 4
The Problem 4
Studies in Chicago 5
5. The Method 9
Source and Collection of Data 9
Hospital Data 9
Climatological Data 11
Air Pollution Data 18
Estimation of Missing Pollution Measurements 22
Calculation of Community Area Exposure Levels 24
Calculation of Disease Specific Average Exposure Levels 29
Multiple Regression Analysis 29
6. Results and Discussions 33
Models for Total Respiratory Diagnoses (TRD) 33
Models for Allergic Conditions and Upper Respiratory 36
Infections (RDl)
Models for Acute Bronchial and Lower Respiratory Infections 36
(RD2)
Models for Total Cardiac Diagnoses (TCD) 39
Models for Hypertension and Vascular Heart Disease (CDl) 41
The Variation Due to the Pollutant, the Days-of-Week Effects 43
and Weather Changes
Comments 46
References 48
Appendix 51
vi
-------
FIGURES
Page
1. Location of air monitoring sites and 76 community areas 19
in Chicago
vii
-------
TABLES
1. List of cardiac and respiratory diagnoses 12
2. Daily Means of Emergency Room Visits for Major 17
Disease Groups
3. City of Chicago, Department of Environmental 21
Control-Chicago Air Sampling Network
4. No. of Days Monitoring Station Shutdown 23
5. Predication equations for TSP missing data 25
6. Alternate equations for TSP missing data 26
7. Prediction equations for SO missing data 27
8. Prediction equations for NO data 28
9. Morbidity ttodels: Total Respiratory Diagnoses (TRD) 34
10. Morbidity Models: Allergic Conditions and Upper 37
Respiratory Infections (RDl)
11. Morbidity Models: Acute Bronchial and Lower 38
Respiratory Infections (RD2)
12. Morbidity Models: Total Cardiac Diagnoses (TCD) 40
13. Morbidity Models: Hypertension and Vascular Heart 42
Disease (CDl)
14. The variation due to the pollutant, days-of-week 44
effects and weather changes in the models selected
from Tables 9-13
15. Correlation coefficients between emergency room visits 52
for cardiac and respiratory conditions and environmental
measurements in phase I analysis
16. Correlation coefficients between emergency room visits 53
for cardiac and respiratory conditions and environmental
measurements in phase II analysis
17. Correlation coefficients between emergency room visits 54
for cardiac and respiratory conditions and environmental
measurements in phase III analysis
viii
-------
TABLES (continued)
Page
18. Correlation coefficients between emergency room visits 55
for cardiac and respiratory conditions and environmental
measurements in phase IV analysis
19. Correlation matrix of environmental measurements in 56
phase I analysis
20. Correlation matrix of environmental measurements in 57
phase II analysis
21. Correlation matrix of environmental measurements in 58
phase III analysis
22. Correlation matrix of environmental measurements in 59
phase IV analysis
23. Distribution by race of study population; April, 1977- 60
April, 1978
24. Distribution by sex of study population; April, 1977- 60
April, 1978
25. Frequency age distribution of total study population, 61
CCH and UIH, April, 1977-April, 1978
26. Frequency of all diagnostic illnesses of study population, 62
April, 1977-April, 1978
IX
-------
ACKNOHLEDGEMNT
Die authors are grateful to Dr. James Haughton, Administrator of Cook
County Hospital, Mr. Robert Reimer, Administrator of Emergency Services at
Cook County Hospital, Mr. Lawrence Knight, Administrator of the Pediatric
Clinic at Cook County and all their support staff for this support and
guidance while «*>?***'*^ing emergency room ft^i*"*! from <->»'ig facility. Mr. William
Hart, of the University of Illinois Hospital, and his hospital clinic
personnel were generous in providing imUmltBd access to their patient files.
Mr. Janes Herman, Director of the Technical Services Division of the
Chicago Department of Environmental Control and his support staff were
generous in providing ^ir pollution data pertinent to the study. Also the
State of Illinois EPA was quite helpful in providing supplementary air
pollution data.
Th& authors would like to express our appreication to Ms. Sharon Kawasaki,
Ms. Kiyoka Koizumi, Ms. Elain Breck and Mr. James Marselle for their assist-
ance in data processing.
A very special acknowledgement is given to Dr. Wilson Riggan, EPA Project
Officer, for his untiring efforts, cooperation, and patience with this project.
-------
SECTION 1
INTRODUCTION
The United States Conaress passed the Clean Air Act in 19G7 and its
7\mendments in 1970 to protect and enhance the quality of the nation's
air resources while promoting the public health and welfare and the pro-
ductive capacity of its population, reducing harmful emissions, and ensuring
that air pollution problems will, in the future, be controlled in a systematic
wav. Also, in 1970 the city of Chicago passed an ordinance which virtually
banned coal and garbage burning in individual households and businesses.
Such legislative efforts to control air pollution had significant
effects on a decrease in the amount of some pollutants in the city of
Chicago; suspended particulate levels had dropped 26 per cent, and sulfur
dioxide levels had been cut by 50 per cent between 1970 and 1975. However,
carbon monoxide and O7,one levels have remained high, in spite o^ Chicago's
voluntary auto emissions control program.
The present study has quantitatively examined the relationship be-
tween human health and ambient air concentrations of the major pollutants
in the city of Chicago. The study has consisted of two major narts; a
morbidity component and a mortality component. This renort describes the
morbiditv analysis in which linear regression models have been developed
to guantitativelv estimate how air pollution affects the number of emeraencv
room visits for cardiac and resniratorv diseases in two hosoitals in the
citv of Chicaao.
-------
SECTION 2
CONCLUSIONS
A total of forty regression models were developed to examine the
strength of the relationship between an individual pollutant and cardiac
and respiratory morbidity, controlling for climatological and davs-of-
week variables. In spite of limited data base on ambient air pollution
measurements, it is likely that air pollution accounts for some portion
of the variation in the incidence of cardiac and respiratory morbidity.
The following conclusions were made:
1. Sulfur dioxide based on patient exposure levels has explained
about 13% of the variation in emergency room visits for acute
bronchial and lower respiratory infections, and about 22% of
the variation in emergency room visits for total cardiac diag-
noses in cook County Hospital and the University of Illinois
Hospital combined.
2. Nitric oxide based on measurements from the closest site to
these hospitals explained about 7% of the variation in visits
for total respiratory diagnoses, 6% of the variation in visits
for allergic conditions and upper respiratory infections, 4%
of the variation in visits for total cardiac diagnoses and 4%
of the variation in visits for hypertension and vascular heart
diseases.
3. Nitrogen dioxide based on measurements from the closest site
to these hospitals explained only less thin 1% of the variation
in visits for total cardiac diagnoses.
4. Total suspended particulate, carbon monoxide and ozone have not
shown significant associations with any disease groups.
-------
SECTION 3
KECOMMENDATIONS
The most critical limitation in the present study was the lack of
reliable daily measurements of ambient air pollutants throughout the city
of Chicago. It is stroncrly recommended that the re-evaluation of the air
pollution monitoring systems in Chicago be conducted as soon as possible
and more accurate, reliable and usable pollution measurements be available
to users, especiallv for purposes of epidemiological research.
Air pollution measurements at only Polk Street,the closest site to
Cook County Hospital and the University of Illinois Hospital, were used
as the alternate of the averaae patient exposure levels because they were
the only available data at the time of the data collection. Air pollution
measurements from other sites should be sought in the future to measure
patient exposure levels more accurately than the site at Polk Ptreet. The
models developed in this studv can be refined and validated only if a
sufficient number of observations are made of reliable ambient air pollution
measurements from different monitoring sites covering the entire Chicago
area.
Emphasis should be placed on the following if meaningful models are
to be developed: (1) vastly improved air pollution data base, (2) exten-
sion of collection period to span a number of years for seasonal analysis,
(3) development of a seasonal model for other pollutants besides ozone,
(4) re-grouping disease diaanoses to examine acute effects of each pol-
lutant, and (5) examining lag-effects after obtaining more pollution
measurements from other sites in addition to those at Polk Street.
-------
SECTION 4
BACKGROUND INFORMATION
THE PROBLEM
Extremely polluted air is known to be a killer of old people with
chronic illnesses and persons with serious illnesses, as shown in episodes
123 4
in the Meuse Valley , Donora , London , and New York City .
In addition to such an acute effect, long term exposure to aggra-
vated air may cause or exacerbate chronic respiratory diseases, as reported
in the Tokyo-Yokohama area , New Orleans , and Yokkaichi City '
In all of these cases, the air pollution level was significantly higher
than the usual level found in most urban areas.
At usually prevailing levels of air pollution, on the other hand,
direct relationships may not be expected between diseases and daily levels
of air pollution by using a simple epidemiological approach in which
frequency distribution of diseases is compared with levels of air pollu-
tion. Multiple regression analysis is one of the methods to obtain
quantitative estimates of the relationship between diseases and environ-
mental factors. Multiple regression methods had been used to examine the
degree of air pollution's contribution to mortality for the past decade by
some researchers like Sprague and Hagstrom , Lave and Seskin ' ,
17 18 19
Hodgeson , Buechley and his co-workers , and Schimmel and Murawski
In particular, Lave and Seskin have carried out an extensive study to
quantitatively measure air pollution effects on mortality. They estimated
that a 50 percent reduction in sulfates and suspended particulate would
account for a 4.7 percent decrease in the unadjusted total mortality rate,
based on their 1960 annual cross-sectional analysis.
While many regression analyses had been done by using mortality data
in relation to air pollution, climatological measurements and socioeconomic
variables, air pollution effects on morbidity had been estimated in terms
of a dose-response relationship using regression analysis, by only a few
21
researchers. Thompson and his co-workers examined relative contributions
of air pollution to morbidity over a period of nine consecutive seasons by
using the stepwise regression method. Their results showed that the re-
gressions differed markedly by season of the year, and that the climatological
-------
variables seemed more related to common cold rates than the air pollutants.
STUDIES IN CHICAGO
For the past ten years, the Chicago Air Pollution Study Group has been
examining the health effects of air pollution, particularly, to define
those individuals in the pollutants at which their health is adversely af-
fected.
22
The Chicago Chronic Bronchopulmonary Disease Registry Study began
in August, 1966. A total of 571 patients from 16 different facilities,
chiefly, pulmonary and allergy clinics, were gathered into a registry.
They were classified according to the severity of disease and each main-
tained a daily record of acute chest illness. Using the continuous air
pollution monitoring network of the city of Chicago, a model was developed
to assign an air pollution level (SO and particulates) to each square mile
of the city for every 15 minutes of the day. Patients' exposures were
determined by the estimated level of pollution in the residence and occu-
pation for each 24 hour period. Illness appeared to be correlated with
levels of sulfur dioxide with increasing illness at each of 7 levels of
SO pollution. At 0.24 ppm for 24 hours, there were more than twice as
many acute chest illnesses as when the level was 0.04 ppm. This occurred
in males 55 and over with advanced bronchitis and to a lesser degree in
those under the age of 55 with similar bronchitis severity. Those with no
bronchitis or mild bronchitis showed no effect at these levels. Further
analysis, which compared the number of sick to well days for each indivi-
dual in relation to his air pollution exposure, showed a significant re-
lationship in two of three months examined for the same age groups. It
appeared that when SO was considered as a pollution index in males 55
and over with advanced bronchitis, there was a relationship between levels
of pollution and frequency of acute chest illnesses.
An additional group of subjects consisted of 51 patients followed in
22
a chronic respiratory disease clinic . In addition to being classified
according to disease severity, each patient maintained a daily record of
increase or decrease in coughs and dyspnea, and also was trained to examine
his sputum every morning. If the sputum was purulent, the individual was
instructed to start on a broad-spectrum antibiotic on which he remained for
-------
at least seven days. The development of purulence and the use of antibio-
tics were also recorded. The daily record shown in the form of a postcard
was returned each week. Then, each patient's exposure level to SO was
estimated by using daily SO- measurements from the air monitoring network
in Chicago. All patients were classified into three groups: (1) 0-0.09
ppm, (2) 0.10-0.19 ppm, and (3) 0.20 ppm and over. Acute illness (in-
creasing purulent sputum, cough, and dyspnea) was reported in 35% of
patients at 0.20 ppm and over, 24% at 0.10-0.19 ppm, and 23% at 0-0.09
ppm in October, 1967. The same tendency was observed in November and
December, 1967. It was concluded that exacerbation of symptoms and preci-
pitation of acute respiratory illness occurred at high daily levels of
S0_ exposure among elderly persons with advanced bronchitis.
23
Lepper and his associates examined deaths from all respiratory causes,
including tuberculosis and lung and bronchial cancer, in the total white
population of the city of Chicago in the years 1964 and 1965 by pollution
levels and socioeconomic status. Census tracts were classified socioeco-
nomically and placed in pollution categories based on annual averages of
sulfur dioxide. The study utilized SO values measured at 21 stations
spread over the city during three 24 hour periods weekly. Annual average
isopleths were drawn from the data for approximately 150 days in which
measurements were made and an average annual exposure concentration was
calculated from these for each census tract. Census tracts were divided
on socioeconomic criteria into high, medium, and low. Each of these three
subpopulations were classified into three groups: (1) the highest level
of SO , (2) the moderate level, and (3) the lowest level - according to
annual exposure concentrations. Only the white population was analyzed
because of the lack of a high socioeconomic group in the nonwhite popula-
tion. The results revealed a significant rise in the total respiratory
death rate from high to low socioeconomic groups in relation to levels of
SO . At the low SO level, significant differences were not present between
classes; while at the moderate SO level, the rate of the low socioeconomic
cell was significantly higher than those of the medium and upper socioecono-
mic groups. This was also true when they were compared at high SO levels.
The groups with the highest death rates were the lowest socioeconomic
groups exposed to the highest pollution levels.
-------
24
Carnow and Feiveson examined the acute effects of SO on mortality
in the city of Chicago during a nine day episode (October 26 through Novem-
ber 3, 1969). Because SO concentrations varied widely throughout the
city during the episode, the city was divided into three sections (the
high SO- section, the moderate, and the low) according to the average
SO concentration during the episode at the TAM station or stations
closest to the area in which the decedent resided. The sharpest increases
in mortality from cardiac disease occurred in males, mainly older males,
who are known to be most susceptible to cardiac difficulties. Furthermore,
these increases occurred only in the areas of the city with highest average
pollution levels during the nine days of the episode. Some striking in-
creases in mortality were also observed in certain subgroups of females in
the high pollution areas; for example, there were 16 deaths during the
episode from rheumatic heart disease in white females as compared to 8 in
the nine days before, 8 following, and 12 expected. In the moderate pol-
lution section, there was also a substantial increase in ischemic heart
disease mortality for white females; 72 deaths during the episode, 48 deaths
before, 55 after, and 54 expected. There were no significant increases
in mortality in the low pollution areas during the episode as compared
to before and after the episode.
