HUMAN BIOCHEMICAL AND PHYSIOLOGIC
RESPONSE TO ACUTE
PHOTOCHEMICAL AIR POLLUTION EXPOSURE
COPLEY INTERNATIONAL CORPORATION
-------
December 1978
HUMAN BIOCHEMICAL AND PHYSIOLOGIC
RESPONSE TO ACUTE
PHOTOCHEMICAL AIR POLLUTION EXPOSURE
by
R. David Flesh, Gordon H. Kubota
and Eric J. Solberg
COPLEY INTERNATIONAL CORPORATION
7817 Herschel Avenue
La Jolla, California 92037
Contributions were made to this report by
Katherine W. Wilson
Contract No. 68-02-1724
Project Officer
Carl G. Hayes
Population Studies Division
Health Effects Research Laboratory
U.S. ENVIRONMENTAL PROTECTION AGENCY
Research Triangle Park, North Carolina 27711
-------
ABSTRACT
The purpose of this study was to collect and analyze data
relating to the susceptibility of four separate and distinct
population groups to acute, high-level exposures of photochem-
ical oxidant. The groups (or panels) were composed of (1)
college cross-country runners, (2) adult asthmatics, (3) adult
bronchitics, and (4) healthy outdoor workers.
The data were collected during the fall smog season of 1974
in the Los Angeles communities of Covina, West Covina, Azusa, and
and Glendora, California. Panel members were evaluated during
periods of high, intermediate, and low oxidant pollution.
The methods employed in this study included measurements of
numerous biochemical and physiologic parameters as well as sub-
jective responses. Members of all four panels were given com-
prehensive clinical examinations before and after the panel ac-
tivities were conducted. The athletes and bronchitics were
subjected to a series of physiologic tests and discomfort symp-
tom interviews on each of 11 days when oxidant levels were
forecasted to be unusually high or low. The asthmatics and outdoor
workers were given pulmonary function tests and discomfort symp-
tom interviews daily for two weeks regardless of oxidant levels.
Of the data collected, the data from the outdoor worker
panel were most clearly related to air pollution. For example,
eye discomfort, chest discomfort, cough, and phlegm reported by
the panel were positively correlated with concentrations of
ozone. Pulmonary function test scores (maximum forced expira-
tory volume at one second) were negatively correlated with
ozone, but statistically significant only for nonsmokers in the
panel. Chi-square tests of association between the pulmonary
function test scores and the air pollution variables and single
variable linear regressions supported these results. Simple
linear probability regressions, simple LOGIT transformations,
and multivariate regressions of outdoor worker panel data showed
eye discomfort and shortness of breath to be significantly re-
lated to concentrations of ozone.
The analyses presented in this report are based on statis-
tical considerations. They do not specifically address the
medical or public health implications of what was found.
-ii-
-------
CONTENTS
Abstract ii
Figures vi
Tables . .vii
Acknowledgments xv
1. Introduction 1
2. Summary and Conclusions 2
Panel Recruitment and Testing 2
Data Analysis 3
Conclusions . 10
3. Panel Recruitment 12
Selection of Study Area 12
Athlete Panel 12
Bronchitis and Asthma Panels 16
Outdoor Worker Panel 21
4. Field Methodology—Testing Schedules 24
Athlete Panel Testing Schedule 24
Bronchitis Panel Testing Schedule 27
Asthma Panel Testing Schedule 29
Outdoor Worker Panel Testing Schedule 31
5. Field Methodology--Physiologic Test Methods 32
Lung Function Tests ..... 32
Heart Function 35
Miscellaneous 36
6. Raw Data Files 38
Physiologic Data 38
Questionnaires 42
Aerometric Data 42
. 7. Data Analysis--Asthma Panel 45
Statistical Description of the Asthma Panel .... 46
Correlation Analysis 52
Measures of Association Between Variables 58
Selected Multivariate Linear Probability
Regressions 66
Between Group Differences 70
Discomfort Symptoms Analyzed by Date of
Measurement 79
Multivariate Analysis of MAXFEV 85
Overall Summary of the Asthma Panel Data
Analysis 94
8. Data Analysis—Outdoor Worker Panel 96
Statistical Description of the Outdoor
Worker Panel 96
-iii-
-------
CONTENTS (continued)
Correlation Analysis 103
Measures of Association Between Variables 116
Selected Multivariate Linear Probability
Regressions 127
Between Group Differences 130
Discomfort Symptoms Analyzed by Date of
Measurement 138
Multivariate Analysis of MAXFEV 144
Overall Summary of the Outdoor Worker Panel
Data Analysis . .153
9. Data Analysis--Bronchitis Panel , 156
Statistical Description of the Bronchitis Panel . .156
Correlation Analysis 159
Measures of Association Between Variables 166
Selected Multivariate Linear Probability
Regressions 168
Discomfort Symptoms Analyzed by Date of
Measurement .168
Overall Summary of the Bronchitis Panel Data
Analysis 174
10. Data Analysis—Athlete Panel .175
Statistical Description of the Athlete Panel . . .175
Correlation Analysis 180
Measures of Association Between Variables 185
Paired Difference Test of Means of Pulmonary
Function Variables 189
Overall Summary of the Athlete Panel Data
Analysis 189
Appendices 192
A. Interviewer Instructions 192
B. Instructions to Panelists 206
C. Electrocardiograph Data Form 209
D. Electrocardiograph Variables 211
E. Blood Sample Form 216
F. Daily Symptom Record 218
G. Clinical Interview Questionnaire 221
H. Aerometric Data Base 230
I. Proportion of Asthma Panel Reporting Discomfort
Symptoms and Weighted Averages of Air Pollution
Levels Charted by Day Number 240
J. Proportion of Outdoor Worker Panel Reporting
Discomfort Symptoms and Weighted Averages of
Air Pollution Levels Charted by Day Number . . . .257
-iv-
-------
CONTENTS (continued)
K. Proportion of Bronchitis Panel Reporting
Discomfort Symptoms and Weighted Averages of
Air Pollution Levels Charted by Day Number . . . .275
L. Athlete Panel Data Collection Schedule 292
-v-
-------
FIGURES
Number Page
1 Study area boundaries for athlete and outdoor
worker panels 13
2 Study area boundaries for asthma and bronchitis
panels 14
-vi-
-------
TABLES
Number Page
1 Results of Asthma and Bronchitis Panels
Recruitment 18
2 Numbers and Ages of Asthmatics Identified
During Recruitment 19
3 Asthma Panel Size at Time of Recruitment and
Actual Testing ..... 19
4 Numbers and Ages of Bronchitics Identified
During Recruitment 20
5 Outdoor Worker Sub-Panels and Number of Panelists . 23
6 Athlete Panel Testing Schedule and Corresponding
Oxidant Levels 26
7 Bronchitis Panel Testing Schedule and
Corresponding Oxidant Levels 30
8 Composition of the Asthma Sub-Panels 47
9A Statistical Profile of Discomfort Symptom
Variables Measured During Asthma Panel
Surveillance 48
9B Statistical Profile of Pulmonary Function Variable
and Age and Height of Panelists Measured During
Asthma Panel Surveillance 50
9C Statistical Profile of Air Pollution Variables
Measured During Asthma Panel Surveillance .... 51
10 Pearson Correlation Coefficients for the Asthma
Panel 53
11 Correlation Matrix of Air Pollution Variables ... 55
12 Spearman Correlation Coefficients for the Asthma
Panel 56
-vii-
-------
TABLES (continued)
Number Page
13 Kendall Correlation Coefficients for the Asthma
Panel 57
14 Simple Regression Results with MAXFEV as the
Dependent Variable 59
15 Simple Linear Probability Regressions with Eye
Discomfort as the Dependent Variable ...... 64
16 Significant Linear Probability Regressions for
the Asthma Panel 65
17 Multivariate Linear Probability Functions for
Eye Discomfort and Headache--All Asthma
Panelists 67
18 Multivariate Linear Probability Functions for
Eye Discomfort and Headache--Asthma Panelists
Without a Cold 69
19 Sample Means and Standard Deviations for MAXFEV
of the Four Asthma Sub-Panels 73
20 Difference in Variances Test on Asthma Sub-Panels . 74
21 Difference in MAXFEV Means Test on Asthma Sub-
Panels Using Separate Variance Estimates .... 74
22 Difference in Variances Test on Asthma Sub-Panels:
Difference in MAXFEV for Panelists not Having
a Cold 76
23 Difference in Means Test on Asthma Sub-Panels:
Difference in MAXFEV for Panelists not Having
a Cold 76
24 Multiple LOGIT Probability Regressions with
Proportion of Eye Discomfort as the Dependent
Variable 83
25 Multiple LOGIT Probability Regressions with
Proportion of Headache as the Dependent
Variable 84
26A Hasselblad Regression of MAXFEV With OZONE 87
-viii-
-------
TABLES (continued)
Number Page
26B Hasselblad Regression on MAXFEV with CO 88
26C Hasselblad Regression on MAXFEV with N02 89
26D Hasselblad Regression on MAXFEV with NO 90
26E Hasselblad Regression on MAXFEV with All Air
Pollution Variables Measured During Daily
Surveillance of the Asthma Panel 91
.27 Multivariabe Regression of MAXFEV with Sub-Panel
(Group) Variables 92
28 Multivariate Regression of MAXFEV with Lagged Air
Pollution Variables 93
29 Composition of the Outdoor Worker Sub-Panels .... 97
30A Statistical Profile of Discomfort Symptom Variables
Measured During Outdoor Worker Surveillance ... 98
30B Statistical Profile of Smoking Variables Measured
During Outdoor Worker Panel Surveillance .... 99
30C Statistical Profile of Pulmonary Function Variables
and Age and Height of Panelists Measured During
Outdoor Worker Panel Surveillance 101
30D Statistical Profile of Air Pollution Variables
Measured During Outdoor Worker Panel
Surveillance 102
31 Pearson Correlation Coefficients for the Outdoor
Worker Panel 104
32 Spearman Correlation Coefficients for the Outdoor
Worker Panel 105
33 Kendall Correlation Coefficients for the Outdoor
Worker Panel 106
34 Pearson Correlation Coefficients for Smokers in
the Outdoor Worker Panel 108
35 Spearman Correlation Coefficients for Smokers in
the Outdoor Worker Panel 110
-ix-
-------
TABLES (continued)
Number Page
36 Kendall Correlation Coefficients for Smokers in
the Outdoor Worker Panel Ill
37 Pearson Correlation Coefficients for Nonsmokers
in the Outdoor Worker Panel 112
38 Spearman Correlation Coefficients for Nonsmokers
in the Outdoor Worker Panel 114
39 Kendall Correlation Coefficients for Nonsmokers
in the Outdoor Worker Panel 115
40 Simple Regressions with MAXFEV as Dependent
Variable for Smokers 117
41 Simple Regressions with MAXFEV as Dependent
Variable for Nonsmokers 117
42 Simple Regressions with FEVl.Q as Dependent
Variable for Smokers 118
43 Simple Regressions with FEVip0 as Dependent
Variable for Nonsmokers . ' 118
44A Contingency Table for MAXFEV and OZONE for the
Outdoor Worker Panel .... 120
44B Contingency Table for MAXFEV and CO for the
Outdoor Worker Panel 121
44C Contingency Table for MAXFEV and N02 for the
Outdoor Worker Panel 122
44D Contingency Table for MAXFEV and NO for the
Outdoor Worker Panel 123
45 Simple Linear Probability Regressions for the
Entire Outdoor Worker Panel 124
46 Simple Linear Probability Regressions for the
Smokers 125
47 Simple Linear Probability Regressions for the
Nonsmokers 126
-x-
-------
TABLES (continued)
Number Page
48 Multivariate Linear Probability Functions for
Eye, Throat, and Chest Discomfort, Shortness
of Breath, Cough, and Phlegm--All Outdoor
Worker Panelists . .' 129
49 Breakdown of the Four Outdoor Worker Sub-Panels
by Absolute and Relative Frequency by
Observations 130
50 Sample Means and Standard Deviations for MAXFEV
of the Four Outdoor Worker Sub-Panels 132
51 Differences in MAXFEV Variable Test on Outdoor
Worker Sub-Panels 132
52 Difference in MAXFEV Means Test on Outdoor Worker
Sub-Panels Using Separate Variance Estimates . . 133
53 Differences in DIFFEV Variances Test on Outdoor
Worker Sub-Panels 134
54 Difference in DIFFEV Mean Test on Outdoor Worker
Sub-Panels 134
55 Difference in FEVi.Q/FVC Variance Test on Outdoor
Worker Sub-Panels . . , 136
56 Difference in FEVi.o/FVC Means Test on Outdoor
Worker Sub-Panels 136
57 Difference in Variances and Means Tests Between
Nonsmokers (NS) and Smokers (S) on Outdoor
Worker Sub-Panels . 137
58 Significant Simple Regressions of Discomfort
Symptom Proportions 139
59 Significant LOGIT Regressions of Discomfort
Symptom Probabilities . 140
60 Multivariate Regressions of Proportions of
Discomfort Symptoms 142
61 Multivariate Regressions of LOGIT Discomfort
Symptoms 143
-Xi-;
-------
TABLES (continued)
Number Page
62 Multivariate LOGIT Regressions Corrected for
Heteroscedasticity: Outdoor Workers 145
63 Hasselblad Regression for MAXFEV with Air
Pollution Variables 147
64 Multivariate Regressions for MAXFEV and FEVi.Q/
FVC--Smokers and Nonsmokers in Outdoor
Worker Panel 148
65 Multivariate Regressions for MAXFEV and FEVi.Q/
FVC--Smokers and Nonsmokers with Sub-Panel 3
Excluded 150
66 Multivariate Regression on MAXFEV and FEVi.o/
FVC--Smokers and Nonsmokers with Dummy
Variables for Sub-Panels 151
67 Multivarite Regressions of MAXFEV with One-Day
Lagged Air Pollution Variables . 152
68 Composition of the Bronchitis Panel ........ 156
69A Statistical Profile of Discomfort Symptoms Reported
by Bronchitis Panelists on 11 Testing Days . . . 158
69B Statistical Profile of Pulmonary Function Variables
and Age and Height of Bronchitis Panelists
Measured on 11 Testing Days 160
6.9C Statistical Profile of Air Pollution Variables
Measured on 11 Bronchitis Panel Testing Days . . 161
70 Pearson Correlation Coefficients for the
Bronchitis Panel 162
71 Spearman Correlation Coefficients for the
Bronchitis Panel 164
72 Kendall Correlation Coefficients for the
Bronchitis Panel 165
73 Simple Linear Probability Regressions for the
Bronchitis Panel 167
74 Multivariate Linear Probability Regressions for
the Bronchitis Panel 169
-xii-
-------
TABLES (continued)
Number Page
75 Significant Simple and LOGIT Regressions of
Discomfort Symptom Proportions .171
76 Multivariate Linear Probability Regressions with
Proportions of Eye Discomfort, Chest
Discomfort, and Shortness of Breath as
Dependent Variables . 172
77 Multivariate LOGIT Probability Regressions with
Proportions of Eye Discomfort, Chest
Discomfort, and Shortness of Breath as
Dependent Variables 173
78 Composition of the Athlete Panel 175
79A Statistical Profile of Discomfort Symptoms
Reported by Athlete Panelists on 11 Testing
Days 177
79B Statistical Profile of Pulmonary Function
Variables and Age and Height of Athlete
Panelists Before and After Running on 11
Days 178
79C Statistical Profile of Distances Run by Athlete
Panelists on 11 Testing Days 179
79D Statistical Profile of Air Pollution Variables
Measured on 11 Athlete Panel Testing Days .... 179
80 Pearson Correlation Coefficients for the Athlete
Panel Before Running 181
81 Pearson Correlation Coefficients for the Athlete
Panel After Running 182
82 Spearman Correlation Coefficients for the Athlete
Panel Before Running 183
83 Kendall Correlation Coefficients for the Athlete
Panel ^Before Running 184
84 Spearman Correlation Coefficients for the Athlete
Panel After Running 186
-Kill-
-------
TABLES (continued)
Number Panel
85 Kendall Correlation Coefficients for the Athlete
Panel After Running 187
86 Paired Difference Test of Mean FVC and FEV Before
and After Running 190
-xiv-
-------
ACKNOWLEDGMENTS
Copley International Corporation (CIC) is grateful to the
many private citizens, students, and public employees of the com-
munities of Covina, West Covina, Azusa, and Glendora, California
who gave generously of their time by voluntarily participating in
clinical tests and personal interviews. Without their coopera-
tion, this study would not have been possible.
