EPA-600/5-77-006
February 1977
Socioeconomic Environmental Studies Series
HEALTH COSTS OF AIR POLLUTION DAMAGES
A Study of Hospitalization Costs
Health Effects Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development. US. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2 Environmental Protection Technology
3. Ecological Research
4 Environmental Monitoring
5 Socioeconomic Environmental Studies
This report has been assigned to the SOCIOECONOMIC ENVIRONMENTAL
STUDIES series. This series includes research on environmental management,
economic analysis, ecological impacts, comprehensive planning and forecast-
ing, and analysis methodologies. Included are tools for determining varying
impacts of alternative policies; analyses of environmental planning techniques at
the regional, state, and local levels; and approaches to measuring environmental
quality perceptions, as well as analysis of ecological and economic impacts of
environmental protection measures. Such topics as urban form, industrial mix,
growth policies, control, and organizational structure are discussed in terms of
optimal environmental performance, These interdisciplinary studies and systems
analyses are presented in forms varying from quantitative relational analyses to
management and policy-oriented reports.
This document is available to the public through the National Technical Informa-
tion Service. Springfield. Virginia 22161
-------
EPA-600/5-77-006
February 1977
HEALTH COSTS OF AIR POLLUTION DAMAGES
A Study of Hospitalization Costs
by
Ben H. Carpenter, D.A. LeSourd,
James R. Chromy, and Walter D. Bach
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, N.C. 27709
Contract No. 68-01-0427
Project Officer
Donald G. Gillette
Criteria and Special Studies Office
U.S. Environmental Protection Agency
Health Effects Research Laboratory
Research Triangle Park, N.C. 27711
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
HEALTH EFFECTS RESEARCH LABORATORY
RESEARCH TRIANGLE PARK, N.C. 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.
-------
FOREWORD
The many benefits of our modern, developing, industrial society are
accompanied by certain hazards. Careful assessment of the relative risk
of existing and new man-made environmental hazards is necessary for the
establishment of sound regulatory policy. These regulations serve to
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,
epidemiology, and clinical studies using human volunteer subjects. These
studies address problems in air pollution, non-ionizing radiation,
environmental carcinogenesis and the toxicology of pesticides as well as
other chemical pollutants. The Laboratory develops and revises air quality
criteria documents on pollutants for which national ambient air quality
standards exist or are proposed, provides the data for registration of new
pesticides or proposed suspension of those already in use, conducts research
on hazardous and toxic materials, and is preparing the health basis for
non-ionizing radiation standards. Direct support to the regulatory function
of the Agency is provided in the form of expert testimony and 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 results of this study show that the hospitalization incidence
rate and cost of certain pollutant related diseases were significantly
greater among populations residing in the more polluted areas of Pittsburgh.
Although the hospitalization costs associated with air pollutants in
Pittsburgh were nearly 10 million dollars in 1972, the total health costs
resulting from air pollution exposure in the Pittsburgh area would be much
greater when non-hospitalization costs are also included.
John H. Knelson, M.D.
Director,
Health Effects Research Laboratory
TM
-------
ABSTRACT
An investigation of the hospitalization costs of exposure to air
pollution in Allegheny County, Pennsylvania was conducted to determine
whether persons exposed to air pollution incurred higher incidences
of hospitalization or additional costs for treatment. A hospitalization
data-base comprising 37,818 total admissions for respiratory, suspect
circulatory diseases, and control diseases was tested in a cross-section
type analysis for relationships between rates of hospitalization, length
of stay, and levels of air quality in the neighborhoods of patients'
residence. Air quality was identified using data from 49 monitoring
stations. Corrections were made in the analysis for race, age, sex,
smoking habits, neighborhood median income, and type of occupation.
Respiratory and suspect circulatory system disease showed statistically
significant increased hospitalization rates and lengths of stay for those
exposed to higher levels of SOo and particulates compared to those from
neighborhoods meeting air quality standards. At average costs per day for
hospitalization in this area in 1972, the total increased costs of hospital-
ization for the 1.6 million persons in the County was estimated at $9.8
million dollars ($9.1 million for increased hospitalization rates and $0.7
million for increased length of stay).
IV
-------
TABLE OF CONTENTS
Abstract jv
List of Figures vii
List of Tables viii
Acknowledgments ix
1 Introduction 1
2 Conclusions 2
3 Recommendations 3
4 Experimental Procedures 4
Technical Approach 4
The Study Area 5
Air Quality Data Base 5
Hospitalization Data Base 6
Population Data Base 7
Sample Selection 9
Sample Design 11
Other Possible Analyses 15
Drawing Conclusions from the Proposed Analyses 16
Drawing the Sample 16
5 Objective Analysis of Air Quality by Census Tract In
Allegheny County Pennsylvania 19
Methodology 20
Analysis Approach 23
6 Effect of Air Pollution on Hospitalization Incidence Rates . 27
Results of Regession Analysis 28
Analysis Based on Sample Design 30
Estimation Procedures 31
Potential Biases 36
Effects on Hospital Use and Costs 38
-------
TABLE OF CONTENTS (cont'd)
Paqe
7 Effect of Air Pollution on Length of Stay in the Hospital . 40
Analysis of the Data 40
Results 41
Estimated Costs 43
8 Total Hospitalization Costs of Air Pollution 45
9 References 46
Appendix 48
-------
LIST OF FIGURES
Number Page
1 Cressman weight function, y, dependence upon relative
distance, r/R 21
2 SOo levels (1972 yearly average), Allegheny County, Pa.,
yg/m3 25
3 Particulates levels (1972 yearly average), Allegheny County,
Pa., ug/m3 25
-------
LIST OF TABLES
Number
Page
1972 annual average concentrations of S02 and particulates
at five selected census tracts in Allegheny County, Pa . . .
I
II Total 1972 hospitalizations in Allegheny County hospitals
for diseases under study 7
III 1970 population data base (in thousands of persons) .... 8
IV Final data base, hospitalizations of Allegheny County Resi-
dents in 1972, by hospital-set and sampling stratum .... 10
V Sample size requirements for specified hypotheses 14
VI Proportionate stratified random sample of cases of hospital-
ization for circulatory system and control diseases .... 18
VII Scan radium (km) by iteration and pollutant 23
VIII Effects of exposure to air pollution, and other factors, on
incidence of hospitalization 29
IX Classification of Allegheny County Population according to
exposure to pollutants, by race, sex, and age 32
X Estimated hospitalization rates, by level of population
exposed to air pollution 33
XI Population distribution within defined areas 34
XII Effects of assigning all unclassified hospitalizations of a
single air-quality subpopulation 37
XIII Estimated excess hospital ization rates, excess hospital days
and excess hospital costs 39
XIV Effects of exposure to air pollution; and of other factors,
on length of stay 42
XV Increased length of stay (in days per patient) under exposure
to various levels of air pollution 43
XVI Hospitalizations in Allegheny County by pollutant levels . . 44
XVII Estimated additional costs of hospitalization due to exposure
to air pollution 44
XVIII Number of cases treated in 1972 by H.U.P. hospitals in
Allegheny County 49
XIX 1972 annual average concentrations of S02 and particulates
at census tracts in Allegheny County, Pa 52
-------
ACKNOWLEDGMENTS
The authors wish to acknowledge the assistance provided during the
course of the study by members of the Environmental Protection Agency staff.
Dr. Donald G. Gillette, Criteria and Special Studies Office, Environmental
Protection Agency served as Project Officer and provided guidance and
direction. Dr. Fred Abel, now with the Energy Research and Development
Administration, initiated the study and identified several of its
methodological concepts. Mr. T. E. Waddell of EPA's Office of Planning and
Review, served as project coordinator and assisted in development of the
hospital data base.
We also wish to recognize the cooperation and help extended by the
representatives of the hospitals contacted during the study.
IX
-------
SECTION 1
INTRODUCTION
Air pollution damage to human health has been investigated by Ridker,
2 "3
Lave and Seskin, and Park. These, and other studies, have been reviewed
and assessed by Waddell. Sterling et al. studied hospitalization rate and
length of stay as related to daily air pollution. ' ' Hospitalization
costs, however, have not been estimated. Outpatient medical costs for
treatment of respiratory diseases were studied by Jacksch and Stoevener.
They found that, while air pollution may have affected the frequency of
outpatient visits, it appeared not to have affected costs per contact with
the medical system.
This study attempted to determine whether persons exposed to air
pollution incurred higher incidences of hospitalization or additional costs
for treatment. The objectives were to develop an air quality data base and
a hospitalization data base which could be merged with population data for
analysis; and to estimate by appropriate methods the effects of exposure
to pollutants on rates of hospitalization, length of stay in the hospital,
and associated costs. Three classes of diseases were studied: respiratory
diseases, heart diseases, and control diseases.
-------
SECTION 2
CONCLUSIONS
The results of this study of the hospitalization costs -of exposure to
levels of S0£ and particulates in excess of prescribed standards indicate that,
in 1972, subpop.ulations of Allegheny County, Pennsylvania so exposed incurred
significantly greater hospitalization costs. Compared to subpopulations of the
county living in clean-air neighborhoods, subpopulations living in polluted-
air neighborhoods incurred increased rates of hospitalization and increased
length of stay for treatment. For the 1.6 million persons in the county, the
conservatively estimated cost of increased rates of hospitalization was $9.1
million dollars; the cost of increased length of stay, $0.7 million. These
add to a total cost of $9.8 million dollars for the year.
The cost estimates were obtained through a comprehensive analysis of
hospitalizations, air-quality and population data. All hospitals within the
county were included in the development of the hospitalization data base.
Both respiratory diseases and circulatory system diseases suspected of
being affected by exposure to these pollutants were studied. Non-suspect
circulatory system diseases were utilized as controls.
Air quality was measured by: 1) the Huey Plate Sulfation Rate method, with
the results reported as SOo based on calibrations made on-site; and 2) by the
Coefficient of Haze with results reported as pg/m suspended particulates
also based on calibrations made on-site.
Estimated hospitalization rates were corrected for differences in age, sex,
and race distributions among the six different air-quality subpopulations of the
county. Effects of other factors were assessed and distinguished from air-
quality effects. For hospitalization rates, these included the subpopulation
median income, fraction below poverty-level income, fraction married, and the
fraction employed in heavy industry. For length of stay, the effects of the
patients' smoking habits, occupation, and type of hospitalization insurance were
considered, along with the median income of his area of residence and the per-
cent occupancy of the hospital.
These cost estimates should relate directly to the cost benefits of clean
air in Allegheny County.
2
-------
SECTION 3
RECOMMENDATIONS
The hospitalization costs of air pollution found in this study are of
sufficient magnitude to merit continued investigation. The study should be
repeated, applying the methodology in another geographical area, to confirm
the findings prior to their extrapolation nationwide.
The possibility of developing a suitable data base to estimate hospitali-
zation costs of exposure to other types of air-pollution involving photo-
chemical smog should be investigated.
The findings presented herein are based on the experience of the population
for an entire year. Further analysis of this data base should be made to assess
the effects of seasonal variations.
The present data base should be examined for selected smaller groups of
diseases to more closely identify costs with specific diseases.
The costs developed herein should be compared with estimated control
costs for the county, to develop cost-benefit information.
-------
SECTION 4
EXPERIMENTAL PROCEDURES
TECHNCIAL APPROACH
A general concept of excess risk of health costs under exposure to
pollutants was defined by the equation:
Health Costs = 4ypici/ij (4-1)
where p. is the number of persons in the subpopulation exposed to pollutant
level i, c.. is the type j cost to a person in subpopulation i, and r. . is the
extra risk to subpopulation i of incurring the cost of type j. Each term of the
health cost equation accounts for the added health costs of selected illnesses
when a specified subpopulation is exposed to air pollution at a specified level.
The exposure levels, i, may be defined as the separate levels of each
pollutant or as the combination of levels of several pollutants to which sub-
populations are exposed. Either current air-quality standards or other can-
didate values might be used to specify clean air if the relationships between
air-quality, c, and r proved to be sufficiently firm.
Under this study, only hospitalization costs were estimated, but other
types of costs can be accomodated by the health costs model.
Before these cost elements could be estimated, it was necessary to establish
that c and r were related to the levels of pollutants. The conceptual health
costs model had to be verified; verification of the model required suitable
data to test appropriate hypotheses.
Two hypotheses to be tested were:
Ho,: Increased concentrations of air pollution result in no increase
in the cost of hospital service per patient admitted and treated.
Ho2: Increased concentrations of air pollution result in no increase
in the rates of admissions to hospitals for treatment of selected
illnesses.
The first hypothesis is for c, the second for r.
-------
The Study Area
Allegheny County, Pennsylvania was selected as the study area. There
were forty-nine locations for measuring sulfur dioxide and twenty-one locations
for measuring atmospheric suspended particulates. These locations had been
chosen for various reasons as sites of opportunity: for their proximity to
emissions sources, for the varied topography, or for "background" measurement.
Sulfur dioxide was monitored by the Huey Plate Sulfation Rate Method and
reported as S02 ppm, the latter being obtained using a regression equation
based on simultaneous measurements at representative locations during the
study period by both the Huey Plate Method and the Environmental Protection
C 7 O
Agency's standard method for SC^. Particulates were measured as the
Coefficient of Haze (COH) or Soiling Index, using modified Unico Model 2800
3
instruments, and converted to ug/m of total suspended particulates (TSP),
using conversion factors determined at the sampling sites. The data exhibited,
in 1972, neighborhoods that met air quality standards and neighborhoods that
did not. The monthly average temperature was below 50°F for six months of
1972, and above 50°F for the remainder of the year.
The county lies in the central portion of the Pittsburgh, Pa SMSA for
which census data are available, providing socio-economic characteristics of
6
the population by census tract. It had 28 hospitals, 26 of which maintained
patient records by consistent entries, and two of which maintained equivalent
records providing corresponding information without revealing any individual's
identity.
Air Quality Data Base
To assess the impact of ambient concentrations of these two pollutants
upon the hospitalization costs to the population of the county it was necessary
to assign particular quantitative values to the air-quality of each of the 498
census tracts therein. The values assigned had to be based upon existing
measurements at nearby locations. Several techniques for assigning values,
i.e., interpolation or extrapolation, were considered (Section 5). After
trying a second order response surface fitted to the existing data, an intera-
tive, weighted average analysis technique, similar to procedures used in
objective analyses of meteorological data was finally adopted. A few selected
values assigned to census tracts are listed in Table I.
-------
Table I. 1972 annual average concentrations of SO? and participates at five
selected census tracts in Allegheny County, Pa.