25
In the Drexel Home Study , 49 normal volunteers in a home for the
aged were studied. They were classified according to severity of bronchi-
tis, and each person was examined at the same time each day, Monday
through Friday for a total of 25 days. At each visit, resting pulse, blood
pressure, and respiratory rate were obtained along with responses to
questions regarding symptoms relating to cardiovascular, respiratory, and
gastrointestinal systems. Spirometer measurements were carried out using
a Jones Pulmonar, 6-liter bellows dry spirometer. Determinations were
then made of the one-second, three-second, and total force expiratory
volume and maximum expiratory flow rate. In addition, peak flow was
measured using a Wright Peak Flow meter. Each member of the study team was
assigned his own panel of patients and performed the initial history,
physical examination, daily recording of symptoms, and pulmonary function.
Variations in levels of SO and particulate were compared with daily
spirometry and peak flow in the individuals in the study. The result
-------
suggested no relationship between the levels of SO and particulate and
pulmonary function in this population during the period of the study.
Other analyses also revealed no significant changes in function or
symptoms. While the mean levels of SO during the study were relatively
2 3
low for Chicago, 0.024 ppm, the mean particulate level was 165 yg/m
with a range of 72 to 309 yg, levels at which health effects were noted
in some segments of the population in a number of other studies. The
results suggest that this nonsmoking urban aged population represented a
hardy survival population.
With the cooperation of the Chicago Lung Association and fourteen
hospitals throughout the city of Chicago, Namekata and Carnow have carried
out a study to examine effects of air pollution on the incidence of car-
diac and respiratory diseases by using individual records of emergency
2fi
room visits in those hospitals . Recently, they have re-examined effects
of sulfur dioxide and total suspended particulate on cardiac and respira-
tory conditions, based on the same morbidity data and estimated patient
27
exposure levels to those pollutants . Their regression model for all
respiratory conditions included sulfur dioxide, daily maximum temperature
and daily average wind speed as independent variables of which all F
statistics are significant. The model for bronchitis included total
suspended particulate, daily minimum temperature, daily average wind
speed and humidity which indicated all significant F statistics. These
findings are meaningful because statistical significance of the partial
regression coefficient of each pollutant implies that such a pollutant may
possibly have critical impact on the disease even after partitioning the
effects of other independent variables included in a equation. However,
their models for asthma and for cardiac conditions did not show significant
results.
-------
SECTION 5
THE METHOD
SOURCE AND COLLECTION OF DATA
Hospital Data
The two hospitals used as the source of morbidity patient data were
Cook County (CCH) and the University of Illinois Hospitals (UIH). CCH was
chosen as the initial collection site for three reasons. First, in a prior
O £<.
and similiar study , it was shown that on a weekly basis CCH registered and
treated as many or more patients as 13 other city hospitals combined.
Secondly, virtually all of the patients registered at CCH were residents of
Chicago and lastly, the same registrants used CCH as a primary care faci-
lity. Total visits at both hospitals combined ranged from 500 to 800
patients daily.
UIH was chosen as an alternative collection center. A doctor's strike
at Cook County Hospital prompted most of the patients using the CCH facility
to be routed to UIH. To be assured of the continued availability of all
necessary patient data, the monitoring process continued at both hospitals.
Within the hospital, data was collected from three areas; the Emergency
Room, Admissions*, and Pediatrics. Information was sought from these clinics
because the majority of the study subjects were non-appointment patients
only who sought medical attention for acute health disorders. It was felt
that the majority of persons suffering from acute health effects of air
pollution would dispense with the regular scheduling of a formal appoint-
ment and seek immediate medical attention. Another equally significant
reason is that data obtained from hospital records in these areas was
considered reliable and valid as a doctor's diagnosis accompanied each re-
gistered patient who had been treated. Data forms were d:- scarded for any
patient who left without a diagnosis, even if the complaint was acceptable
to the study.
The Chicago Lung Association was initially subcontracted to acquire
the necessary patient information and to work with project personnel in the
*The admission of patients into UIH was not the sole function of the UIH
Admission Clinics. Staff physicians also handled scheduled appointments
for checkup, rechecks, etc. and treated patients with unscheduled appoint-
ments .
-------
development of a collection system. A collection sheet entitled "Hospital
Record of Patients with Respiratory and Cardiac Conditions" (hereafter
referred to as the data form) was drafted.
After a month of collection, the arrangement with the Chicago Lung
Association was found to be unneeded and the contract was terminated. A
morbidity coordinator was added to the research team to supervise the data
collection procedures and the subsequent processing of data. Clerical
personnel were added to the project and trained to collect data from each
hospital thrice weekly.
Data collection officially began April 12, 1977, and spanned one year,
terminating on April 14, 1978. The morbidity collection team gathered data
for every Tuesday, Wednesday and Thursday of each week from both hospitals.
By omitting collection of data for weekends, days surrounding weekends
(Fridays and Mondays), and holidays, a fairly regular pattern of visitation
(in regards to type and number of patients) was established.
Data was collected at each hospital site by a 2 member team. Data
forms were filled out to record information for each person whose illness
was applicable to the study. Patient information transcribed on each form
consisted of data and time of visit, address, age, race, sex, complaint,
physician's diagnosis and whether patient was or was not a hospital ad-
mission. The patient's name was only written on the form if this was the
only means of defining the sex.
Immediately after the data had been collected, the team validated
their work by switching hospital and data forms. In such a manner they
challenged each other's work and corrected any mistakes or irregularities.
Validation completed the on-site hospital collection procedure.
After data had been collected, number codes were assigned for community
area, census tract, patient's complaints and physician's diagnosis. Each
form was then stamped with a six-digit sequence/reference number, codes
were checked and sent out to be keypunched. When the keypunched cards and
computer lists were returned, they were verified against the original data
form for possible error. The data forms, cards, and lists were filed in
sequential order for later analysis.
The only patient records utilized from CCH Pediatrics were log sheets
on which general information was listed as a registration process. The
10
-------
sex of the patient and the physician's diagnoses were not specified and
formal patient records, which would have included these two items, could
not be made accessable to the collection team. In light of the unavaila-
bility of complete and valid patient information, CCH Pediatrics was ter-
minated as a source of data in December, 1977. All data collected to this
point was deleted from the study. CCH Emergency Room and Admissions account
for all CCH patients included in this study whose age was 15 years or older.
Two major groups of diseases and illnesses that could be considered
as probable health consequences of existing air pollution levels were
first compiled by key project personnel prior to the start of official data
collections. All of these health disorders (listed in Table 1) were
medically diagnosable, and of a respiratory or cardiovascular nature.
Numerous past studies on air pollution versus health have shown that most
highly urbanized populations experience a constant barrage of assorted
ambient pollutants and as a result will suffer more severly and more
frequently from these two categories of illnesses than from any other major
disease grouping(s).
A code number was assigned to each diagnosis. This number served as
a general indicator of type (respiratory or cardiovascular) and severity
(acute or chronic) of the diagnosis. This number system also proved to
be a useful training and reference guide for hospital data collectors.
After computing daily averages of patients for eight disease groups
(Table 2), three groups were excluded from the final analysis because of
too few cases to obtain stable statistical measurements. As a result, the
following disease groups were included in the final regression models:
1. All respiratory diagnoses (Group 8)-TRD
2. Allergic conditions and upper respiratory infections (Group 4)-RDl
3. Acute bronchial and lower respiratory infections (Group 5)-RD2
4. All cardiac diagnoses (Group 3)-TCD
5. Hypertension and vascular heart disease (Group 2)-CDl
Climatological Data
The U.S. Department of Commerce, National Oceanic and Atmospheric
Administration (NOAA) maintains a local Climatological data bank comprising
Chicago-area stations. The data bank currently monitors meteorological
variables at 3 stations within the city of Chicago; on the University of
11
-------
TA'3I£ 1
I.IST OF CAHDIAC JV.JD HKf.
Group 1. Organic, Arteriosclerotic and Heart Failure
Diagnostic
code 'lo.
Diagnostic
code 'lo.
niaonosis
02 Angina ^
02 Angina Pectoris 21
02 Impending Myocardial Infarction 21
02 Unstable Angina 21
03 Cardiac Arrhythmias 22
03 Arrhythmias 22
03 Heart Flutter 22
03 Atrial Fibrillation 36
03 Bradycardia 36
03 Fibrillation - 36
03 Flutter 36
03 Premature Ventricular Constriction 36
03 Tachycardia 36
03 Ventricular Fibrillation 36
03 W.P.W. Syndrome 36
10 Cardiac Arrest 36
10 Cardiac Decompensation 36
10 C.H.F. 36
10 Possible Heart Failure 36
10 Pulmonary Edema 36
10 Ventricular Failure 40
11 Ischemic Heart Disease 43
11 Myoscrdial Ischemia 44
19 Coronary Occlusion 46
19 Myocardial Infarction/Occlusion 47
21 Palpitations 52
Abnorznal
Cardiomyopathy
Congential Heart Disease
Other Cardiac Conditions
Pericarditis
Bacterial Pericarditis
Viral Pericarditis
Aortic Incompetence (Insufficiency)
Aortic Stenosis
Chronic Rheumatic Heart Disease
Click
Click Syndrome
Systolic Click
Valvular Pathology
Valvular Syndrome
Mitral Regurgitation
Mitral Stenosis
Mitral Valve
Murmvr
Rheumatic Heart Disease
ASHD/CAD
A.V. Block
Coronary Pulmonale
Endocarditis
Coronary Heart Disease
Unspecified Cardiac Problems
Group 2
Mvnertensive and Vascular Heart Disease
"Diagnostic'
Diagnostic
code '!o.
Diaqnosis
code 'lo.
Diaqnosis
15 Hypertension
15 Hypertensive Heart,Disease
33 Dizziness/Vertigo
33 Fainting
33 Vasovaqal Attack
Jjbnycope
37 Vascular Disease
37 Vascular Insufficiency
50 Cerebral Vascular Attack
50 Stroke
12
-------
o 3. Total Cardiac Diagnoses - In Mohabetical <*>rder
1-21 .Yc -r-rmal EKG 2-:0
1-02 A.-/;ina 2-; 3
1-02 Anjina Pectons 1-16
1-36 Aortic Inccripetence (Insuff.) 1~03
1-36 Aortic Stenosis 1-52
1-03 Arrhythmias 1-02
1-';:. ASHD/CAD 1-36
1-43 A.V. Block 1-36
1-03 Atrial Fibrillation 2-37
1-22 B3--terial Pericarditis 2-37
1-03 Brady cardia 2-33
1-10 Cardiac Arrest 1-10
1-C3 Cardiac Arrhytiimia 1"03
1-10 Cardiac Decompensation 1~22
1-C1 CardiOMyopachy 1-03
2-50 Cerebral Vascular Attack
1-10 C.H.F.
1~36 Chronic Rheumatic Heart Disease
1-36 click
1"36 Click Syndrome
1-?1 Congential Heart Disease
1-47 Coronary Heart Disease
1-19 Coronary Occlusion
1-44 Coronarv Pulmonale
2-33 Dizziness/vertiyo
1-»'6 Endocarditis
2-33 Fainting
1-03 Fibrillation
1-03 Flatter
1-C3 Heart Flutter
2-15 Hypertension
2-15 Ky^3rtensive Her-rt Disease
1-02 In pending Myocartiial .Infarction
1-11 Isrhemic Heart Disease
1-36 Mitral .Regurgitation _
1-36 Mitral Srenosis
1- i.C iiici-al Valve _
1- .^6 Muriur -
1-19 Hyocardial Infarction/Occlusion _
1- 11 llyorardial Ischemia
1- 21 Ot.-i3r Cardiac Conditions
1- 21 Palpitations
1- ?!' Pericarditis
1-10 Pcssible Heart Failure
1- 3j Prcnature Ventricular Contractions (P.V.C.)
1- i(J Pulnonary Eder.a
1- 36 Rheumatic Hsart Disease
Stroke
Sy.,?ope
Syijolic Click
lacnycardia
Unsperifitd Cardiac PruL-l>_-MS
Unstable Angina
Valvular Pathology
Valvular Syi'drome
Vascular Disease
Vascular Intufliciency
Vasovaqal Attack
Ventricular Failure
Ventricular Fibrillation
Viral Pericarditis
W.P.W. Syndrome
13
-------
Group 4. Allergic Conditions and Acute Respiratory Infections.
Diagnostic
Code No.
01
01
01
17
17
17
30
30
32
Diagnosis
Allergic Cold
Allergic Rhinitis
Ha yf ever
Laryngitis
Pharyngitis
Sore Throat
Paranasal Sinusitis
Sinusitis
Strep Throat
Diagnostic
Code No.
34
34
34
35
35
25
35
42
Diagnosis
Tonsillitis
Follicular Tonsillitis
Hypertrot>hic Tonsillitis
Cold
Coryza
Cough
Upper Respiratory Infection (URI)
Allergy (Croup)
Group 5. Acute Bronchial and Lower Respiratory Infections
Diagnostic
Code No
04
04
06
06
06
Diagnostic
Diagnosis Code Ho.
Asthma
Bronchial Asthma
Bronchitis
Acute Bronchitis
Tracheo bronchitis
Group 6. Other Related Respiratory
Diagnostic
Code No.
08
03
03
08
25
23
23
26
26
Disorders
Diagnostic
Diagnosis Code No.
Chest Pain
Atypical Chest Pain
Husculoskeletal Chest
Pain
Unsoecified Chest Pain
Pulmanarv insufficiency
25
27
27
31
Diagnosis
Pleurisy
Pleuritis
Broncheolitis
Pneumonia
Diagnosis
Respiratory Distress (Failure)
Respiratory Alkalosis
Hyperventilation Syndrome
Dypsnea (Shortness of Breath)
14
-------
7 r SPIR'YTORY DW-WSIS CON ' T
Group 7- Chronic Restrictive Respiratory Infections and other Pulnojiary
Cor/Jit ions
Diagnostic
Code No
Diagnosis
Diagnostic
Code No.