CIC was assisted by several technical groups throughout the
data collection phase of the study. The Lung Association of Los
Angeles County performed all testing of the athlete and bronchi-
tis panels and the detailed clinical studies of the asthma and
outdoor worker panel members. Much of the data tabulation was
also performed by the Lung Association. This work was directed
by Dr. Stanley N. Rokaw, Medical Director, and Dr. A. Lloyd
Anderson, Director of the Lung Association's Breathmobile Pro-
ject. Drs. Rokaw and Anderson were assisted by Mary Jo
Masteller, Project Nurse, and Edward L. Otoupalik, Chief Techni-
cian of the Breathmobile.
Recruitment and daily testing of the outdoor workers panel
was carried out by the Statewide Air Pollution Research Center
of the University of California, Riverside. Dr. C. Ray Thompson,
Research Biochemist, directed the work. He was assisted by
Robert M. Bullock and Maida Rockliff.
Forecasts of oxidant levels were provided by Dr. George
Lauer, Manager of Advanced Programs, Rockwell International Air
Monitoring Center. Air quality data were obtained from monitor-
ing stations in the communities of Covina and Glendora,
California. The monitoring stations were operated by Rockwell
International under contract with the U.S. Environmental Protec-
tion Agency (EPA). Additional air quality data were obtained
from Los Angeles County Air Pollution Control District and the
California Air Resources Board.
Recruitment of the athlete, asthma, and bronchitis panels
was done by Copley International Corporation. Daily testing of
asthma panel members also was the responsibility of CIC.
Dr. Katherine W. Wilson, Director of Air Quality Studies, di-
rected this work. She was assisted by Dorothy A. Vincent,
Beatrice L. Wicks, and Joyce G. Revlett, Project Coordinator.
-xv-
-------
The data analysis phase of the study was directed by Drs.
Gordon H. Kubota, Director of Quantitative Methods, and Eric J.
Solberg, Senior Statistician at CIC. Dr. Kubota was assisted by
Dr. Michael D. Lebowitz, Project Consultant in Epidemiology and
Biostatistics, Charles V. Wagner, Manager of Technical Services
at Copley Computer Services, Inc., and Elizabeth L. Lakinger of
the CIC statistical staff.
Copley International Corporation is especially grateful to
Drs. Carl G. Hayes, EPA Project Officer, Robert S. Chapman, and
Victor H. Hasselblad of the Health Effects Research Laboratory
for their support and suggestions during the data collection and
data analysis phases of the study.
The overall responsibility of the study rests with R. David
Flesh, Group Director of Environmental Sciences, at CIC. Ques-
tions concerning the methods employed in performing the work or
the results achieved should be directed to Mr. Flesh.
-xvi-
-------
SECTION 1
INTRODUCTION
The purpose of this study was to collect and analyze data
relating to the susceptibility of certain population groups to
acute, high-level exposures of photochemical oxidant. Four popu-
lation groups (or panels) were examined:
1. Trained runners, who may be most resistant to
adverse health effects, but who may show phys-
iologic or biochemical impairments when phys-
ically stressed.
2. Documented adult asthmatics, who may be un-
usually sensitive.
3. Chronic bronchitics, who may also exhibit above
average sensitivity.
4. Healthy outdoor workers, who because of their
constant outdoor exposure may respond to acute
oxidant pollution more readily than other
healthy segments of the population.
The data were collected during the fall smog season of 1974
in the Los Angeles communities of Covina, West Covina, Azusa, and
Glendora. Panel members were evaluated during periods of high,
intermediate, and low oxidant pollution.
Previous studies of health effects associated with acute
photochemical oxidant exposure have indicated increased eye irri-
tation and coughing at ambient concentrations above 200 yg/m^
(0.10 ppm). These studies employed data collection methods which
relied.heavily on judgements made by recruited subjects. The
knowledge gained was limited by the precision and reproducibility
of the subjective responses and the lack of objective measure-
ments .
The methods employed in this study included measurements of
numerous biochemical and physiologic parameters as well as sub-
jective responses. Data were obtained by means of lung function
tests, electrocardiograms, blood samples, nasal swabs, tear sam-
ples, and questionnaires. The methods employed and the results
obtained are provided in this report.
-1-
-------
SECTION 2
SUMMARY AND CONCLUSIONS
PANEL RECRUITMENT AND TESTING
A college cross-country team was recruited to serve as the
athlete panel for this study. Seventeen males agreed to partic-
ipate, although an average of only eight athletes appeared for
testing on each of the 11 testing days.
Door-to-door interviewing was used to locate and recruit
groups of females to participate on the bronchitis and asthma
panels. Fifty-four bronchitics were selected to participate.
All were in the 35 to 55 age bracket and all smoked cigarettes.
All reported symptoms of chronic bronchitis (defined as more than
50 days of cough and phlegm per year) during the recruitment in-
terviews . The bronchitis panel was tested on 12 days over a
period of seven weeks. Only 38 bronchitics attended testing ses-
tions regularly enough for sufficient test results to be entered
into the data file for analysis.
Sixty-two asthmatics were recruited. All were in the 21 to
50 age bracket and were nonsmokers. All reported active cases of
asthma (defined as 2 to 100 attacks per year with wheezing and
shortness of breath) during the^recruitment interviews. The
asthma panel was divided into four sub-panels which were tested
on four successive two-week periods. Only 41 asthmatics partic-
ipated throughout their assigned two-week periods.
Post Office employees and city workers were recruited for
the outdoor worker panel. All were males in the 21 to 50 age
bracket. Slightly less than half were smokers. Like the asthma
panel, the outdoor worker panel was divided into four sub-panels
which were tested on four successive two-week periods. Although
95 workers were selected for the panel, the analysis covers only
85 workers for whom there were sufficient data and no indication
of preexisting lung disease.
Members of all four panels were given comprehensive clinical
examinations before and after the panel activities were con-
ducted. All were given clinical interviews at the time of the
second set of clinical examinations. These interviews replicated
the symptom questions asked during recruitment and gathered other
health and background information about the panelists.
-2-
-------
The athlete and bronchitis panels were subjected to a
series of physiologic tests and discomfort svmptom interviews on
each of the testing days. The athletes undertook the test.
both before and after they subjected themselves to stress, i.e.,
before and after they ran distances of at least two miles. The
testing days were scheduled based on forecasts of unusually
low or high photochemical air pollution (oxidant levels). The
asthmatics and outdoor workers were given pulmonary function
tests and discomfort symptom interviews daily for two weeks re-
gardless of oxidant levels.
DATA ANALYSIS
Data from the four panels were analyzed in decreasing order
of the amount of information gained. The asthma and outdoor
worker panels provided the most information. The asthmatics
were expected to be the most sensitive of these two groups;
thus, the asthma panel data were analyzed first. A summary of
the results of the asthma panel is presented first in this re-
port. The analysis of the outdoor worker panel is presented
second. The order of the two remaining panels was suggested
mostly by the number of observations available for analysis. A
summary of the results of the bronchitis panel is presented
third. The analysis of the athlete panel is presented last.
Of the data collected, the data from the outdoor worker
panel were most clearly related to the air pollution levels. A
likely explanation of this finding is that panel members were
breathing outdoor air for several hours while engaged in physi-
cal work before they reported for testing. Daily testing was
conducted in the afternoon, at the end of the panel members' work
shifts. The asthmatics may have been the most sensitive panel,
but for a variety of reasons such as placing self-restrictions
on activities and remaining indoors during periods of the day
when air pollution levels could be expected to be high, the re-
sults of analyzing the data from this panel were not as signifi-
cant. The number of observations obtained from members of the
bronchitis and athlete panels were far less than those obtained
from the asthma and outdoor worker panels. The bronchitics,
like the asthmatics, probably were careful to avoid strenuous
activities during periods of the day when air pollution levels
were high.
Very few significant results were obtained from the athlete
panel. One explanation for this outcome is that the ambient
concentrations of air pollution on the days when the panel was
tested were not high enough to cause measurable biochemical and
physiologic responses in the panel members. An equally plaus-
ible explanation, however, is that the data collection was
centered on the wrong variables and/or was conducted under the
wrong circumstances.
-3-
-------
It was probably a mistake to collect data during practice
instead of competition and to depend entirely on health data in-
stead of both health and performance data to assess the impact
of air pollution. A study conducted by other investigators in
which the performance measurements of cross-country runners were
used and relationships between air pollution and performance
were found is discussed in the overall summary of Section 10
(see Pages 189 and 191). By collecting data during practice in-
stead of competition as was done in the present study, there was
no assurance that the panel members achieved maximal exercise on
any of the testing days. In addition, the runners' times were
not consistently recorded in the present study, so an attempt to
relate concentrations of air pollution to performance was not
possible.
Other factors which likely reduced the possibility of ob-
serving any effects of air pollution on panel members were the
following:
• Concentrations of photochemical oxidant were not
particularly high on the days when panel data
were collected.
• Bronchitis and athlete panel members had to be
scheduled for physiologic tests one at a time
due to a lack of duplicate sets of test equip-
ment. Consequently, the tests had to be sched-
uled from mid-morning (hours before photochemical
oxidant peaked in the study area) to late after-
noon (hours after the peaks had occurred). Such
peaks usually occurred around 1 p.m. Not all
panel members could be tested during the hours
of peak exposure on any testing day.
Three types of variables were included (or at least examined
for usefulness) in the analysis of each panel: discomfort symp-
tom variables, physiologic test variables, and air pollution var-
iables. The discomfort variables were qualitative in nature,
with panelists answering either "Yes" or "No" to questions about
symptoms. The most useful of the physiologic test variables were
the pulmonary function variables, maximum forced expiratory vol-
ume at one second (maximum FEV^ Q or MAXFEV) and maximum forced
vital capacity (maximum FVC or MAXFVC). The term "air pollution
variables" is used to cover the weather variables, relative hu-
midity and temperature, as well as the air pollutants measured at
the times the panel data were collected. Data on ozone (repre-
senting photochemical oxidant), nitrogen dioxide, nitric oxide,
and carbon monoxide were obtained from the Los Angeles County Air
Pollution Control District and used in the analysis. An explana-
tion of why certain air pollutants were included in the analysis,
while others were not included is given in Section 6.
-4-
-------
The study was primarily concerned with human response to
photochemical oxidant. Consequently, the emphasis in presenting
the results is on the relationships found between the human re-
sponse variables and ozone.
Asthma Panel Results
Pearson and non-parametric (Spearman and Kendall) correla-
tion analyses were performed on all four panels. For the asthma
panel, the results showed that only eye discomfort (reported at
the time of the daily interviews) was consistently and signifi-
cantly correlated with the air pollution variables, especially
with ozone. Pulmonary function variable MAXFEV was not directly
correlated with the air pollution variables.
Three methods of association between variables were also ap-
plied to the asthma panel data. Simple linear regressions of
MAXFEV indicated a significant relationship with relative humid-
ity and temperature, but not with the air pollutants measured.
Use of contingency tables showed dependence of having a cold and
reporting throat discomfort, headache, nausea, or other discom-
fort (not specified) at the time of the interview or headache or
cough that day. There was no evidence of dependency between
having a cold and eye discomfort or chest discomfort at the time
of the interview or having a cough that day. Contingency tables
also showed no strong evidence between MAXFEV and the air pollu-
tion variables.
Simple linear probability regressions indicated that eye
discomfort was significantly related to ozone. Selected multi-
variate linear probability regressions also indicated that eye
discomfort was significantly related to ozone.
A sensitivity analysis for asthma panelists having or not
having a cold was performed to test whether there was any statis-
tically significant reaction to air pollution. The subset of
asthmatics having a cold was examined first. Those not having
a cold were examined second.
For the subset having a cold, two primary differences with
respect to the whole panel were found. First, the correlations
were generally larger, but not significant. Second, the correla-
tion between cough and the concentration of ozone was large and
significant, but the correlation was negative. Scattergrams did
not indicate any obvious nonlinear relationships.
Comparison of linear probability regressions for the subset
having a cold to the linear probability regressions of the whole
asthma panel led to an observation that the percentage of varia-
tion explained was much larger for the restricted sample. The
asthmatics having a cold were more sensitive to ozone in experi-
encing eye discomfort and headache than the whole panel.
-5-
-------
Correlations and linear probability functions were estimated
for the asthma subset not having a cold. The estimates were al-
most identical to those computed for the unrestricted panel.
As stated earlier, the asthma panel was divided into four
sub-panels, which were tested over different two-week periods.
Differences between the sub-panels could have existed because of
differences in their composition due to small sample or self-
selection bias. A series of tests was performed to determine
whether such differences existed which could not be controlled
statistically. In these tests, the dependent variable was either
MAXFEV or MAXFEV adjusted for age and height. Difference in var-
iance tests, difference in means tests, and a two-way analysis of
variance were the methods employed.