Census
Tract
101
1.604
2003
4980
5514
UTM
Easting
548.6
587.9
579.3
597.2
597.8
Coordinates
Northing
4476.7
4474.3
4478.3
4464.4
4466.2
SOz
yg/m
85.8
141
40.6
155
97.9
Particulates
(yg/m3)
87.9
84.9
74.9
88.5
124.7
Census tracts were sorted into three SOo and three particulate levels to
establish subpopulations for comparison. Level 1 tracts met the standard. Level
2 tracts exceeded the standard enough to have a possible effect. Level 3 tracts
exceeded the bound of Level 2. Class levels were:
SO? Particulates
level
yg/m
<80
80 - 99.3
>99.3
ppb
30
30 - 37.2
>37.2
1 (low)
2 (medium)
3 (high)
yq/m3
<76
76 - 115
>115
level
1 (low)
2 (medium)
3 (high)
Hospitalization Data Base
Three classes of diseases were considered: respiratory diseases, suspect
circulatory diseases, and control circulatory diseases. All diseases were
identified by the ICDA-8 Code. Respiratory diseases included in the data
base were ICDA-8 numbers 462 through 515.9, except 508.1 (polyp of vocal
cords or larynx). Suspect circulatory system diseases were Mos. 410-414.9;
1?
427-429.9; 435; 435.9; and 436.9. These are the ischemic heart diseases,
the symptomatic heart diseases, transient cerebral ischemia, and acute cerebro-
vascular disease. Control circulatory system diseases were Nos. 390-404; 420-
426; 430-434.9; 436; 437-448; 450-458.9; and 580-480.5. These are rheumatic
fever, chronic rheumatic heart diseases, hypertensive diseases, certain cere-
brovascular diseases; diseases of arteries, arterioles, and capillaries, and
-------
diseases of veins. The last category identified nephritis and nephrosis.
There are only a few cases of these diseases, and they were included in the
control diseases.*
Records (with case history but without names) were obtained from the
twenty-eight hospitals in Allegheny County. The total number of records was
37,818, which included all cases treated in 1972.* Table II shows the breakdown
of these into the three classes under study. It also shows the number of
records from each of two files, the first a data base from HUP**, the second
an equivalent data base compiled by two hospitals not in the HUP file.
Table II. Total 1972 hospitalizations in Allegheny County hospitals for
diseases under study.
Disease Class
Respiratory
Suspect Circulatory
Controls
Total
HUP*
Hospitals
11,550
21,133
5,049
Number of Hospitalizations
Non-HUP
Hospitals
30
56
Total
11,550
21,163
5,105
37,818
*Hospital Utilization Project, Pittsburgh, Pa.
Population Data Base
The population base was developed from 1970 Census data. Table III
shows the race-age-sex distribution of the population. Here the population is
further classified according to the pollutant levels in their neighborhoods
of residence. Six S02-particulate classifications are shown. These are identi-
fied by letters. For example, there was 72,260 white males age 1-44 in the
high S02-medium particulates neighborhoods (HM).
The Census data were the source of additional population characteristics
utilized in this study. For each of the 498 census tracts of Allegheny County,
the median income, the fraction of the population with income below poverty
*A breakdown of all diseases by diagnosis and number of cases is included in
the Appendix.
**Hospital Utilization Project.
-------
Table III. 1970 Population data base (in thousands of persons).
CO
Stratum
Race
White
White
White
White
White
White
White
White
Black
Black
Black
Black
Black
Black
Black
Black
TOTAL
& Other
& Other
& Other
& Other
& Other
& Other
& Other
& Other
Sex
Male
Male
Male
Male
Female
Female
Fema 1 e
Female
Male
Male
Male
Male
Female
Female
Female
Female
Age
1 to
45 to
65 to
75 or
1 to
45 to
65 to
75 or
1 to
45 to
65 to
75 or
1 to
45 to
65 to
75 or
44
64
74
Older
44
64
74
Older
44
64
74
Older
44
64
74
Older
LL
45.566
16.042
3.097
2.110
46.941
16.971
3.829
3.524
0.941
0.261
0.145
0.112
1.034
0.307
0.125
0.178
141.183
Pollutant Classification*
LM
125.707
49.282
11.011
6.231
130.084
54.675
14.621
10.248
5.974
1.473
0.477
0.221
6.455
1.616
0.451
0.236
418.852
ML
20.293
5.444
0.964
0.620
20.635
5.628
1.290
0.821
0.441
0.084
0.009
0.021
0.378
0.073
0.030
0.015
56.746
MC
179.964
69.628
18.059
10.317
189.213
81.353
26.160
16.993
27.647
8.541
2.880
1.304
32.055
10.049
3.069
1.427
679.259
HL
2.015
0.920
0.258
0.140
2.013
0.973
0.372
0.245
0.177
0.023
0.001
0.014
0.136
0.044
0.011
0.001
7.343
HM
72.260
29.607
7.740
4.680
76.186
34.082
10.799
6.758
12.352
3.241
1.329
0.560
14.640
3.904
1.451
0.772
280.370
Total
445.80
170.92
41.73
24.20
465.07
193.68
57.07
38.59
47.53
13.62
4.84
2.23
54.70
15.99
5.14
2.63
1583.75
The S02 level is indicated by the first letter; the particulates level, by the second: L = low, M =
medium; H = high; C = combined M + H. There were no LH or HH census tracts. MC = MM + MH.
-------
level, the fraction married (and maintaining a home together) and the fraction
employed in heavy occupations were considered. Heavy occupations were Bureau
of Census classes: 5 (Craftsmen and Kindred Workers); 6 (Operatives, except
Transport); 7 (Transport Equipment Operatives); and 8 (Laborers, except
Farm). The remaining classes, including unemployed, retired, and housewives,
were considered to be light occupations.
Sample Selection
Additional information was required for each patient in the data base to
locate his census tract of residence. This was obtained for all respiratory
disease cases. It was obtained for a sample of the circulatory system disease
and control disease cases. The sampling procedure was developed to limit
the probability of error in assessing differences in rates of hospitalization
between neighborhoods of differing air quality. Sample sizes sufficient to
assess differences in the rates of hospitalization were considered adequate
to assess the differences in length of stay. Random sampling within the
strata defined by race, sex, and age was employed. Sample allocation to
strata was proportional to the number of cases in each stratum. A 100 percent
sample was utilized for respiratory disease cases and also for all three types
of illnesses in two hospitals that were not included in the HUP data base.
Cases so selected were assigned to the proper census tract (or identified
as living outside the county) with the help of the hospitals. This assignment
identified the pollutant exposure level for each case. Considerable hand
effort was required to resolve some 1800 addresses properly. Some cases were
lost from the sample because hospitals could not locate their records. The
final data base is shown in Table IV. There were 15,833 hospitalizations of
which 12,420 were from Allegheny County. Set 1 comprises the 26 HUP hospitals;
set 2 the two non-HUP hospitals which were completely enumerated.
This data base is believed to be the best ever used for the purpose of
estimating hospitalization costs associated with air pollution. The details
of the sample design are given in the next section.
-------
Table IV. Final data base, hospitalizations of Allegheny County Residents in 1972, by hospital-set and
sampling stratum.
Stratum
Race-Sex
Age
W Males
1-44
45-65
65-74
75+
W Females
1-44
45-64
65-74
75+
NW Males
1-44
45-64
65-74
75+
NW Females
1-44
45-64
65-74
75+
Subtotal
Total
Respiratory Disease
i.e.
1489
1325
716
654
1437
1060
575
504
451
202
102
52
373
128
42
34
9,144
Both Sets
O.C.
457
353
145
85
427
270
94
65
21
3
2
2
12
6
4
2
1,948
11,550
Unk.
80
36
43
31
78
46
22
28
24
12
6
3
30
8
9
2
458
Heart Disease
i.e.
103
542
265
242
33
280
258
367
10
46
28
20
13
42
30
30
2,311
Set 1
O.C.
39
235
55
34
9
65
48
32
1
2
1
0
0
0
1
1
523
2,999
Unk.
12
60
19
16
2
16
19
16
0
0
2
1
0
1
1
0
165
Set 2
*
i.e.
1
3
5
2
0
2
4
4
1
4
0
2
0
1
1
0
30
30
i.e.
24
144
115
107
17
128
124
163
3
19
14
7
8
22
12
10
917
Control Diseases
Set 1
O.C.
10
60
22
19
18
37
21
27
2
2
1
0
0
0
0
3
222
1,198
Unk.
2
14
9
6
4
4
4
6
1
3
0
0
1
2
3
0
59
i.e.
6
9
4
1
3
8
3
7
0
0
2
1
1
0
1
2
48
Set 2
O.C.
1
1
0
0
1
2
1
1
0
0
0
0
0
0
0
0
7
56
Unk.
0
1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
1
W - White; NW - Nonwhite; I.C. - In County; O.C. - Outside County; Unk. - Unknown; *No O.C.s or Unk.s in Set 2.
-------
Sample Design
For determining sample size requirements, a simplified analysis scheme
based on random samples and the normal approximation was employed. The normal
approximation to the true distribution of the test statistics should be
adequate for the large sample sizes considered.
Null (H_) and alternative (H ) hypotheses were written as:
o a
HQ : D(ii'j) = 0, and
Ha : D(ii'j) = D , where
a o
D(ii'j) = the difference between the hospitalizations for disease class j
per 1000 persons at risk in subpopulations i and i'. This difference is
defined by:
D(ii'j) = N(ij)/X(i) - N(i'j)/X(i'), where
N(ij) = the number of 1972 hospitalizations from subpopulation i for condition j,
N(i'j)= the number for subpopulation i',
X(i) = the number of persons, in thousands, for subpopulation i,
X(i') = the number of persons, in thousands, in subpopulation i'.
Data collected from the sample were to be used to estimate D(ii'j) by
D(ii'j), which may be expressed by:
D(ii'j) = N(+j)p(ij)/X(i) - N(+
where p(ij) is the sample estimate of p(ij) and
Assuming that the normal approximation is valid, the statistical test
may be written as:
Reject HQ if D(ii'j)/(V[D(ii'j)])°-5 LI
Accept H otherwise.
The term V[D(ii'j)] is the variance of D(ii'j). The value of I is chosen
to control the probability, a, of rejecting the null hypothesis when it is
true. This value may be obtained from a standard normal probability table as
the value of the normal deviate corresponding to a. To determine sample size,
another value of 1 is similarily chosen to control the probability, B, of
accepting the null hypothesis when the alternative is true.
11
-------
To determine the required minimum sample sizes, values a and 6 were
chosen along with a particular alternative hypothesis (specified by the
value of D ) and then the number of hospitalizations, n, was calculated so
that a <_ aQ and 6 £ B for the selected D . Certain additional notations
and assumptions were required to develop the computations. Let x(i) be
the proportion of the population belonging to subpopulation i, i.e.,
The term x(i') is similarly defined for subpopulation i1. It was assumed
that subpopulations considered do not overlap, i.e., no person can be a member
of both subpopulations i and i'. Further, let the hospital ization rates for
condition j and population i be
recalling that X(i) is expressed in thousands of persons. Let the rate for the
overall population of Allegheny County for condition j be
Note that R(.j) can be determined for each j on the basis of data already
available.
It was assumed that the estimates p(ij) and p(i'j) are subject to binomial
variation and therefore, after some simplification:
VCD(ii'j)] = n-1[R(.j]2{p(ij)[l-p(ij)]/x2(i)
+ p(rj)[l-p(rj)]A2(r)
+ 2 (ij) (i'
The sample size required for particular selected values a , BQ, and D were
stated approximately [if n is shown to be large] as
}O
-
where Z(aQ) and Z(BQ) are the normal deviates corresponding to aQ and e res-
pectively and V[D(ii'j)jH ] and V[D(ii 'j) |Ha] are the variances of O(ii'j) given
the null or the alternative hypothesis, respectively.
In order to compute values for the variances, realistic assumptions had
to be made about the respective values of p(ij) and p(i'j) associated with a
particular rate difference, D . A reasonable, but now always true, assumption
12
-------
is that the hospitalization rate for persons not in subpopulation i and i1
is the same as the overall hospitalization rate R(.j) for condition j. Under
this assumption p(ij) and p(i'j) may be expressed in terms of R(.j), A(i),
x(i ' ), and D as
DQ X(i) x(i')
p(ij) = X(1) + R(.j)Lx(1) + x(i')j'
and
D_ A(i) x(i')
In light of the preceding discussion and the stated assumptions, sample
size requirements were determined as a function of the parameter R(.j), D ,
A(i), x(i'), a , and eQ. Recall that these parameters have been defined as
follows:
R(.j) = county-wide hospitalization rate for condition j expressed
in hospital izations per thousand persons at risk;
D = the hypothesized difference between hospitalization rates
for subpopulation i and i1 under the alternative hypothesis;
x(i) = proportion of the county population belonging to subpopu-
lation i [x(i') is similarly defined for subpopulation i1];
a = the specified probability of rejecting the null hypothesis
(the hypothesis of no difference) when it is true; and
3Q = the specified probability of accepting the null hypothesis
when the alternative is true.
Table V shows sample size requirements for some possible values of these para-
meters.
The examples presented in Table V were chosen to illustrate sample size
requirements for some possible hypotheses. The values of x(i) and x(i') in
examples 1 through 7 correspond to those appropriate for the total population
residing in the high pollution area and the total population residing in the
low pollution area. The values of x(i) and x(i') in example 8 correspond to
those appropriate for males 65 and over residing in high pollution and low
pollution areas, respectively; this subpopulation was chosen since it illustrates
sample size requirements for some of the smaller subpopulations that might have
been considered in the study. It may be noted that larger overall sample sizes
are required if attention is to be focused on small subpopulations. The values
of R(.j) correspond to those for circulatory diseases possibly pollutant-related,
13
-------
Table V. Sample size requirements for specified hypotheses.
Estimated Minimum Sample
Example
Suspect
1
2
3
4
5
Control
6
7
Control
8
R(.j)
Do
Circulatory
13.2
13.2
13.2
13.2
13.2
Diseases
3.2
3.2
Diseases
3.2
0.3
1.
2.
3.
5.
1.
0.5
X
ID
*t
i1)
ao
B0= 0.1
= 0.05
BQ= 0.05
ao :
SQ= 0.05
, n
= 0.01
V
0.01
Diseases
0.
0.
0.
0.
0.
0.
0.
248
248
248
248
248
248
248
0.
0.
0.
0.
0.
0.
0.
341
341
341
341
341
341
341
115,919
10,471
2,629
1,172
424
622
2,487
145,292
13,130
3,298
1,469
532
781
3,123
212,369
19,178
4,814
2,144
775
1,138
4,553
292
26
6
2
1
1
6
,262
,413
,635
,957
,071
,571
,282
, males 1 65
2.
0.
012
0.