Diagnosis
09
09
09
13
13
13
13
COPD 28
Emphysema 28
Restrictive Lung 28
Pattern 28
Granulomatous Disease 28
Pneuraoconioses 39
Sarcoidosis 51
Tuberculosis (TB)
Lung Abscess
Pulrtonary Congestion
Purulent Congestion
Respiratory Tract Infection
Viral Syndrome
Pleural Effusion
Unknown Pulmonary Problem
15
-------
Group 8. Total Respiratory Diagnoses - In Alphabetical Order
5-06 Acute Bronchitis
4-01 Allergic Cold
4-01 Allergic Rhinitis
4-42 Allergy (Croup)
5-04 Asthma
6-08 Atypical Chest Pain
5-26 Broncheolitis
5-04 Bronchial Asthma
5-06 Bronchitis
6-08 Chest Pain
4-35 Cold
7-09 COPD
4-35 Coryza
4-35 Cough
6-31 Dypsnea (Shortness of Breath)
7-09 Emphysema
4-34 Follicular Tonsillitis
7-13 Granulomatous Disease
4-01 Hayfever
4-34 Hypertrophic Tonsillitis
6-27 Hyperventilation Syndrome
4-17 Laryngitis
7-28 Lung Abscess
6-08 Musculoskeletal Chest Pain
4-30 Paranasal Sinusitis
4-17 Pharyngitis
7-39 Pleural Effusion
5-23 Pleurisy
5-23 Pleuritis
7-13 Pneumoconioses
5-26 Pneumonia
7-28 Pulmonary Congestion
6-25 Pulmonary Insufficiency
7-28 Purulent Congestion
6-27 Respiratory Alkalosis
6-25 Respiratory Distress (Failure)
7-28 Respiratory Tract Infection
7-09 Restrictive Lung Pattern
7-13 Sarcoidosis
4-30 Sinusitis
4-17 Sore Throat
4-32 Strep Throat
4-34 Tonsillitis
5-06 Tracheobronchitis
7-13 Tuberculosis (TB)
7-51 Unknown Pulmonary Problems
6-08 Unspecified Chest Pain
4-35 Upper Respiratory Infection
7-28 Viral Syndrome
16
-------
ir
C
3
£
tn
rH
C
CN
rc
E-
tn
4J
H
in
rH
e
0)
M
U-
c
tn
(0
0!
rl
(0
o
*
0)
tn
(0
e
H
vC
vO
^
r*»
0
01
CO
ro
CO
CO
m
rH
ro
CO
^^
00
G
3
0
tr
(ft
oi
o
c
Cr
rH
rH
r_j
m
ro
£>
rH
VC
<*
r-
CO
m
m
rH
m
m
^
^
0
4J
rd
(^
rl
Q^
tn
Q)
JH
y^
0)
c.
c
3
C
(0
tn
c
0
H
4J
H
tS
c
o
o
o
H
Cr
rl
0)
H
3
^
^
P
3
0
C
**^
in
c
O
rl
4J
U
O
in
c
rH
O*.
ro
rH
rH
*sT
vC
rH
r-
VC
»
vD
rH
O
r~
VC
rH
_^
in
o
CJ
en
o
H
4-1
O
CD
14-1
q
H
v,
XH
0
ro
rl
C,
in
0)
r«-|
01
rH
0
c
1C
rH
rd
rH
^
U
C
0
VH
l-C
CD
4J
3
U
CN
r-
<5r
VO
CTi
t
r*^
r-
c
*
CO
in
r-{
r-
,^^
ic
g.
o
Vl
e
tn
^4
O
^_j
0
in
H
T
t>.
i--f
0
4J
fl3
M
H
c
w
^1
c
0)
4J
(C
r 1
GJ
^l
^
ro
in
rH
Cji
^J4
in
rH
^r
TD
c
ro
in
C
O
H
4J
CJ
0)
M-
C
H
^°*
y^
O
4J
rO
^4
H
C-
tn
0)
^4
Oi
rl
O
H
V4
\ \
tn
CJ
^
CJ
H
C
0
6
ro
O
rH
rH
in
*
rH
CN
i pr^
jr h
r-
^
3
s
C
tn
c
0
rH
4J
rl
c
c
0
O
&
(0
C
£
iH
O
C-
V-t
0>
4J
O
ro
rH
CN
in
CN
0
fN
^^
ro
CX
^
o
V4
o
tn
01
tn
0
&
ro
H
t!
U
rfl
H
t3
rl
rfl
O
rH
i-H
in
CN
tf
CN
CN
*
in
CN
C
in
in
r-
Cr
&
01
rH
a
H
rH
3
5!
0)
01
IT,
*.
01
4-1
0
17
-------
Chicago campus; at Midway Airport; and near Buckingham Fountain in the loop.
However, the Midway Airport site presented the most concise set of daily data
as maintained by a complete and precision monitoring system. For this
purpose, daily climatological variables in the model are represented by
Midway Airport (as published by U.S. Department of Commerce, NOAA).
The daily measurements used in the analysis were average temperature
in degrees fahrenheit, total water precepitation in inches, average wind
speed in miles per hour, average relative humidity in per cent, hours of
possible sunshine in per cent, and sky cover in tenths.
Air Pollution Data
The city of Chicago maintains a large and complex air pollutant
monitoring network. A total of 34 sites have been created and maintained
over the past fourteen years. Although each site is within the boundaries
of the city, the governing responsibility of each site might lie with any
one of four agencies (the Federal Environmental Protection Agency, the
State of Illinois EPA, the City of Chicago Department of Environmental
Control (DEC) and the U.S. Public Health Services). Name, location and
site-codes for each station may be found in Figure 1. It might be noted
that although all sites are registered on the chart, over a period of time,
there have been stations shut-downs, full closures and alternations in
the equipment utilized. However, this is the most complete and concise
listing of what sources are available in metropolitan Chicago.
A variety of pollutant parameters are measured within the network,
including suspended particulate (total, benzene soluble, soiling index,
etc.), and elements as tin, nickel, lead, chromium, barium and arsenic,
and gases such as carbon monoxide, sulfur dioxide, nitric oxide, nitrogen
dioxide, total oxidants, ozone, methane and hydrocarbons. As expected,
a variety of collection and analysis methods are utilized in each process,
offering an overlap of data files. It is quite possible for more than
one agency to maintain surveillance of these air contaminants at one site.
Albeit confusing, these data analysis results should be more meaningful
in tandem with concurrent results from several monitors operated by in-
dependent agencies.
As previously mentioned, most of the city's air quality data is recorded
18
-------
Van Steutjrn (^
Taft (03)
Sullivan (2(1)
Edgewnter (37)
Ijikrvlrw ((14)
l-ogan Square (28)
Austin West (36)
Crane (13)
Polk (33)
)'J i E-PA (01)
G.S.A. (05)
(02)
Clay (19)
Figure 1. Location of air monitoring sites and 76 community areas in Chicago
19
-------
by the Department of Environmental Control. The DEC maintains two networks:
(1) the 26 station Chicago Air Sampling Network (CASN) and (2) 8 station
Telemetric Air Monitoring Network (TAMN). Monitored by the Technical
Services Division, the air contaminants currently monitored are sulfur
dioxide, particulate, carbon monoxide, nitrogen dioxide, nitric oxide and
ozone. The CASN network currently monitors 3 air contaminants; SO concen-
trations measured by West-Gaeke method on a 24 hour sample every six days;
TSP concentrations using a hi-vol sampler for 24 hours every three days; and
NO- concentrations using a modified Christie method every six days.
The TSMN network measures the remaining air contaminants on a daily
basis, tabulating 96 values per day (reported on a 15-minute basis) : again
SO , utilizing a conductivity analyzer every 15 minutes, every day; CO
concentrations measured with a non-dispersive infarred analyzer and NO and
NO (again) concentrations using the colorimetric method analysis.
Supplementary data was later provided by one site, Polk Street, at the
University of Illinois Medical Center monitoring all above noted pollutants,
plus ozone. However, this site is monitored by the State of Illinois
Environmental Protection Agency on a daily basis.
The Chicago Department of Environmental Control (DEC) was petitioned
for all Chicago air contaminant data monitored from the time period April
1, 1977 through April 30, 1978. This utilized both the 26 station CASN
network and the 8-station TAMN network. Data for the TAMN stations was
marked on monthly log sheets, having been recorded hourly for CO, O , NO
and SO_. CASN stations were recorded in Hie data format published by the
DEC. An example is displayed in Table 3. Unfortunately there was to be
little uniformity of collection periods; TSP was monitored every 3 days,
while SO and NO were monitored every six days (the same day). Of course,
all TAMN stations monitor the environment on a daily basis. Due to this
infrequency of pollutant monitoring, of 150 study days for which emergency
room visit records were collected, SO and NO were measured on 20 of these
^ £
days, and TSP on 42 of the days.
Separate master files were created for each pollutant, TSP, S0_ and
NO_ from the CASN network, accordingly for each day they were monitored.
However, after examining more air data for CO, O and SO (TAMN data) a
decision was made to eliminate these pollutant files for a variety of
20
-------
Table
CITY OF CHICAGO, DEPARTMENT OF ENVIRONMENTAL CONTROL
CHICAGO AIR SAMPLING NETWORK
Sulfur Dioxide S02 PPM April, 1977
DATE
A -
B -
C -
D -
F -
H -
I -
J -
K -
M -
N -
O -
P -
Q -
R -
S -
T -
V -
w -
03
04
11
12
05
06
14
15
07
09
16
17
10
18
13
25
20
28
21
Taft
Lakeview
Steinmetz
Cooley
GSA
Austin
Parr
Kelly
Lindblom
Stevenson
Calumet
CVS
Fenger
Carver
Clay
Kenwood
Sullivan
Logan Sq.
Hale
1
.024
.014
.039
.020
.023
.018
-
.022
.023
.018
.013
.051
.021
.022
M
.019
.020
-
.015
7
.007
.003
.020
.062
.026
.016
-
.021
.018
.017
.011
.018
.018
.024
M
.021
.028
-
.024
13
.009
.022
.013
.022
.028
.015
-
.019
.014
.014
.011
.010
.010
.010
M
.009
.021
-
.043
19
.015
.012
.014
.017
.019
.019
-
.008
.007
.007
.006
.006
.006
.011
M
.025
.017
-
.020
25
.008
.010
.014
.010
.010
.008
-
.010
.011
.011
.007
.013
.013
.012
M
.013
.008
-
.006
NO.
5
5
5
5
5
5
-
5
5
5
5
5
5
5
M
5
5
-
5
AVERAGE
.023
.021
.017
.014
.010
80
U -
YY-
E -
OO-
y -
29
22
32
30
31
34
37
Von Steuben
Washington
S.W.F.P.
Anthony
Adams
CRIB
Edgewater
AVERAGE
-
.019
.010
.009
.031
.011
.016
.021
-
.017
.011
.009
.014
.013
.020
.019
-
.027
.008
.040
.030
K
.017
.109
-
.007
.006
.025
.039
K
.013
.015
-
.012
.005
.006
.018
K
.011
.010
-
5
5
5
5
2
5
107
21
-------
reasons. This design would have been ideal to study, allowing us to use
nearly all the hospital data (TAMN data is daily data, versus 20 days usage
with solely the CASH network). However, the TAMN data could not be used
because of the disproportionately large amount of station shut-downs with-
in the study frame. There were significant periods of time during the study
period in which only 2-3 out of the 8 stations were open (for one parameter).
In addition, TAMN data was often subject to a careless rbunding-off process
onto the log sheet. For example, all measurements for SO,, were reported to
one significant digit, whereas most stations will show a. difference in SO
levels only when at least two significant digits are reported.
In order to supplement the CASN station data, data under the auspices
of the State of Illinois Environmental Protection Agency was added to the
statistical model. This site, Polk Street, at the University of Illinois
Medical Center was chosen since it was the closest to the Cook County -
University of Illinois Medical Center complex. Polk Street maintains daily
files of SO_, CO, NO, NO- and O., and was not plagued by shut-downs or
alterations in management. Therefore, again separate files, by date, were
created for each pollutant.
After all air contaminant files were transcribed, keyed on-line and
verified, a process was initiated to create community area exposure levels.
This technique was appropriate with CASN data only however, since
readings were obtained from only one station in the city, community areas
exposure levels could not be used for the Polk Street data. Consequently,
the assumption was made that this one reading was representative of the
entire city.
ESTIMATION OF MISSING POLLUTION MEASUREMENTS
The major problem faced with utilizing measurements from the city air
pollution network was the fairly large amount of missing data. As shown
in Table 4, out of 26 stations measuring TSP, 4 were closed down more than
25% of the time and therefore completely eliminated from the study. They
were Farr (14), Clay (19), Logan Square (28) and Crib (34). Similarly,
5 out of 26 sites measuring SO and NO were eliminated (Farr, Clay, Logan
Square, Von Steuben (29) and Crib).
For the remaining stations, the percentage of days shut-down ranged
22
-------
TWT.r 4 '
No. of Days Monitoring Station Shutdown
Station
Camp - 2
Taft - 3
Lakeview - 4
GSA - 5
Austin - 6
Lindblom - 7
Stevenson - 9
Fenger - 10
Steinmetz - 11
Cooley - 12
Farr - 14
Kelly - 15
Calumet - 16
CVS - 17
Carver - 18
Clay - 19
Sullivan - 20
Hale - 21
Washington - 22
Kenwood - 25
Logan Square - 28
Von Steuben -29
Anthony - 30
Adams - 31
SWFP - 32
Crib - 34
Edgewater - 37
Total Days Measured 132 66 66
* Stations Eliminated from Study
TSP
23
11
8
23
22
15
16
11
12
128*
16
13
17
24
132*
32
15
17
16
73*
11
9
15
24
70*
20
so2
4
0
5
5
3
4
1
2
0
66*
2
0
1
2
66*
3
0
3
0
66*
66*
8
2
1
30*
3
N02
8
4
0
5
5
3
4
1
2
0
66*
2
0
1
2
66*
3
0
3
0
66*
66*
8
2
1
31*
3
23
-------
from 0 to 25% for the duration of the study. (See Table 4). Since for
these sites the amount of missing observations was fairly low, the data was
supplemented with estimates of SO , NO and TSP levels.
This estimation was done by multiple regression methods. For each
site, the four closest sites were identified. Then, using monitoring days
on which all 5 sites were open,a predictive equation was developed in which
levels at the 4 surrounding sites were used as predictors. This was done
for each site and each pollutant, since there was no evidence that a predic-
tion equation at one site would be the same for all pollutants. In a num-
ber of instances, so many sites were closed down on a given day including
a predictor station that alternate regression equations were developed and
used in the manner described above. Tables 5,6,7 and 8 list the regression
2
equations used and their R values.
CALCULATION OF COMMUNITY AREA EXPOSURE LEVELS
To determine if there is a relationship between exposure to air pol-
lution and morbidity, a method to measure exposure must be established.
Obviously, to get the most accurate measure of a person's total exposure to
air pollution on a certain day, the person's location throughout the day
must be known and also the corresponding air pollutant levels at each loca-
tion. This was not feasible utilizing the available resources.
However, the residential address of emergency room patients was available.
Under the assumption that the average person spends a majority of his time
in or near his home, the community area in which the patient resides was
designated as his location. One major limitation to this process is that
a number of people who travel long distances to work spend a large portion
of their day away from home. Associated with daily absence from residence
is the consideration of exposure to pollutants caused by a person's occupa-
tion. In the first case, it was noticed that most people do not work more
than 12 hours a day, and so their residence would be most representative
of their location for the whole day. In the second case, no means were
available of obtaining the patient's occupation and work address.