These methods indicated that Sub-panels 1, 2, and 3 were not
different from each other, but Sub-panel 4 was distinct. When
MAXFEV was adjusted for age and height, the conclusions remained
unaltered. Sub-panel 4 was not grouped with the other three sub-
panels in subsequent analysis. The evidence was strong, however,
for treating the other three sub-panels homogeneously.
The discomfort symptoms were analyzed by computing the pro-
portion of respondents reporting each discomfort symptom on each
of the 40 different test dates. The average and maximum values
of the air pollution variables were calculated for each date. A
LOGIT specification was then estimated using the discomfort symp-
tom as the dependent variable. This functional form is logically
consistent with the interpretation of the dependent variables as
a proportion since that variable was constrained to fall within
the unit interval. Eye discomfort was chosen for analysis be-
cause it showed greater correlation to the air pollution vari-
ables than the other symptoms. The primary conclusion of the
analysis was that the LOGIT provides more realistic predictions
at the extremes of the explanatory variables, but the linear '
probability specification provides reasonable predictions
throughout the observed ranges of the air pollution variables.
Multiple LOGIT regressions were estimated using eye discom-
fort and headache as alternative dependent variables. The multi-
LOGIT approach to estimating the probability of these variables
was a useful extension of the simple LOGIT specification. There
was no evidence of serial correlation when Sub-panel 4 was ex-
cluded or when dichotomous explanatory variables were included
to control for between group differences. Ozone was directly
related to eye discomfort.
Two multivariate regression models were used to explain var-
iations in MAXFEV, while controlling for differences between in-
dividuals. First, a model called the Hasselblad model was
estimated. That model uses dummy variables to control for dif-
ferences between individuals. Second, a standard linear
-6-
-------
regression model was estimated which used dummy variables to con-
trol for differences between sub-panels. In both cases, none of
the air pollution variables had estimated coefficients that were
statistically different from zero. The conclusion is that most
of the variation in MAXFEV was explained by differences between
the individuals in the asthma panel.
Multivariate regression was also applied which used lagged
air pollution variables as explanatory variables. None of the
lagged air pollution variables contributed to a reduction in the
explained variation in MAXFEV.
Outdoor Worker Panel Results
Correlation analyses were performed on the data of the out-
door worker panel. Both parametric and non-parametric correla-
tions were computed and tested for statistical significance.
MAXFEV was not significantly correlated with any of the air pol-
lution variables. However, chest discomfort, cough, and phlegm
were positively correlated with ozone. Eye discomfort again
showed the greatest senstivity to the air pollution variables; it
also was positively correlated with ozone.
The correlation analysis was replicated separately for
smokers and nonsmokers of the outdoor worker panel. There were
more significant correlations for smokers than for the outdoor
worker panel as a whole. Headache was not significant for the
entire panel but was positively correlated with ozone for
smokers. The total number of significant correlations for non-
smokers was smaller than for the entire panel. This was partic-
ularly true for the discomfort symptoms, shortness of breath,
cough, and phlegm. However, MAXFEV was negatively correlated
with ozone and statistically significant only for the nonsmokers.
The nonsmoking outdoor workers seemed to be more responsive in
terms of decreased lung function to the air pollution variables
than any other group examined. These differences between
smokers and nonsmokers in the outdooor workers panel indicated
that separate treatment of the two groups was necessary.
Single variable linear regressions were estimated with
MAXFEV or MAXFEV/MAXFVC as the dependent variable. Most of the
air pollution variables were significantly related to MAXFEV for
smokers. However, the sign of the coefficent for ozone was posi-
tive. For nonsmokers, only ozone was statistically significant
and the sign of the coefficient was negative.
Contingency tables were constructed and chi-square tests of
association were made between MAXFEV and the air pollution vari-
ables. These tests supported the hypothesis of dependence be-
tween MAXFEV and the air pollution variables for the outdoor
worker panel.
-7-
-------
Linear probability functions were estimated using the di-
chotomous discomfort variables as dependent and the air pollution
variables as explanatory. The relationship of eye discomfort
with ozone was stronger for smokers than for nonsmokers. Eye
discomfort, throat discomfort, chest discomfort, shortness of
breath, and cough were all directly related to ozone. There were
fewer significant linear probability regressions for nonsmokers.
However, discomfort was significantly related to all the air pol-
lution variables for the nonsmokers.
Multivariate linear probability regressions were estimated
for the whole outdoor worker panel. The data were restricted to
those outdoor workers who reported not having a cold and not tak-
ing medication. Those outdoor workers exhibited significant
response to the air pollution variables in reporting eye discom-
fort, chest discomfort, shortness of breath, cough, and phlegm.
The outdoor worker panel was divided into four sub-panels
prior to daily surveillance like the asthma panel. Tests were
performed to check for small sample or self-selection bias be-
tween the four sub-panels. Three different variables were ap-
plied in these tests: recorded MAXFEV; MAXFEV normalized by age,
height, and whether or not the individual was a smoker; and the
ratio MAXFEV/MAXFVC. Tests for differences in sample variances
and sample means were performed. The results indicated that only
Sub-panel 3 was.statistically different from the other three sub-
panels .
Difference in variances and means tests were also performed
between smokers and nonsmokers. The sample was restricted to
panelists not having a cold and not taking medication. The com-
parison of MAXFEV, MAXFVC, and the ratio MAXFEV/MAXFVC between
smokers and nonsmokers indicated there was a significant differ-
ence between the two groups. Smokers had lower recorded MAXFEV,
MAXFVC, and MAXFEV/MAXFVC scores than nonsmokers. This demon-
strated further that the influence of smoking was an important
attribute to consider in the analysis of the data.
The discomfort symptoms were analyzed by computing propor-
tions of the symptoms and the means of the air pollution vari-
ables for each of the 54 dates of surveillance. Single variable
and multivariate regressions were performed using both the linear
probability and LOGIT probability specifications.
The simple linear probability regressions resulted in
several statistically significant associations. Eye discomfort,
throat discomfort, shortness of breath, and cough were signifi-
cantly related to ozone. Simple LOGIT transformations resulted
in significant relationships between eye discomfort and ozone and
between shortness of breath and ozone. Multivariate regressions
resulted in only eye discomfort and shortness of breath having
-8-
-------
consistently significant estimates after a correction for hetero-
scedasticity was applied. And, only ozone remained statistically
significant as an explanatory variable.
Several multivariate linear regression models were applied
which used MAXFEV and MAXFEV/MAXFVC as dependent variables. The
Hasselblad model included dummy variables to control for indi-
vidual differences. The standard linear specification included
age and height or utilized the MAXFEV/MAXFVC transformation to
control for individual differences. A dummy variable technique
was also applied to control for differences between the four sub-
panels. In addition, separate regressions were estimated which
utilized lagged averages of the air pollution variables. The re-
sults of the alternative multivariate regressions showed that
differences in MAXFEV are explained quite well by individual dif-
ferences. But MAXFEV was not statistically sensitive to the air
pollution variables. Moreover, the application of lags resulted
in progressively weaker results the longer the lags.
Bronchitis Panel Results
Both parametric and non-parametric correlation coefficients
were computed for the bronchitis panel and subjected to statis-
tical tests of significance. The pulmonary function variables
MAXFEV and MAXFVC were not significantly correlated with any of
the air pollution variables for the bronchitis panel. Moreover,
when contingency tables were constructed and chi-square tests of
association were applied, this finding was reinforced. The cor-
relation analysis between the qualitative discomfort variables
and the air pollution variables revealed that eye discomfort was
significantly related to ozone.
Single variable and multivariate linear probability func-
tions were estimated for each of the discomfort symptom vari-
ables. The results from the single variable regressions were
generally consistent with those obtained from the correlation
analysis. The multivariate regressions showed no significant re-
lationships between the discomfort variables and the air pollu-
tion variables.
The proportion of panelists reporting each discomfort symp-
tom was computed for each of the 12 dates of testing. The aver-
age level of the air pollution variables was also computed for
each date. The LOGIT specification was applied to selected dis-
comfort symptom proportions. Eye discomfort was significantly
related to ozone, but not to any of the other air pollution vari-
ables. The remaining discomfort variables were not significantly
related to ozone.
Multivariate linear probability regressions using the same
proportions indicated that ozone was positively related to eye
-9-
-------
discomfort, chest discomfort, and shortness of breath. Multi-
variate LOGIT regressions produced the same results.
Athlete Panel Results
Correlation analysis was applied to the data of the athlete
panel. Two sets of correlations were computed, one set before
and a second set after the athletes experienced the stress re-
sulting from a practice run of at least two miles. Before run-
ning, eye discomfort and other discomfort (not specified) were
significantly correlated with ozone. After running, eye discom-
fort and headache were significantly correlated with ozone.
No significant relationships between MAXFEV and MAXFVC and
the air pollution variables were found either before or after
running. Correlation analysis, contingency tables, and scatter-
grams did not indicate any discernable association between the
pulmonary variables and the air pollution variables. Paired dif-
ference tests for the means of FVC before and after running and
the means of FEV]_o before and after running were made. No sig-
nificant decrease in either FVC or FEV]^Q occurred as a result of
running at least two miles.
Linear probability functions were estimated for the discom-
fort symptom variables. Eye discomfort before and after running
was the only discomfort symptom significantly related to ozone.
Further analysis of the athlete panel data did not seem
warranted, particularly in light of the small sample size.
CONCLUSIONS
The following major conclusions are supported by the
analyses:
• Smoking is an important factor to consider in an
analysis of the effects of air pollution on human
subjects.
• Outdoor workers are among the most sensitive
groups to photochemical oxidant when tested after
hours of exposure.
• Outdoor workers who smoke exhibit sensitivity to
air pollution differently from outdoor workers
who do not smoke.
• Eye discomfort is the most significantly related
symptom to photochemical oxidant.
-10-
-------
• The study gave little statistical support to the
sensitivity of asthmatics, bronchitics, or col-
lege long distance runners.
• Nontraditional regression methods such as LOGIT
provide greater analytical flexibility when con-
sidering qualitative discomfort symptoms.
• Throughout the panels, most of the variations in
MAXFEV can be explained by differences between
individuals.
• In analyzing panel data, it is important to con-
trol for differences between sub-panels.
• Lagged air pollution information did not signifi-
cantly contribute to explaining variations in
lung function.
-11-
-------
SECTION 3
PANEL RECRUITMENT
SELECTION OF STUDY AREA
The study was performed near EPA air monitoring stations in
Covina and Glendora. These communities experience high levels of
photochemical oxidant pollution and could be expected to provide
a range of oxidant levels from below 160 yg/m^ (0.08 ppm) to
above 480 yg/m^ (0.24 ppm). All panelists lived within six kilo-
meters of the Covina or Glendora monitoring stations as shown in
Figure 1. Oxidant levels at the panelists' residences were ex-
pected to be similar to those recorded at the monitoring
stations.
During^the data collection phase of this study, the EPA was
conducting environmental health studies unrelated to this project
in areas within 3.2 kilometers (2.0 miles) of the monitoring sta-
tions as shown in Figure 2. Residents of these areas who were
participating in the EPA studies (or who were eligible to partic-
ipate in them) were not recruited for this project. Specifi-
cally, asthmatics and bronchitics were recruited from outside the
two 3.2 km areas shown in Figure 2. This was done to maintain
total separation of the two study programs. However, outdoor
workers and athletes were recruited from the entire area within
6.0 km of the monitoring stations.
ATHLETE PANEL
The members of this panel were required to have the follow-
ing qualifications:
1. Be trained runners who run one mile or longer
distances.
2. Be in proper physical condition for distance
running at the beginning of the test program,
so that no training effect would be superim-
posed upon the results.
3. Practice during the afternoon when high oxidant
levels would normally occur.
-12-
-------
EPA Air Monitoring Stations
«L.A. County APCD Air Monitoring
Station
SCALE
Figure 1. Study area boundaries for athlete and outdoor worker panels.
0
1
0
1
1
(In miles)
1 1
2
1
3 4
1 1
(in kilometers)
-13-
-------
Figure 2. Study area boundaries for asthma and bronchitis panels.
Study Area
EPA Air Monitoring Stations
«L.A. County APCD Air Monitoring
Station
SCALE
(in miles)
012
I I I
(in kilometers)
-14-
-------
4. Be willing and able to give the time required
for the physiologic test-run-physiologic test
sequence which was expected to take approxi-
mately one hour each time it was administered.
In addition, it was expected that the panel had to be composed of
consenting adults, because they would be required to run during
periods for which the Los Angeles County Air Pollution Control
District had issued School Health Warnings. School Health Warn-
ings advocate that students from elementary grades through high
school be excused from strenuous indoor and outdoor activities
and could influence colleges to prevent enrolled minors from par-
ticipating in such activities also. These warnings are called
when instantaneous oxidant levels reach 0.35 ppm. The program
specified that a series of physiologic tests be given about 20
minutes before each running exercise and another series be given
about 20 minutes afterwards. Under an optimum schedule, athletes
could start the tesing program at 12 to 15 minute intervals, and
the desired panel of 20 athletes could be tested in four hours
and 40 minutes. This elapsed time was longer than the normal
practice session for most track teams; consequently, the size of
the athlete panel had to be reduced to a number that could be ac-
commodated in a three-hour period.
It was-considered essential for the testing program to be
conducted between late August and the end of October because high
oxidant levels are most probable at that time of year. It was
known that high school and college track teams do not train in an
organized way during the summer months, so a year-round joggers'
club seemed preferable. However, local joggers' clubs proved to
be made up of businessmen who jogged for less than an hour in the
early morning or later afternoon, and as a result could not op-
timally meet the requirements of the testing program. High
school athletes could not be used due to the School Health Warn-
ings mentioned previously. A continued search finally produced
an excellent possibility at a community college located in the
study area. A track team was found that closely met the require-
ments of the study. It was learned that if the athletes agreed
to participate, they would not be restricted by School Health
Warnings.
The athlete director and the track coach at Citrus College
in Azusa agreed to cooperate in the study and recommended the use
of the college cross-country team. The athletes were supposed to
train generally at their own convenience during the summer and
were expected to be in proper physical condition when the fall
semester began. Practice sessions were loosely structured but
usually took place in the afternoons. The track coach insisted
that the entire physiologic testing.schedule be prepared in ad-
vance and that no testing be done on days of track meets or on
days immediately preceding track meets. The athletes themselves
were interviewed when they reported for practice one week before
-15-
-------
fall classes began, and all agreed to participate. The athlete
panel was thus composed of all cross-country team athletes who
reported for practice on the days when physiologic testing,was
scheduled. Seventeen athletes participated, and attendance
ranged from 7 to 11 with an average daily attendance of 8. Con-
sequently, limiting the size of the panel, so that time would be
available to test all of them, proved unnecessary.