013
3,555
4,462
6,511
8
,975
R(.j) County-wide hospitalization rate per 1000
D Hypothesized difference in hospitalization rates for subpopulations i and i1
x(i) Estimated fraction of population in high pollutant areas, based on pre-
liminary air quality analysis
x(i') Estimated fraction of population of low pollutant areas, based on preliminary
air quality analysis
a Risk of claiming a difference > 0 when it's really 0
6 Risk of claiming a difference = 0 when it's really D .
14
-------
and control diseases, respectively. The values of D are what appeared at
first consideration to be reasonable candidates for specifying the alternative
hypotheses; other values might have been considered.
This analysis permitted some tentative conclusions to be drawn about
sample size requirements.
1. Enumeration of all respiratory disease cases in the file would be
necessary to permit meaningful comparisons to be made between small subpopu-
lations such as males over 65 residing in different areas.
2. A sample of 3000 circulatory diseases suspected of association with
air pollution should provide good control of the a and 3 risks for a D of 3,
and fairly good control for a D of 2. Assuming similar types of comparisons
are to be made for control diseases, and considering the lower R(.j) value for
this category, a D slightly less than 1 should be controlled (aQ = 0.01,
B = 0.05) with a sample size of about 1200. The total for both categories
4200, was attainable within the budget for data acquisition.
3. A supplemental sample of hospital records from peripheral counties
might do more to reduce total errors in subsequent estimates than heavier
sampling in the 28 Allegheny County hospitals. The seventeen hospitals
in the neighboring counties (Beaver, Butler, Washington, and Westmoreland)
were contacted. Fourteen of these were in the HUP data base. Beaver County
records showed only a few patients from Allegheny County. Butler County
record librarians found essentially no Allegheny cases. Washington and
Westmoreland County hospital librarians reported essentially no cases from
Allegheny County.
Other Possible Analyses
The analyses discussed in considering sample size requirements were based
on a simple comparison of rates. Control diseases have been included in the
study to test the hypothesis
HQ : D(ii'j) = D(ii'j');
i.e., the differences between subpopulations i and i1 are the same for condi-
tion j, say pollutant related circulatory system diseases, as they are for
condition j', control diseases.
15
-------
Drawing Conclusions from the Proposed Analyses
The methods described above permit the analyst to decide, within controlled
error levels, whether subpopulation hospitalization rates for various conditions
are similar or different. Since the study is primarily addressed at determining
the relation to air pollution, one may wish to assert that any detected
differences among subpopulations are directly attributable to the level of air
pollution experienced by these populations. Such an assertion or conclusion
may not be based solely on the statistical tests discussed above since the
subpopulations may possess a number of other known or unknown characteristics,
which were in no way experimentally controlled to allow the difference to be
associated solely with the pollutant level.
The inability to conclude causation even though association has been
proven (with possible error) is a weakness of all observational studies
o
and particularly of most studies involving human populations. Many previous
health studies, however, have been made which, collectively, help to establish
that the respiratory diseases under study increase in prevalence or incidence
among populations exposed to SOo and particulate air contamination. Additional
studi'es are showing the biological impact of such exposure on living tissues.
Given these findings, this study is concerned only with estimation of those
increases in hospitalization rates for these diseases that may be occurring
among exposed subpopulations and the translation of such increases to costs.
For this purpose, it is necessary only to prove association, which is being
done through application of statistical tests.
Lesser evidence supports the effects on circulatory system diseases
and to that extent this study itself is a source of further evidence, and
employs controls as a necessary element of the methodology.
Drawing the Sample
The previous section described the sizing of the samples, based upon
the need for precision in comparisons of different subpopulations. A sample
size of 3000 pollutant-suspect circulatory diseases and 1200 control diseases
was selected. Respiratory diseases were not to be sampled. Instead, all
cases were being utilized for analysis.
16
-------
A proportionate stratified random sample of the required size was
selected. Race, sex, and age were used to identify the strata. Samples were
drawn separately for the population-suspect-circulatory diseases and the control
diseases. Table VI shows the numbers prescribed for the strata, and the total
number of cases from which they were drawn. These numbers provided weighting
factors equal to the ratio: total numbers in stratum/number in stratum sample.
The selection process began by reading a record from the computer tape.
Records on the tape were rank-ordered by hospital, which were coded by number.
Each record was read, screened, and discarded if the patient was less than a
year old. If retained, it was then categorized by identifying its stratum.
The stratum was checked to see if its quota of the sample had been filled.
If not, a random number, R(O-l), was generated and subtracted from the ratio
A/B, which equaled the number of samples yet to be drawn from the stratum
divided by the number of records in the stratum remaining unconsidered. If
the differences were greater than or equal to zero, the record was taken into
the sample, and the elements of the ratio reduced by one. If the differences
were less than zero, the record was discarded and the stratum count reduced
by one. Accepted records were printed onto forms by computer, with space for
hospitals to record additional data.
This procedure resulted in the selection of 3000 circulatory disease
records and 1198 controls, as shown in Table VI.
17
-------
Table VI. Proportionate stratified random sample of cases of hospitalization
for circulatory system and control diseases.
Ci
rculatory
Pollution Suspect
Stratum
W Males under 45
W Males 45 to 64
W Males 65 to 74
W Males 75 and over
W Females under 45
W Females 45 to 64
W Females 65 to 74
W Females 75 and over
NW Males under 45
NW Males 45 to 64
NW Males 65 to 74
NW Males 75 and over
NW Females under 45
NW Females 45 to 64
NW Females 65 to 74
NW Females 75 and over
Total
Population
1085
5894
2387
2057
315
2556
2292
2933
79
342
217
147
89
300
222
218
21,133
Samp!
154
837
339
292
45
363
325
415
11
48
31
21
13
43
32
31
3,000
System Diseases
Control
e Population
150
919
617
556
163
718
627
824
25
103
65
28
36
100
62
56
3,049 1
Sample
36
218
146
132
39
169
149
196
6
24
15
7
9
24
15
13
,198
W - White
NW - Non White
This table includes population and sample sizes for Set 1 (HUP) hopsitals.
Refer to Table IV for population and sample sizes for the totally enumerated
cases and Set 2 (non-HUP) hospitals for all three types of illnesses.
18
-------
SECTION 5
OBJECTIVE ANALYSIS OF AIR QUALITY BY CENSUS TRACT
IN ALLEGHENY COUNTY PENNSYLVANIA
In order to assess the impact of the ambient concentrations of participate
and S02 upon the hospitalizations of the population of Allegheny County,
Pennsylvania, it was necessary to assign quantitative values to the air quality
in each particular area. The values assigned should be based upon the existing
measurements at nearby monitoring sites. Several available techniques for
assigning values were tested. In one attempt, a second-order response
surface was fitted to the existing data, but the error at points of measure-
ment was large. Three separate surfaces were then fitted to data from three
parts of the county with some success. When the three analyses were combined,
however, extreme gradients of concentrations occurred at the boundaries of
the areas, making the total analysis unacceptable. An iterative, weighted
average analysis technique, similar to procedures used in objective analyses
of meteorological data, was finally adopted.
In Allegheny County, S02 is measured at 49 locations and atmospheric
particulates are measured at 21 locations. The locations were chosen for various
reasons: proximity to emissions sources, varied topography, or "background
concentrations area". As a result, air monitoring sites are distributed rather
inhomogeneously about the county. Some are rather tightly clustered and others
are rather remote. The following technique was employed to develop S02 and
particulates levels for each census tract using the air quality data base. As a
preliminary to application of the technique, the locations of all monitoring
stations, and the centroid of all county census tracts were identified by their
Universal Transverse Mercater (UTM) coordinates.
19
-------
METHODOLOGY
The analysis technique assumes that the contribution of an observed
pollutant concentration to the estimated pollutant concentration within
a census tract decreases as a function of the distance between the two
locations, until the separation exceeds a specified distance, R, called the
radius of influence of the observed pollutant concentration. Thus, all
observations within a radius R of a given location (census tract) are
accorded a weight in computing the interpolated pollutant concentration for
that location. The computation begins by determining the distances, r..,
' J
between a particular census tract (the centroid of which is identified by
coordinates x^, y^) and measurement site j (located at coordinates x,, y.,
j = 1,2..., n). The following relationship is applied:
The weighted average of sufficiently nearby measured pollutant concentrations
is then calculated to give the estimated pollutant concentration for the census
tract, $.
? W.. d>.
$. = J iJ J (5_2)
j ij'
where W. . is a weight function dependent upon r.. and R.
* \j ' J
There are many possible choices of function form for W The form
1C ' J
used by Cressman,
R2 - r*
' J
(5"3)
(5-4)
was chosen based upon successes in previous interpolations of atmospheric
variables. This function, normalized to the interval (0,1), is shown in
Figure 1 .
20
-------
1.0
w
0.5 -
0.0
0.5
r/R
1.0
Figure 1. Cressman weight function, H, as a function of relative
distance, r/R.
21
-------
If R is large, many observations will be included in the estimate of
. Even if several observations are very near the census track, <£ tends
toward an average value in R. In the resulting analysis, maxima and minima
values are reduced, the mean of the estimated values, -'s, is nearly equal
to the mean of the observed values, -'s is much
smaller than the variance of .'s.
J ^
If R is small, the variance of .'s will be more nearly equal to the
variance of the 4>-'s. However, if R is too small there may be some locations
J
which do not have r..'s less than R. Thus no interpolation is possible.
' J
A three-step iterative procedure was adopted to analyze the entire area
of Allegheny County, maintaining the spatial variability of the measurements.
On the first iteration, a large value of R, R(l), was used so that at least
three observations were included in each interpolation. On the second iteration,
a smaller value, R(2) was used. R(2) depended upon the average density of
observation over the analysis area. The results of the first iteration were
also incorporated into the second analysis by a preassigned weight. Finally,
R(3), a very small value, was used with the previous iteration to help
regain the observed values within the analysis field.
Letting v indicate the iteration, W. .(v) denotes the Cressman weight
function using R(v). The expression for $.. can be generalized to:
W *,- (v) + i W..(v) +,
' ,_! 'J J
^.(v+1) = ^^ (5-5)
W + I W..(v)
j=l 1J
where .(0) = 0, and W is a small positive constant.
On the first iteration, equation 5-5 estimates a
-------
ANALYSIS APPROACH
The computations were carried out using census tract and monitoring
station locations mapped into the square grid centered on the city of Pittsburgh
and normalized by the grid length, 2 km. The average concentration of SO^
and of particulate measurements for 1972 from each monitoring site were used as
inputs for the analyses. Two interpolations were done for each pollutant.
The first analysis used the concentration, C, as the dependent variable j
The second used the natural logarithm of the concentration as that variable,
i.e., (j». = InC. The census tract values of concentration were obtained by
exponentiating the results.
The spatial representations were examined for continuity and agreement
with observed values. In both cases, the analysis using the logarithms showed
a more pronounced peak near the monitoring stations with little gradient in
other areas. The analyses using observed concentrations showed a smoother
transition with increasing distance from a monitoring station. Reasoning that
over a year's time, with many different wind and dispersion conditions, a
smooth transition between data points would be more likely, the analysis using
the observed data was accepted as preferred.
The scan radii, R, were different for the particulate estimations and the
S02 estimations, because of the different data density. Table VII shows the
R(v)'s which were used. The greatest difference is in R(2) which was due to
the difference in number of observations available. Letting A be the overall
area of the Allegheny County interpolation grid (3600 km2) and N the number
of observations, then
R(2) =
Tab!e VII. Scan radius (km) by iteration and pollutant.
V
1
2
3
Particulates
30.0
13.0
1.6
Sulfur Dioxide
24.0
5.0
1.6
23
-------
Figure 2 shows approximate contours of levels of S02 drawn on a map of
Allegheny County, using the estimated levels by census tracts. Figure 3 shows
approximate contours of levels of particulates. The effects of uneven terrain
and known emission sources are quite evident. A complete listing of the
estimated air quality by census tract is given in the Appendix.
24
-------
ro
en
Figure 2. SO,, levels (1972 yearly average), Allegheny County, Pa.,
-------
ro
en
figure 3. Participates levels (1972 yearly average), Allegheny County,
Pa.,
-------
SECTION 6
EFFECT OF AIR POLLUTION ON HOSPITALIZATION INCIDENCE RATES
The data were examined first by multiple regression analysis techniques
to determine whether the relationship between hospitalization incidence
rates (hospitalizations per year per 1000 persons) for respiratory and
circulatory system diseases and the levels of pollution were statistically
significant, and whether such a relationship was not significant in control
disease data. Such effects were indeed established, except for the control
diseases. Accordingly, the data were further analyzed by methods which
utilized the sample design more fully in disclosing the nature of the effects.
A regression model was used in which the unit of observation was the
census tract, and the population considered was that defined by the Census.
Incidence rates data were merged with air-quality data and population data for
the analysis. The rates (in hospitalizations per 1,000 population) were then
examined for their relation to several factors, and treated simultaneously as
independent variables. Air-quality variables were: the census tract S09 level
o ^
(ppb), the particulates level (yg/m ), and their product (interaction).
Population characteristics included were: the census tract fraction males,
fraction white race, fraction married (and living with spouse), fraction with
income less than or equal to the poverty level (as defined by the Census, with
consideration of the family status, etc.), the median income, the fraction
employed in heavy industry, and the fractions in three age classes (65-74
years, 45-64, and 75 or over).
In development of the model, all variables were treated as continuous
variables. The analysis imposed the assumption that the smoking habits of the
population did not vary significantly with census tract pollutant level, since
data for this characteristic of the population were not available. This
assumption seems reasonable.
27
-------
Three regression analyses were made, one for respiratory diseases, one
for suspect circulatory system diseases, and one for control diseases. Ideally,
the control diseases would reconfirm that the effects of air pollution are
disease specific. Several additional benefits from the use of controls are
14
discussed by MacMahon.
RESULTS OF REGRESSION ANALYSIS
Of the 498 census tracts in the county, required Census data were complete
for 493, and these were used in the analysis. The effects estimated by
regression analysis are shown in Table VIII. Their significance is also
indicated. All effects are listed. However, effects such as those of age are
not unexpected and were estimated primarily to minimize their influence on the
effects of pollutants.
Exposure to pollution appears not to have affected the incidence rates
for control diseases judged by the large significance levels. Such exposure
seems definitely to have increased incidence rates for respiratory diseases,
judged by the low significance levels. The effects on incidence rates for
circulatory diseases are positive (hospitalization rates increased with exposure
to higher pollution levels); the significance levels are not quite so low as
those for respiratory diseases, but are too low to be ignored. In comparison
to control diseases, the evidence is strong that hospitalization rates for the
circulatory system diseases under consideration here are increased by exposure
to the higher pollutant levels. (It is certain that they did increase in
Allegheny County in 1972).