By using measurements from Chicago's 26 air quality monitoring sta-
tions, a procedure was developed to estimate the level of pollutants in
each community area. To accomplish this, the following criteria were
24
-------
TA' :^ 5
PREDICATION EQUATIOt-'S FOR TSP MISSING DATA
0.30 Taft =0.908 x Von Steuben + 22.306
0.72 Lakeview =0.427 x Edgewater +0.396 x Sullivan -t-0.343 x Steinmetz -5.128
0.58 GSA *0.347 x Cooley +0.262 x Kenwood +28.692
0.58 Austin = 0.653 x Steinmetz +0.225 x i olley +19.39
0.73 Lindblom «-0.548 x Calumet +0.258 x Kenwood +0.131 x Kelly +7.479
0.34 Stevenson =0.484 x Fenger +0.344 x Calumet +13.821
0.84 Fenger =0.483 x Carver +0.336 x Stevenson +0.145 x CVS -5.15
0.74 Steinmetz =-0.463 x Von Steuben +0.310 x Austin +9.631
0.55 Colley '0.373 x Lakeview +0.450 x GSA +0.395 x Edgewater +8.189
0.30 Kelly =0.756 x Lindblom +24.389
0.81 Calurae-t =0.784 x Lindblom +0.208 x CVS -0.070
0.81 CVS =0.476 x Anthony +0.241 x Calumet +0.108 x Adams +0.259
0.70 Carver =0.501 x Anthony +0.465 x Fenger+12.665
0.62 Sullivan =0.329 x Lakeview +0.339 x Steinmetz +0.156 x Edgewater +3.593
0.57 Hale =0.906 x Calumet +23.222
0.80 Washington =0.787 x Carver +110.32
0.68 Kenwood =0.547 x Calumet +0,262 x SWFP +5.445
0.64 Von Steuben =0.269 x Lakeview +0.367 x Edgewater +0.149 x Taft +8.675
0.80 Anthony =9.773 x CVS +0.344 x Fenger +9.190
0.52 Adams =0.573 x CVS =0.434 x Fenger +41.445
0.54 SWFP "1.148 x Kenwood +4.645
0.65 Edgewater =0.462 x Lakeview +0.339 x Von Steuben +16.031
25
-------
TAHLF 6
ALTERNATE EQUATIONS FOR TSP MISSING DATA
R2
0.64 Lakeview =0.510 x Steinmetz +0.202 x Cooley +0.187 x Taft -0.684
0.63 Lindblom =0.327 x Kenwood +0.232 x Kelly +0.177 x Hale +0.108
x Stevenson +10.968
0.32 Stevenson =0.621 x Calumet +0.193 x Hale +13.261
0.58 Calumet =0.346 x Fenger +0.263 x CVS +0.208 x Kelly +11.596
0.77 Von Steuben =0.540 x Steinmetz +0.175 x Cooley +0.154 x Taft -1.621
0.53 Fenger =0.454 x Calumet +0.311 x CVS +0.042 x Carver +12.259
0.58 Lakeview =0.884 x Edgewater +9.1767
0.51 Austin =0.893 x Steinmetz +25.83
0.72 Lindblom =0.619 x Calumet +0.280 x Kenwood +11.937
0.67 Lindblom =0.859 x Calumet +12.269
0.82 Fenger =0.461 x Carver +0.483 x Stevenson -1.897'
0.66 Steinmetz =0.738 x Von Steuben +18.904
0.52 Cooley =0.555 x Lakeview +0.552 x GSA +15.700
0.47 Cooley = 0.976 Lakeview +30.683
0.78 CVS = 0.531 x Anthony +0.244 x Calumet.+7.253
0.74 CVS =0.683 x Anthony +12.155
0.64 Carver =0.847 x Anthony +12.100
0.53 Von Steuben =0.620 x Lakeview +22.282
26
-------
TABLE 7
PREDICTION EQUATIONS FOR SO MISSING DATA
0.51 Taft =0.339 x Steiimetz +0.209 x Edgewood +0.001
0.64 Lakeviow =0.478 x Kdgewater +0.253 x Steinmetz +0.001
0.22 GSA =0.790 x Kelly +0.784 x SWFP +0.005
0.51 Austin =0.415 x Steinmetz +0.194 x Cooley 0.147 x Hale +0.001
0.65 Lindblom =0.517 x Kelly +0.140 x Kenwood + ..301 x Stevenson
0.71 Stevenson =0.862 x Calumet +0.243 x Lindblom +0.006 x Kile
0.66 Fenger =0.191 x CVS +0.176 x Stevenson +0.343 x Calumet +0.127 x Carver
0.61 Steinmetz =0.588 x Lakeview +0.656 x Taft
0.66 Cooley =3.436 x Edgewater -0.810 x Lakeview +0.335 x Austin +0.001
0.58 Kelly =0.623 x Lindblom +0.071 x Hale +0.001
0.72 Calumet =0.360 x S- evenson +0.976 x CVS +0.124 x Fenger
0.50 CVS =1.298 x Calumet +0.245 x Adams +0.001
0.47 Carver =0.714 x Fenger +0.147 x Adams +0.001
0.65 Sullivan =0.544 x Edgewater +0.339 x Steinmetz +0.001
0.30 Hale=0.703 x Stevenson +0.828 x Calumet +0.004
0.32 Washington =0.334 x Adams +0.244 x Anthony +0.003
0.26 Kenwood =0.726 x Lindblora +0.007
0.28 Anthony =0.273 x Adams +0.243 Y. Washington +0.001
0.47 Adams =0.452 x CVS +0.355 x Washington +0.002
0.23 SWFP =0.258 x Lindblom +0.100 x GSA +0.002
0.70 Edgewater = 0.432 x Lakeview +0.372 x Sullivan +0.250 x Taft
27
-------
. 8
PREDICTION EQUATIONS FOR NO DATA
0.27 Camp =1.103 x Cooley +0.004
0.38 Taft =0.676 x Lakeview +0.005
0.70 takeview =0.351 x Steimetz +0.249 x Taft +0.267 x Sullivan +0.176
x Edgewater +0.006
0.29 GSA =0.687 x Cooley +0.286 x SWFP +0.008
0.35 Austin =0.488 x Steimetz +0.469 x Cooley +0.006
0.52 Lindblom =0.480 x Kelly +0.430 x Calumet +0.003
0.14 Stevenson =0.394 x Hale +0.018
0.45 Fenger 0.287 x Calumet +0.227 x CVS -0.074 x Carver +0.120 x Stevenson
+0.005
0.57 Steinmetz =0.587 x Lakeview +0.181 x Austin +0.004
0.66 Cooley =0.422 x Lakeview +0.157 x GSA +0.100 x Austin +0.012
0.57 Kelly =0.510 x Hale +0403 x Lindblom +0.006
0.52 Caluraet =0.296 x Lindblom +0.254 x Fenger +0.193 x Stevenson +0.200
X CVS +0.007
0.35 CVS =0.355 x Calumet +0.273 x Adams +0.160 x Anthony +0.008
0.09 Carver =0.485 x Anthony -0.653 x Washington +0.285 x Adams +0.030
0.46 Sullivan =0.477 x Lakeview +0.263 x Steinmetz
0.65 Hale =0.473 x Calumet +0.315 x Kelly +0.152 x Stevenson +0.003
0.40 Washington =0.339 x Anthony +0.300 x Adams -0.093 x Carver +0.016
0.41 Kenvraod =0.612 x SWFP +0.242 x Kelly +0.010
0.30 Anthony =0.556 x Washi-.gton +0.286 x CVS +0.003
0.33 Adams =0.390 x CVS +0.413 x Washington +0.007
0.50 SWFP =0.445 x Kenwood +0.211 x GSA +0.209 x Crib -0.166 x Kelly +0.006
0.32 Edgewater =0.399 x Lakeview +0.243 x Steinmetz +0.013
ALTERNATE EQUATIONS FOR NO MISSING DATA
0.28 Washington =0.288 x Adams +0.427 x Fenger +0.013
0.21 Anthony =0.586 x Fenger +0.248 x CVS +0.008
28
-------
used:
1. If a station existed in the community area, that station's measure-
ments were used for the community area.
2. If two or more stations were located in a community area, their
daily measurements were averaged.
3. If no station existed in the community area, the closest stations
were located and their distances to the center of the community
area were estimated. The community area daily measurements were
calculated as the weighted average (inversely proportionate to the
distance of each site from the CA) of all the surrounding sites.
Thus, for each of the three pollutants the exposure levels of 76 com-
munity areas were generated for each day by a Fortran IV program.
CALCULATION OF DISEASE SPECIFIC AVERAGE EXPOSURE LEVELS
Once a set of community area exposure levels was completed for each of
the three pollutants (TSP, SO,, and NO ) the next step was to obtain personal
exposure levels for each of the 21,000 patients' records. This was done
by identifying the date on which the patient visited the emergency room
and the community area in which the subject resided. The community ex-
posure levels for that day then were added to the patient record.
The exposure levels of all patients in each specific disease category
were combined for each pollutant. Then an average exposure level was ob-
tained by dividing the sum of the individual exposure levels by the number
of patients in the disease category. If each patient's exposure level is
X. and the number of patients in any specific disease category is n, an
average exposure level (AEL) to a specific pollutant for that disease
category can be expressed as follows:
(i)
MULTIPLE REGRESSION ANALYSIS
Multiple regression techniques were used to examine if any specific
kinds of air pollutants have significant impact on the variation of emer-
gency room visits for cardiac and respiratory conditions, even after
29
-------
controlling for daily climatologic changes and days-of-week effects.
The dependent variable was the percentage of excess visits for cardiac
or respiratory disease, which was calculated as follows:
100 (X. - X)
% of excess visits = + 100 (ii)
X
where X. is the number of emergency room visits for a specific disease on
the i-th day and X is the average number of visits for that disease during
the study period. To avoid a negative sign on per cent of excess visits,
100 was added to the right side of the equation. If X. is greater than the
average, % of excess visits becomes greater than 100%. If X. is smaller
than the average, its percentage becomes less than 100%.
The justifications to use per cent of excess visits as a morbidity
index are as follows: (1) a majority of people who utilize both Cook
County Hospital and University of Illinois Hospital are persons from low-
income families whose health condition may be more affected by air pollution
than others, and (2) it is assumed that per cent of excess visits would
not be influenced by the size of the risk population (although unknown)
because the risk population would not be changed during the one-year study
period unless there were great changes in policies and administrations for
patient care services in these two hospitals. Frequency distributions of
patients in both hospitals are shown according to race, sex, age and diag-
nostic group in Tables 23-26 in the Appendix.
Three types of independent variables were used; (1) air pollutants,
(2) climatological measurements and (3) days of the week as dummy variables.
As described in Air Pollution Data, the number of observations was different
from one pollutant to the others, so that only one pollutant was included
in each equation. Accordingly, four different data sets were prepared and
regression analysis was done in four phases as follows:
Phase I. Only these days on which SC> and NO? were monitored by
the CASN were used (i.e., 20 study days). Two models
were developed:
SO -I (based on average exposure levels of patients)
NO,-,-1 (based on average exposure levels of patients)
30
-------
Phase II. Only those days on which TSP was monitored by the CASH
were used (i.e., 42 study days). Only one model was
developed:
TSP-II (based on average exposure levels of patients)
Phase III. Only those days on which pollution measurements from the
University of Illinois Medical Center at Polk Street
were recorded were used (i.e., 131 study days). Four
models for each disease group were developed:
SO -III (based on daily averages at Polk Street).
NO?-III (based on daily averages at Polk Street).
CO -III (based on daily averages at Polk Street).
NO -III (based on daily averages at Polk Street).
Phase IV. A separate analysis of ozone in each disease group was
done by including all study days between May 17, 1977
and August 31, 1977, the period when daily maximum
temperature was usually higher than 75 F, This allowed
36 study days to be used. Only one model for each
disease group was developed:
O -IV (based on daily averages at Polk Street).
Daily climatological measurements used for analysis were (1) average
temperature in degrees fahrenheit, (2) total water precipitation in inches,
(3) wind speed in miles per hour, (4) humidity in per cent, (5) possible
sunshine in per cent and (6) sky cover in tenths. Finally, days of the
week (Tuesday, Wednesday and Thursday) were included in all models as dummy
variables because frequency of hospital visits may be higher on a specific
day of the week than on other days.
The general model to be tested in this study is expressed as follows:
Y=BflX0 + 3iXi + BzX2 + £3X3 + 3«*Xl» + ...+ BnXn + £ (iii)
where XQ= 1 for all observations, and where Xj and X2 are used for day-of-
week identification as follows:
Xj X2
i o = Tuesday
o i = Wednesday
o o = Thursday
and where 3o, 3i/ , 3n = regression coefficients (Bo is constant since
31
-------
since Xo = D
X = a snecific pollutant
X., ..-, X = climatoloqical variables included in the equation
4 n
n=3+m (m is the number of climatoloqical variables added to the
equation.)
e = random error term, and
Y = % of excess visits in a snecific disease group.
In comnutation procedures, a specific nollutant was introduced in each
reqression analysis as a first sten, then two dummy variables (Tuesdav and
r\7ednesdav) as a second step, and climatological variables according to the
stenwise method. Since our interest was centered on the air pollution
variable, only climatoloqical variables which had significant or meaninqful
influences on the reqression model were included according to the criterion
that the ^ value of the regression coefficient is areater than or close to a
significant level (n<.10).
32
-------
SECTION 6
RESULTS AND DISCUSSION
MODELS FOR TOTAL RESPIRATORY DIAGNOSES (TRD)
Eight regression models were created to examine air pollution effects
on total respiratory diagnoses (TRD) as shown in Table 9. The orooortion
of the variation in percent of excess visits for TRD ranged from 19% in
model 4 (NO -I) to 36% in model 2 (SO -I). According to the F ratios, the
overall test for goodness of fit of the regression equation "as statisti-
cally significant in models 1, 3, 5, 6, 7 and 8. All regression coefficients
of the pollutants were not statistically significant excent NO in model 7
(NO-III). Humiditv and sunshine as climatological variables were not in-
cluded in anv of the models. Wind speed was included only in model 8
(0--IV). Daily average temperature had the most significant impact on TRD
in models 1, 3, 5, 6 and 7. That is, as the temperature decreases, the
number of patients with respiratory conditions increases. Only the last
model for ozone had significant days-of-week effects of TRD.
Bv paying attention to each model, the strength of the relationship
between a specific pollutant and TRD can be examined. Model 1 slyjws that
the partial regression coefficient of TSP, .230, was greater than its
standard error, .170, but was not significant, controllina for average
temperature and nrecipitation which were meaningful additions to the nodel.
Models 2 and 3 examined SO effects on TRD, although their SO indices
were different. The regression coefficient of SO in model 2 was considerably
greater than its standard error, but not statistically significant because
the samole size was too small. The regression coefficient of SO in model
3, on the other hand, indicated no association between SO and TRD. SO
£ £,
measurements at the University of Illinois Medical Center on Polk Street
were used in model 3 as an alternative of the average exposure levels (AEL)
of patients, since there were only 20 observations of SO measurements from
the Chicago Air Sampling Network (CASN). According to the correlation co-
efficient between SO from the CASN and SO from Polk Street, 0.477, as
provided in Table 19 in the Aopendix, SO levels at Polk Street did not
reflect well the average exposure levels of patients. Also, differences in
the regression coefficient of SO between models 2 and 3 can be attributed
33
-------
^^
D
S
E-
*
tn
W
cS
^
H
Q
>!