Although this cross-county team satisfied many of the re-
quirements of this study, there were several shortcomings. The
athletes as a whole were extremely independent and were accus-
tomed to modifying their practice schedules to suit their indi-
vidual situations. Apparently, this is a normal attitude for
the distance runners. It was not possible to get them to at-
tend the testing sessions when they felt that some other prac-
tice routine would improve their performance in an upcoming
meet. The track coach was interested in winning meets and did
not insist that his athletes attend the testing sessions. If a
similar panel is required for future studies, it might be advan-
tageous to consider other groups such as military recruits or
civil servants who are in good physical condition, but are not
restricted because of athletic competition.
BRONCHITIS AND ASTHMA PANELS
These panels are discussed together because the same tech-
niques were used to recruit both. The material presented in
Appendix A was used to recruit people who might have been el-
igible for asthma and bronchitis panel membership. Confirmation
of eligibility was based on clinical examinations and in-depth
interviews given later by the Lung Association at the Associa-
tion's Breathmobile. (The "Breathmobile" is a large, totally
enclosed trailer, equipped with physiologic test equipment.)
Panelists were required to live outside the 3.2 kilometer
areas shown in Figure 2, but within 6.0 km of the Glendora or
Covina air monitoring stations. Bronchitis panelists were re-
quired to be females between the ages of 35 and 55, to be cigar-
ette smokers, and to report themselves to have chronic bron-
chitis as evidenced by more than 50 days of cough and phlegm per
year (see Appendix A, Bronchitis Panel Questionnaire). A panel
of 50 qualified individuals was required. . Asthma panelists
were required to be females between the ages of 21 and 50, to be
nonsmokers, and to have active asthma (2 to 100 attacks the pre-
vious year) which had been diagnosed by a doctor (see Appendix
A, Asthma Panel Questionnaire). A panel of 60 qualified asth-
matics was required. Originally, both of these panels were
planned to have equal numbers of males and females, but the
quotas of males could not be filled. This was due primarily to
the fact that panelists were required to be available for test-
ing on weekday afternoons, when most males in the desired age
-16-
-------
categories were at work. Rather than use a very small number of
males which might impair statistical significance, the recruiting
requirements were changed and all-female panels were used."
Panelists were recruited primarily by door-to-door inter-
viewing. A list of 90 referrals was received from a group of
cooperating physicians; however, only 11 of the 90 lived in the
designated study area and none of the 11 met all of the require-
ments for panelists. Local hospitals were cooperative, but
supplied lists of individuals who were considered by the project
team to be too ill to participate in the study. Most panelists
were recruited by a systematic interviewing effort which covered
every residence in the study area. The interviewers sought to
determine whether the resident's self-reported medical condi-
tions, age, and sex qualified her for the panel and whether s.he
was willing and able to participate in the tests. No clinical
or socioeconomic information was collected during the door-to-
door interviews.
Each interviewer was supplied with the following materials:
• Interviewer instructions
• Map of the study area
• Interviewer record sheets including a sample that
was properly filled out
• Asthma panel questionnaires
• Memoranda to volunteers for-asthma panel
• Bronchitis panel questionnaires
• Memoranda .to volunteers for bronchitis panel
• Reminder memos to female bronchitis panelists
Copies of all these materials are included in Appendix A. Those
pertaining to the asthma panel were printed on blue paper and
those pertaining to the bronchitis panel were on yellow paper in
an attempt to avoid confusion. The memoranda, which contained a
summary of the study test program, were left with each volunteer
panelist to remind her of her agreement and to supply her with a
telephone number to call if she had any questions. In addition,
the memoranda served to remind the interviewers of the details
of each testing program.
Two teams of four interviewers each were used, and the re-
cruiting effort lasted for six weeks. Two interviewer team
leaders were responsible for dividing the study area among their
-17-
-------
'o
team members, and for assuring that all households were ap-
proached. The team leaders also made sure that interviews
within the city boundaries of Glendora and Covina were conducted
only by those interviewers who were licensed to work in those
communities. (No licenses were required in Azusa or in the un-
incorporated areas.) The overall results of the panel recruit-
ing effort are summarized in Table 1. In this tabulation any
conversation with a knowledgeable adult is counted as a response.
Second contacts were attempted only when a neighbor or friend of
the household suggested that it would be worthwhile. This ex-
plains the higher percentage of successful second contacts.
TABLE 1. RESULTS OF ASTHMA AND BRONCHITIS PANELS RECRUITMENT
Sequence of Contacts Number/Percentage
Doors approached 13,396
Individuals responding to first contact 5,497
Percentage of successful first contacts 41%
Second contacts attempted 1,286
Individuals responding to second contact 645
Percentage of successful second contacts 50%
Overall percentage of successful contacts 42%
The following table shows the numbers of reported asth-
matics who were identified during the recruitment. According
to these figures, one of every ten contacts resulted in the iden-
tification of an asthmatic, and 1 of every 37 contacts led to
the identification of an asthmatic in the 21 to 50 age bracket.
This is not to suggest that there is an asthmatic in one of
every ten households in the general population. Each person
contacted was asked for names of friends who might qualify for
the panel, and special efforts were made to follow up on all
these leads. Most asthmatics seemed eager to participate in the
study and made a genuine effort to be available to the inter-
viewer. A much lower number of asthmatics per household would
be determined by a survey designed to give results representing
the general population.
-18-
-------
TABLE 2. NUMBERS AND AGES OF ASTHMATICS IDENTIFIED
DURING RECRUITMENT
Category
Number Identified
Asthmatics, male and female, age 21 to 50
Asthmatics, male, age 21 to 50
Asthmatics, female, age 21 to 50
Asthmatics, male and female, under age 21
or over age 50
Total asthmatics identified
Total households successfully contacted
164
66
98
465
629
6,142
From the 98 female asthmatics in the 21 to 50 age bracket,
a panel of 62 was selected. In general, all nonsmokers with 2
to 100 attacks per year were enrolled if they stated they had
wheezing and shortness of breath with asthma attacks, if their
asthma had been diagnosed by a doctor, and if they were willing
and able to report for the scheduled tests. The panel was
divided into four sub-panels which were tested for four succes-
sive two-week periods. The panelist attrition rate increased
noticeably with the interval between recruitment and testing as
shown in Table 3 below, and several panelists withdrew before
the sub-panels were formed and testing began.
TABLE 3. ASTHMA PANEL SIZE AT TIME OF RECRUITMENT
AND ACTUAL TESTING
Number Originally Number Reporting
Sub-Panel Recruited for Testing
Time Between
Recruiting
and Testing
1
2
3
4
14
15
14
15
13
12
10
6
1 week
3 weeks
5 weeks
7 weeks
-19-
-------
Panelists dropped out for a variety of reasons such as
illness, moving out of the area, and changing jobs or working
hours, but very few refused without a specific explanation.
If a similar study is conducted in the future, it would be
advantageous to compensate for the expected attrition by
assigning more panelists to the sub-panels scheduled for the
last part of the program or by recruiting later for those
sub-panels.
Bronchitics were easier to recruit than asthmatics, and
54 qualified panelists were identified after four and one-half
weeks of interviewing (compared with six weeks for asthmatics).
Table 4 shows the numbers of reported bronchitics identified
during the recruitment. These figures cannot be used as an
indication of the distribution of bronchitics in the general
population. The decision to use an all-female panel was made
early in the recruiting program, and, after that time, no spe-
cific inquiries were made concerning male members of the house-
holds. Nearly all of the respondents were female, since the
interviewers were instructed to record information on male
bronchitics only when it was volunteered.
TABLE 4. NUMBERS AND AGES OF BRONCHITICS IDENTIFIED
DURING RECRUITMENT
Category Number Identified
Bronchitics, male and female, age 35 to 55 109
Bronchitics, male, age 35 to 55 27
Bronchitics, female, age 35 to 55 82
Bronchitics, male and female under age 35
or over age 55 318
Total bronchitics identified 427
Total households successfully contacted 5,640
Each of the bronchitis panelists were tested on 11 of the
12 testing days over a period of seven weeks. Attendance declin-
ed somewhat as the season progressed, but 38 out of the original
54 panelists attended regularly enough for sufficient - test results
to be entered into the data file for analysis.
Both the asthma and bronchitis panelists were required to
report to testing locations as far as four miles from their
-20-
-------
homes. At the time they were recruited, they were asked whether
they preferred to drive their own cars or have transportation
provided. About half of the panelists wished to be transported.
Some panelists—particularly the asthmatics—had young children
who usually accompanied them to the testing sessions. The test-
ing facilities were deliberately set up to provide a place for
children to play without disturbing the tests. This arrangement
was explained to the panelists at the time they were recruited.
OUTDOOR WORKER PANEL
The members of this panel were required to be healthy
males, aged 25 to 50 years, who lived and worked in or near the
communities of Covina, West Covina, Azusa, and Glendora, Cali-
fornia. The geographic area is shown in Figure 1. A panel of
80 workers was sought which was divided into four sub-panels of
20 workers each to be tested during four successive two-week
periods.
It was hoped that field workers from a large landscape
nursery located in the study area could be used, but suitable
arrangements could not be made. As an alternative, postal
letter carriers and outdoor municipal workers were recruited in
the four above-named communities. The postmasters (or their
assistants) were most cooperative and recruited the following
numbers of male letter carriers: Glendora, 23; Azusa, 12;
Covina, 11; and West Covina, 12. All were in the desired age
bracket. The city managers of Azusa, Covina, and West Covina
gave permission for panelists to be recruited from among their
outdoor city workers. The recruits were distributed as follows:
• Azusa 12 (3 each from parks, water, streets,
and electric power departments)
• Covina 15 (7 from refuse, 4 from water, and
4 from streets departments)
• West Covina 10 (from various city departments)
This gave a total of 95 panelists who were then divided into
four sub-panels.
Ideally, sub-panels should be balanced with equal distribu-
tions of ages, work tasks, and work locations. However, the
study protocol specified that the workers be tested at the end
of the working day, and they had also been promised that no
more than 10 or 15 minutes would be required. To accomplish
this, all the Glendora workers were, assigned to one sub-panel,
all Azusa workers another, and so on. Tests were conducted
simultaneously at two locations on the work premises — at post
offices and at city maintenance yards. This method of grouping
-21-
-------
workers into sub-panels was not ideal, but most workers were un-
willing to volunteer for a program which required them to drive
four or five miles to a central test location, and insufficient
equipment and personnel were available to test simultaneously at
all seven work locations. The final distribution of the sub-
panels is shown in Table 5.
The cooperating postmasters and city managers not only gave
their permission for the recruiting of panelists but also made
available space for conducting the daily tests. In addition,
they permitted employees to take time off from work for the
clinical examinations that were conducted before and after each
two-week testing session. It would have been difficult to con-
duct these examinations after normal working hours, and some
workers would have been reluctant to donate the necessary amount
of their own time.
The letter carriers and some of the maintenance workers
were on staggered shifts to permit them to cover Saturdays and
Sundays. In addition, the schedule of holidays was different
for Federal and city workers; in fact, holidays differed among
the three cities. Tests were required to be conducted at the
end of the working day, so they were appropriately scheduled
even to accommodate irregular work shifts. An attempt was made
to test each panelist ten times, and the testing period was ex-
tended until this was accomplished whenever possible. The ob-
stacles were not serious, but they should be taken into account
by those planning similar studies for the future when partici-
pants do not work conventional five-day weeks.
Similar studies which utilize employees in the private sec-
tor might have to be set up differently. For example, employers
may not wish to pay their employees while they participate in
the tests. However, they may allow their employees to have time
off without pay and allow the tests to be conducted in a van
parked on company premises. Under these circumstances, to com-
pensate for lost wages, it would be necessary to pay workers who
participate for the time they spend taking preliminary and post
study examinations and two weeks of pulmonary function tests.
-22-
-------
TABLE 5. OUTDOOR WORKER SUB-PANELS AND NUMBERS OF PANELISTS
i
N>
CO
I
Panel City Worker Category
1 Glendora Letter carriers
2 Azusa Letter carriers
Park maintenance
Water department
Street maintenance
Electric power
3 Covina Letter carriers
Refuse workers
Water maintenance
Street maintenance
4 West Covina Letter carriers
City mainentance
Number of Workers
Tested in Each
Category
23
12
3
3
3
3
11
7
4
4
12
10
Total Number of
Workers in Each
Sub -Panel
23
24
26
22
-------
SECTION 4
FIELD METHODOLOGY--TESTING SCHEDULES
The entire discussion of field methodology is divided into
two sections. The first section is an explanation of testing
schedules which were different for each of the four study
panels. The second section is a discussion of the physiologic
test methods. Many of the same tests were run on all panels, so
the second section is organized by tests rather than by panels.
ATHLETE PANEL TESTING SCHEDULE
The study protocol required that the following items of
data be obtained from each member of the athlete panel:
1. Before and After Running:
a. Single breath oxygen tests
b. Volume-time tracing of FVC maneuver
c. 12-lead electrocardiograms
d. Blood pressure
e. Heart rate
f. Nasal swabs for white cell counts (taken on
four days with the option of adding three
more days)
g. Questionnaire information on eye discomfort,
throat discomfort, chest discomfort, cough,
shortness of breath, and headache
h. Tear samples for lysozyme determinations
2. Only After Running
a. Running distances
b. Running times
3. On Five Occasions After Running
a. Peripheral venous blood samples for:
i. Total white cell counts
ii. Differential white cell counts
(including eosinophils)
iii. Immunoglobulin determinations
-24-
-------
4. On One Occasion Before Running:
a. Comprehensive clinical interview
The tests were conducted in the Lung Association's Breath-
mobile and in a supplementary trailer unit both of which were
parked on the Citrus College athletic field. As each athlete
reported for testing the review of discomfort symptoms for the
day of testing was completed. An electrocardiograph was secured
with the athlete in the supine position and pulse and blood
pressure were determined. Following this, the athlete entered
the Breathmobile where anthropometric measurements were made by
the technician. The nurse secured measurements of the tear
volume, employing the Schirmer strip and, at the completion of
this test, obtained swabs of nasal secretions for direct slide
streaking. The athlete then entered a second cubicle in the
Breathmobile where single-breath oxygen tests and forced expira-
tory (spirometry) tests were performed. The time of completion
of this series of tests was recorded.