The regression analysis was helpful in assessing several population
characterisites as factors that might have affected incidence rates. According
to the analysis, the fraction males was not a significant one. The effect of
race, represented by the fraction white in the analysis, was significant: whites
appear to have lower hospitalization rates for respiratory diseases and higher
rates for suspect circulatory system diseases, compared to other races. Rates
for control diseases do not appear to be race dependent. Married persons
appear to incur lower rates of hospitalization for both respiratory and suspect
circulatory system diseases, but not for the control diseases. Persons with
incomes at or below poverty level showed significantly higher rates for respira-
tory and co'ntrol diseases, but not for suspect circulatory diseases. Admission
28
-------
-J2
Table VIII. Effects of exposure to air pollution, and other factors, on incidence of
hospitalization.
Control Diseases
Respiratory Diseases
Effect Significance* Effect Si
Intercept
SOo level*, ppb
Participates level, pg/m3
S02 x Parti culates
Fraction males
Fraction white
Fraction married
Fraction <_ poverty level
i ncome
Median income, $/yr
Fraction in heavy industry
Fraction age 65-74
Fraction age 45-64
Fraction age 75 or more
-8.7
-0.008
0.012
0.00009
5.97
-1.39
1.99
74.1
0.001
6.2
1.01
7.63
32.6
0.41
0.98
0.92
0.98
0.38
0.20
0.49
0.005
0.098
0.019
0.915
0.11
0.003
-15.4
0.71
0.27
-0.008
-1.52
-3.47
-10.8
92.1
0.00002
8.6
-11.1
19.7
15.6
gnificance
0.20
0.053
0.049
0.054
0.85
0.005
0.001
0.002
0.79
0.005
0.30
0.0003
0.215
Suspect Circulatory
System Diseases
Effect Significance
-41.8
1.34
0.58
-0.015
1.86
5.41
-24.9
35.4
0.00006
6.95
7.0
37.4
21.0
0.13
0.12
0.069
0.12
0.92
0.06
0.0013
0.42
0.78
0.32
0.78
0.003
0.47
The significance is the theoretical probability of obtaining an absolute value of t (t = estimate
of effect v standard error of the estimate) as large or larger than that exhibited by the data
under the hypothesis of a zero effect.
-------
rates appeared to increase with census tract median income only for control
diseases. They also increased with percentage employed in heavy industry,
except in the case of suspect circulatory diseases.
While the regression analysis serves to show that there are significant
effects of air pollution on hospitalization incidence rates, the model is
inadequate to provide estimates to those effects. As indicated in Table
VIII, both S02 and particulates levels were entered into the calculations
of the regression model in their basic units, so that the effects are per
3
ug/m for particulates and per ppb for S02- While the model could be used
to estimate a hospitalization rate for each census tract, the errors of such
estimates would be relatively large, and the distinction between populations
exposed to prescribed air quality and polluted air would be affected. The
population pollutant exposure experience breaks down into six of the nine
possible combinations of "levels" of S02 and particulates. There were no
census tracts with low S02 and high particulate levels, nor any with both
pollutants at the high levels. Very little of the population experienced
exposure to medium SOo-high particulates, and therefore, this subpopulation
was combined with the medium S02-medium particulates subpopulation to avoid
inflated error estimates that were obtained using the small population
separately. With six distinct pollutant-exposure levels, it is doubtful
that three parameters in the regression model can provide the desired com-
parisons. To obviate these difficulties, the incidence rate data were
analyzed as a proportionate stratified random sample.
ANALYSIS BASED ON THE SAMPLE DESIGN
In addition to the regression analysis discussed above, an analysis
of hospitalization rates based on six levels or categories of exposure to air
pollutants was conducted. Annual hospitalization rates per thousand persons
were computed using the 1972 sample data on hospitalizations and 1970 Census
base data. The link between the two data sets was achieved by identifying
the residence of each patienL by census tract. Air quality measures were
assigned to census tracts as discussed in Section 5; then air quality
measures were associated both with the hospitalization estimates and with
the population base counts by age, sex, and race.
Ideally, both hospitalization and base data should be based on the same
year. Since 1972 is only two years from 1970, it has been assumed that popu-
lation changes and shifts occurring over that short period of time would not
materially effect the results of the analysis.
30
-------
Sixteen age-sex-race categories were utilized in stratifying the
hospital cases prior to sample selection. These sixteen categories were
collapsed to twelve in the analysis by combining the age classes "65 to 74"
and "75 and older" into a single "65 and older" category. This was done
primarily to eliminate all zero cells in the population data base. Population
bases in thousands of persons are shown in Table IX.
Total hospitalizations were then estimated for each exposure level, and
converted to rates per 1000 persons. These estimates are shown in Table X.
Age-sex-race adjusted rates are shown at the bottom of Table X.
Part of the observed differences in hospitalization rates among the six
areas might be attributed to factors other than air pollution- Due to differences
in age, sex, and race distribution (Table XI), estimates of hospitalization
rates for each area were also computed using the total SMSA population as a
standardizing distribution. These adjusted rates are also shown in Table X,
along with their standard errors. The final entries in Table X show the
median pollutant level for both S02 and particulates, and the median family
income.
Estimation Procedures
Standard estimation procedures for stratified samples were utilized in
computing estimates of hospitalizations by pollutant exposure areas. Since
the inferences drawn from the analysis are intended to extend beyond the
finite list of 1972 hospitalizations in Allegheny County, no finite population
correction factors were used in any of the variance estimates. This is not
meant to suggest a claim of external validity to any particular population
defined statewide or nationally. Ignoring the finite connection factors
treats the occurrence of hospitalizations as a random process taking place
under conditions defined in terms of the existing population characterized
by age, race, sex, and residence in a defined air quality area.
Estimates of hospitalization rates, R (dp), for disease d in pollutant
level p were computed as a weighted average
12
R (dp) = E W (pi) R(dpi)
a i=l a
where
a = 1 or 2 for type of weighting distribution;
31
-------
Table IX. Classification of Allegheny County Population according to exposure to pollutants,
by race, sex, and age. (In thousands of persons).
co
ro
POLLUTANT LEVELS*
Race
White
White
White
White
White
White
Black and
Other
Black and
Other
Black and
Other
Black and
Other
Black and
Other
Black and
Other
Total
Sex
Male
Male
Male
Female
Female
Fema 1 e
Male
Male
Male
Female
Female
Female
Age
1 to 44
45 to 64
65 or Older
1 to 44
45 to 64
65 or Older
1 to 44
45 to 64
65 or Older
1 to 44
45 to 64
65 or Older
LL
45.566
16.042
5.207
45.941
16.971
7.353
0.941
0.261
0.257
1.034
0.307
0.303
141.183
LM
125.707
49.282
17.332
130.084
54.675
24.869
5.974
1.473
0.698
6.455
1.616
0.687
418.852
ML
20.293
5.444
1.584
20.635
5.628
2.111
0.441
0.084
0.030
0.378
0.073
0.045
56.746
MC
179.964
69.628
28.976
189.213
81.353
43.153
27.647
8.541
4.184
32.055
10.049
4.496
679.259
HL
2.015
0.920
0.398
2.013
0.973
0.617
0.177
0.023
0.015
0.136
0.044
0.012
7.343
HM
72.260
29.607
12.429
76.186
34.082
17.557
12.352
3.241
1.889
14.640
3.904
2.223
280.370
TOTAL
445.80
170.92
65.93
465.07
193.68
96.66
47.53
13.62
7.07
54.70
15.99
7.77
1583.75
The first letter represents the S02 level; the second, the particulates level. LL = low S0£, low
particulates. MC = medium S02, combined medium and high particulates, etc.
-------
Table X. Estimated hospitalization rates, by .level of population exposed
to air pollution.
CO
00
Subpopulation Exposed (1000's)
Total Hospital izations, 1972
Respiratory Diseases
Suspect Circ. Sys. Diseases*
Control Diseases
Hospitalization Rates (per 1000)
Respiratory Diseases
Suspect C1rc. Sys. Diseases
Control Diseases
Hospital ization Rates (per 1000)
(Age-Sex and Race Adjusted)
Respiratory Diseases
Suspect Circ. Sys. Diseases
Control Diseases
Median S0? Level, pg/m
£ 3
Median Participates Level, iig/m
Median Family Income, $
LL
111.183
518
(22)**
951.
(80)
184.
(27)
3.
(0-
6.
(0.
1.
(0.
4.
(0.
7.
(0.
1.
(0.
66
64.
7343
5
3
88
16)
74
67)
31
19)
20
20)
24
64)
56
26)
4
LM
418.852
2216
(42)
4292
(155)
1015
(58)
5
(0
10
(0
2
(0
5
(0
10
(o
2
(0
66
84
7096
.29
.10)
.25
.37)
.48
-H)
.67
.12)
.74
-41)
.50
-15)
.6
ML
b6.746
281
(17)
331 . 2
(48)
76.7
(18)
4.95
(0.29)
5.84
(0.84)
1.35
(0.32)
5.98
(0.42)
7.59
(0.12)
2.28
(0.62)
83
72.7
7312
MC
679.259
4191
(51)
7276.9
(179)
1896
(70)
6
(0
10
(0
2
(0
5
(0
10
(0
2
(0
83
85
7107
.9
.17
.08)
.72
.26)
.79
.10)
.95
.07)
.25
.26)
.67
.10)
.1
.3
1IL
MM
7.343 280.37
78 1830
(9) (39)
119.9
(29)
25.2
(10)
10
(1
16
(3
3
(1
9
(1
14
(4
3
(1
104
73
7055
.62
.20)
.32
.93)
.43
.37)
.82
.16)
.91
.13)
.73
.58)
.9
.5
3343.5
(58)
713.9
(50)
6.53
(0.14)
11.93
(0.50)
2.55
(0.18)
6.19
(0.13)
11.42
(0.48)
2.38
(0.17)
105. 1
87.3
7358
Total
1583.75
9144
(43)
16315
(157.8)
3913
(60)
5.77
(0.03)
10.30
(0.10)
2.47
(0.02)
5.77
(0.03)
10.30
(0.10)
2.47
(0.04)
The fractional values result from estimating, using the sample data.
**
( ) denotes standard deviation of tlie value entered iuniedtately above.
-------
Table XI. Population distribution within defined areas.
PERCENT OF
Population
Group
White Male
1-44
45-64
65+
White Female
1-44
45-64
65+
Non-White Male
1-44
45-64
65+
Non-White Female
1-44
45-64
65+
White
Non- White
Males
Females
Persons aged 1-44
Persons aged 45-64
Persons aged 65+
<80
<76
32.3
11.4
3.7
33.2
12.0
5.2
.7
.2
.2
.7
.2
.2
97.8
2.2
48.4
51.6
66.9
23.8
9.3
TOTAL POPULATION BY POLLUTANT LEVEL
so2 (
ug/m3)
80-99.3 > 99
Parti culates (Mq/m3)
176
30.0
11.8
4.1
31.1
13.1
5.9
1.4
.4
.2
1.5
.4
.2
96.0
4.0
47.9
52.1
64.0
25.6
10.4
<76
35.8
9.6
2.8
35.4
9.9
3.7
.8
.1
.1
.7
.1
.1
98.1
1.9
49.1
50.9
73.6
19.8
6.6
I76
26.5
10.3
4.3
27.9
12.0
6.4
4.1
1.3
.6
4.7
1.5
.7
87.2
12.8
47.0
53.0
63.1
25.0
11.9
<76
27.4
12.5
5.4
27.4
13.3
8.4
2.4
.3
.2
1.9
.6
.2
94.5
5.5
48.3
51.7
59.1
26.7
14.2
.3
I76
25.8
10.6
4.4
27.2
12.2
6.3
4.4
1.2
.7
5.2
1.4
.8
86.4
13.6
47.0
53.0
62.6
25.2
12.2
All
Allegheny
County
28.1
10.8
4.2
29.4
12.2
6.0
3.0
.9
.4
3.5
1.0
.5
90.7
9.3
47.4
52.6
63.9
25.0
11.1
34
-------
d = 1, 2, or 3 for respiratory diseases, suspect circulatory diseases,
or control diseases, respectively;
p = 1, 2, , 6 for the six defined pollutant levels as shown in
Table IX;
i = 1, 2, , 12 for the 12 race-sex-age categories shown in Table IX;
W (pi) = weighting function for rates specific to pollutant level p and age-
a sex-race category i; and
R(dpi) = estimated hospitalization rate for disease d, air pollutant level p,
and race-sex-age category i.
Race-sex-age specific estimates, ft(dpi), were computed as
R(dpi) = { I P(hdpi) N(hdi)}/X(pi)
h=l
for i = 1, 2, 4, 5, 7, 8, 10, and 11 where
h = 1 or 2 for the sampled or enumerated hospitals, respectively;
P(hdpi) = proportion of studied hospitalization cases from hospital group
h, disease d, and race-sex-age group i that are associated with
pollutant p;
N(hdi) = total number of 1972 hospitalizations from hospital group h, disease
d, and race-sex-age group i, and
X(pi) = 1970 Census population of pollutant level area p and race-sex-age
group i.
For race-sex-age specific estimates involving over 65 (i.e., i = 3, 6,
9, and 12), estimates were computed as
R(dpi) = { i z P(hdp(i,j)) N(hd(i,j))}/X(pi)
h=l j=l
where the subscript j identifies the age groups 65-75 (j = 1) and 75 and over
(j = 2) used as strata in the sample design and collapsed into a single age
group (65 and over) for analysis purposes.
The weighting function W (pi) was computed in two ways depending on
whether a standardizing race-sex-age distribution was employed. Non-standard-
ized estimates (a = 1) were based on weights appropriate to the individual
pollutant level defined areas, i.e.,
/ i
i=l
35
-------
Standardized estimates (a = 2) were based on weights determined from the
race-sex-age distribution of the entire county, i.e.,
6 612
VUpi) = i X(pi)/ z I. X(pi).
p=l p=l i=l
Variance estimates were computed for the rate estimates as:
12 9
var[R (dp)] = z lr(pi) var [R(dpi)]
a i=l a
where
var[R(dpi)] = {x i var[P(hdp(i,j))] N2(hd(ij))}/X2(pi),
h j
var[P(hdp(i,j))] = P(hdp(i,j)) [l-P(hdp(i,j
and
n(hd(i,j)) = sample size for stratum hd(i,j).
The subscript j and the summation over j may be eliminated from the formulae
for those race-sex-age categories whose age is under 65 (i.e., for i = 1, 2,
4, 5, 7, 8, 10, and 11).
Potential Biases
Some cases were lost from the sample because hospitals could not locate
their records. Most of the hospitals felt that the patients involved were most
likely from outside the county, therefore, they were omitted in the analysis of
the data. The worst possible bias that these missing patients could introduced
is the bias resulting from their having all belonged to a single air-quality
subpopulation of Allegheny County. Although this was considered quite unlikely,
its effect was examined. Table XII shows the increases in hospitalization
rates that would result if all unclassified cases were added to cases from a
single air-quality subpopuletion. For example, the 458 unclassified cases out
of 11,550 respiratory diseases cases, would increase the rate in the LL sub-
population from 3.88 to 7.12. If added to the HL subpopulation, the rate
would increase from 10.62 to 72.99, which is unreasonably high.