O
EH
E
C-
tn
1^
p.
O
E"*
..
a
Q
o
>*
H
S
eg
c
S
I
0
^^
-*
i*-r
.r-s,
vfl
*- *
.. «
n
31
^_^
ro
N
rH
£>
M
t
ro
O
M
H
H
1
O
H
M
H
1
8
H
H
M
1
CN
O
H
1
CN
O
M
H
H
1
O
*>
H
1
tn
H
H
1
1
-*
*
T^ UO rH O
tn m (N ro
ro m
*
*
*
* ro in <»o
in ro ro CN
H
in *
rH *J
*
*
*
* ro M m
in ro rs (N
* * * rH
(N -
rH in
*
*
co ^r T T
in ro ^r c\j
. rH
o -
rH ^0
^tf C) co \n
^T rH CO p-H
*
*
cc ro o ^
n ro CN CN
* » rH
0 *
rH vD
c \p m ^r
O ro m rH
* »
rH in
*
n IN kO ro
CN m
D^
3 o
rCN 'd
+J 0* ro *
3
2; CM "C
ro ** o
r* ci ro
ro »
*» ro rH
i
p~ LO ro
CN rH LO
rH rH
in ci O
co 01
Cl *
rH rO CN
rH
in » CN
ro O vc
cs ro ro
rH
o
o co ro
rH IO LO
1
LO
ro no
CN Cl »
rH CN
CN ro rH
in ro
CN O ro
r- CN rH
r- o o
O CN rH
rH
rH
en a
CD ,
fi *
CO 4->
U 3
in rH
8 g
^,
ro
CO
*J.
^^
m
^r
i
n
^^
r~
ro
Cl
f-v
CO
*-^
,_^
CO
Ci
-__
in
J,
CO
ro
CN
^
fN
CO
rH
[l-
fc
* CO
^D
m
1 *T
4t
* ro
in r-
10 *
e>
* 00
CO .
* *3"
1 ^*
* r-
* CO
CO
LO ?
r
* «
* ro
co r~
. r-
Cn
Cl
rl U,
3
id
I
#
r-s * ^*
CN ci r* ci
r^ o rH tn
CN CN "? LO
1 ^^ » '
in ro
P 01 0)
0 -U -rl > C
in -rl Cl O M
73 -rl O R
a -ri o n
c E ci >^ n
rl 3 rl X 3
S E c, tn 10
k
r^ f-~
in en
ro \n
CN ^
M1 O
CN O
*-r
CN CN
ro o
i
^^
in o
in o
^*
CN ^s
VO Cl
. in
LO
rH ^~
1
^^
r** rH
ro 0
^-
10 ~
Cl rH
^
O
1
Cl
rH in
CN
in
p
a;
n t,
in
u
U~(
a>
QJ RJ
CO
^j< *,-^
VD Ci
O CO
t
in
o r~^
^ v£
«
^3" ^f
iH *-^
0
ro .
^^
o m
^ r^
^. *^
^* ^^.
CN in
* ^J
rH
H ---
ro ,--,
cr> ro
O
rH
tr
in t,
>,
0
«
O
-§
o
CM
T3
§
J;
rH
C
0
JJ
C
r3
^J
3
rH
rH
O
o.
M
0
MH
"*^
rl
0
rl
rl
O
o
rl
CO
t3
§
4J
tn
J
d
.,_!
u
rH
UH
O
O
u
0
rl
tn
in
u
n
CP
o
rl
*
Cl
rH
*S
rH
>
JJ
C
o
13
Cl
O,
Cl
c
H
6
O
v^
£
r7
Cl
4J
§
rH
O
u'
£
«
«
«
T3
C
CO
^
in
>, o
rH
2 k
^j
o «
Cl *
a
tn
rt O
rH
£ ^
TJ
9
0 A!
W -H
rl
- 11
tji tn
O "3
IH
in
UH «
O
-O
tn u
Cl X
C rl
rl «
rH E
c tn
O -rl
O Ci
o o
£§
Cl CJ
tn -H
Cl **H
tH -H
VI -H
tn
Cl
« o
O rH
H G
U >
in o
4J
fO Cl
p £
tn F
CN
34
-------
to the difference in correlation coefficients between TRD and SO from the
CASN (0.364) and between TRD and SO from Polk Street (-0.035), as shown in
Table 15 in the Appendix. Using air pollution measurements from Polk Street
as an alternative of the AEL was probably the greatest limitation in this
study, although only air pollution data from Polk Street were available at
the time of data collection.
Association of NO with TRD was not significant in both models 4 and
5. Most of the variation of TRD in model 5 (34%) was explained by average
temperature, sky cover and precipitation. The correlation coefficient be-
tween NO from the CASN and NO from Polk Street was only 0.179 (Table 19
in the Appendix). Correlation coefficients between NO, and TRD were -0.172
based on the CASN data and 0.011 based on the data from Polk Street (Table
15 in the Appendix).
Association between CO and TRD was not significant, after partitioning
effects of temperature and sky cover which were more strongly related to
TRD than CO was. However, the standard error of CO was smaller than its
regression coefficient.
Model 7 shows significant association between NO and TRD, controlling
for average temperature and days-of-week effects. Other climatological
variables were not included in the model because of no meaningful contri-
butions to the increase in the variation or F ratio by adding them to the
equation. In short, the regression coefficient of NO, 116.8, was significant
at p<.10 (3.43> F (.10, 1, 120) = 2.75). Thirty-three per cent of the
variation in TRD was accounted for by average temperature, NO and days-of-
week variables.
Model 8 indicated no significant association between O and TRD. The F
statistic of O. was only 0.836 which was far from the significant level, F
(.10, 1, 30) " 2.88. Climatological variables included in the model were
sky cover (significant at P< .05) and wind speed. F statistics of day-of-
week variables were both statistically significant; 5.30 > F (.05, 1, 30)
» 4.17 on Tuesday and 9.40 > F (.01, 1, 30) » 7.56 on Wednesday. This
means that the number of emergency room visits for TRD varied significantly
among the days of the week. No reason was found for this.
35
-------
MODELS FOR ALLERGIC CONDITIONS AND UPPER RESPIRATORY INFECTIONS (RDl)
Table 10 is a summary of eight models for allergic conditions and
upper respiratory infections (RDl). The variation of RDl explained by
the regression models ranged from 29% in models 1 and 4 to 40% in model 2.
According to the F ratio used to test the overall goodness of fit of the
regression equation, statistically significant levels were observed in
models 1, 3, 5, 6, 7 and 8, but not in models 2 and 4 from phase I analysis
which had only 20 observation days.
Out of eight models, only model 7 (NO-III) indicated a significant
impact of the pollutant, or NO, on RDl. The standard error of NO, 85.4,
was about half the size of its regression coefficient, 161.5. Its F statis-
tic, 3.58, was greater than F (.10, 1, 120) =2.75 but smaller than F
(.05, 1, 120) = 3.92. Average temperature had the strongest association
with RDl in which the F statistic of its coefficient was 31.88 > F (.01,
1, 120) = 6.85. Precipitation, sky cover and wind speed were included in
the model because of their considerable influences on both the dependent
and other independent variables. The days-of-week variables were not
significant at all as shown in their F statistics, 0.16 and 0.46. Thirty-
four per cent of the variation of RDl was explained by the linear model
including NO, average temperature, wind speed, precipitation, sky cover and
the days-of-week variables, with F ratio, 9.25> F (.01, 1, 120) = 6.85.
Among seven other models, coefficients of SO in both models 2 and
3 were considerably greater than their standard errors, although their
F values did not reach a significant level. There is little reason to
support the strong association between RDl and other pollutants, such as
TSP, NO , CO, and O according to models 1, 4, 5, 6 and 8.
MODELS FOR ACUTE BRONCHIAL AND LOWER RESPIRATORY INFECTIONS (RD2)
As shown in Table 11, the models for acute bronchial and lower res-
piratory infections (RD2) were less reliable since the F ratio to test
the overall goodness of fit of the regression equation was found significant
in model 8 (O -IV) only. Variation in emergency room visits for RD2 ex-
plained by the regression equations fluctuated between 0% in model 4 and
39% in model 2.
Only models 2 and 3 exhibit a statistically significant coefficient
36
-------
^
t *
g
z
f~
C
B
t
M
2
o
EH
5
H
C,
£
«
H
{3
O
g
vv:
M
M
O
^
U
M
J3
H
r4
^
a
u
Q
Q
£
M
G
M
8
g
0
rH
a
§
£-.
CD
*"
-^
r**
^^
VD
in
*_rf
2.
^^
CO
CM
^^
rH
^
M
1
M
M
H
4
2
H
M
H
A
o
M
H
M
1
CM
O
z
M
CM
O
H
H
H
1
c
w
H
CM
O
W
H
H
P<
B
*
!"* fO C4! O
in fo CQ PO
(N in
J
O"> ^* if) **">
in ro o* cr» CM
» rH
O *
VD
^f eft n m
in
in r^ o^ ro
» < »
M m
Pi
rH O
0- -H
4-» K nJ
z: fc. -o
c
en
ro
en
0
f^
c-
rH
CO
er>
CM
rH
r-
in
CO
S
c
in
CM
en
rH
rH
*J
!
01
c
o
u
*» in ^^
. . rH
rH O O
in in
in
4t r»>
in T co
rH in
vD CO CO
rH ~
rH CM
O CM O
rH
rH CO »
CM rH r^
CO O rH
C
rH VD ^
^^
r- n 5
CO ro
CO v£) rH
1 *--
^*,
r-
rH in co
rH VD *
CO CM rH
^_^
^f
CO CM CO
O en
VO CO rH
rH rH
CO rH ?0
CO 00 O
rH rH .
rH
D1 0)
S oi fr.
4J
1
rH
rH
g
«
« CO T rH
CO CO rH VD
VD rH rH rH
CO *-^
1 ^
41
He ^-^
« CM
rH rH
O
r- CM
1 -^
« VD
CM O
rH .
r- 0
vr
I *~"
* «
« r-
VD O
vD O
. CO
1 *-*
^
rH in
vD vD
T CM
«
CM P
in in
VD
CO
v> tr en
a b s *. a t
n
IX "O
3 VV
u »! >,
n o. u
U 01 .rH
0) "O
C, -0 .rl
F. C |
^
en
rH
rH
1
CM
ID
rH
H
1
in
r-
rH
rH
|
*
m
>o
CM
(O
m
rH
rH
*
r»*
O
CM
1
*
CM
VO
r-
rH
1
tr
vV
S
U
a
u
U
H
0
HI
D.
^
4 r^
^T in VD
CO O M1
in vo
^^ ^^
VD in en
vO rH CO
CM rH CM
«
«» r~*
r- n CM
f CM VD
CM rH CM
CO O 1-
in CN in
CM rH CM
in
en
CO
f-^ ^*
vo co o
^r rH ro
CM rH CM
^^
in
0
i
^^
CO
o
^^
Cn tr
tK rl fc4 K
rl
O O
> c
8 2
n
5 §
t/l Ul
^^ + ^^
in « o
CM t CO
CM CM r-
' CO ^^
0
rH
CM
1 ^-
^^
Ol rH
\O O
1 **^
r*- rH;
VO 0
1 *"*
in _
o
in m
*H
^^
in fH
vo o
i **
CM
VO ^
en
rH CO
rH t
1 ~
^
Cn rH
VO rH
CO -'
tr
t, rl b
a
u
0)
UH
w
.* >.
U fvl
0 -0
S 01
in p
o f<
1
I
ft
* ,r-,
* tH
O 0^
^. r-.
ro *
rH
C» §
^ *
*""
^^
^T H
CO ^T
CO ^->
rH P
CO T
CO ^^
CO --,
CM
CM O
1 i
^^
rH Cn
r^ co
CO
ro
o
0
^
C rH
ro o
rH --'
tn
St.
^,
O
0)
o
b.
0
§ .
"£
rH ,-^
§ S
^ R
C ^
5 """
3 «
rH «
rH *
O
p. -o
rl Ifl
O
in
rl 0
s^,;
S^s
H
*O 4J *
rl O *
itj fl;
O & >
§01 r~
f> O
1 1 c i
+^ M rH
01
» tT 8
C -O
0 C *
rl (3
u
;
fl) 01 -rH
O k c
> c «
w u
P 01 M
C O MH
O iH 'H
o c, c
C fl) D1
0) V4 *rt
d< 0)
tt) (U
^ H VH
C Q O
H
O rH
-53 §1
(0 0) Q)
«,.rj M
rl « O
^^ *-x
rH CM
H)
4J
O
-------
CM
|
w
0
\j
E
z.
H
^
8
g
M
Cu
in
W
3.
i
§
0
S
a
H
6
1
m
tl
t(
»3
fj*
«.
a
§
o
s^
S
H
g
O
at
.
a
§
^.^
CO
^»
^^
~
,-,
ID
n
^^
-*
co
-
^_^
N
^^
rH
M
1
CO
O
H
rH
1
O
H
H
(H
t
8
H
M
M
I'M
O
X
M
(N
O
z
n
H
1
cT
CO
H
t
CM
O
l-t
M
t
&4
B
ft
*
CO *3" 0"i O
m n o m
i * *
m in
^ vo r- rr
CN O (N CN
H
rH -
*r ^ ^ *r
CM O CM CN
. rH
rH »
VO
CN m ^o \o
M o m CN
' * rM *"i
V
H O O VO
O O O rH
ro
r- co ^r m
^ ^
CO CO
in r--
in co
rH rH
1
^T CO
CO 0
CO cn
.
*
CO rH
rH CN
^f
1
4
*
in rH
in v
CO rH
in IN
CM O
CN cs
rH CM
CP 09
M m
4J
S
-p
S
rH
rH
o
f*
^-^
*^
rH
v_^
x-x
*3*
O
^
^-v
rH
O
^^.
c^
CN
rH
>-
^
0
o
w
^_^
o
r-
*?
\£>
CN
V£5
*"*
^
rH
ro
"
fc,
ft ft
* ^^ ft ^^
in r^ cn ^r cn
o cn CM CN co
r^ o in in CM
I ^^ ^*
rH x^ .