The athletes then proceeded to their scheduled run, which
in most instances was 3.2 km (2.0 miles), with stopwatch con-
trol. After the athletes returned from their run, the clock
time was recorded and the athletes immediately underwent elec-
trocardiography, pulse and blood pressure measurement. The tear
test was repeated and additional nasal swabs were taken. The
review of discomfort symptoms was also repeated. On several
visits, but not all, a blood chemistry and blood count determin-
ation was secured by the nurse at this point in the routine.
The subjects again underwent single-breath oxygen tests and
forced expiratory (spirometry) tests. Finally, the time of com-
pletion of the second series of tests were recorded.
The study protocol supplied by the Environmental Protection
Agency required that tests be performed on five weekdays when »
the maximum hourly average oxidant was expected to be 160 yg/m
(0.08 ppm) or less, and on six weekdays when the level was ex-
pected to exceed 480 yg/nP (0.24 ppm). It was assumed that
tests should be conducted during the one or two hours of the day
when these oxidant levels persisted rather than being scheduled
at other times in the day. The athlete panel, however, consist-
ed of a college cross-country team in the midst of their competi-
tive season, and as a result, the coach gave the testing schedule
third priority behind cross-country meets and academic require-
ments. The actual test schedule and corresponding oxidant levels
are given in Table 6.
As can be seen in Table 6, high oxidant levels occurred
less frequently than predicted during the period when tests
could be scheduled. In 1974 the highest oxidant days occurred
early in the season--before the college semester began. Nasal
swabs and blood sampling were planned for 5 of the 12 testing
-25-
-------
TABLE 6. ATHLETE PANEL TESTING SCHEDULE AND CORRESPONDING OXIDANT LEVELS
I
ro
Date
9/9
9/10
9/12
9/19
9/26
10/3
10/10
10/17
10/24
10/29
11/6
Time of Day
9-11
9-11
1-3 p
1-3 p
1-3 p
1-3 p
1-3 p
Noon- 3
Noon- 3
Noon- 3
Noon- 3
a.m.
a.m.
.m.
.m.
.m.
.m.
.m.
p.m.
p.m.
p.m.
p.m.
Oxidant Levels in ppm
at Beginning and End
of Testing Periodt
0.02-0. 08 (no forecast)
0
0
0
0
. 0
0
0
0
0
0
.05-0
.17-0
.26-0
.10-0
.10-0
.09-0
.08-0
.09-0
.02-0
.04-0
.09
.18
.33
.13
.11
.10
.15
.10
.02
.05
(0
(0
(0
(0
(0
(0
(0
(0
(0
(0
.28)
.25)
.26)
.18)
.12)
.10)
.15)
.08)
.02)
.12)
No. of Athletes
Tested Each Day
7
10
7
11
7
9
5
8
9
7
7
Nasal Swab
Scheduled?
yes
yes
no
yes
no
no
yes
no
no
yes
no
Blood
Samples
Scheduled?
yes
yes
yes
no
no
yes
no
no
yes
no
no
tNumbers in parentheses indicate forecasted maximum hourly average.
-------
days as indicated on the schedule. Originally, this sampling
was planned for two low oxidant and three high oxidant days.
The schedule had to be based on forecasted oxidant levels be-
cause validated oxidant measurements from the EPA air monitoring
stations were not available until several months after the field
testing was completed. In many instances the actual oxidant
levels were lower than the forecasted levels, and only one of
the testing days turned out to be a truly high oxidant day.
The hourly schedule for each testing session was somewhat
loosely structured because the athletes refused to make definite
commitments. Generally, the testing facilities were ready to
receive the first athlete at least 15 minutes before the cross-
country workout was scheduled to begin. The athletes did not
report as a group for the workout but appeared individually over
a two to three hour period depending on their schedules or moti-
vations for the day. As soon as an athlete reported he was
given the first series of tests. He then ran two miles and
reported back for the second series of tests. Electrocardio-
grams could be run simultaneously on two individuals, so it was
rarely necessary to keep an athlete waiting to begin the tests.
The testing facilities remained open until the track coach indi-
cated that no more athletes were expected to appear on that
particular day.
Two athletes reported for all the testing sessions and were
evaluated as scheduled. For the other 15 athletes (who missed
from three to ten sessions each), make-up samples were collected
to compensate for absences on days when nasal swabs and blood
samples were scheduled.
BRONCHITIS PANEL TESTING SCHEDULE
The study work plan required that the following items of
data be obtained from each member of the bronchitis panel:
1. On Each Day of Testing:
a. Single-breath oxygen test
b. Volume-time tracing of FVC maneuver
c. 12-lead electrocardiograms
d. Blood pressure
e. Heart rate
f. Nasal swabs for white cell counts (taken on
four days with the option of adding three
more days)
g. Questionnaire information on eye discomfort,
throat discomfort, chest discomfort, cough,
shortness of breath, and headache
h. Tear samples for lysozyme determinations
-27-
-------
2. On Five Occasions During the Testing Period:
a. Peripheral venous blood samples for:
i. Total white cell counts
ii. Differential white cell counts
(including eosinophils)
iii. Immunoglobulin determinations
3. On One Occasion Upon Entry to the Program
a. Comprehensive clinical interview
The tests were conducted in the Lung Association's Breath-
mobile, which was parked in a central location. The panelists
were given appointments for their tests at ten minute intervals
between the hours of 11 a.m. and 2:50 p.m. and 4 p.m. to
7:50 p.m. The late time period was necessary to accommodate
panelists who worked during the day and could only come in the
evenings. Panelists were furnished transportation to the
Breathmobile unless they preferred to drive their own cars.
Each panelist was sent a letter explaining the scheduling of the
tests. A copy of this letter appears as Appendix B to this
report. When the panelist arrived at the Breathmobile, a
trained interviewer administered the questionnaire for symptoms
of discomfort pertaining to that day. An electrocardiograph was
then performed with the panelist in the supine position, and
pulse rate and blood pressure were recorded. Next, anthropo-
metric measurements were made. The nurse secured tear samples
and nasal swabs. On five of the visits to the unit, bronchitic
subjects also underwent blood sampling for WBC and differential
counts, and immunoglobulin determinations.
The study protocol required that tests be performed during
the week, on five days when the maximum hourly average oxidant
was expected to be 160 yg/m3 (0.08 ppm) or less, and on six days
when the level was expected to exceed 480 yg/m^ (0.24 ppm).
Panelists were notified of each test by telephone during the
afternoon or evening of the day preceeding the test. To accom-
plish this it was necessary to make an oxidant forecast at noon
for the maximum level to be reached in the study area on the
following day. Most routine forecasts, such as those made by
the Los Angeles County Air Pollution Control District, are made
at 9 a.m. and give the expected maximum oxidant level for that
day in an area which is considerably larger than the area of in-
terest in this study. Thus, the routine Los Angeles County
forecasts could not be used, and special forecasts had to be
made under these more demanding conditions. The situation was
complicated even further because real time or next day validated
oxidant data were not available from the EPA air monitoring
stations in the study area to use in checking the forecasts.
-28-
-------
Nonetheless, plans proceeded to choose testing days based on ox-
idant forecasts specially prepared for the project.
Tests for the bronchitis panel were scheduled only for
Tuesdays, Wednesdays, and Fridays. Mondays could not be used
because some of the meterological information on which the fore-
casts were based was not available on weekends. The test equip-
ment was used for the athlete panel on Thursdays. The actual
testing schedule and corresponding oxidant levels are given in
Table 7. A comparison of the forecasted and actual maximum
hourly oxidant levels shows that as in the case of the athlete
panel, the forecasts were only marginally helpful in selecting
appropriate testing days. In future studies of this type, it
would probably be just as effective to set up the entire testing
schedule in advance and not bother with forecasts and the ac-
companying problems of having to give panelists such short
notice for testing. Also, testing should probably be restricted
to three or four hours in the afternoon when high oxidant levels
are experienced. At the maximum testing rate of six persons per
hour, a panel of 18 to 24 could be tested in this key period.
If a larger panel is required, the test equipment could be du-
plicated to permit simultaneous testing of two persons, or per-
haps sub-panels could be tested on different days. Either of
these modifications would, of course, almost double the cost of
working with a bronchitis panel.
ASTHMA PANEL TESTING SCHEDULE
The study protocol required that the asthma panel be
divided into four groups. These groups were tested over suc-
cessive two-week periods, one group per period, during the
season of highest photochemical oxidant exposures. A panelist
was tested once every weekday over the two-week period of cover-
age for his group. The daily tests consisted of three measure-
ments with a Sted-Wells Spirometer of one-second forced expira-
tory volume (FEV^ Q) and a series of interview questions
relating to symptoms of discomfort. The testing equipment was
located in a central facility and panelists reported there at
assigned times. Transportation was provided unless the panel-
ists wished to drive their own cars. Tests were conducted
between the hours of 10 a.m. and 7 p.m. with as many as possible
being scheduled between 1 p.m. and 4 p.m. when oxidant levels
were expected to peak. Up to 12 persons per hour could be
tested and attempts were made to schedule 12 persons per hour
during the peak oxidant period. However, many panelists had
commitments that prevented them from reporting during the pre-
ferred testing period. If all 62 panelists could have reported
during the preferred testing period, additional equipment would
have been required to accommodate them.
-29-
-------
TABLE 7. BRONCHITIS PANEL TESTING SCHEDULE AND CORRESPONDING OXIDANT LEVELS
1
OJ
o
1
Date
9/4tt
9/5tt
9/18
9/24
9/25
10/1
10/9
10/15
10/16
10/18
10/22
10/25
Time of Day
Noon-7:30 p.m.
Noon-8:30 p.m.
11:30
11:30
11:00
11:30
11:00
11:00
11:00
11:00
10:00
11:30
a.m. -8:00
a.m. -7:30
a.m. -8:00
a.m. -7:30
a.m. -8:00
a.m. -8:00
a.m. -7:30
a.m. -8:00
a.m. -8:00
a.m. -7:30
p.m.
p.m.
p.m.
p.m.
p.m.
p.m.
p.m.
p .m.
p.m.
p.m.
Oxidant Levels in ppn
at Beginning and End
of Testing Periodt
0
0
0
0
0
0
0
' 0
0
0
0
0
.13-0.29
.20-0.34
.09-0.34
.01-0.17
.02-0.19
.01-0.19
.05-0.13
.05-0.13
.02-0.14
.01-0.15
.01-0.13
.01-0.14
(0
(0
(0
(0
(0
(0
(0
(0
(0
(0
(0
(0
.30)
.33)
.32)
.35)
.26)
.35)
.08)
.19)
.22)
.09) -
.07)
.12)
No. of
Bronchitics
Tested Each Day
22
19
33
37
• 38
31
29
30
29
29
34
31
Nasal Swab
Scheduled?
yes
yes
yes .
no
no
no
yes
no
yes
no
yes
yes
Blood
Samples
Scheduled?
yes
yes
yes
no
no
no
yes
• no
yes
no
yes
yes
tNumbers in parentheses indicate forecasted maximum hourly average.
ttHalf of'panel tested on each of these days; most of panel tested on remaining days.
-------
Detailed clinical examinations of each panelist were con-
ducted before and after her two-week examination period. The
examinations were conducted in the Lung Association's Breath-
mobile which was parked in a central location. Anthropometric
measurements and a pulmonary function evaluation consisting of
spirometry, single-breath oxygen test, and body plethysmography
were included. A questionnaire for symptoms of discomfort was
also completed at that time. The study protocol did not specify
that any particular oxidant levels should prevail during these
detailed clinical studies, .but to the extent possible, they were
conducted on low oxidant days so that, for baseline setting
purposes, health effects due to air pollution might be mini-
mized. A clinical interview questionnaire was administered
after the field studies were completed.
OUTDOOR WORKER PANEL TESTING SCHEDULE
The study work plan required that the outdoor worker panel
be tested in the same way as the asthma panel (see above) with
one exception—that the tests be conducted at the end of the
work day. The only feasible way of accomplishing this was to
set up the test equipment on the work premises so that the
workers could be tested immediately on returning from the field.
The workers would not have volunteered to participate if they
were required to take much extra time to report to a remote test
location. Testing was performed simultaneously at two different
locations, and the testing locations were different for each of
the four study groups. Tests were done between 10 a.m. and
6 p.m. with most being performed between 2 p.m. and 4 p.m. It
was necessary to test six days a week because many workers were
on irregular shifts so that Saturdays were covered. Because of
this fact, not all panelists in each study group were tested on
the same days. The testing was continued until each panelist in
the group had been tested ten times; then the equipment was
moved to a new location for the convenience of the next group.
Detailed clinical examinations of each panelist were con-
ducted before and after the two-week examination period. Pro-
cedures were identical to those described above for the asthma
panel. The workers were permitted by their employers to take
time off from their jobs to go to a central location for these
examinations. No more than six panelists could be handled in
an hour so it would have been impossible to schedule everyone
for clinical examinations at the end of the working day.
-31-
-------
SECTION 5
FIELD METHODOLOGY--PHYSIOLOGIC TEST METHODS
Generally, the same physiologic test methods were used on
all test subjects regardless of which panel they belonged to.
For this reason this section is organized by tests rather than
panels. The single exception was spirometry which was done by
one method for athletes and bronchitics and by another method
for asthmatics and outdoor workers. Both methods are described.
LUNG FUNCTION TESTS
Single-Breath Oxygen Test
This test was performed on each study day for the athletes
and bronchitics and as part of the detailed clinical examina-
tions, before and after each two-week testing period, for the
asthmatics and outdoor workers. Each subject performed the
single-breath oxygen test in the standing position with a nose
clamp in place; a flanged valve was introduced into the mouth.
Each subject was asked to breathe normally for a few breaths
through a low dead space 9-way valve which was opened to room
air. The needle valve assembly of a rapidly responding nitrogen
analyzer was interposed between the mouthpiece and the 9-way
valve. After stablizing, the subject was asked to exhale com-
pletely, and hold his breath briefly while the valve was
switched, opening the inspiratory port to a reservoir containing
100 percent oxygen. The subject was then instructed to inspire
fully and deeply, within ten seconds, to his total lung capa-
city. He then expired through the expiratory port, now opened
to an Ohio 780 electronic spirometer, at a slow but constant
rate (generally between 0.5 and 0.8 liters/second as monitored
on a recording device) through the next ten seconds to the
residual volume position. Expired volume and expired nitrogen
concentration were simultaneously recorded on the horizontal and
vertical axis of a Hewlett-Packard XYZ ink recorder. Readjust-
ment of the nitrogen concentration baseline permitted recording
of subsequent trials on the same graph paper. The technician
observed the tests as they were completed, to select the best
determination of the three or to request further trials if
needed. The tracings were examined by the chief technician;
values for the Delta N2 (750 to 1,250 ml) of exhalation, the
-32-
-------
volume exhaled at the Phase III inflection point, and the total
volume exhaled were then recorded for later data processing.