36
-------
Table XII. Effects of assigning all unclassified hospitalizations
a single air-quality subpopulation.
to
CO
--J
Item
Total Sample
Number of Unclassified
Estimated hospitalization
Air quality L
L
M
M
H
H
Respiratory
Diseases
rate per 1000 persons*
, L
, M
, L
, c
, L
, M
3
5
4
6
10
6
11
.88
.29
.95
.17
.62
.53
550
458
( 7
( 6
(13
( 6
(72
( 8
Circulatory
System
Diseases
2999
.12)
.38)
,02)
.84)
.99)
.16)
6
10
5
10
16
11
1
.74
.25
.84
.71
.32
.93
65
( 7.
(10.
( 8.
(10.
(38.
(12.
Control
Diseases
1198
248
90)
64)
75)
95)
79)
52)
1.30
2.43
1.35
2.79
3.43
2.55
(3.06)
(3.02)
(5.72)
(3.16)
(3.72)
(3.43)
The first estimate is the rate without including unclassified hospitalizations; the second, in paren-
theses, is the rate obtained by adding all unclassifieds to the single air quality level.
-------
A second potential bias could have resulted from persons in Allegheny
County entering hospitals in neighboring counties, and thus not being included
in the data. Hospitals in neighboring counties were contacted concerning
this possibility; all indicated that to their knowledge, essentially no such
cases had occurred. On the other hand, substantial numbers of persons in
neighboring counties did enter Allegheny County hospitals, and the data
analysis took this into account.
Effects on Hospital Use and Costs
The total number of excess hospitalizations for each of the three disease
conditions was estimated by comparing the hospitalization rates in the area of the
3 q
county meeting air quality standards (S02 <80 yg/rrr and particulates <76 ug/m )
with the remaining parts of the county. These comparisons are shown in
Table XIII. The difference in rates is applied to the population residing
in areas that did not comply with standards. Estimates of 3,000, 5,650, and
1,832 excess hospitalizations for respiratory, suspect circulatory, and con-
trol diseases associated with noncomplying areas are shown. Standard error
estimates are given in parenthesis below each estimated number.
The cost of excess hospitalizations was calculated assuming that length
of stay and cost per day did not vary with level of pollution. (These assump-
tions are not entirely correct; the costs of extra length of stay are estimated
separately in Section 7.) Average lengths of stay and average cost per day
were computed from data provided by 26 of the 28 hospitals in the study. Average
lengths of stay were: 9.4 days for respiratory diseases; 13.7 days for sus-
pect circulatory system diseases; and 15.4 days for control diseases.
Corresponding average audited costs per day were: $68.10, $67.20, and $71.30.
These costs differ because of the slightly different distribution of the dif-
ferent diseases among the hospitals.
Using these length of stay and cost per day estimates, total excess
hospital days and total excess costs were computed, and are given in Table XIII,
along with their standard errors.
38
-------
Table XIII. Estimated excess hospitalization rates, excess hospital days and excess hospital costs
CO
HOSPITALIZATION RATES
Disease
Respiratory
Diseases
Suspect
Circ. Sys.
Diseases
Control
Diseases
Total
Air
Pollutants
At or
Below
Standards
3.88
(0.16)
6.74
(0.57)
1.31
(0.19)
Air
Pollutants
Above
Standards
5.96
(0.03)
10.65
(0.11)
2.58
(0.03)
Excess
Hospital izations
Per
1000 Persons
1 yr or older
2.08
(0.17)
3.91
(0.58)
1.27
(0.20)
7.26
(0.64)
Total Excess
Hospitalizations
3,000
(254)
5,640
(838)
1,832
(300)
10,472
(926)
Total* Excess
Hospital Days
28,205
(2,386)
77,274
(11,478)
28,214
(4,615)
133,693
(12,599)
Total*
Excess
Costs
$1,920,769
(162,490)
5,192,823
(771,309)
2,011,642
(329,119)
9,125,241
(854,190)
*Based on a population of 1,442,570 persons 1 year old and over residing in those areas not meeting pre-
scribed standards for air quality; assumed average lengths of stay equal to 9.4 days for respiratory
diseases, 13.7 days for suspect circulatory diseases, and 15.4 days for control diseases; and assumed
average costs per day of $68.10 for respiratory diseases; $67.20 for suspect circulatory diseases; and
$71.30 for control diseases.
-------
SECTION 7
EFFECT OF AIR POLLUTION ON LENGTH OF STAY IN THE HOSPITAL
ANALYSIS OF THE DATA
The data were examined by multiple regression analysis techniques to
determine whether the relationship between the patient's length of stay and
the level of pollution at his neighborhood residence was statistically
significant. If so, then attendant costs could be estimated with justification.
A regression model was used in which the unit of observation was the individual
patient, and the population considered was the population of hospitalized
persons. His length of stay was examined for its relation to several factors,
treated simultaneously as independent variables. Personal characteristics
included the patients sex, race, age, smoking habits, occupation, how his
hospital bill was paid, and whether he had had surgery. His exposure to
pollutants S02 and particulates was based upon the levels for his census tract.
The hospital load, or percent occupancy, was taken from hospital records. The
median income of his census tract was taken from census data.
In the development of the regression model, median income and hospital
load were treated as continuous variables. The others were treated as class
variables; differences between specific levels and the average over all levels
were computed and tested for statistical significance.
The occupation effect was computed as half the difference in average
length of stay between "light" occupations and "heavy" occupations. The
rationale for this comparison was that the "light" occupations were less likely
to impose additional pollutant burdens on the persons than the "heavy" occu-
pations.
Smoking habits were defined using four codes: 0 = unknown (for those whose
habits were not recorded); 1 = non-smokers (including those who had quit);
2 = light smokers (one pack or less a day); 3 = heavy smokers (more than a pack
a day, or recorded as "heavy" smoker). The effects were computed by comparing,
in sequence, the unknowns, the non-smokers, and the light smokers with the
average over all four classes.
40
-------
Four age classes were established: 1 = patients 1-44 years; 2 = 45-65;
3 = 65-74; and 4 = >74. Each of the first three classes were compared with the
average of all classes. Patients less than one year old were excluded because
of their limited exposure to air pollution.
Five payment classes were established: 1 = those using Medicare, Medicaid,
or Government Insurance; 2 = Blue Cross or Commercial Insurance; 3 = Workmans
Compensation or UMW Insurance; 4 = other types of insurance; 5 = selfpayment.
Each of the first four classes were compared with the average of all classes.
Because pollutant levels for SC^ were so often closely correlated with
levels of particulates, four classes of air quality were established: 1 =
low S02 and low particulates; 2 = low S02 and medium particulates; 3 = high,
or medium SC^ and low particulates; 4 = high or medium SCL, and high or medium
particulates. These classes covered all patient exposures. Each of the first
three classes were compared with the average of all four.
The effect of sex was determined as half the difference between males and
females (a negative effect thus would indicate a greater length of stay for
females). The effect of race was calculated as half the difference between
whites and other races.
Three analyses were made, one for respiratory diseases, one for suspect
circulatory system diseases, and one for control diseases. Ideally, control
diseases could reconfirm that the effects of air pollution are disease specific
Several additional benefits from the use of controls are discussed by MacMahonJ4
RESULTS
The effects estimated by regression analysis are shown in Table XIV.
Their significance is also indicated. All effects are listed. However, effects
such as those of age and smoking were not unexpected and were estimated pri-
marily to minimize their influence on the effects of pollutants.
Exposure to air pollution appears not to have affected length of stay for
control diseases. The comparisons for control diseases are not significant,
judged by the large values in the significance column (0.50, 0.84, and 0.77).
The comparisons for respiratory diseases, on the other hand, show that length
of stay for pollutant levels 1 and 2 are significantly different from the aver-
age. The comparisons for suspect circulatory diseases show that length of stay
for pollutant level 3 is very significant, greater than the average, and that
length of stay for pollutant levels 1 and 2 is significant at a sufficient level
to merit further study.
41
-------
TabVe XIV. Effects of exposure to air pollution; and of other factors, on length of stay.
ro
Control Diseases
Respiratory Disease
Suspect Circulatory
System Diseases.
Intercept
Sex 0.5 (males -females)
Race 0.5 (whites-others)
Surgery 0.5 (no surgery-surgery)
Median Income, $ Astay/A$
Occupation 0.5 (light-heavy)
Smoking: Unknown -avg.
Nonsmokers - avg.
Light smokers - avg.
Pollutant exposure
1-avg.
2-avg.
3-avg.
Age
(1-44) - avg.
(45-64) - avg.
(65-74) - avg.
Payment class
1 - avg.
2 - avg.
3 - avg.
4 - avg.
Hospital % of occupancy Astay/A%
Number of Cases
Effect Si
28.01
- 0.23
0.21
- 3.2
0.000
- 1.0
0.15
1.02
0.22
- 1.8
0.32
1.0
- 2.4
0.96
- 0.19
2.3
0.01
- 8.7
8.6
- 0.16
951
gnificance*
0.003
0.70
0.83
0.0001
0.85
0.34
0.89
0.38
0.93
0.50
0.84
0.77
0.25
0.51
0.90
0.54
0.99
0.39
0.32
0.11
Effect
6.6
-0.2
-0.12
-0.31
0.00004
0.05
-0.27
-0.69
0.39
-0.63
0.39
-0.28
-4.29
-0.02
1.82
0.12
-0.43
0.20
0.70
0.05
8,
Significance
0.0001
0.015
0.33
0.0002
0.07
0.71
0.09
0.0001
0.24
0.030
0.033
0.38
0.0001
0.89
0.0001
0.77
0.29
0.90
0.19
0.0004
993
Effect Si
11.8
- 0.17
- 0.22
- 0.29
- 0.0000
0.04
- 0.70
0.25
0.18
- 1.16
- 0.69
2.4
- 2.5
- 0.14
0.97
1.25
1.54
- 0.24
- 6.99
0.01
2,31
gnificance
0.0002
0.39
0.51
0.15
0.99
0.92
0.04
0.50
0.81
0.11
0.14
0.009
0.001
0.77
0.05
0.34
0.20
0.94
0.05
0.72
2
The significance is the theoretical probability of obtaining an absolute value of t (t = estimate of
effect :- standard error of the estimate) as large or larger than that exhibited by the data under the
hypothesis of a zero effect.
-------
Table XV shows the expected values of increased length of stay at each
of the classes of pollutant levels studied along with the standard errors.
These were derived from the comparisons shown in the previous table. The
Table XV. Increased length of stay (in days per patient) under exposure to
various levels of air pollution.
AIR POLLUTION LEVELS
Diseases
Respiratory
Suspect Circulatory
Controls
At or Below
Standards
0
0
0
Low S02
Med Part.
1.02
(0.39)
0.47
(0.93)
2.12
(3.29)
High S02
Low Part.
+0.35
(0.55)
3.46
(1.46)
2.80
(5.40)
High S02
High Part.
+1.15
(0.37)
0.61
(0.90)
2.28
(3.20)
increased stays exceeded their standard error except for the low particulates
level. For suspect circulatory diseases, the high S02~low particulates
exposure showed the greatest effect on length of stay, an increase of 3.5
days. This suggests a strong effect of high S02 at moderate particulates
levels. The controls showed no significant increase in length of stay with
increases in pollutants; sample size was, of course, smaller for controls.
ESTIMATED COSTS
The additional costs to persons in Allegheny County for 1972 was estimated
in two steps: the number of hospitalizations for respiratory diseases and for
circulatory diseases at each of the three pollutant levels greater than
standards (Table XVI) was multiplied by the extra length of stay (Table XV)
if significantly greater than zero, to obtain the total extra days of hospital-
ization (Table XVII). The total was 10,744 days in 1972. This was then
converted to 1972 dollars by multiplying by the audited average cost per day.
Such costs were available to the study for 26 of the 28 hospitals. The average
daily cost was $68.10 for respiratory diseases and $67.20 for suspect circula-
tory system diseases. The total estimated cost to residents of Allegheny County
in 1972 due to increased length of stay in the hospital was thus found to be
$731,697.
43
-------
Table XVI. Hospitalizations in Allegheny County by pollutant levels.
AIR POLLUTION LEVELS
Diseases
Respiratory
Suspect Circulatory
Control
At or Below
Standards
548
(23)
951
(80)
184
(27)
Low S02
Med Part.
2216
(42)
4292
(155)
1016
(58)
Med & High
S02
Low Part.
359
(19)
451
(56)
102
(20)
High S02
Med & High
Part.
6021
(64)
10620
(227)
2611
(86)
Total
9144
(85)
16314
(292)
3913
(109)
( ) standard deviation
Table XVII. Estimated additional costs of hospitalization due to exposure
to air pollution.
AIR POLLUTION LEVELS
At or Below
Diseases Standards
Respiratory
Estimated additional 0
days
Estimated additional 0
cost, $
Suspect Circulatory
Estimated additional 0
days
Estimated additional 0
cost, $
Total additional days
Total additional costs,!
Low S02
Med Part.
2260
(865)
153906
(58960)
0
0
2260
(865)
153906
(58960)
Med & High
S02
Low Part.
0
0
1560
(686)
106267
(46125)
1560
(686)
106267
(46125)
High S02
Med & High
Part.
6924
(2228)
471524
(151700) (
0
0
6924
(2228)
471524
(151700) (
Total
9184
(2390)
625430
162755)
1560
(686)
106267
(46125)
10744
(2331)
731697
169164)
( ) standard deviation
44
-------
SECTION 8
TOTAL HOSPITALIZATION COSTS OF AIR POLLUTION
This study provides estimates of additional hospitalization costs to
subpopulations living in the neighborhoods of Allegheny County, Pennsylvania
which show levels of S0£ and particulates in the ambient air that exceed
present air-quality standards. There are two types of costs. One is the
cost of increased hospitalization rates for relevant diseases; the other,
the additional cost to admitted patients due to a greater length of stay
related to their having lived in areas of excessive air pollution. Estimates
of both types of costs were developed for the year 1972. They are as
follows:
Cost of increased rates of hospitalization $9,125,000
Cost of additional length of stay $ 731,700
Total Cost $9,856,700
45
-------
SECTION 9
REFERENCES
1. R.G. Ridker, Economic Costs of Air Pollution, Frederic A. Praeger,
New York, 1967. pp. 31-56.
2. L.B. Lave and E.P. Seskin, "Air Pollution, Climate and Home Heating,"
American Journal of Public Health, 62:909 (1972).