CO O CO IN O
. rH rH rj cn
1 IN 1 CM T
~ ^
rH CO rH CO CO CM O
cn r» in co H *r H
rH CM 1 rH 1 CM in
| -^ ^ <~.
ft
ft »^ cn
rH in ^f
T in
. . vo
IN <»
in
rH
rH
«
« ^
CM *r m IN T
cn cn rH IN IN
IN co r^ vo
rH r*- co ^f cn cn cn
o cn r^ o vo c^ rH
rH CO VO CO *J" fN r^
*^ 1 ._ ~^
ft
vo rH co cn vo
V \O CO VO CM
» t
rH lO 1 CO CM
I rH
cn 0-. fJ> Cn CP Cf D>
d> 0) QJ Q) ft) O O
[fl
O
o
VH
HI
o x >*
S-H «3 id
rQ 4J 11*0
3 tu id H S n
4Ji)>i4Jg)oTo)
« &4J-H ? et-iS
M
^^
m
ro
*-*
»-»
f"«
ro
^^
0
^^
o
o
r~
in
0
* *
^^
IN
C-
lu
^_,
CO CO
co cn
O rH
CM
^^
CM CM
O rH
O rH
rH *~^
O CN
rH
cn
rH
^
cn 10
in in
O rH
rH ,-x
O O
"
^^
v cn
CM V
0 rH
r~ co
CO rH
r^ CM
CO
o ^~
co m
rH
in
tr
S CM
N
O
in
I
t3
§
^^
^
rH
8
4J
s
^J
*H
rH
R
O
l_(
o
tu
E
id
D
§
4J
to
^
tu
-tH
U
Ij
8
o
c
o
H
in
V
n
tr>
4)
v<
fc
r (
"S
-H
rJ
10
4J
C
1
§
QJ
T3
C
H
id
u
M
o
-
5
§.
^^
O
«
«
«
a
3
. ^
>* in
rH O
> »'
O
H ii
rH H
id
o E
to
tu
£8
w S
to ro
0) O
r4 *rH
a) ^H
^ §
ffi -H
1J
o o
r-l
4J rH
10 O
jj 0)
a rH
to 0)
t P
38
-------
for the pollutant included (SO in both cases). However, the coefficient
of SO in model 3 was negative. This is contradictory to the common know-
ledge that SO is harmful to the respiratory tracts. Model 3 is not
appropriate to estimate effects of SO on RD2. Despite the small sample
size in model 2 (of which the SO variable was based on the average ex-
posure level of patients), the standard error of SO , 2141, was less than
half of its regression coefficient, 5355, with a significant F statistic
of 6.26 > F (.05, 1, 13) = 4.67. Other independent variables included in
model 2 were average temperature, wind speed, sky cover and days-of-week.
MODELS FOR TOTAL CARDIAC DIAGNOSES (TCD)
Table 12 shows the regression models for total cardiac diagnoses (TCD)
The variations of TCD explained by the equations were fairly small as
compared with those of TRD in Table 9, except in model 2 which explains
2
48%. In short, R 's were 12% in both models 3 and 6, 13% in 7, 24% in
1, 25% in 8, 26% in 4, and 48% in 2. Significant F ratios were produced
in models 3, 5, 6 and 7. Tt is interesting that all models for TCD had
a significant day-of-week effect to some extent, while only model 8
(O -N) for TRD did.
Three pollutants exhibited significant regression coefficients; SO
in model 2, NO_ in 5, and NO in 7, In model 2, the regression coefficient
and standard error of SO were 3869 and 17^7, respectively with a signifi-
cant F value, 5.13 > (.05, 1, 13) = 4.67. Wind speed, sky cover and sun-
shine were included in the model because they substantially improved the
significance of the total model and the coefficient of SO . The days-of-
week variables were significant (F values, 4.29 and 3.16 > F (.10, 1, 193)
= 3.14). Forty-eight per cent of the variation of TCD was explained in
model 2, but its F ratio did not reach the significant level because of
the small sample size. In model 5 (NO -III) , the regression coefficient,
standard error and F statistic of NO were 224, 115 and 3.80 ( > F (.10,
1, 120) = 2.75, or < F (.05, 1, 120) = 3.92), respectively. Only 14% of
the variation of TCD was explained by this model which included sunshine,
temperature and precipitation, in addition to the days-of-week variables.
The overall goodness of fit of the model was significant, as indicated by
its F ratio, 3.35 > F (.01, 6, 120) = 2.96.
39
-------
u
S3
Q
M
§
a
9
. «. H
CO 1
* ro
O
H
H
--^ H
~ 0
2
M
H
^ M
- 8
M
H
n 1
in r>j
0
10
H
(N fN
O
to
H
H
1
CM
d 8
o m i*- o
rH in
+
I
vO ro ro in
ro rH CO rN
ro *
tn
*
in fN ro *r
ro fH eft (N
. rH
fN *
^0
«
ro t~t ro CN
« rH
ro -
vD
^»
X ^ *
ro o *
rH (N ^
tN ^r co
1 £
J
*«-^ VO *"* * *'^>
in rH ro CN in rH
r- co cr> r- fN
rH rH rH (N CTt
| ^ | ^ ^
«.
CO CTv rH * * * .
c^ cj» wo in co
rH * 0^ fN VD
1 » rH rH 1 CO
| ^
O ^f
^- VD
\n M
1 «-*
*
*
O r~* * *-^
ro r*» CO C^
^3* * *
rH fN fN O
1 *^ >H
*~^
^, .
m P § 5 r- ro
00 H rH
7 ci ~" ^
*
f»* \£l (N CO
o m o^
CJi * *
rH fO. CN i *J (U O
i-l W -H D( O -H
fl) 13 M O ^
C. T3 M t) W
E H i Jl ^ §
F *3 :c P< in to
^. * ^-s.
C^- fN k£) (N
m CN VD ^
fN *~-r ro *^-^
cr^ n co in
in co in -31
O^ rH CO rH
,
/< >< -o
0) «3 tn
a) -a o
?tn c
a) tJ
o P ^
>.
10
a
c
B
"^
"c
O' rH
O
S 6
JJ ^»
rH *
rH *
° *
o
° s
^-' 1 K
-^ ^>
^1 rH ID
0 O O
VJ -H II
m -u e
0
TJ O
ra w *
a si
i ^ -
W '0
VI k, rH
« "0 II*
4J C S
C 10 -'
Cl
H *
o a
H
in m
H | »j
C) - W
u tr n
o o
C rJ 4J
O Cl
H u-i ra
M O
in to
o m a
g, g TJ
CJ --H S
^ H M
" c e
rH 01
£ -C -H
ra o
H JJ O
to C 0
10 0) c
> IT r3
0 0
4J rJ -H
C C 'H
1) 0) H
"g ^ &
O CJ -H
a, ^ 01
0) Cj
coo
iJ rH
X tn a)
0 -H >
13 .P O
0) (t rH
^i tn o
o g
^x ^
rH fN
..
O
o
7.
40
-------
Model 7 showed a siemificant association between NO and TCD. The
pollutant's regression coefficient, standard error and F statistic were
155.9, 81.81 and 3.63 ( > F (.10, 1, 120) = 2.75, or < F (.05, 1, 120)
= 3.92), respectively. Only 13% of the variation of TCD was explained
by this model which included sunshine, precipitation and the days-of-
week variables. Its F ratio to test the goodness of fit of the model
was 3.85, indicating statistical significance (i.e., 3.83 > F (.01, 5,
120) = 3.17).
MODELS FOR HYPERTENSION AND VASCULAR HEART DISEASE (GDI)
Table 13 is a list of regression models for hypertension and vas-
cular heart disease (CD1). The variation of GDI explained in each model
ranged from 10% in model 3 to 46% in model 2. Significant F ratios
appeared in models 2, 3, 5, 6, 7 and 8. Some models show significant
day-of-week effects, but not as strong as TCD.
Two models indicated significant regression coefficients for the pol-
lutants, but only one model, model 7, was valid since the other, model 8,
showed a negative regression coefficient for O . This negative effect of
O on CD1 was caused by its inverse relationship with GDI (-0.299), as
provided in Table 18 in Appendix. Since O would not reduce a risk of
having CD1, this regression model was unacceptable for O -CDl analysis.
In model 7, standard error of NO, 94.4, was half the size of its regres-
sion coefficient, 183.3, with a significant F value, 3.77>F(.10, 1, 120)
= 2.75, or < F(.05, 1, 120)= 3.92. Sunshine was chosen as an additional
independent variable because it significantly improved the relationship
between CDl and NO. Although the coefficient of the pollutant in each of
six other models did not attain the significance level of 10%, models 2
and 5 showed considerably smaller standard errors for SO and NO , res-
pectively, than their regression coefficients. In model 2, 46% of the
variation of CDl was explained by the regression equation including pre-
cipitation, sky cover and the days-of-week variables, in addition to the
SO variable (regression coefficient = 2206, standard error = 1581 and F
value = 1.95 F(.10, 5, 14) » 2.31. In model 5, only
13% of the variation was explained by the equation including sunshine,
41
-------
^J
g
"-'
L;
a
K
V,
O
K
H
2j
oi
5
B
S
>
c
JJ
rtl
S?:
0
H
BUI
g
U
p.
~
a
Q
2
B
H
D
M
I
P-l
a
Isj
,»-x
00
* *
f+i
r-
*-^
In
*-*.
*_>
^^
w-
(^N
CN
*-%
H
^
H
I
ro
*-"
w
M
M
A
S
M
H
H
1
8
H
I-H
H
'
0
H
1
O
Z.
H
H
M
1
rs
O
CO
H
1
CM
O
«
H
H
1
P
*
<
in c ^ fN
^T fN \O CO r^1
» *
rN n o
«
*
*
ro rH m fN m
"? - fN
' S
*
*
f- "tf & ro
co I-H r*» fN i£
(N * m
r- o
*
*
^D CO
m o o ^o
rf fN co m vO
* *
»H m ro
p^
I .2 S
rTM , iJ 4J
? c 2 *i 1
r (k. T3 U
* ^^
* 1^
^ fN O
o o
^ ^o ^r
1
ro *j" ff^.
. . r-
00 O
rH ro
*"'
^
r* ro co
v0 CO rO
If) *f r-l
o ^r .-»
p^ ^r J
CD ro
-H rH rH
*"^
rH v£) ^^
^r
in CN rH
CM r- ^>
ro m P
in fi ifi
fN ro
vD ^ in
o co o
CN in *
CN rH rH
i-* r* (n
O rH W1
H CN rN
* * 1 *
tf i
S .'»,
U
R
i)
3
rH
rH
g
« '-s
# ro
01 O
(N
rH
1 rH
~ ~ 0 ,- *
vDf'-tOOro'j' inc1'
T >r» ^r m in CN P-
H rH rH rH CN 1 ^T
^
f ^>,
f^. f*^ f f1^
> 4-* dj flj
18 O, 4J -H > C
rl 0) -rl D, O -rl
<1) "0 -H O £
0 TJ -rl O Ul
EC E V >< C
01 -H 3 >H A: 3
PS E: CL w in
* ^^.
CO *-. ro ro
rH in (N o
0 rH r/'
-, V
O ^C t** ro
o in m
rH
*- O
ro ro CN ro
in CO in
r- -- ci
^-» f~*.
r^ c^ ro r-
r- T CN o»
m ^. co
r^ cr» *
r- i1 o o
o
CO CN *
^
--. CT^
\D *S* rH CN
in co m
rH
r- ^ en
C^ *H ro ^-^
vO m CN o
c
C^ «N 1
CN -** -^
r^ ro r^ *
rH CO CO C\
O
rH CN ^
CN ^ rH -*
tn Cn
5) i
X >t *§
i m
O m
rl > I
rl -H S
0 i)
O
D D *
10 to
O 0 -
« * S
« t.""1
-"'i
c § ~
V *
rl
O c
rH "
UH U
8 tr ffl
(D ±J
C -H B
O 10
tn C to
n Q
£1 0]
M O -O
D1 C O
S3*
10
- c e
0) O
rH tf}
^ *O -H
13 1)
rH -P O
rl C U
*0 V C
> n TO
o o
4J M -rl
C O, tH
flj 0) -H
1st
T3 >4-l
coo
""' 'i> rH
X 0) O
0 -rl >
10 4J (1J
O 10 rH
n in 01
£ ti S
M (N
5
o
42
-------
precipitation, temperature and the days-of-week variables, in addition to
the NO variable (regression coefficient « 187.0, standard error « 134.4
and F value * 1.94 < F (.10, 1, 120) » 2.75). The F ratio was 3.02 > F
(.01, 6, 120) - 2.96.
THE VARIATION DUE TO THE POLLUTANT, THE DAYS-OF-WEEK EFFECTS AND
WEATHER CHANGES
The variation in each disease group explained by the regression model
was divided into three portions; (1) due to the pollutant (R?) , (2) due
to the days-of-week variables (R§), (3) due to the climatological factors
(Rj). Fourteen models were selected from Tables 9-13 according to F
values or standard errors of the pollutants which were considered to be
a significant or meaningful contribution to the regression model. Table
14 shows percentages of the variation due to the pollutant, the days-of-
week variables and the climatological percentage factors, R (sum of Rj,
R| and R|) and (1-R2) which is the percentage of variance unexplained or
attributable to the residuals. The following path diagram explains the
possible relationship between cardiac and respiratory diseases and three
factors:
weather changes
air pollution I
days of the week
cardiac and
respiratory
diseases
RI (the variation due to the pollutant) included the variation shared by
the dependent variable (the disease) and the pollutant without partition-
ing out the variation due to the days-of-week and weather variables, be-
cause weather changes have direct influences on air pollution levels
which are also affected by human activities (i.e., industrial operations,
traffic conditions, etc.) according to the days of the week.
In the column of R? (the variation due to the pollutant), model 2
(based on the average exposure levels of S0_) showed the highest per cent
of variation in the disease groups, TRD (13.28%), TCD (21.72%), and GDI
43
-------
TABLE 14
THE VARIATION DUE TO THE POLLUTANT, DAYS-OF-WEEK ErrECTS
AND WEATHER CHANGES IN THE MODELS SELECTED FROM TABLES 9-13
Due to
Disease Model Pollutant
TRD
RD1
RD2
TCD
CDl
(l)TSP-II
(2)S00-I
(6) CO-III
(7)NO-III
(2)S02-I
(3)S02-III
(7)NO-III
(2)S02-I
(2)S02-I
(5)N02-III
(7)NO-III
(2)S02-I
(5)N02-III
(7)NO-III
0.29
13.28
3.35
7.60
9.62
11.69
6.50
13.45
21.72
0.62
4.36
21.48
0.12
3.84
Total Variation (%)
Due to days-of- Due to weather
week effects changes R 1-R
(# <*23>
1.14
8.14
0.57
0.73
8.05
0.30
0.55
0.77
12.82
2.37
2.01
14.44
1.19
1.05
25.58
14.20
28.89
24.43
22.21
21.31
27.43
24.50
13.55
10.94
6.92
10.48
11.45
7.65
27.01
35.62
32.81
32.76
39.88
33.30
34.48
38.72
48.09
13.93
13.29
46.40
12.76
12.54
72.99
64.38
67.19
67.24
60.12
66.70
65.52
61.28
51.91
86.07
86.71
53.60
87.24
87.46
44
-------
(21.48%). In RDl, the variation due to SO from the Universitv of Illinois
Medical Center in model 3, 11.69%, was greater than that based on the
average exposure levels in model 2, 9.62%. In RD2, only model 2 was chosen
and SO_ explained 13.45% of the variance. TSP accounted for only 0.29%
of the variation in TRD. CO showed 3.35% of the variation in TRD. The
variation due to NO was 7.60% in TRD, 6.50% in RDl, 4.36% in TCD and 3.84%
in GDI. The variation due to NO was very small, 0.62 in TCD and 0.12%
in GDI.