Volume-Time Tracing of FVC Maneuver
This test was performed on all panelists. For the athletes
and bronchitics, the test was performed on each study day. For
the asthmatics and outdoor workers, the test was performed
during the detailed clinical examinations before and afer each
two-week testing period.
The test was performed with the panelist in the standing
position and a clean disposable mouthpiece was employed. The
subject was instructed to reach maximum inspiration. Then while
placing the mouthpiece between his lips and closing them tightly
on the tube, he was instructed to exhale as rapidly and force-
fully as possible, until he could express no more air. The test
was repeated at least four additional times until the technician
felt the three maximal expiratory efforts had been achieved.
Volumes and flow rates identified by this procedure were on mag-
netic tape as well as on oscillographic paper. In the analysis
of tape outputs, the "best breath" was selected (Massey format)
for the recording of spirometric values included in the final
printout.
FEV1.0
This measurement was made on the outdoor workers and asth-
matics using Sted-Wells type spriometers manufactured by
Collins. Before the start of the testing program the spirom-
eters were compared with each other and with the Ohio 780
electronic spirometer in the Breathmobile to make sure that all
gave equivalent readings. Agreement among them was within ex-
perimental error, and no correction factors had to be devel-
oped. Values for FEV^g for three trials were hand calculated
from the instrument charts using conventional techniques and
were subsequently corrected to BTPS. Measurement of FVC was not
specified in the study protocol, so the test subjects were not
instructed to finish their expiration completely so that values
for FVC could also be calculated. For all tests, the spirom-
eters were located indoors in air conditioned rooms which re-
mained at approximately the same temperature for the entire two-
week testing period.
Body Plethysmography
This test employed the Ohio 3100 system. Flow volume loops
defining box pressure versus mouth pressure, and box pressure
versus airflow were generated, according to techniques standard-
ized by Du Bois. Subjects were seated inside the plethysmograph
with a nose clip in place and asked to breath calmly through the
-33-
-------
mouthpiece-pneumotachograph assembly. After venting the box
several times, over one to two minutes, box pressure equilibra-.
tion was achieved as indicated by the superimposition of a few
tidal excursions of box pressure on the oscilloscope. Subjects
were then asked to pant, with the shutter open at the end ex-
piratory breathing position. After several panting demonstra-
tions, one or two acceptable loops were retained on the lower
half of the storage oscilloscope. The shutter was then automat-
ically closed at the end of expiration, and one or two box pres-
sure versus airway pressure loops were retained on the upper
half of the oscilloscope. The shutter was then reopened. The
tangents of the angles formed between the horizontal axis and
the loops stored on the oscilloscope were then determined by
aligning parallel lines of a plastic ruled overlay device
(swung in front of the oscilloscope) to the appropriate portion
of the loops. After appropriate alignment, the tangents were
read directly off the plastic device. Each series of panting
maneuvers was repeated through four additional tests. The
readings of the five tangents, with the shutter open and closed,
were then recorded. Airway resistance and thoracic gas volume
at functional residual capacity were calculated from each set of
tangents read from the oscilloscope. The independently measured
box calibration factors, apparatus dead space, and apparatus
resistance were also entered into the equations. The box pres-
sure calibration was corrected for a flow volume displacement of
the subject, using the subject's weight in.kilograms and assum-
ing a tissue density of one gram per cubic centimeter. For each
set of measurements, specific airway conductance was calculated.
In the analysis, the calculated airway resistance and thoracic
gas volume for the five attempts were averaged.
Calibration of Pulmonary Equipment
The nitrogen analyzer was calibrated, employing pure oxygen
(07o nitrogen) and humidified room air (78.9% nitrogen). The
spirometers were calibrated for volume with a special syringe
(1.5 liter volume). Flow calibrations were performed with an
external rotameter while air was supplied to the spirometer at
ten liters per second by means of another spirometer and a
Scotch Yoke mechanism.
The body plethysmograph was calibrated according to the in-
structions of the Ohio 3100 Operations Manual. A reciprocating
pump delivered a calibrated 50 cc stroke for box pressure veri-
fication. Electronic calibrations for mouth pressure and flow
were externally validated with a U-tube water manometer and a
flow-generating device in conjunction with a Fischer-Porter
rotameter.
-34-
-------
HEART FUNCTION
Blood pressures, heart rates, and 12-lead electrocardio-
grams were taken using standard techniques on a Hewlett-Packard
electrocardiograph. The unit was secured with subjects in the
supine position. Calibration and electrode placement were per-
formed according to standard instructions. (Calibration and
testing were performed by the same technician on almost all oc-
casions.) In the instance of athlete studies, the time delay
from completion of the exercise until completion of the post-
exercise electrocardiogram was recorded.
Interpretation of the electrocardiograms was completed by
a board certified cardiologist who prepared a report according
to the format shown in Appendix C. This report was subsequently
coded and tabulated in digital form. The coding system used in
translating the cardiologist's report to digital form is provided
as Appendix D.
HEMATOLOGY
CBC and Differential
Subjects were seated and the nurse performed a standard
venipuncture in the antecubital area, after skin cleansing.
Specimens were secured by the B.D. vacutainer system. One tube
with anticoagulant was sequestered for cooling, and was used for
the determination of blood count and differential count. The
second tube was allowed to clot and then was centrifuged. The
serum was decanted into a separate identified container. These
were frozen and transferred to the CLMG lab for determination of
immunoglobulins, which was performed by methodology described
below. CBC's and differentials were performed by the hematology
laboratory of the St. Frances Hospital in Lynwood, California,
using Coulter Model S counter for WBC, RBC, Hgb, Hct, MCV, MCH,
and MCHC. Differential counts were made in the standard way
using the hemocytometer. The reporting format is reproduced in
Appendix E.
Nasal Smears
Subjects were seated and the nurse applied a swab to the
nasal membranes, securing secretions from one or both nares.
These were then applied to a clean glass slide which had been
identified with the subject's I.D. number, the date of collec-
tion, and whether the specimen was secured before or after out-
door exercise. A standard fixative was then applied to the
slides, and they were packaged and transported to the laboratory
of the St. Frances Hospital for staining and examination for
eosinophils. Findings were reported as "negative," "few"
(<5%), "moderate" (5 to 20%), and "many" (>20%).
-35-
-------
Immimoglobulins
Serum samples were obtained as described above under '"CBC
and Differential." Quantitative analyses were performed by the
Bio-Sciences Laboratories of Los Angeles. Immunoglobulins A, G,
M, and D were determined by radial immunodiffusion using the
method perfected by Fahey, and immunoglobulin E was determined
by a radio immunoassay method using ]>25 as the labeling agent.
A kit obtained from Pharmacia Diagnostics was used for the IgE
determination.
MISCELLANEOUS
Tear Samples
The lysozytne concentration of tears was determined by the
Proctor Foundation for Research in Ophthalmology at the San
Francisco Medical Center .of the University of California. Tear
samples were obtained according to directions supplied by them. .
Test subjects were seated quietly in the Breathmobile. A mea-
sured strip of test paper (Schirmer Strip) was carefully in-
serted in the conjunctival sac of each eye. The nurse observed
the saturation of the test strip through a five minute period of
time. The strips were then removed (earlier if fully satur-
ated) , and the distance of saturation (and the time to complete
this if less than five minutes) was recorded for each eye. The
removed strips were divided at the line of saturation, and a
strip placed into a glass vial and sealed with a screwtop lid.
Each vial was labeled to indicate whether taken from the right
or left eye and the time of day, as well as being identified as
to the subject's practice of wearing glasses during usual activ-
ity. The sealed vials were then mailed to the Proctor Founda-
tion in batches. Enzyme content was determined by standard
methodology under the direction of Dr. Ernest K. Goodner.
Reports of Discomfort
On each testing day every panelist was questioned about
symptoms of discomfort, amount of smoking, and respiratory ill-
ness. Responses were recorded on the form shown in Appendix F.
The questions were asked at the beginning of the testing period
except in the case of the athletes who were questioned twice--
before and after the race. Generally, a single interviewer or
a pair of interviewers recorded the responses for each panel.
Efforts were made to keep techniques as uniform as possible
among interviewers, but it is possible that slight differences
might have occurred between panels. It is believed that no
biases due to questioning techniques occurred within a given
panel.
-36-
-------
Clinical Interview Questionnaire
A sample questionnaire is shown in Appendix G. This .ques-
tionnaire was administered to all panelists by a trained inter-
viewer at the end of the field testing portion of the study.
The study protocol originally required that the questionnaire be
administered at the first comprehensive clinical examination,
but questionnaire clearance was not received in time for this to
be done. It is believed that the change in the time of adminis-
tration did not affect the accuracy of the responses.
-37-
-------
SECTION 6
RAW DATA FILES
During the course of this study, raw data were obtained in
a variety of ways—on magnetic tape, from instrument charts or
oscilloscope screens, from handwritten records, or on printouts,
from computers that were integral components of automated analy-
tical systems. Ultimately all data items were put into digital
format and tabulated. The collected tabulations were bound into
a Raw Data Book. One copy of this book was sent to the EPA Pro-
ject Officer for the Agency's reference. This chapter of the
final report lists the items included in the data book and,
where appropriate, explains how the data were obtained.
PHYSIOLOGIC DATA
Anthropometric and Miscellaneous
The following items are included:
• Height (inches)
• Weight (pounds)
• Systolic BP
• Diastolic BP
• Pulse rate
e Running time for race (minutes) - athletes only
• Date and time of day measurements were made
For each member of the asthma and outdoor worker panels there
are two sets of measurements--before and after the two-week
period of daily testing. For each member of the athlete and
bronchitis panels there are measurements for all of the testing
days or for as many of the days as the subject was present. In
addition, for the athlete panel, blood pressure values and pulse
rates were measured before and after each race.
-38-
-------
Body Plethysmography
The data file includes the average five attempts for 'the
following:
• Thoracic gas volume (liters)
• Airway resistance (cmH^O/liters/second)
• Date and time of day measurements were made
These measurements were made on the asthma and outdoor worker
panelists during the clinical examinations which were performed
before and after the two-week periods of daily testing. There
are, therefore, two sets of measurements for each member of the
asthma and outdoor worker panels. These measurements were not
made on the athletes or the bronchitics, however.
Closing Volume
The following items were tabulated for the best of three
efforts:
• Volume (ml) at the beginning of the plateau
• Difference in nitrogen percentage at 1,250 ml
and 750 ml
• Vital capacity minus closing volume divided by
vital capacity /VC-CV\ expressed as a percent
lvc r
• Vital capacity (ml), labeled "slow VCM to dif-
ferentiate it from VC
• Medication during past four hours - yes or no
• Ease of reading closing volume
a Quality of tracing
• Date and time of day measurements were made
In instances where the ease of reading the closing volume was
recorded as "poor," the VC-CV values were judged to be invalid
VC
and were not included in the data file. There are two sets of
measurements for each member of the asthma and outdoor worker
panels which were made before and after their two-week daily
testing periods. For each member of the athlete and bronchitis
panels, there are measurements for each of the testing days
-39-
-------
when the subject was in attendance. Measurements were made
before and after the race for the athlete panel.
Pulmonary Function
The data file contains the following items for the asthma
and outdoor worker panels:
• Corrected FEV, ~ (liters) - three maneuvers
• Date and time of day measurements were made
Measurements were made once on each day of the two-week daily
testing period.
. For the athlete and bronchitis panels, more sophisticated
electronic equipment was used and the following items were tabu-
lated for the best of three maneuvers:
e Forced expiratory flows averaged over the fol-
lowing volume fractions: 0.2-1.2%, 25-75%,
75-85%, 75-90%
• FVC (liters)
o FEV-L 0 (liters)
0 (liters)
• FEV,
. FEV3J
• Maximum forced expiratory flow rate
• Time of maximum flow (seconds)
o Forced expiratory flow rates at the following
volumes.: 25%, 50%, 75%
• Date and time of day measurements were made
Measurements were made at every testing session that the
athletes and bronchitics attended. For the athletes, measure-
ments were made before and after each race.
Electrocardiographs
Twelve-lead electrocardiograms were run at every testing
session for the athletes and bronchitics. In addition, the
athletes we/re tested before and after each race. After being
interpreted by a cardiologist, these results were reported on
-40-
-------
the form shown in Appendix C and converted to digital form in
accordance with the coding sheet shown in Appendix D. All these
items were tabulated in the raw data files. Provision was made
for recording two conclusions from the cardiologist; however, in
most cases there was only one conclusion. No electrocardiograms
were obtained on members of the asthmatic or outdoor worker
panels.
Hematology
Venous blood samples were obtained from each member of the
athlete and bronchitis panels on five of the testing days.
The following measurements were performed and are included in
the data files:
• White blood count
• Red blood Count
• Hemoglobin (gm)
• Hematocrit (7o)
3
• Mean Corpuscular Volume (y )
• Mean Corpuscular Hemoglobin (yyg)
• Mean Corpuscular Hemoglobin Concentration (70)
• Differential white cell count including segmented
neutrophiles, band forms, metamyelocytes, myelo-
cytes, total neutrophiles, eosinpphiles, basophiles,
lymphocytes, and monocytes
• Absolute white cell counts as calculated from
WBC and differential
o Platelets (normal, increased, decreased)
• RBC (normal, anisocytosis, hypochromia,
poikilocytosis, polychromasia)
• Immunoglobulins A, G, M, and D (mg70)
• Immunoglobulin E (units/ml)
• Eosinophiles from nasal smears (negative, few,
moderate, many)
• Date and time of day measurements were made
-41-
-------
Lysozyme in Tears
Tear samples were obtained from each member of the athlete
and bronchitis panels on 5 of the 11 testing days. Samples from
the athletes were obtained before and after the race. The fol-
lowing items are tabulated:
• Schirmer, right eye (mm)
• Schirmer, left eye (mm)
• Lysozyme, right eye (yg/ml)
• Lysozyme, left eye (yg/ml)
• Date and time of day measurements were made
QUESTIONNAIRES
Daily Symptom
Daily symptom records were maintained for each subject for
each testing day and were recorded before and after the race for
the athletes. Samples of the record forms have been given pre-
viously in Appendix F. All of these records are included in the
raw data files.
Clinical Interview
The clinical interview questonnaire (Appendix G) was admin-
istered once to each subject. Responses to all 55 questions are
included in the data files along with the subject's birthdate,
age, sex, and height.