3. W.R. Park, The Economic Impact of SO? Emissions in Ohio, Midwest Research
Institute, Kansas City, Mo.
4. I.E. Waddell, The Economic Damages of Air Pollution, Environmental
Protection Agency, Report No. EPA-600/5-74-012, 1974. Chapter 5.
5. J.A. Jaksch and H.H. Stoevener, Outpatient Medical Costs Related to Air
Pollution in the Portland, OregorTArea, EPA Office of Research and
Development, Report EPA-600/5-74-017, 1974.
6. N.A. Huey, "The Lead Dioxide Estimation of Sulfur Dioxide Pollution,"
J.A.P.C.A., 18(9): 610 (1968).
7. M. Boulerice and W. Brabant, "New PbOp Support for the Measurement of
Sulfation," J.A.P.C.A., 19(6): 432(1969).
8. Allegheny County Bureau of Air Pollution Control, Pittsburgh, Pa.
9. U.S. Bureau of Census, Census of Population and Housing, 1970, Census
Tracts, Final Report PHC-(1)-162, Pittsburgh, Pa. SMSA (1972).
10. Hospital Utilization Project, 400 Penn Center Blvd., Pittsburgh, Pa.
11. Eight Revision, International Classification of Diseases, U.S. Dept. HEW,
Public Health Service, Publication No. 1693, December 1968.
12. Consultation with Dr. Edward Haag, M.D., Human Effects Research Branch,
Environmental Protection AGency.
13. U.S. Bureau of Census, General Social and Economic Characteristics,
U.S. Summary, PC(1)-C1, Appendix B, pages APP-29-32 (1970).
14. B. MacMahon, T.F. Pugh, and 0. Ipsen, Epidemiolocpc Methods, Little,
Brown and Company, Boston, 1960.
15. G.P. Cressman, "An Operational Objective Analysis System," Mon. Wea. Rev.
87, 367-374 (1959).
46
-------
16. T.W. Sager, "Relating Spatial Distributions of Pollutants to Health
Effects", Proceedings 9th International Biometric Conference, Vol. II,
Biometric Society,Box 5962, Raleigh, N.C., p. 35-58, 1976.
17. T.D. Sterling, J.J. Phair, S.V. Pollack, D.A. Schunsky, I. DeGroot,
"Urban Morbidity and Air Pollution", Archives of Environmental Health,
13, August 1966, p. 158-170.
18. T.D. Sterling, S.V. Pollack, and J.J. Phair, "Urban Hospital Morbidity
and Air Pollution," Archives of Environmental Health, 15, September 1967,
p. 362-374.
19. T.D. Sterling, S.V. Pollack, and J. Weinham, "Measuring the Effects
of Air Pollution on Urban Morbidity," Archives of Environmental Health,
18, April 1969, p. 485-494.
47
-------
APPENDIX
48
-------
Table XVIII. Number of cases
hospitals in Al
Disease
Respiratory Diseases
Acute pharyngitis
Acute laryngitis and tracheitis
Acute upper resp. infect.
Acute bronchitis and brochiolitis
Influenza, unqualified
Influenza with pneumonia
Influenza w/other resp. manifestations
Influenza w/digestive manifestations
Influenza w/nervous manifestations
Viral pneumonia
Pnemococcal pneumonia
Other bacterial pneumonia
Pneumonia due to other specified organism
Acute interstitial pneumonia
Bronchopneumonia, unspecified
Pneumonia, unspecified
Bronchitis, unqualified
Chronic bronchitis
Emphysema
Asthma
Chronic pharynaitis & nasopharyngitis
Chronic sinusitis
Chronic laryngitis
Hay fever
Other diseases of upper resp. tract
Empyema
Pleurisy
Spontaneous pneumothorax
Pulmonary congestion & hypostasis
Pneumoconiosis due to silica & silicates
less children under 1 year
total data for the study
Suspect Circulatory System Diseases
Acute myocardial infarction
Other acute & subacute forms of ischeiiric
treated
leqheny
ICDA
-8 No
462
464
465
466
470
471
472
473
474
480
481
482
483
484
485
486
490
491
492
493
502
403
506
507
508
510
511
512
514
515
410
411
in 1972 by H.U.P.
County.
Number of Cases
Primary Diagnosis
201
185
558
1382
250
87
88
17
3
302
344
71
90
14
726
2201
647
430
775
1395
12
238
55
32
965
43
302
175
137
53
11778
228
11550
4126
1094
heart diseases
continued
49
-------
Table XVIII .(continued)
Disease
Chronic ischemic heart disease
Angina pectoris
Asymptomatic ischemis heart disease
Symptomatic heart disease
Other myocardial insufficiency
111 -defined heart disease
Transient cerebral ischemia
Acute but ill-defined cerebrovascular
disease
less children under 1 yr.
total data for the study
Control Circulatory System Diseases
Rheumatic fever w/o heart involvement
Rheumatic fever w/heart involvement
Diseases of pericardium
Diseases of mitral valve
Diseases of aortic valve
Diseases of mitral & aortic valve
Diseases of other endocardia! structures
Other heart diseases, -specified
as rheumatic
Malignant hypertension
Essential benign hypertension
Hypertensive heart disease
Hypertensive renal disease
Hypertensive heart & renal disease
Acute pericarditis, nonrheumatic
Acute & subacute endocarditis
Acute myocarditis
Chronic disease of pericardium, nonrheu.
Chronic disease of endocardium
Cordiomyopathy
Pulmonary heart disease
Subarachnoid hemorrhage
Cerebral hemorrhage
Occlusion of precerebral arteries
Cerebral thrombosis
Cerebral embolism
ICDA
-8 No.
412
413
414
427
428
429
435
436
390
391
393
394
395
396
397
398
400
401
402
403
404
420
421
422
423
424
425
426
430
431
432
433
434
Number of Cases
Primary Diagnosis
11308
219
6
2936
73
117
392
881
21152
19
21133
4
71
1
248
101
98
12
224
17
331
259
36
16
14
23
5
15
89
100
53
27
106
100
623
49
continued
50
-------
Table XVIII (continued)
ICDA Number of Cases
Disease -8 No. Primary Diagnosis
Acute but ill-defined cerebrovascular dis. 436 140
Generalized ischemic cerbrovascular dis. 437 462
Other & ill-defined cerebrovascular dis. 438 91
439 1
Arteriosclerosis 440 398
Aortic aneurysm (nonsyphilitic) 441 125
Other aneurysm 442 22
Ohter peripheral vascular dis. 443 46
Arterial embolism & thrombosis 444 170
Gangrene 445 93
Polyarteritis nodbsa & allied conditions 446 13
Other dis. of arteries & arterioles 447 19
Diseases of capillaries 448 7
Pulmonary embolism & infarction 450 303
Phlebitis & thrombophlebitis 451 193
Other venous embolism & thrombosis 453 11
Varicose veins of lower extremities 454 104
Hemorrhoids 455 104
Varicose veins of other sites 456 6
Nonifective dis. of lymphatic channels 457 1
Other diseases of circulatory system 458 51
Acute nephritis 580 69
5051
less children under 1 yr. 2
total data for the study 5049
51
-------
Table XIX
1972 ANNUAL AV:2?.AGE CONCZNT SAT IONS OF 502 ANE P AF
CENSUS
TRACT
101
102
201
202
301
302
303
304
a 01
402
i-03
404
405
406
407
4v,8
501
502
503
504
5G5
506
507
503
509
601
602
c 03
604
635
7 C 1
702
703
704
705
AT CFliSUS IF AC IS
UTM COCF
EASTING
584.575
586.040
584.962
586. C5C
566.040
585.560
586. 102
536.348
587.310
537.930
536.683
589. 110
588.490
588. 73C
566.735
587. 79C
536.664
537. 150
566.820
587.470
567.980
588. 172
568.750
587. 595
567.040
587.033
5d7.253
587.660
588.093
5S7.960
589.490
590. 130
59b. 380
590.615
590.613
IN ALLZGHZNY
DINATSS
NORTHING
4476. 695
4476.578
4477. 066
4478. 109
4477.547
4 '.'.76. 840
4477. 258
4476.891
4U76.797
4U77.035
4477.69 1
4477.520
U476.73 1
4476.547
Ha 76 .008
4476.402
4477. ^88
4477.72^
4476. 918
4477.258
4477.719
UU7B.4U 5
4478. 516
4476.096
4478.063
4479.031
a (i 7 8. 62 5
4U79.727
4L79 . 50"-
4478. 922
4478.059
4476.453
4476.199
4 ti 75. 88 3
4478.566
COUNTY, EA.
*
S02 PA
(FP3) (
31. 1
32.9
31. 1
33.2
33.5
31. 2
33.8
36.4
39. 1
39.1
39.0
39.2
39.2
39.3
39.7
39.3
35. 1
38. 3
38.9
39.0
38.9
30.9
31. 3
38.1
33.8
3C . 5
30.9
30. 1
30. 1
3'". 3
37.5
34. 2
35.3
30.9
32.3
TICULATS3
r: TIC HLAT^S
OG/.1**3)
87.9
87 .3
87.9
86.7
86.4
87.3
86 .3
36.0
1 o 0 . 7
101 .2
100.2
99. 5
1 C 1 . 1
1TC.9
1C" .2
101 .0
86. 2
98.5
93.6
K? . 8
1 C C . 5
86. 1
es. 9
98.0
n < . 4
36.9
s. 6 . 6
5^ . 1
So. 3
35,5
? = .':
85. 3
3C . ?
c 5 . 1
85.0
ppb x 2.667 = ug/M3
52
-------
lab] e_ XIX. Jc^n tinyed)_ _
1972 ANNUAL AVL-'AGL CONCENTRATIONS OF 502 AMD PAS7ICUL A T.^5
AT CENSUS TRACTS IN ALLEGHENY COUNTY, EA.
CENSUS
TFACT
706
707
708
601
602
803
804
805
8C6
807
608
9C1
902
903
1001
1002
1003
10C4
1005
1006
1007
1 101
1 102
1 103
1 104
1 105
1 106
1 107
1 108
1201
12C2
1203
1204
120-5
1206
UTM COCF
EASTING
591.085
591. U20
591.600
589.455
588.865
589. 190
589.530
589.260
589.873
590.3^6
590.760
588.070
538.382
588.735
568.955
590.050
590.460
590.890
589.870
ego c^r
589.706
590.910
591.410
591.690
590.597
59 1.480
591.81 0
591. 155
591.026
593.505
593.967
593.080
592. 326
591.992
592.290
DINATIS
N OF THING
U478.422
UU79.016
4478.6r5
4479. 633
4479.176
4478.547
4478.863
U479. 313
4479.215
4479.281
4479.21 1
"^80.648
U480. 137
4(t79.625
aaai .523
4461.938
^482. 020
4
-------
Table XIX (cpntinued) .
1972 ANNUAL ?.VE^AGF CCN CZNTF AT 10 K S OF ?02 A>TD PAFTICULAT?S
AT CENSUS TCACTS IN ALLZGKZMY COUNTY, Pi.
CENSUS
TFACT
1207
13C1
13C2
13C3
1304
1305
13C6
11C1
1402
1103
1 4 OU
1405
1406
1U07
1 U08
14C9
1410
1411
15C1
15C2
15C3
1 5C4
1505
15C6
15C7
16C1
16C2
U03
1604
16C5
17C1
1702
1703
1801
16G2
UTM COORDINATES
EASTING NOFTHING
592. 9CC
593.620
593.831
593. 543
59". 292
59". 720
595. 2C6
59C .25^
59 1. CSC
55 1. 554
592.012
592. 872
593. 3CO
590. 4CC
591.775
591.493
593.560
593.210
569. 63C
589.445
59C. 143
59C.99L
589.877
589.5^0
59C.U92
587.232
5o7.248
587.296
587. 9UC
586.657
564.79C
586. 355
586.C52
584.722
585.665
4^79.070
ii479.297
ua79.C27
1178.590
4478. 313
4U78. 9^5
4478.438
li477.227
iiti77.699
4477. 168
4ii78.C27
4^78.309
4477.59^
4476.3tiC
4476.227
4475. 32C
"476.250
447^. 96 1
U474.266
1173.21 9
4473.168
4473.297
4473. 238
4475. 184
4475.03 1
4175. 52C
UU7U.777
i:^7U. 195
H474.30S
4473.637
4476. COO
4475.570
4475. 02C
4U75.293
4474.89 1
S02 PAETICOLATiS
(PP3) (UG/.M.**3)
31.5
31 .9
31.7
30. 1
3C.3
31. 3
31.2
4G . 2
36.4
37.2
31.9
29.5
29.2
43.2
22.9
22. t
32.8
31.8
68.6
6C.C
61.3
63.5
61.3
77.3
76.3
45.5
49.2
5C.2
51 .1
47.8
31.2
42.2
"2.6
31.6
40. 5
83.8
83.7
83.9
84. 1
84.4
84. 1
84 .6
Si. 9
84.8
85,0
8 '4. 5
84 .4
84.8
85.2
85.4
85.6
BO. 5
6C. 3
89.fi
«9.o
P. 9. 7
89.0
89.3
85.2
85.5
36.6
8". 9
84 .7
an . 9
P :.. . 6
87. 9
a5.4
S5.2
P7.u
55. 3
ppb x 2.667 = ug/M3
54
-------
1972 ANNUAL A VS.TAG.:: CCNCINT F.ATIONS OF S02 ?.NT P AF. TICUL A Ti!
AT CENSUS IF?.CIS IN ALL2GHSUY COUNTY, EA.
C2HSUS
T 5 A CT
1 603
1304
18C5
1 606
1901
19C2
1903
1 904
1905
1906
19C7
19C8
1909
19 1C
2C01
20 02
20C3
2 C 04
2005
2CC6
20 07
2 COS
2009
2010
20 11
2012
2101
21C2
2103
2104
21C5
2201
2202
2203
22 04
IE
ppb x 2.667 =
UTM COCFDINATZS
EASTING
585. 250
58U.628
584.603
58fc. 800
582.757
583.505
583.600
583.926
583.852
582.921
582.691
583.622
583. C2C
583. 292
580.260
579.250
579. 33C
580.075
580. "7C
580.290
581.575
581.286
581. 970
581 .442
581.727
582.700
582.556
582. 650
582.695
582.010
582.797
583.406
583.723
584.260
564. 385
yg/M3
NORTHING
4^74.773
4474.574
447/1. K8
tfc73. 270
U476.379
1476.273
4475.887
4475. 480
wU74. 641
4473.875
H472.922
4472.527
4U71.9U5
4471. 231
4479.313
4U78.666
U478.27C
4 i' 78. 8 13
4476.668
4^78.0^7
4477.426
4476.953
tti77.059
4473.145
ti"76.0G8
4U75.629
M78.797
4479. 160
4478.441
4478.930
4477.953
4476.086
4478.598
4^78.703
U478. 270
W
£02*
(FEE)
37.4
32.7
32.3
32.2
29.7
31.0
31.0
31.2
31.3
30.«
31.4
31.3
31.7
31.4
21.4
12.3
14.7
20.4
21.2
24.2
24. t
23.8
26.0
28.7
26. 1
2Q.7
28.3
29.7
28.5
26. 2
29.0
3C.5
31.0
31.9
32.0
PAR TIC OLA TES
(OG/M**3)
85. U
85.5
85. 3
8U.8
86.7
P7.8
87.7
87.5
85.7
85.6
85.3
6U.8
84.7
84.3
86.9
76 .8
74.9
86.7
87.0
86.8
87.0
S6.9
86.9
85.4
96.6
86. 11
87. u
37.5
37.3
£7. u
87. 2
8^.2
87. u
87. t
87.4
-------
Table XIX (continued)
1972 ANNUAL AVERAGE CONCENTRATIONS OF ?G2 AND PAF.TICl'L ATSS
AT CENSUS '13. ACTS IN ALLEGHENY COUNTY, PA.