The variation due to the days-of-week variables (R|) was small in all
models except in model 2 which might be due to its small sample size. The
9 7
variation due to weather changes (Rj) ranged from 6.92% to 28.89%. RS was
greater than RI in all models except model 2 for TCD and for CD1.
45
-------
7 7
In the previous study by Namekata and Carnow , the regression model for
all respiratory conditions were shown as follows:
% of excess visits = 90.24 + 557,42 SC>2 - 0.389 Max temp. + 1.797 Windspeed
(326.69) (0.188) (0.945)
2.91 4.27 3.62
where % of excess visits was the same definition as the one in the present
study. SO was the average patient exposure level (ppm), and max temp, and
wind speed were daily maximum temperature (°F) and daily average wind speed
(miles per hour), respectively. Figures under each coefficient indicate a
standard error in parentheses and an F value with an underline. The days-of-
the-week variables were not included because only Tuesday's data were used.
Thirty-six per cent of the variation of this disease category was explained
by the above model, while model 2(SO~-I) for total respiratory diagnoses in
2
Table 9 had also a R of 0.36. SO in the above model was significant at the
.10 level while SO in model 2 in Table 9 approached significance. Although
the previous study used data from 14 hospitals in Chicago, the strength of
the relationship between SO2 and respiratory disease was similar to the one
in the present study. Despite the limited data base this similarity implies
that sulfur dioxide might have acute effects on respiratory disease at fairly
low levels which are usually found in the Chicago area.
27
Another important finding in the previous study was the significant
relationship between total suspended particulate (the daily average of patient
exposure levels) and bronchitis, as shown in the following model:
% of excess visits=-5.51+0.404TSP-1. 234 min. temp.+5. 490 wind speed+0.,345 Humidity
(0.203) (0.273) (1.585) (0.429)
3.95 20.40 12.00 3.875
where % of excess visits for bronchitis was the same as the one defined before,
TSP was the average patient exposure level (ug/m ), and min.temp., wind speed
and humidity were daily measurements of minimum temperature ( F), wind speed
(miles per hour) and humidity (%), respectively. Figures of each independent
variable were defined in the model for all respiratory conditions. The
coefficient of TSP w^s significant at the .1.0 level and close to the .05 level
(3.95 > F(.10,l,40)=2.84, but 3.95 < F(.05,1,40)=4.08). Forty-seven per cent
46
-------
of the variation of bronchitis patient visits was explained by this model.
A comparison was not possible between this model and model 1 (TSP-II) in
Table 11 - models for acute bronchial and lower respiratory infections (RD2)
in the present study, because the disease group for RD2 consisted of nine
respiratory diagnoses; asthma, bronchial astham, bronchitis, acute bronchitis,
tracheobronchitis, pleurisy, pleuritis, broncheolitis and pneumonia. The
insignificant association between TSP and RD2 in the present study might be
caused by this disease classification, since the regression model for asthma
in the previous study did not show a significant association with either SO
or TSP, while the model for bronchitis had a significant coefficient of TSP
as described above.
47
-------
REFERENCES
1. Firket, M.: "Sur Les causes des accidents survenus dans la vallee de la
Meuse, lors des brouillards de decembre, 1930." Bulletin de L'Academie
Royale De Medecine De Belgigue, Vol. XI, pp. 683-741.
2. Schrenk, H.H., Heinmann, H., Clayton, G.D., et al: "Air Pollution in
Donora, Pennsylvania. Epidemiology of the unusual smog episode of
October, 1948," Preliminary Report, Public Health Bulletin 306, 1949.
3. Morbidity and Mortality During the London Fog of December, 1952, Reports
of Public Health Medical Service No. 95, London: Her Majesty's
Stationary Office, 1954.
4. Greenburg, Leonard; Jacobs, Morris B.; Drolette, Bernadette M.; Field,
Franklyn and Braverman, M.M.: Report of an Air Pollution Incident in
New York City, November, 1953." Public Health Reports Vol. 77, pp. 7-
16, January, 1962.
5. Huber, T.; and others: "New Environmental Respiratory Disease (Yoko-
hama Asthma)", AMA Arch. Industr. Hy. 10: 399-408, November, 1954.
6. Phelps, Harvey W.; Sobel, Gerald W.; and Fisher, Neal E.: "Air Pollution
Asthma Among Military Personnel in Japan." J.A.M.A. pp. 142-145,
March 18, 1961.
7. Phelps, Harvey W.; and Koike, Shigeo: "Tokyo-Yokohama Asthma, (The Rapid
Development of Respiratory Distress Presumably Due to Air Pollution)."
American Review of Respiratory Diseases,Vol. 86, No. 1, pp. 55-63, July,
1962.
8. Ladd, Barry; Phelps, Harvey: "Incidence of Air Pollution Bronchitis in
Military Personnel in Japan." Diseases of the Chest, Vol. 43, No. 2,
pp. 151-154, February, 1963.
9. Phelps, Harvey W.: Follow-Up Studies in Tokyo-Yokohama Respiratory
Disease." Arch. Environ. Health, Vol. 10, pp. 143-147, February, 1965.
10. Meyer, George W.: "Environmental Respiratory Disease (Tokyo-Yokohama
Asthma): (The Case For Allergy)." Paper presented at Air Pollution
Medical Research Conference of the American Medical Association, San
Francisco, December 6, 1974.
11, Weill, Hans; Ziskind, Morton M.; Derbes, V.J.; Horton, Robert J.; McCaldin
Roy 0., and Dicerson, Richard C.: "Recent Developments in New Orleans
Asthma." Arch. Environ. Health, Vol. 10, pp. 146-151, February, 1965.
48
-------
12. Yoshida, Katsumi; Oshima, Hidehiko; Imai, Masayuki: "Air Pollution and
Asthma in Yokkaichi," Arch. Environ. Health, Vol. 13, pp. 763-768,
Dec. 1966.
13. Huddle, Norie; Reich, Michael and Stiskin, Nahum: Chapter 2. Yokkaichi
Asthma, Island of Dreams-Environmental Crisis in Japan, Autumn Press,
N.Y., 1975.
14. Sprague, Homer A.; Hagstrom, Ruth: "The Nashville Air Pollution Study:
Mortality Multiple Regression." Arch. Environ. Health, Vol. 18, pp.
503-507, Apr. 1969.
15. Lave, Lester B.; and Seskin, Eugene P.: "Air Pollution and Human Health."
Science, Vol. 169, No. 3947, pp. 723-732, August 21, 1970.
16. Lave, Lester B. and Seskin, Eugene P.: An Analysis of the Association
Between U.S. Mortality and Air Pollution, Journal of Am. Stat. Ass.,
Vol. 68, No. 6, June 1973, pp. 284-290.
17. Hodgson, Thomas A.: "Short-Term Effects of Air Pollution on Mortality
in New York City." Environmental Science and Technology, Vol. 4, No. 7,
pp. 589-597, July, 1970.
18. Buechley, Robert W.; Riggan, Wilson B. et al.: "SO Levels and Pertur-
bations in Mortality - A Study in the New York - New Jersey Metropolis,"
Arch. Environ. Health, Vol. 27, pp. 134-137, Sept. 1973.
19. Schimmel, Herbert; Murawaki, Thaddens J.: "The Relation of Air Pollution
to Mortality," Journal of Occupational Medicine, Vol. 18, No. 5, pp.
316-333, May 1976.
20. Lave, Lester B.: Seskin, Eugene P.: Air Pollution and Human Health,
The Johns Hopkins University Press, Baltimore, 1977.
21. Thompson, Donovan J.: Lebovitz, Michael; et al.: "Health and the Urban
Environment, VIII. Air Pollution, Weather, and the Common Cold,"
American Journal of Public Health, Vol. 60, No. 4, pp. 731-734, April,
1970.
22. Carnow, Bertram w., Lepper, Mark H., et al: Chicago Air Pollution Study-
S0? Levels and Acute Illness in Patients with Chronic Bronchopulmonary
Disease, Arch. Environ. Health, Vol. 18, May, 1969, pp. 768-776.
23. Lepper, Mark H., Shioura, N., Carnow, Bertram W., et al: Respiratory
Disease in an Urban Environment, Ind. Med., Vol. 38, No. 4, April,
1969, pp. 126-131.
24. Carnow, Bertram W., Feiveson, Sandra: Morbidity and Mortality During the
Chicago, 1969 Air Pollution Episode, unpublished paper.
49
-------
25. Carnow, Bertram W., Carnow, Virginia: Air Pollution, Morbidity, and
Concept of No Threshold, Advances in Environment Science and Tech-
nology Vol. 3, pp. 127-156, edited by James N. Pitts, Jr. et al, John
Wiley & Sons, Inc., 1974.
26. Namekata, Tsukasa; Carnow, Bertram W.: Impact of Multiple Pollutants on
Emergency Room Admissions, IIEQ Document No. 77/02, 1976.
27. Nam«kata, Tsukasa; Carnow, Bertram W.: Models to Estimate Air Pollution
Effects on Cardiac and Respiratory Diseases in Chicago, To be published
in the near future.
50
-------
APPENDIX
51
-------
TABLE 15 CORRELATION COEFFICIENTS BETWEEN EMriRGE'TCY ROO:-1! VISITS
FOR CARDIAC AMD RESPIRATORY CONDITIONS *rr> rNV
TVIT^ IN H7\SE I ANALYSIS
(1)
(2)
(3)
(4)
(3)
Allerqic ^cute Hypertension
Total conditions bronchial Total and
resniratory and urjoer and lower Cardiac vascular
diannoses resoiratorv resniratorv Diagnoses 'teart
infections infections Disease
TSP-CN
SO -CN
NO -CM
SO -Polk St.
MO., -Polk St.
CO- polk St.
JO- Polk St.
Temperature
Wind sneed
Humidity
Precipitation
Skv cover
Sunshine
-0.028
0.364
-0.172
-0.035
0.011
-0.146
0.093
-0.192
0.1^6
0.143
-0.348
0.1^9
-0.193
-0.121
C.31'i
-0 . 269
-0.053
-0.142
-0.16"
-0.079
-0.210
0.306
0.152
-0.426
-0.020
0.122
0.045
0.367
-0.005
0.124
0.114
-0.094
0.248
0.072
-0.148
-0.101
0.024
0.103
-o.oqo
-0.011
0.46C
-0.228
0.242
0.056
-0.134
0.273
-0.212
-0.126
0.051
-0 . 30 1
0.151
-0.133
-0.074
0.463
-0.248
0.233
0.027
-0.155
0.278
-0.161
-0.137
0.110
-0 . 30 3
0.242
-0.190
Samnle size M=20
Critical values of the correlation coefficient; 0.444 (ci=.05),
0.561 (nt=.01)
CN after the pollutants indicates the measurements from the city
air sampling network which were used to estimate the patient exposure
levels.
Polk St. after the pollutants indicates the measurements from the
Illinois E.P.A. sampling station on Polk Street at the University
of Illinois Medical Center.
-------
TABLE 16 CORRELATION COEFFICIENTS BETWEEN EMERGENCY ROOM
'rrgiTS FOR CARDIAC AND RE.^IPATORY CONDITIONS AMD
ENVIRONMENTAL MEASUREMENTS IN "HA^E II .ANALYSIS
TSP-CN
SO -Polk St.
NO -Polk St.
CO- Polk St.
NO- Polk St.
Temoerature
Wind soeed
Humidity
Precipitation
Sky cover
Sunshine
(1)
Total
Respiratory
Diaqnoses
0.054
0.210
-0.041
0.110
0.209
-0.407
-0.016
0.038
-0.323
0.052
-0.131
(2)
Allerqi c
Conditions
and utjoer
respiratory
infections
-0.010
0.283
-0.144
0.084
0.190
-0.448
-0.006
0.119
-0.345
0.019
-0.108
(3)
Acute
bronchial
and lower
resniratory
infections
0.070
0.011
0.072
0.053
0.119
0.001
-0.111
-0.266
-0.069
0.062
-0.061
(4)
Total
cardiac
diagnoses
-0.011
0.242
0.056
-0.134
0.273
-0.212
0.126
0.051
-0.301
0.151
-0.133
(5)
Hypertension
and
vascular
heart
disease
0.014
0.256
-0.023
-0.120
0.213
-0.117
-0.010
0.146
-0.216
0.214
-0.204
Note: Sample size N=42
Critical values of the correlation coefficient; ,304 (o=.05),
.393 (ot=.01)
CN after the pollutants indicates the measurements from the city
air sampling network which were used to estimate the patient exposure
levels.
Polk St. after the pollutants indicates the measurements from the
Illinois E.P.A. sampling station on Polk Street at the University
of Illinois Medical Center.
53
-------
TABLE 17 CORRELATION COEFFICIENTS BETWEEN EMERGENCY ROOM
VISITS FOR CARDIAC AND RESPIRATORY CONDITIONS AND
ENVIRONMENTAL MEASUREMENTS IN PHASE III ANALYSIS
r - - - _ .
so2~ polk st.
NO - Polk St.
CO- Polk St.
NO- Polk St.
Temperature
Wind sneed
Humiditv
Precipitation
Skv cover
Sunshine
(1)
Total
respiratory
diagnoses
0.294
-0.128
0 . 1 83
0.276
-0.552
-0.050
0.170
-0.155
0.180
-0.255
(2)
Allergic
conditions
and upper
respiratory
infections
0.342
-0.183
0.129
0.255
-0.549
0.030
0.145
-0.176
0.157
-0.217
(3)
Acute
bronchial
and lower
respiratory
infections
-0.055
-0.028
0.065
0.094
-0.049
-0.163
-0.039
-0.009
0.058
-0.086
(4)
(5)
Hypertension
Total and
cardiac vascular
diagnoses heart
disease
0.123
0.078
0.040
0.209
-0.185
-0.095
0.17G
-0.073
0.253
-0.259
0.117
0.035
-0.028
0.196
-0.168
-0.103
0.240
-0.037
0.278
-0.2^8
Note: Sample size N=131
Critical values of the correlation coefficient in
.195 (n=.05), .254 (ct=.01)
= 100;
CN after the pollutants indicates the measurements from the city
air sampling network which were used to estimate the patient exposure
levels.
Polk St. after the pollutants indicates the measurements from the
Illinois E.P.A. sampling station on Polk Street at the University
of Illinois Medical Center.
54
-------
18 CORRELATION COEFFICIENTS BETWEEN EMERGENCY ROOM
VISITS FOR CARDIAC AMD RESPIRATORY CONDITIONS AND
ENVIRONMENTAL MEASUREMENTS IN PHASE IV ANALYSIS
(1)
(2)
(3)
(4)
(5)
Allergic Acute Hypertension
Total conditions bronchial Total and
respiratory and upper and lower cardiac vascular
diagnoses respiratory respiratory diagnoses heart
infections infections disease
O3-Polk St.