AEROMETRIC DATA
The aerometric data file contains relevant aerometric data
obtained from Federal, state, and county agencies operating air
monitoring stations in or near the study area. Specifically,
the file contains data collected by the following air monitoring
stations:
• EPA station 0841 at Glendora, California
• EPA station 0842 at Covina, California
• California Air Resources Board station at
Temple City, California
-42-
-------
• Los Angeles County Air Pollution Control
District station at Azusa, California
The Glendora, Covina, and Azusa stations are located within
the study area (refer again to Figure 1). The Temple City sta-
tion is located several miles to the west. Based on these loca-
tions and on comparisons of the data collected by the four
stations, it was decided to rely primarily on the aerometric
data from the Azusa station in the analysis of the health infor-
mation. Two factors led to this decision. The Azusa station
offered the largest number of observations of the air pollutants
of interest to this study. In addition, the levels of oxidant
and oxides of nitrogen measured at the Azusa station tended to
fall between those measured at the Glendora and Covina stations.
A further discussion of these factors and an explanation of the
oxidant correction factor is given in Appendix H of this report.
The air pollution variables included in the analysis were
ozone (03), nitrogen dioxide (N02), and nitric oxide (NO). The
values were obtained as hourly averages measured in parts per
hundred million and converted to parts per million (ppm) before
being entered into the raw data files. Ozone provided the usual
estimate of oxidant. Both prominent oxides of nitrogen were
included; nitrogen dioxide, which was known to be toxic to the
lungs, and nitric oxide, which was considered to be much less
toxic than N0£, but which served as a control for spurious re-
sults. Carbon monoxide (CO) was also included in the analysis
because of the commonality of its sources in the study area with
those of oxidant and oxides of nitrogen. The values of CO were
obtained as hourly averages measured in ppm; however, these
values were inadvertently entered into the raw data files in
terms of ppm x 100.
Twenty-four hour averages for sulfur dioxide, total sus-
pended particulates, suspended sulfates, and suspended nitrates
were entered into the raw data files, but were not included in
the analysis. The values for these pollutants were generally
below the air quality standards for 24 hours and, therefore,
were not expected to cause measurable biochemical or physiologic
response.
Relative humidity data were not available from any monitor-
ing station in the study area. Values in the raw data files
were estimated by assuming that the dew points in the study area
were identical with dew points measured at Ontario International
Airport, which is located about 17 miles southeast of the study
area. Relative humidities were calculated from temperatures
obtained in the study area and from the assumed dew points.
Values are recorded as "high" (>75%), "medium" (50 to 75%) and
"low" (<507o). Temperature data were obtained as hourly averages
measured in degrees Fahrenheit from one of the two EPA air moni-
toring stations.
-43-
-------
The aerometric data were obtained (and relative humidity
was estimated) for each hour between 7 a.m. and 9 p.m. for each
day health information was collected. As a prelude to the, anal-
ysis of the health information, a correlation matrix was
developed for the air pollution variables in order to determine
the extent of collinearity between these variables. The matrix
is presented in Table 11 in Section 7.
-44-
-------
SECTION 7
DATA ANALYSIS--ASTHMA PANEL
The analyses presented in this section and in Sections 8,
9, and 10 are based on statistical considerations and do not
specifically address the medical or health aspects of the work.
The results obtained should provide starting points for inter-
pretation by persons trained in medicine and epidemiology. The
analyses of the four panels are presented in decreasing order
of the information gained:
• Section 7
• Section 8
- Data analysis of the asthma panel
- Data analysis of the outdoor worker
panel
• Section 9 - Data analysis of the bronchitis
panel
• Section 10 - Data analysis of the athlete panel
The asthma and outdoor worker panels provided similar levels of
information. Although the asthmatics were considered to be po-
tentially the more sensitive of the two groups to the effects
of air pollution, it might be said that more interesting re-
sults came from comparing the effects of air pollution on
smokers versus the effects of air pollution on non-smokers in
the outdoor worker panel. The order of the bronchitis and
athlete panels was suggested mostly by the number of observa-
tions available for analysis.
STATISTICAL DESCRIPTION OF THE ASTHMA PANEL
It is recalled that the asthma panel was composed of non-
smoking females in the 21 to 50 age bracket. Although 62
females were selected for participation, four dropped out before
daily surveillance began. The remaining individuals were
divided into four sub-panels which were tested for four succes-
sive two-week periods. Table 8 summarizes the composition of
the sub-panels. Of the 58 females assigned to the sub-panels,
only 41 participated throughout the testing periods.
-45-
-------
The analysis of the asthma panel data is restricted to the
discomfort symptom, pulmonary function, and air pollution vari-
ables measured during daily surveillance. Comprehensive clini-
cal examinations of each panelist were conducted before and
after the panelist's testing period and a clinical interview
questionnaire was administered after the field studies were com-
pleted. The information obtained from the clinical examinations
and the interview was used to verify the presence of asthma
symptoms and was not meant to be a part of the daily surveil-
lance data base. The information indicated that all of the
panelists actually had symptoms of asthma as they reported dur-
ing recruitment.
Description of the Variables
The discomfort symptom variables were measured with the
form shown in Appendix F. These variables were qualitative and,
consequently, had to be coded for analysis. "Yes" responses
were coded to equal unity and "No" responses were coded with
zero. A statistical profile of the discomfort symptom variables
is given in Table 9A. The variables were:
9 EYES (eye discomfort)
o THROAT (throat discomfort)
e> CHEST (chest discomfort)
® HEADACHE
o NAUSEA
0 OTHER (other discomfort)
9 HEADACHE EARLIER
© BREATH (shortness of
breath)
& COUGH
© PHLEGM
e COLD (bad cold)
e MEDICINE
The smoking variable included on the form was not applicable to
the asthma panel, since all of the panelists were nonsmokers .
The pulmonary function variable used in the analysis was
MAXFEV, which was the maximum FEV-, 0 score of three maneuvers
attained each day by each panelist. The AGE and HEIGHT of each
panelist were recorded as age in years and standing height in
inches. These data were used to adjust (or normalize) the
MAXFEV scores prior to analysis. A profile of the MAXFEV, AGE,
and HEIGHT variables is given in Table 9B. MAXFEV is shown in
hundred liters (liters x 100).
-46-
-------
TABLE 8. COMPOSITION OF THE ASTHMA SUB-PANELS
I
-p-
Characteristics
Subjects enrolled in panel
(all females)
Current cigarette smokers
Subjects 21 to 30 years old
Subjects 31 to 40 years old
Subjects 41 to 50 years old
Subjects with 12 or more
years of school completed
Race other than white
Sub-
Panel 1
13
0
7
5
1
10
3
Sub-
Panel 2
12
0
6
2
4
9
2
Sub-
Panel 3
10
0
2
6
2
8
1
Sub-
Panel 4t
6
0
2
2
1
5
1
Total
41
0
17
15
8
32
7
tAge missing for one panelist in Sub-Panel 4.
-------
TABLE 9A. STATISTICAL PROFILE OF DISCOMFORT SYMPTOM VARIABLES MEASURED DURING ASTHMA PANEL SURVEILLANCE
oo
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEVNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
EYES = Eye discomfort now
- 0.338
- 0.224
= -1.525
- 0.474
- 0.688
THROAT -
*> 0.325
= 0.220
= -1.434
- 0.469
- 0.751
CHEST -
- 0.401
- 0.241
- -1.834
= 0.491
- 0.405
HEADACHE
- 0.219
= 0.171
- -0.146
- 0.414
- 1.363
NAUSEA =
- 0.042
- 0.041
- 18.789
- 0.201
•» 4.565
OTHER =
- 0.409
- 0.242
« -1.860
- 0.492
- 0.371
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
Throat discomfort now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS -
Chest discomfort now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS -
- Headache now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
Nausea now
RANGE
MINIMUM
MAXIMUM
VALID
MISSING OBS =
Other discomfort now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
1.000
0.000
1.000
379
0
1.000
0.000
1.000
379
0
1.000
0.000
1.000
379
0
1.000
0.000
1.000'
379
0
1.000
0.000
1.000
379
0
1.000
0.000
1.000
379
0
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
. . . . SKEWNESS
HEADACHE EARLIER = Headache earlier today
= 0.288
- 0.205
• -1.114
• 0.453
- 0.941
BREATH
- 0.317
- 0.217
- -1.374
- 0.466
= 0.791
COUGH =
- 0.507
- 0.251
" -1.997
- 0.501
- -0.026
' PHLEGM
= 0.753
- 0.187
- -0.615
- 0.433
• -1.178
COLD =
•> 0.050
- 0.048
- 15.048
• 0.219
- 4.134
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
• Shortness of breath
RANGE
MINIMUM
' MAXIMUM
VALID OBS
MISSING OBS
Cough today
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
•* Phlegm today
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
Bad cold today
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
- 1.000
= 0.000
- 1.000
- 379
- 0
today
- 1.000
= 0.000
=• 1.000
- 379
- 0
- 1.000
- 0.000
= 1.000
- 379
- 0
- 1.000
- 0.000
- 1.000
- 186
- 6
- 1.000
- 0.000
- 1.000
- 379
- 0
MEDICINE - Medicine today
". 0.530
- 0.250
- -1.983
- 0.500
- -0.122
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
- 1.000
- 0.000
- 1.000
- 379
- 0
-------
Available aerometric data were selected to correspond to
the times of daily surveillance. A profile of the air pollu-
tation and weather variables used in the asthma panel analysis
is given in Table 9C. The variables were:
*
• OZONE (estimate for e No (nitric oxide)
oxidant)
• CO (carbon monoxide) • HUMID (relative
humidity)
0 N0£ (nitrogen dioxide) • TEMP (temperature)
The air pollution variables were obtained in hourly aver-
ages measured in ppm (CO in ppm/100). Relative humidity was
estimated by hour and expressed in percent. Temperature was
obtained as hourly averages measured in degrees Fahrenheit.
Missing health and aerometric observations were few and
not serious. N02 readings contained the greatest number of
missing observations and, where only one hour was missing,
values were estimated by averaging the N0£ values for the hour-
before-test and hour-after-test.
A problem arose due to the lack of enough equipment to
test all members of the asthma panel simultaneously. Panel
members had to be scheduled for testing from mid-morning
through late afternoon. Aerometric values varied significantly
over this period. In order to account for the effects on
health associated with exposure to air pollutants and to humid-
ity and temperature:at the time of the testing, weighted:.aver-
ages had to be developed. In doing so the hour at which each
panelist reported for testing was noted and the aerometric values
for that hour were included in the weighted average for that day.
This approach is expected to have been far more sensitive than
using the maximum hourly average or an average of the hourly
values over an arbitrary portion of the day.
The proportions of asthma panelists who reported discomfort
symptoms are shown in Appendix I for each of the 40 days when
symptom data were collected. The weighted averages of the air
pollutants, relative humidity, and temperature are also given in
Appendix I for each of the same days. The charts are provided
to facilitate comparisons between the proportions of panelists
who reported symptoms and the weighted averages of air pollution.
CORRELATION ANALYSIS
Pearson and two non-parametric correlation analyses were
performed on the asthma panel data; The results are reported
on the following pages.
-49-
-------
TABLE 9B. STATISTICAL PROFILE OF PULMONARY FUNCTION
VARIABLE AND AGE AND HEIGHT OF PANELISTS MEASURED
DURING ASTHMA PANEL SUREVEILLANCE
VARIABLE: MAXFEV = Maximum FEV1>0 (in liters x 100)
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
284.085
6,186.592
0.422
78.655
-0.345
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
480.000
59.000
539.000
378
1
VARIABLE: AGE - Age of panelist (in years)
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
32.780
60.476
626
7.777
0.704
-0
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
27
22
49
40
1
VARIABLE: HEIGHT = Height of panelist (in inches)
MEAN
VARIANCE
KURTOSIS
STD DEV.
SKEWNESS
64.075
9.097
-0.476
3.016
-0.394
RANGE = 11
MINIMUM ,-* = 58
MAXIMUM • = 69
VALID OBS = 41
MISSING OBS = 0
-50-
-------
TABLE 9C. STATISTICAL PROFILE OF AIR POLLUTION
VARIABLES MEASURED DURING ASTHMA PANEL SURVEILLANCE
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
OZONE = Estimate for oxide (in ppm)
0.090
= 0.005
0.242
0.069
1.027.
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
0.290
0.010
0.300
379
0
CO = Carbon monoxide (in ppm/100)
= 0.038
0.000
= - -0.230
0.013
0.536
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
0.060
0.020
0.080
378
1
N02 = Nitrogen dioxide (in ppm)
0.077
0.002
4.625
0.040
1.510
NO = Nitric oxide
0.014
0.000
= 10.601
0.008
= 2.877
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
(in ppm)
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
0.260
0.010
0.270
353
26
0.050
0.010
0.060
353
26
HUMID = Humidity (in percent)
= 62.164
= 82.264
= -1.086
9.070
0.091
TEMP = Temperature
= 74.211
= 89.305
= -0.698
9.450
= 0.126
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
25.000
50.000
75.000
379
0
(in degrees Fahrenheit)
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
42.000
53.000
95.000
379
0
-51-
-------
Pearson Correlation
The Pearson correlation coefficient (R) was used to measure
the strength of relationship between two interval scale vari-
ables. R measures both the goodness of fit of a linear regres-
sion line and provides a test of a hypothesis of independence
between two variables. The null hypothesis of interest is
HQ: p = 0
where p is the population correlation coefficient. If two vari-
ables are linearly independent, then p = 0. Rejection of the
null hypothesis leads to tentative acceptance of the alternative
hypothesis of linear dependence between the two variables.
A test of this hypothesis in a bivariate normal population
is given by the t-ratio, t = R [(N-2)/(1-R2)]%, with N-2 degrees
of freedom. A two-tailed test of statistical significance was
used.
The pearson correlation coefficients for the asthma panel
are reported in Table 10. The significance levels are included
at a = 0.01, 0.05, and 0.10. A positive correlation indicates
.a direct relationship between the two variables, while a negative
correlation indicates an inverse relationship. If a one-tailed
test is preferred, merely interpret the levels of significance
shown as a/2 rather than a.
The relationship between MAXFEV and the air pollution vari-
ables was expected to be inverse; therefore, the correlation
coefficients were expected to be negative. The estimated cor-
relations are all of the wrong sign..
None of the coefficients for THROAT, GHEST, NAUSEA, OR OTHER
are significantly different from zero. The coefficients for
the variables BREATH and PHLEGM are significant only with re-
spect to OZONE.
Air pollution variable NO is not significantly correlated
with any of the discomfort symptoms. The variable HUMID is
significantly negatively correlated with EYES, HEADACHE, HEAD-
ACHE EARLIER, and MAXFEV. HUMID is not significantly corre-
lated with any of the other qualitative variables. The variable
TEMP is significantly positively correlated with EYES, HEADACHE,
HEADACHE EARLIER, and MAXFEV.