CENSUS
TF ACT
2205
2301
2302
2303
2401
24 J2
24C3
2 501
2502
2503
2504
2505
26C1
2602
2603
2604
2605
2605
26C7
26C8
2fC9
261C
2611
2701
2702
2703
2704
2705
27C6
2707
2801
28C2
2303
28C'4
2805
UTM COORDINATES
EASTING
584.210
584.956
585. COC
585. 140
585. 221
565.833
586. 525
583.473
583.225
533. 717
584. 260
584. 415
532.660
583. 190
583.050
583.760
583.737
584.380
583. 792
584. 24C
584. 40C
585. 320
583.730
531. 135
581.905
581. 555
581.67C
582. 05C
582. 290
582.245
578. 296
580.100
579.970
579,787
578.513
1IOETHING
4477.785
4^78.754
4478.445
4478.093
4479.391
4479. 27 C
4480.090
tit 79. 391
4478.99 2
4478.863
U479. 375
4478. 96 1
4482.453
4482.652
4^80.020
4479.852
4480.46^
4480. 105
4481. 531
4K82.152
4^80.809
4480.09«
4U83.25C
4431 .742
4481 .91 8
4480.930
4"79.797
4^80. 352
4 UPC . 3'', 8
4U79.668
4478.723
4477.566
4«;76.39B
4475. S33.
4 4 7t . 1 7 2
502 I
(FF3)
31. 1
32.3
32.5
31.8
32.3
32.5
32. 1
31.5
30. 8
31. 1
31.9
32. C
34.8
34.9
32.4
32.1
32.7
31.9
33.8
34. 1
32.6
32. 1
35.0
32.5
33. 1
31,2
27.3
3C.8
32.9
29.3
8.9
26.3
26.6
26 .4
22. C
?A? TICUL.&T2S
(UG/M**3)
87. 9
87.4
87. 2
87.5
87.5
87.4
87.5
87.6
87.5
87.4
87. 6
67 .5
38.9
39.0
87.3
87. -7
87.9
87.3
88."
83.7
88.1
8^.7
88.9
87 . 3
83. 3
83 .0
«7.5
87 .9
33.1
87.6
72.2
36.6
86. 2
95.!:
c2. 3
*ppb x 2.667 = ug/M3
56
-------
1972 ANNUAL AV35AGZ CON CE!i T?AT IONS OF SC2 AND P APTICU LAT E:
AT CINSUS TT.ACTS IN ALL-GHENY COUNTY, FA.
CENSUS
IE ACT
2806
29C1
2902
2903
3001
3101
3102
3103
3201
3202
3203
3204
4011
4012
40 13
4021
40 22
4031
4C32
4T33
4040
4050
4060
4070
"0 80
4C90
410C
41 10
4 120
4 131
4132
M33
41 34
4135
i» 141
ppb x 2.667
UTM COGF
EASTING
580.713
585.600
586. 167
586.333
585.420
590. 35C
592. 313
591.578
584.760
585. 432
564.570
564.620
608.690
607.570
607. 1^7
606. 390
606. 645
605.700
6 C 5 . " 8 0
60 u. 758
603. 385
602. 295
604.783
596. 126
588. 49^
56C.990
577.860
574.660
576. 395
581.396
580.181
581.223
584.055
564. 221
587.846
= yg/M3
DILATES
NCI-THING
""75.945
""72. 293
4472.336
"470.973
t"73. 938
""71 .520
""69. 168
'-471 .641
4470.590
""70. 992
4471. 762
4469.727
"499.520
""96. 53^
41-97. 375
4495. 980
""95. 449
at 95. 117
""95. "77
4495.004
4"93.563
Ui'93.672
"500.453
4498. 242
4*99.230
""98,633
4 n 98. 70 3
4499.887
4493. 261
4"93 .852
"492.599
4"90.188
4 "90. 125
4491 . S45
""93.926
*
S02
(PPB)
25.5
39.2
43.8
31,8
39.9
50."
38.5
^3.0
31.2
31 .8
31.2
31.2
2". 8
24. 0
24 .0
24. 0
2" . 1
2". 5
24. 5
24 .6
36.5
26.3
25.7
23.0
26.2
22. 1
25.0
26.0
26 .0
20 . 3
23.9
25.3
25.2
24.9
26.7
Pi\?.TICULAT:S
(UG/a**3)
86.5
84. 1
84 .0
8U.O
84.9
86. 3
87.3
86.7
83.2
83."
64. 1
82 .9
74.0
79.2
79.1
79.3
79.8
6D. 1
80. 1
30.3
80.7
81 .0
7S.2
83.3
"5. 4
411 ,7
'^5. 1
46.6
55.7
5C . 3
cb>3
55.4
62.2
54.5
U7. 3
57
-------
1972 ANNUAL AVZSAGE CONCENTKA TICKS CF SC2 AMD F AF.T ICU 1 AT ~Z2
;,T CFN'SUS TFAC1S IH ALLiGHESY COUNTY, PA.
CENSUS
TRACT
4142
4150
4 160
" 171
"172
4180
"1 90
"200
"211
4212
"220
4230
"241
4242
4243
4251
"252
4261
4262
"263
4264
4265
"266
4271
4272
"231
U282
"291
4292
4293
"29"
4295
4295
t'297
"311
ppb x 2.667 =
HTM COC
EASTING
588.420
593.662
601.985
602.950
603.070
601.410
598.761
596. 520
596.266
591.793
59". 046
592.943
591. 170
590.330
590. 05C
5 8 8 . 1 3 C
538.775
589.960
589.143
586. 826
588. 170
586.515
587.695
587.020
586. 945
586.273
585. 390
583.600
53". 865
560. 485
582. 302
583.050
580.635
581.765
582. 550
ug/M3
bDINATTS
rOFTHIN G
4490.598
4"90. 509
4 "89. 67 6
4438.641
4"83. 133
£iH68.572
""89. 125
4482.855
4483.695
"484.578
"485.855
4432. 863
4 "33. 020
4483. 195
U482.977
4"83. 129
4484.03 1
4 "63 .75 6
4467.766
""88.617
4"85.820
4 " 8 5 . 138
4^82.852
4431. 977
4481.230
""80.60 2
4461 .969
4484. 163
""87. 160
4487.809
"487. 14 1
"437. 387
4" 84. 03 5
"483.055
4485.063
58
*
SO 2
(PPB)
29.0
29. 1
33.0
30. 1
29.3
31.8
24. 1
22.2
19.2
21.8
15. 1
2". 4
26.0
26.0
26.0
26. 3
26.9
25.9
29.0
28.2
27.2
32.3
28.5
29.7
30.5
31.9
32.9
35.0
32.7
28.6
33. 1
31.4
32.7
34. 4
3". 3
PAFTICULATES
(UG/M** 3)
58. 9
81.0
64.3
7". 6
32.0
79.7
86.3
31.2
31.5
86.5
83.5
8". 6
8f .9
87. 7
87.8
38.3
69.2
38. '\
79.5
70.5
86. 9
37.1
?3, 5
88.2
87.9
37.3
33.5
87.7
79.6
76.2
79. o
78. 1
3 £ l>
38.3
35. 3
-------
Table XIX (continued)
1 972 ;\;-JNDAL AVIPAGI CONCENTRATIONS OF SO 2 AND PAF TICOLAT3S
.M C1NSUS T5ACTS IN A1LDGHENY COUNTY, E?.
CENSUS
TEACT
4302
4311
4312
4313
4314
4321
4322
4323
ii330
4340
4350
4360
4370
4380
439C
4t: Of,
44 10
(if: 20
4 a 30
uutic
4451
4452
4453
UU6G
4470
4"7C
4480
449C
45C1
45C2
45C3
4504
45^5
4511
^512
UTM COCt
CASTING
581 .4
-------
1972 ANNUAL AVIrtAG7: CCI1CE NT ?.AT IONS OF SO 2 AN C PARTICULARS
AT CF^ISUS IF.ACTS IS ALLZGHENY CC'JSTY, FA.
CENSUS
TSACT
4520
4530
4540
4550
4 5 60
4571
4572
45SO
4591
4592
4600
1-6 10
4621
4622
4623
4624
4625
4631
4632
4633
4634
*635
4636
4fc4i
"642
4651
4652
4653
4654
4655
4660
4670
4681
4662
t£63
ppb x 2.6,67 =
DTK COCF.DIFATES
EASTING
561.466
564.950
565. 580
569. C9C
572.409
575.213
575. 967
575.111
572.897
573.739
576.098
574. 276
578.492
578. 190
578. C3C
578. 60C
578.885
579.407
580.010
579.810
579. 36C
579. COO
578.427
579.440
578.690
578. 800
579.390
579.560
578.830
578.410
577.860
577. 25C
577. 336
576.790
577. 170
yg/M3"
MOFTHING
4479.633
4473.270
4U68.984
4472. 129
4467.277
4U67.977
4467.402
4471 ,090
"481. 547
4477. 105
4480.80 1
"464.586
4481 .51 6
4480.813
4480.512
4480. 145
4480 .488
"4 80. 410
4480.53 1
4 4 79. 68 C
4 4 79. 75 C
4 n 7 9 . 8 3 6
4"79.813
4477. 395
4477.346
4476.719
4476 .598
4475. 969
4476.105
4'i?5. 453
u"76. 1C5
4474.758
4473.824
4472.896
"472.750
fin
*
S02 PARTICULARS
(EP3)
28. 2
27.2
32.9
33.7
25.2
24. 1
23.7
27.3
25.5
13.6
19.8
25 .0
29.4
22.4
21 .3
21. 1
22.4
23. C
25.0
21.5
21.1
21. C
20. 1
25.5
16.7
23.9
26 . 6
26.5
26.3
19.2
17.9
19. 2
22.8
27.4
27.3
(UG/M**3)
68.0
61 . 9
59. 1
62.0
61 .7
64.4
65.0
68.7
78. 3
74 .4
82.8
93.7
92.4
P5.4
65.2
65.6
85. 9
36. 4
66 .?
86. 6
86.2
36.:
8,5 . 4
78.8
72. 2
73. 1
8r . 2
85.2
84. 1
52.°
8 2 . u
BO.*
79.3
76. 6
77.6
-------
1972 ANNUAL AV.3FAGL CONCZ MT 3 AI IONS OF S02.ANC ? ARTICUL A T"3
AT CENSUS TRACTS IN ALLFGHENY COUNTY, E.A.
CENSUS
TFACT
4684
^685
4686
a 6 90
4701
4702
4703
47C4
4710
4721
4722
4723
4724
^731
4732
4733
4734
t'735
"736
4741
4742
4751
4752
4753
4754
4761
4762
U771
"772
4773
U761
4782
4790
"8G1
Ji F02
UTM COORDINATES
EASTING
577.670
577.700
578.475
580.321
578.687
577. 143
578.036
577.6^8
577. 115
581.682
581.406
562.050
531.625
580.635
58C.756
5o2. 104
581.287
58C . 313
57$. 505
579.240
577. 383
561.910
580 .243
581.970
583.21-0
583.050
583.622
58U.540
585. 582
566.695
586.550
5S7.138
583.632
585.781
S87.8U-3
NOETHING
4U72.813
4473. 4 8 'I
HU72.91 8
4U7U. 105
4471.953
4U72. 172
4470. 145
1U68. 734
4471.438
4u72.324
«.' 4 7 1.72 3
4471 .551
4471.043
4471.367
"470.375
""70.090
4U68.809
aa68. 516
4469.895
U466.50H
4464.906
4466.344
^464.020
4463. 594
U464.266
^''68.84''
4468.41 8
4467. 6^8
4U68. 605
t 4 67. 66 8
4U69.676
44 6 9. 57 '4
"470. 254
u 4 6 6 . G 2 3
4467.43 3
S02* PA.-.TICULAT.IS
(FFB)
27.3
24.4
27.3
24.2
27.3
27.3
27.0
26. 9
27. 1
29.6
28.4
30.4
28.4
27.7
27.7
29.4
2S.2
27.7
27.0
25.3
20. 8
32.3
32. 1
32.2
32. 1
31.7
31.6
31.7
31 .3
32.7
31 .0
31.1
31. 3
32.0
35. r
(rjG/.i**3)
79.0
79.9
81. 2
85.3
80. 7
76.6
75.8
71.0
75.2
35. 1
3!'- .6
84. 8
84. u
83.9
83. 3
84. 1
83.0
RO. 6
80 .0
73. 3
53.3
SI. 7
72.3
75.3
79.1
d3. 3
82. 9
32. 6
'33.5
3". 7
« 4 . «
84.9
33. 5.
?3 .-
85.6
-*
ppb x 2.667 H-,, .. .,
b I
-------
1972 ANNUAL AVE5.AGE CONCENTRATIONS CF SC2 AND PAP.TICUI AT ES
AT CENSUS TRACTS IN ALLEGHENY COUNTY, FA.
CZKSUS UTK COORDINATES
T^.ACT
a&03
4804
48 11
4612
U613
4821
4822
4823
4831
4832
4833
4834
4635
4641
4842
ufct3
4844
< 8 4 5
4850
4861
4862
4863
U864
4865
4866
4570
4881
U882
4663
4884
4865
4886
a 8 90
4900
45 11
*
ppb x
ZASTING
587.315
588.287
586.016
586.096
585.710
591.230
592.030
592. 400
592.370
592. 470
592.700
592.935
592.410
593. 9SO
593. 160
593. 130
593,373
592. 973
594. 3C6
59b.80C
597.430
597. 360
597.990
597.950
597.360
594.064
594.438
595. 121
593.950
596. 390
590.017
592. 138
588. 177
584. 837
590. 173
2.667 = ug/M3
NORTHING
4 u 7 1 . 8 1 3
4470. 516
4474. 238
'1473.695
4473. 379
4472.047
U472.8B7
4471 . 590
4473.387
"473.008
at 73. 188
U472. Qb 8
4472.652
^72 .852
au73. 223
4472.50?