Temperature
Wind speed
Humidity
Precipitation
Sky cover
Sunshine
-0.147
-0.131
-0.253
0.290
-0.016
0.261
-0.080
-0.116
-0.069
-0.089
0.213
-0.091
0.172
0.019
-0.215
-0.238
-0.407
0.214
0.121
0.301
-0.233
-0.247
-0.163
-0.202
0.104
0.001
0.180
-0.025
-0.299
-0.264
-0.136
0.132
0.024
0.162
-0.057
Note: Sample size N=36
Critical values of the correlation coefficient; .330 (ot=.05),
.424 (0=.01)
CN after the pollutants indicates the measurements fron the city
air sampling network which were used to estimate the patient exposure
levels.
Polk St. after the pollutants indicates the measurements from the
Illinois E.P.A. sampling station on Polk Street at the University
of Illinois Medical Center.
55
-------
u:
M
<
5
^ i
M
U
§
C-
z
H
1
~
a
^
EH
1
5
g
M
>
U
t.
X
rH
r*
!^
jr
§
B
a
0
u
m
rH
a
§
EH
O
1 C
C .H
(7 in
M
1 (I)
> >
£ 8
r)
D C
H 0
U -H
>H *0
t JJ
1
Jj >
S 4J
~ -o
a
o o
.5 S
"
.
x
r-i
O
P4
'
r
[ rH
P.
1
^
0
fr
1
c"
"
rH
0
p4
1
C
U
i
c
5
i
0
IV!
D
c.
S
t^
r~ 3\
vC in
vC CM
C 0
1
o. «r
C in
VC rH
C C
1
rH in
rH rH
O C
1 '
in in
CM CM
P- rH
O 0
1
in in
cc r-
ro o
C? O
vS 1-
CM rH
^ *r
O 0
i
en ro
p~ en
rH 0
c- c
8 S
in rH
c c
o m
VJ? rH
ro c
o o
1
p- r-
0 T
c o
1
vD ro
CO rH
in o
C 0
I
V0 C
t- o
c c
C rH
C VD
C P-
o o
rH C
o 5
1 1
C (N
tr 0
EJ u:
i
OJ *T
m t-i
0 0
1
& 1*
0^ ro
CN C
0 C
1
rH CM
O ro
C rH
C C
1 1
CC CM
C in
in rH
0 0
i i
VTl rH
P- T
vo *r
o c
f vD
C v£
C O
(T CN
CM CO
^f rH
C C
CM P"
»H VD
in *?
0 0
01 lO
r- o
r-t CM
0 C
1
P- C
CM C
^J* C
O rH
O P-
C CM
O f
rH C
rH P*
C f
o o
O P"
ro ^*
in c
c c
1
5 f
fM (N
c c
C* CM
rH CO
c o
ro in
rH ro
0 0
1 1
0 rH
C ro
0 C
1 1
rH P*
0 0
1 1
vD ro
CO O
ro ^3*
C 0
1 t
C CM
TJ* X
ro o
c c
1
rH rH
*! 01
in rH
O C-
CM O
ro O
O O
0 rH
1
O CM
C ro
O O
rH 0
1
V£3 P*
C* VD
n f
c o
i
P- rH
rH in
0 C
in ro
rH VO
C rH
C O
1
0 C
0 CO
ro LO
O O
C
r Q
CM 1
c c
r- c
vc O
c m
O 0
CM CO
C CO
rH ro
0 0
1
CM C
co r~
rH rH
0 0
1
ro CN
CM ro
rH *J/
0 0
1
^ O
rH CN
VO rH
C 0
1
CO 0
*r o
rH O
O rH
1
O CO
O T
0 rH
H O
1
rH CM
en co
rH O
0 0
I
rH O
in ro
o o
CM VC
CO *T
rH VC
0 0
1
CM ro
^r o
o o
1
ro t**
C>' rH
O T
0 0
1
1
vC
CO
ro
C
1
rH
^*
**.
O
1
C"
P-
vo
o
1
m
c
c
in
CO
ro
C
1
0
TJ 01
C Q)
H C
? in
vD
P-
O
1
O
CO
P-
o
ro
rH
0
o
o
o
rH
cn
01
ro
d
CM
ro
c
o
1
ro
CM
rH
O
S
ro
0
1
TT
rH
CN
C
1
CN
in
rH
O
1
CO
in
0
i
in
CM
rH
0
in
CM
P-
C
1
1
3 ^
S~. 'O
CO
CO
CM
0
1
p.
f^
CM
0
c
c
c
rH
rH
o
rH
CO
rH
O
o
rH
O
CN
CO
rH
d
1
VO
rH
ro
C
1
t^
VC
o
d
i
CN
CO
rH
C
1
rH
0
o
o
1
in
rH
O
1
rH
r»
rH
O
1
1
r)
D C
H 0
O -H
P ra
C- JJ
ro
tn
o
i
o
0
o
rH
P-
CM
O
O
CO
P-
c
^*
CN
CM
C
CO
CO
ro
i
CM
C
rH
d
m
CO
o
1
ro
n
rH
O
1
rH
CO
C
O
g
CN
C
1
in
rH
C
ro
c
VC
c
i
V,
o
±'. O
V U
o
o
o
rH
vC
ro
O
d
i
m
CC
CM
o"
1
S
c
1
in
0
ro
O
1
O
c
in
d
rr
vo
c
o
vo
CM
ro
d
ro
01
rH
C
TT
H
W -H
nj O
rH -H
O JJ
1 0
V
4J
O
56
-------
tt
t-l
i
i
H
t-l
£
g
I
S
I
i
i
i
i
B
K
B
i
I
*
i
5
3
a
8
i£
«n n
h
^t
SB
§ J
12
9 ^4
a B
si
X
1
t
&
A
i
£
A
8
.K
4
£
£
i
E
«
o
CD
rt
?
m
t-l
O
*
a
rt
?
in
?
1
o
s
fi m
5 s
0 «p
Ch CD
n tn
- o
o o
1
g I
? ?
Pt w
n r-
n «-t
0 ?
n to
n «r
o o
o o
n o
O r*
a o
Sft
n
*4 o
i i
? °
^ tfi
0 0
1 §
to en
3 n
H rt
O O
1
? S
O O
o ej>
5 §
o n
o o
1
S JS
H >H
O O
n 10
s g
? ?
in o
S g
? "
§ s
o rt
H o
'
tn t-
v -<
o o
t
fn rt
0 O
in CD
0 0
m *
0 0
II
=
o
'
P;
O
0
o
m
o
o
o
rt
o
g
o
**
1
?
3
«P
0
i
*
o
4
?
4
in
7
in
s
o
i
i!
tH
<0
n
OD
n
O
n
0
g
o
**
o
o
5
o
1
3
o
OD
in
o
o
B
o
3
4
O
rt
O
t
1
t-t
ft
n
o
'
5
m
O
|
4
ff
r«
O
O
m
O
a
s
0
5
o
Q
'
n
o
Q
a
o
in
o
O
I
o
o
1
Procipi
tation
i
o
T
g
o
"
w
n
o
03
rt
«B
O
C3
O
M
O
i
" s
o
o
to
ft
Q
rt
n
e
?
3
CJ
CD
N
rt
0
i!
!
ri
S
09
^
H
n
?
tn
r-l
S
3
s
«
I
i
£
J? B
* 1
8 S
«4
a «-4
5 ^
g. *»
tn o
1
57
-------
H
in
is
§
M
tH
IH
EC
a
E
H
e
§
s
D
(^
g
,J
£
K
jr
S
^_(
2
H
fe
X
s
EH
g
g
£*
5
g
oi
8
CM
s
EM
D
C
H
£
M
§
t/1
1
u
.*
en
i
H
c c
H O
£4 irj
CU 4J
4J
H
o
H
§
=
0
C 0
c o
H C-
E: to
1
a
£-1
,_-)
£
I
X
rH
o
i
CJ
H
£
i
(N
C
rH
O
CM
i
(N
in
in
r-
rH
O
1
(N
rH
C
1
rH
CO
C
o
1
rH
rH
rH
0
r-
r-
n
C
1
c*
vo
c
1
5
vo
o
CO
c
c
o
c
1
c
c
o
rH
c
r
i
(N
C
IT.
(M
CO
rH
d
O
c
1
vO
0
c
m
c
o
P)
n
o
i
t--
m
c
5
m
C
rH
0
O
0
o
rH
VO
C
C
c
1
i
(N
O
03 CO
rH O
0 rH
O O
1 1
ro c
^ rH
C rH
O O
1
r- m
rH r*
rH O
O O
1 '
H n
O iH
c o
CO VD
0-
E -u
7 -H
s: TJ
rH
VO
O
1
0^
01
(N
C
o
0
o
rH
1-
OJ
d
(N
in
o
d
o
r-
rH
O
P
O
o
1
I-
rH
H
O
1
*
0
0
tH
CO
C
d
i
i
rH
.2-8
O -H
HI 4J
£3
rH
00
CO
O
1
O
0
o
rH
d
01
d
in
c
in
d
vO
O
d
i
0^
rH
0
1
s
i-H
C
n
VD
0
d
i
§
0
o
1
r-
H
0! rH
IT]
O U
H 'H
E -H
Id rH
V. 0
..
IV
0
58
-------
2
£^
Z
1
D CO
p^ 1 1
H in
Z S
fo rt
O
^
X M
BK
a? K
§*
H H
EH
S8
K ^i
P£ Ex3
OH «£?
8£
D
CM 8
H
f.
U)
C
p
IT.
0)
1
^.
o
tr
C
0
H
4J
(0
+J
H
H
0
s
IX
£
iH
o
1
tt
o
0)
0)
&
o
c
rl
g:
0)
Vf
3
JS
w
VO O
ro
U 0)
a o>
o
rH
0 «
N >
H
M rH
18
01 U
H -H
O< 4J
E *H
* cJ
*
0)
K
59
-------
co
r^
en
H
I
en
C
O
H
fi
13
5
M-l
O
0)
o
G
-S
-p
3
XI
H
M
P
tn
rH
Q
fH
dP
CO U
> G
H 0)
fd ftS o"*
4-) rH CO
p CD n
EH « fa
0) 0
-P C
3 CU
H rH 3
rtj O &1
4J W 0)
83£
dP
0) O
> G
H Q)
Jj r^
rd 51
W rH c
H (I)
K rH CO
U CO M
O BJ fa
Q) U
-P G
o 5*
K in Q)
O XI rH
o <2 fa
CO
CJ
£
co en o n
4
O <£> CN O
rH CO
^D O -^* CN
01 ro O U)
CN r^
rH
n en rH r^
t
r- co n o
CN VD
in r^- <3* CN
r*" rH "vj1 CO
CN CN H
rH rn
CTl -"31 r* CN
t
U1 CN rH O
CTi
H m o o
CTI ^r CN
^
rH
U
CD A; vi co
4J O <1) Cli
H ra x! tn
x: rH 4J c
o
o
0
rH
CN
en
CN
0
CN
o
o
o
rH
co
VO
vo
^}t
o
t
0
o
rH
^*
CN
l-O
M
CO
en
rH
Sf
I
r--
cn
rH
M
Oi
fi
3
c
o
H
+J
-P
U3
rH
Q
co
rH
XI
Q) O
> a
rH 0)
rH 4-> 3
tl rtj CT
P rH (U
p cu s-i
EH Pi fa
Q) O
-P G
3 0)
rH rH 3
rd O u1
jj in cu
0 -9 *H
EH ^ fa
0) O
> C
H (D
-P P
rcj &
ffi rH 0)
H CO S-i
D cr; fa
a) o
P G
3 O
rH 3
tc in cu
rH XI Vl
D 33 fa
dP
0) O
> c
H Q)
4J J3
W H CO
cj co s-i
u rt fa
0) U
4J G
3 CO
rH 2
O CJ1
W «) d)
C_) n j^|
O ^ fa
X
0)
CO
r-- H CN
ft
^o n o
3< LO
oo r^ r^
^ f^ ^
en o
rH
n ^< n
^D OO O
^ in
n H ^"
V£> CTl rH
CN CN
CO O CN
*
a> m o
sj" IT)
l_O V^ (-T)
rH 00 CN
m CN
r- co
0) U
rH d)
cu cd pj
rH g U)
S fa D
0
O
o
rH
CN
en
CN
0
CN
O
O
O
rH
00
IO
-------
Pi £°
< U,
> o
I! 3.
oor-ot-r-<*i*o-r
^o r*- PJ in pi f\j INI
tNJ . CT^ rr .M O O
oinr-mcoo ^
o cr
I » o
« jj t.
U o
U (X!
o
o
Z G
I 2
u 2
u <;
«o
-------
Table 26 . Frequency of all Diagnostic Illnesses
of Study Population April, 1977 - April, 1978
Diagnostic Absolute Relative
Group Frequency Frequency (%)
Respiratory 13985 68.9
Cardiac 3294 16.2
Omissions 3013 14.9
202.92 100.0
62
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/1-79-024
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Model for measuring the health impact from changing
levels of ambient air pollution: Morbidity study
5. REPORT DATE
_Aiigusfc_1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Tsukasa Namekata, Bertram W. Carnow, Zanet-Flourney-Gill
Eileen O'Farrell, and Domenic J. Reda
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Dept. of Occupational and Environmental Medicine
School of Public Health
University of Illinois
Chicago, IL 60680
10. PROGRAM ELEMENT NO.
1AA816
11. CONTRACT/GRANT NO.
68-02-2492
12. SPONSORING AGENCY NAME AND ADDRESS
Health Effects Research Lab
Office of Research & Development
U.S. EPA
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/11
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This study quantitatively examines the relationship between human health and
ambient air concentrations of the major pollutants in the city of Chicago. This
report describes the morbidity analysis in which linear regression models have been
developed to quantitatively estimate the degree of the air pollution contribution to
emergency room visits for cardiac and respiratory diseases in two major hospitals in
the city of Chicago.
Based on the significant associations between the pollutants and the disease
groups, holding climatological and days-of-week variables constant, tne variation due
to the pollutant is estimated. According to the results, sulfur dioxide based on
patient exposure levels can account for about 13% of the variation of emergency room
visits for acute bronchial and lower respiratory infections and about 22% for total
cardiac diagnoses. Nitric oxide based on measurements from the closest site to the
hospitals can account for about 7% of the variation of visits for total respiratory
diagnoses, 6% for allergic conditions and upper respiratory infections, 4% for total
cardiac diagnoses and 4% for hypertension and vascular heart diseases. Total suspended
particulate, carbon monoxide and ozone do not show significant associations with any
disease groups.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATl Field/Group
Air Pollution, Cardiac/Respiratory Diseases
Health Effects, Regression Analysis
06F
18. DISTRIBUTION STATEMENT
PUBLIC
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
73
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. 4-77)
PREVIOUS EDITION IS OBSOLETE
63
------- |