4471. 512
4470.477
4472.359
4^70 . 69 5
4469.672
4^69. 180
u 4 69. 379
4466.676
4468.563
4 4 6 7 . 0 1-' -1
Ut71. 363
4471.563
4^69.367
4U68.73C
au67. 84 8
4466. 1U5
4 4 6 5 . 1 '4 5
4u62. 109
4^60.719
CO
302
(PPB)
47.3
33.0
43.9
44.8
42. 3
49.2
4u. 8
42.9
39.2
42.3
40 .9
39.6
42.8
32.7
36.5
38.3
34. 9
36.2
29.3
22.6
38.5
38. 3
38.7
37.6
36. a
40. 3
27.7
23.2
31.7
33.5
40.5
40.0
36.3
32. 2
30.5
PAFTICULATSS
(UG/3--3)
84. 6
85.6
85.0
84.7
84.7
86. 5
86.5
87. 1
74. 2
36. 6
rt 6 . 6
86.8
86.7
87.2
71.8
87.0
87.5
87,7
87.6
93. 7
9U. 3
97.9
91 .7
99.3
107. 1
39.3
88. 1
83 .'>'
P3.5
95.7
73. 7
77.9
35.3
81.0
87.8
-------
1972 A MNUAL AVSFAGE CONCENT BAT IONS OF S02 ANE PARTICIPATES
AT CENSUS T.-AC1S IK ALLEGHENY COUNTY, EA.
CENSUS UT1 COORDINATES S02*
T?ACT EASTING NORTHING
4912
U921
U922
U923
4924
4925
<- 9 26
U927
U930
49"C
4950
4961
"962
^970
a 9 60
U991
"992
a 9 S3
5001
5002
50 10
5020
5C30
5041
5042
5C43
5050
5060
5C70
5081
5C82
5091
5 092
5093
5 1C 1
ppb x 2.667
592.222
593.830
594.330
595.457
594.795
593.875
594.410
595.015
596.060
596.796
596.483
598.506
601. 189
598. 240
597.175
594. 393
594.213
594.502
596.018
595.540
599. 125
601.640
601.217
599.912
601.167
603. 243
fa 04. 250
602.930
60 1. 15C
600.863
60 1.055
600.248
600. 100
59 9. 40 3
598.810
= yg/M3
4i'-62.C63
4462.602
4462.043
"461. 055
4 « 6 0.820
^460.320
4459.945
4460.336
445S. 438
4457.770
a45u.235
U457.906
UU61.949
4461.563
4^64.352
4^63.71 1
4464.367
U 1 6 ii . 9 3 C
4465. 3^0
4466.152
4463. 355
4461 .250
4466.^69
4 4 70. 941
4468.641
4471 .777
4470.605
UU71.9JC
4U7C.926
4472.367
ua71.723
4473.645
au72. 90 2
4U73. nil
4472.535
£3
36.8
31.6
38.7
43. 9
40.9
37.0
37.0
40. 1
35.8
36.6
35.6
36.4
25. 1
40.9
56. 1
27.8
28.5
36.9
55.2
41.5
2". 7
25.7
32.0
33.9
37.0
31.3
35.4
29.8
37.0
35.2
36. a
30 . 1
32.5
30.4
30.6
89. 5
83.1
93.4
96.5
96,4
Q0.5
90.8
94. 4
91 .6
92.0
«8.8
93. 3
95.5
93.ii
88.5
77.7
78. 3
85.7
88.5
90. 5
94 .2
95.9
95 . 1
91 .3
9U. 1
95 .7
97. 3
95. 1
93.0
91.9
92.5
90. 5
90.8
69.9
90. 1
-------
1972 ANNUAL AVi-AG?: CON CIN TBATICKS OF SC2 AND PART IC'J LAT **£
AT CEMSUS TFACTS IN ALLiGHSNY COUNTY, FA.
CENSUS
I?- ACT
51C2
51 1C
5121
5122
5123
5124
5125
5 126
5127
5131
5132
5133
5 134
5135
5UO
5142
5113
5 151
5152
5 153
515U
5155
5161
5162
5 17C
51 80
5 190
5200
5211
5212
5213
5214
52.15
5221
5222
UTK COORDINATES
SA STING
598.265
596. 49C
598.010
596. 260
c- 9 7 . C 1 C
596.550
596.790
596. 433
596. C70
595.950
595.760
5 96.400
596. 385
595. 79C
595.ii23
5S5. 13 C
594. 63C
594.876
594.917
594. 125
593.773
5S3.73C
594.137
594.672
595. 99C
597. 4UQ
597*. 510
599.662
603.672
601.902
6 T 3. 919
6C6.666
606.917
603.870
603. 630
NCrTHING
4472.223
4i:73. 638
4473. 199
4472.703
4472. 445
4 1 72. 93 3
4^73.250
4473.71 1
Uii73 . 73 8
4^73. 438
4473.227
4U72. 637
4472.203
4U72.828
4U73.727
4 c 73. 9^5
4u74. 195
4^74.637
4^75. 352
4474. 765
4475. 160
4475. 945
u 4 7 6 . 1 2 9
4!i76.023
4474. 833
Ut?u. 988
U477. 348
^475.285
ii^78. C27
U474. 89?
4U74. 598
i:474.73C
4477.602
4473.477
u. U72 . 9U 9
SC2* PAPTICUL
(PPB)
30.2
28.7
28.8
29.7
23.4
28.6
2?. 5
28.1
27.9
27.9
27.9
23. 5
22.6
28.1
27. 6
27.7
29.0
28.3
28.7
34.7
34.9
3^ . 4
29.9
29.4
28.3
27.5
34,0
26.7
25.2
25.7
25.6
26.4
26.2
26. 1
26 .4
(DG/M**;
89.9
89.2
89.2
89.7
89. C
88.5
Sb. 4
88. C
87.8
87.9
87.9
88. 6
88.8
88. 1
67.5
87. 2
86. 8
86. 7
86 . 4
61.9
61.1
65 .'i
35. 3
86. C
87. 0
37. 7
86 .0
39. 1
83.6
91 . U
93. 8
96. 'j
63. 1
95.1
95.
-------
1972 ANNUAL AVSF.AGi, CCriCSNTfiATICNS OF S02 ANT PARTICULATES
AT C2MSUS TF.AC1S IN ALLiGHSNY CCUNTY, FA.
CENSUS
T? ACT
5231
5232
5233
5234
5235
5236
5237
5238
5241
5242
5251
5252
5253
5261
5262
5263
5501
5502
55C3
5504
5505
55 06
5507
5513
55C9
55 10
55 11
5512
55 13
5514
5515
5515
5517
5513
56C1
UTK COORDINATES
EASTING
595. 159
596. 977
598.917
596.757
601.515
6 0 C . 5 1 C
60 1. 191
59b. 575
598. 065
598. 305
598.850
593. C25
596. 566
6C6.39C
605. 326
609. 380
595.480
595. 380
596. 5^5
596. 315
596. 615
596.620
597. 502
598. 6«0
598. 77C
599. 320
599.155
593.930
598. 225
597.790
c.97. 34C
598. 143
5 9- -3. 9 30
599. 260
596.437
NORTHING
4480. 141
4 479. 50 C
4480.512
4U78.906
4U8C.379
fcu78. 676
4483.828
4482.316
u 4 8 4 . 0 4 7
4^84.563
4485.477
^485. 473
4486. 418
4U86.004
44 61 .184
4480. 77C
4466.629
ti«i66. 977
4 467. 09 'A
4^66.547
4U66.512
"466 .633
4466. 926
4466.855
4^67.80 9
^467. 531
at; 66 .020
4^66. d59
^466. 145
4466.145
4465. 98
-------
!lbJe_XIX._(_contJ_nu.edl_
1972 ANNUAL AV^s.G^ CONG Si: T5A TIONS OF SC2 AND FAETICULAT "5
AT C2NSUS TIACTS IN ALLEGHENY COUNTY, FA.
CENSUS
I^ACT
5602
5603
560"
5605
56C6
DTK COC
ZASTING
595. 396
59U.695
59U.C92
59U. 026
59^. 740
:SDINAT3S
NOHTHING
UU77.793
UU77.73G
4U77.383
UU76. 75C
L\H 77 . cgt;
S02*
(PPB)
3C.8
30.1
29.5
29.2
29,8
PAP.TICTJLATES
(UG/M**3)
85.0
8U . 9
85.0
85. H
P5.3
56C-7 594.95C Uu77.379 3C. 1 85.2
5-XS 595.52C iia77.Jii8 3C.6 85.3
56C9 595.74C 4^76.828 30.5 85.8
5610 595.080 UU76.793 29.9 85.6
"pob x 2.667 = ug/M3
66
-------
TECHNICAL REPORT DATA
(Please read Inaructiuiis on the reverse before completing)
. REPORT NO.
EPA-600/5-77-006
4. TITLE AND SUBTITLE
HEALTH COSTS OF AIR POLLUTION DAMAGES
A Study of Hospitalization Costs
3. RECIPIENT'S ACCESSION" NO.
5. REPORT DATE
February 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Ben H. Carpenter, D.A. LeSourd, James R. Chromy, and
Walter D. Bach
8. PERFORMING ORGANIZATION REPORT NO.
RTI 41U-768
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, N.C. 27711
10. PROGRAM ELEMENT NO.
1AA601
11. CONTRACT/GRANT NO.
68-01-0427
12. SPONSORING AGENCY NAME AND ADDRESS
Health Effects Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park. N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
ABSTRACT
An investigation of the hospitalization costs of exposure to air pollution in
Allegheny County, Pennsylvania was conducted to determine whether persons exposed to
air pollution incurred higher incidences of hospitalization or additional costs for
treatment. A hospitalization data-base comprising 37,818 total admissions for
respiratory, suspect circulatory diseases, and control diseases was tested in a
cross-section type analysis for relationships between rates of hospitalization, length
of stay, and levels of air quality in the neighborhoods of patients' residence.
Air quality was identified using data from 49 monitoring stations. Corrections were
made in the analysis for race, age, sex.
Respiratory and suspect circulatory system disease showed statistically
significant increased hospitalization rates and lengths of stay for those exposed
to higher levels of S02 and particulates compared to those from neighborhoods
meeting air quality standards. At average costs per day for hospitalization in this
area in 1972, the total increased costs per day for hospitalization for the 1.6
million persons in the County was estimated at $9.8 million dollars ($9.1 million
for increased hospitalization rates and $0.7 million for increased length of stay).
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI lield'Group
Economic analysis
Benefit Cost Analysis
Air Pollution
Health
Economic Effects, Health
05 C
06 S
13. DISTRIBUTES STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
21. NO. OF PAGES
UNCLASSIFIED
76
20. SECURITY CLASS (Thispage!
22. PRICE
EPA Form 2220-1 (9-73)
67
-------
INSTRUCTIONS
1. REPORT NUMBER
Insert the EPA report number as it appears on the cover of the publication.
2. LEAVE BLANK
3. RECIPIENTS ACCESSION NUMBER
Reserved for use by each report recipient.
4. TITLE AND SUBTITLE
Title should indicate clearly and briefly the subject coverage of the report, and be displayed prominently. Set subtitle, if used, in smaller
type or otherwise subordinate it to main title. When a report is prepared in more than one volume, repeat the primary title, add volume
number and include subtitle for the specific title.
5. REPORT DATE
Each report shall carry a date indicating at least month and year. Indicate the basis on which it was selected (e.g., date of issue, date of
approval, date of preparation, etc,}.
6. PERFORMING ORGANIZATION CODE
Leave blank.
7. AUTHOR(S)
Give name(s) in conventional order (John R. Doe, J. Robert Doe, etc.). List author's affiliation if it differs from the performing organi-
zation.
8. PERFORMING ORGANIZATION REPORT NUMBER
Insert if performing organization wishes to assign this number.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Give name, street, city, state, and ZIP code. List no more than two levels of an organizational hirearchy.
10. PROGRAM ELEMENT NUMBER
Use the program element number under which the report was prepared. Subordinate numbers may be included in parentheses.
11. CONTRACT/GRANT NUMBER
Insert contract or grant number under which report was prepared.
12. SPONSORING AGENCY NAME AND ADDRESS
Include ZIP code.
13. TYPE OF REPORT AND PERIOD COVERED
Indicate interim final, etc., and if applicable, dates covered.
14. SPONSORING AGENCY CODE
Leave blank.
15. SUPPLEMENTARY NOTES
Enter information not included elsewhere but useful, such as: Prepared in cooperation with, Translation of. Presented at conference of.
To be published in, Supersedes, Supplements, etc.
16. ABSTRACT
Include a brief (200 words or less] factual summary of the most significant information contained in the report. If the report contains a
significant bibliography or literature survey, mention it here.
17. KEY WORDS AND DOCUMENT ANALYSIS
(a) DESCRIPTORS Select from the Thesaurus of Engineering and Scientific Terms the proper authorized terms that identify the major
concept of the research and are sufficiently specific and precise to be used as index entries for cataloging.
(b) IDENTIFIERS AND OPEN-ENDED TERMS Use identifiers for project names, code names, equipment designators, etc. Use open-
ended terms written in descriptor form for those subjects for which no descriptor exists.
(c) COSATI FIELD GROUP - Field and group assignments are to be taken from the 1965 COSATI Subject Category List. Since the ma-
jority of documents are multidisciplinary in nature, the Primary Field/Group assignment(s) will be specific discipline, area of human
endeavor, or type of physical object. The application(s) will be cross-referenced with secondary Field/Group assignments that will t'ollou
the primary posting(s).
18. DISTRIBUTION STATEMENT
Denote releasability to the public or limitation for reasons other than security for example "Release Unlimited." Cite any availability to
the public, with address and price.
19. & 20. SECURITY CLASSIFICATION
DO NOT submit classified reports to the National Technical Information service.
21. NUMBER OF PAGES
Insert the total number of pages, including this one and unnumbered pages, but exclude distribution list, if any.
22. PRICE
Insert the price set by the National Technical Information Service or the Government Printing Office, if known.
A Form 2220-1 (9-73) (Reverse)
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Technical Information Staff
Cincinnati, Ohio 45268
OFFICIAL BUSINESS
PENALTY FOR PRIVATE USE, $3OO
AN EQUAL OPPORTUNITY EMPLOYER
POSTAGE AND FEES PAID
US ENVIRONMENTAL PROTECTION AGENCY
EPA-335
Special Fourth-Class Rate
Book
\
UJ
O
/f yo
------- |