-52-
-------
TABLE 10. PEARSON CORRELATION COEFFICIENTS FOR THE ASTHMA PANEL
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
MAXFEV
OZONE
0.25***
-0.11
0.02
0.13**
-0.04
0.03
-0.02
0 . 11***
-0.09*
0.15**
0.02
CO
0 . 16***
-0.02
0.04
0 . is***
-0.04
-0.01
0 . 10**
0.01
-0.03
-0.03
;0.10**
N02
0 . 16***
-0.02
0.08
0 . 10*
-0.01
0.03
0.09*
-0.01
-0.02
0.04
0. 13**
.NO. ....
0.01
0.06
0.11
-0.06 .
-0.08
-0.02
-0.01
0.04
0.07
-0.05
0.05
.... HUMID .
0 . 21***
0.06
-0.03
-0.15***
-0.07
-0.06
-0.12**
-0.04
0.01
-0.04
-0.08*
. TEMP
0.21***
-0.02
0.04
0 . 20***
0.04
0.01
0.15***
0.08
-0.05
0.11
0.12**
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
Out of all the discomfort symptom variables, EYES shows the
most significant correlation with the air pollution variables.
NO is the only air pollution variable which is not significantly
correlated with EYES. Among the air pollution variables OZONE
has more statistically significant positive coefficients than
the others. The next most significant air pollution variables
are CO, and N02, which are also significantly correlated with
MAXFEV.
Correlation coefficients for the discomfort symptoms and
air pollution variables are reported in Tables 10 and 11. Note
that the correlation between HUMID and TEMP is -0.81 and is sig-
nificant at a = 0.001. This indicates possible collinearity
between HUMID and TEMP. Note further, in Table 10, that the
significant correlations for HUMID and TEMP and the discomfort
symptoms are always opposite in sign and similar in magnitude.
This analysis leads to the conclusion that HUMID and TEMP should
not be included in any multiple regression specification at the
same time.
Non-Parametric Correlations
Spearman's rank correlation coefficient (Rs) and Kendall's
tau (T) are both non-parametric; neither depends on a normal
distribution. The Spearman's coefficient (Rs) is corrected for
the occurrence of tied ranks. The significance of RS can be de-
termined by the same t-statistic used for the Pearson coeffi-
cient with Rs substituted for R. The Kendall's tau coefficient
(T) is also corrected for tied ranks. The significance of T is
determined by comparing T to a normal distribution with a
standard deviation equal to [(4N + 10)/9N(N - 1)]% where N is
the number of observations.
The non-parametric correlation coefficients for the discom-
fort symptoms and the air pollution variables are reported in
Tables 12 and 13. The results tend to confirm those derived
from the Pearson correlation coefficients.
The qualitative variable. EYES is the only symptom variable
which is consistently significant with respect to the air pollu-
tion variables. The next most significantly correlated variable
is HEADACHE. The qualitative variables THROAT, CHEST, NAUSEA,
OTHER, HEADACHE EARLIER, and COUGH are not strongly correlated
with air pollution variables OZONE, CO, and N02- Variable NO
has stronger correlations than indicated by the Pearson correla-
tions, but the coefficient with NAUSEA is of the negative sign.
Negative correlation of discomfort symptoms with NO should be
expected when the symptom variables show positive correlation
with OZONE. NO is usually negatively correlated with OZONE.
Table 11 shows such a relationship in this study.
-54-
-------
TABLE 11. CORRELATION MATRIX OF AIR POLLUTION VARIABLES'
I
Ul
Ul
I
OZONE
CO
N02
NO
HUMID
TEMP
OZONE
1.0000
(0)
S = 0.001
0.3921
(378)
S = 0.001
0.0645
(353)
S = 0.227
-0.2517
(353)
S = 0.001
-0.4915
(379)
S = 0.001
0.5608
(379)
S = 0.001
CO
0.3921
(378)
S = 0.001
1.0000
.(0)
S = 0.001
0.7216
(352)
S = 0.001
0.2285
(352)
S = 0.001
-0.1714
(378)
S = 0.001
0.1983
(378)
S = 0.001
N02
0.0645
(353) ,
S = 0.227
0.7216
(352)
S' = 0.001
1.0000
(0)
S = 0.001
0.3822
(347)
S = 0.001
-0.2026
(353)
S = 0.001
0.1705
(353)
S = 0.001
... NO .
-0.2517
(353)
S = 0.001
0.2285
(352)
S = 0.001
0.3822
(347)
S = 0.001
1.0000
(0)
S = 0.001
0.1206
(353)
S = 0.023
-0.1839
(353)
. S = 0..001
HUMID
-0.4915
(379)
S = 0.001
-0.1714
(378)
S = 0.001
-0.2026
(353)
S = 0.001
0.1206
(353)
S = 0.023
1.0000
(0)
S = 0.01
-0.8099
(379)
S = 0.001
TEMP
0.5608
(379)
S = 0.001
0.1983
(378)
S = 0.001
0.1705
(353)
S = 0.001
-0.1839
(353)
S = 0.001
-0.8099
(379)
S = 0.001
1.0000
(0)
S = 0.001
tNumber of cases is in parentheses; S = level of significance.
-------
TABLE 12. SPEARMAN CORRELATION COEFFICIENTS FOR THE ASTHMA PANEL
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
OZONE
0.26***
0.04
0.01
0.13***
-0.04
0.03
-0.01
0.12**
-0.08*
0.13**
CO
0.15***
0.05
0.03
0 . is***
-0.04
-0.92
0.07*
-0.12
-0.04 .
-0.05
N02 .
0.12***
0.01
0.01
0.10**
-0.01
-0.06
0.06
-0.03
-0.05
0.03
NO
0.06**
0 . 08*
0.12**
-0.01
-0.10**
-0.04
0.01
0.03
0.09*
-0.07
HUMID
-0.21***
-0.09***
-0.04
-0.15***
-0.07*
-0.06
-0.12***
-0.04
0.01
-0.04
TEMP
0. 21***
0.11**
0.05
0.20***
0.03
-0.01
0 . 14***
0.08*
-0.06
0.11*
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
TABLE 13. KENDALL CORRELATION COEFFICIENTS FOR THE ASTHMA PANEL
1
Ln
-J
1
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
OZONE
0.22***
0.03
0.01
0.11***
-0.03
0.02
-0.01
0 . 10**
-0.06*
0.11**
CO
0.13***
0.04
0.02
0.16***
-0.04
-0.02
0.07*
-0.02
-0.03 .
-0.05
N02
0.11***
0.01
0.01
0.08**
-0.01
-0.05 •
0.05
-0.02
-0.04
0.03 .
NO
0.06**
0 . 08*
0 . 11**
-0.01
-0.09**
-0.04
0.01
0.03
0.08
-0.-07
HUMID
-0.20***
-0.08**
-0.03
-0.14***
-0.07*
-0.05
-0.12***
-0.04
0.01
-0.04
TEMP
0.18***
0.09**
0.04
0 . 16***
0.02
-0.01
0 . 12***
0.06*
-0.05
0.09*
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
Summary of the Correlation Analysis
The only discomfort symptom which is consistently and, sig-
nificantly correlated with the air pollution variables is EYES.
The only other qualitative variables which appear to be con-
nected with air pollution are HEADACHE and BREATH. MAXFEV does
not appear to be linearly related to the air pollution data for
the asthma panel.
The air pollution variable OZONE appears to be the most ex-
planatory. OZONE is positively correlated with EYES, HEADACHE,
BREATH, and PHLEGM. But the missing reports of PHLEGM lead to
skepticism in drawing any conclusion from the associated corre-
lation coefficients. The other air pollution variable which ex-
hibits significant correlation with some of the discomfort
symptoms is CO. NO does not appear to be strongly correlated
with any of the discomfort symptoms. Furthermore, N02 seems
only weakly correlated with the discomfort symptoms. Therefore,
NO and N02 are omitted in further analysis of the asthma panel.
Finally, the levels of humidity and temperature are highly
correlated with each other. And the correlations with the dis-
comfort symptoms indicate they have an effect which is opposite
in sign but equivalent from a statistical viewpoint. A multi-
variate analysis which uses several explanatory variables, like
multiple regression, should not use both humidity and tempera-
ture at the same time, since multicollinearity seems to exist
between the two explanatory variables. The distributions of
estimated regression parameters are quite sensitive to multi-
collinearity.
MEASURES OF ASSOCIATION BETWEEN VARIABLES
The following pages describe the associations found between
most of the variables included in this study. For example, the
association between MAXFEV and the air pollution variables and
the dependency of the discomfort symptoms on having a cold are
examined. In all, four measures of association are applied.
Simple Linear Regressions of FEV^ Q
Simple linear regression equations of the form
Y = a + bX
were computed with MAXFEV as the dependent variable and the air
pollution variables as the explanatory variables. The intercept
(a) and slope (b) were computed by ordinary least-squares re-
gression. The results are reported in Table 14.
-58-
-------
TABLE 14. SIMPLE REGRESSION RESULTS WITH MAXFEV AS THE
DEPENDENT VARIABLE
EXPLANATORY: OZONE
CORRELATION (R) = 0.02233
R SQUARED = 0.00050
SIGNIFICANCE = 0.66514
STD ERR OF EST = 78.73977
INTERCEPT (a) = 281.80560
SLOPE (b) = 25.44250
EXPLANATORY: CO
CORRELATION
R SQUARED
SIGNIFICANCE
(R) =
0.10295
0.01060
0.04577
STD ERR OF EST
INTERCEPT (a)
SLOPE (b)
78.22557
260.41677
625.38935
EXPLANATORY: N02
CORRELATION (R) = 0.09306
R SQUARED = 0.00866
SIGNIFICANCE = 0.08039
STD ERR OF EST
INTERCEPT (a)
SLOPE (b)
78.62108
276.09287
126.04091
EXPLANATORY: NO
CORRELATION (R) = 0.05480
R SQUARED = 0.00300
SIGNIFICANCE = 0.30528
STD ERR OF EST
INTERCEPT (a)
SLOPE (b)
78.04043
279.27642
554.59520
EXPLANATORY: HUMID
CORRELATION (R) = -0.08631
R SQUARED = 0.00745
SIGNIFICANCE = 0.09381
STD ERR OF EST
INTERCEPT (a)
SLOPE (b)
78.46550
330.55249
-0.74750
EXPLANATORY: TEMP
CORRELATION (R) = 0.12223
R SQUARED = 0.01494
SIGNIFICANCE = 0.01743
STD ERR OF EST
INTERCEPT (a)
SLOPE (b)
78.16886
208.66602
1,01616
-59-
-------
The statistics which describe these simple regressions in-
clude the correlation coefficient (R) and the coefficient of de-
termination (R.2) , which measures the percentage of the var-iance
of the dependent variable explained by regression. The test of
a significant linear relationship between MAXFEV and the explan-
atory variable involves a two-tailed test that R is signifi-
cantly different from zero. The level of significance (a) at
which that hypothesis can be rejected is also reported in
Table 14.
MAXFEV is not significantly linearly related to OZONE.
MAXFEV is significantly related to CO at a level of significance
greater than a = 0.05, but the sign of the slope term is not
consistent with expectations. The results are probably spur-
ious. The same can be said of the regression with N02 and NO as
the explanatory variables.
No expectations of the signs of the slope coefficients of
HUMID and TEMP were formulated. The regression with HUMID as
the explanatory variable is significant at a level greater than
a = 0.10. The sign of the estimated coefficient is negative,
which indicates that MAXFEV is inversely related to the level of
humidity. The largest coefficient of determination is provided
with TEMP as the explanatory variable, and the estimated slope
has positive sign and is significant at a = 0.017.
In summary, there are no strong inverse linear relation-
ships between MAXFEV and the air pollution variables. Only the
variables HUMID and TEMP yield results which are both reason-
able in terms of sign and statistically signficant. Correlation
analysis indicated that HUMID and TEMP are probably collinear.
Inspection of scattergrams (reproduced in the Data Analysis
Supplement) indicated no clear non-linear relationship between
MAXFEV and any of the air pollution variables for the asthma
panel taken as a whole.
Contingency Tables Between Discomfort Symptoms and Having a Cold
Asthma panelists were asked whether or not they experienced
eye discomfort, throat discomfort, chest discomfort, headache,
nausea, other discomfort (unspecified), headache earlier, short-
ness of breath, cough, and phlegm. The panelists were also asked
asked whether or not they had a bad cold or were taking any med-
ication on each day tested. It was expected that the response to
to the discomfort symptom questions were dependent on the
presence of a cold or taking medication.
In the following analysis of association between the
symptom attributes and the cold attribute, the contingency tables
tables are (2x2) and the degrees of freedom are (2-1)•(2-1) = 1.
For one degree of freedom, the x value needed to reject a
-60-
-------
hypothesis of independence at the a = 0.01 level is 6.63. Re-
jection of the hypothesis of independence allows one to say that
some statistical association exists between the two attributes.
If the hypothesis is rejected, one may feel confident in con-
cluding that the two attributes are in some way related. The
X2 required to reject the hypothesis of indpendence at a = 0.05
is x = 3.84, and at a = 0.10 is x2 = 2.70. (The contingency
tables are included in the Data Analysis Supplement.)
The computed x2 value for COLD and EYES is x2 = 2.35, which
is not large enough to reject the hypothesis of independence at
the level of significance of a = 0.10. Thus, there is no sta-
tistical evidence of dependence between eye discomfort and the
presence of a bad cold for the asthma panel. The computed x
value for COLD and THROAT is x = 22.02, which is large enough .
to reject the hypothesis of independence at a = 0.01. This
leads to the tentative acceptance of the alternative hypothesis
of dependence between COLD and THROAT. Throat discomfort seems
to be dependent on the presence of a cold. The computed x
value for COLD and CHEST is x = 0.81, which is not large enough
to reject the hypothesis of independence at a = 0.10. The re-
lationship between COLD and HEADACHE is indicated by a computed
X2 of 3.61. This is large enough to reject the hypothesis of
independence at a = 0.10, but not at a = 0.05. The relationship
between COLD and NAUSEA is indicated by a computed x2 of 3.95.
This is large enough to reject the hypothesis of independence at
a = 0.05, but not at a = 0.01. The presence of headache or
nausea and having a cold appears to be dependent, but that de-
pendence is not strong. The computed x2 f°r COLD and OTHER is
X2 = 7.53, which is large enough to reject the hypothesis of
independence at a = 0.01. The presence of other discomfort and
having a cold also appears to be dependent. The alternative hy-
pothesis of dependence is supported between COLD and HEADACHE
EARLIER. The computed x is 17.46, which is large enough to re-
ject the hypothesis of indpendence at a = 0.01. Thus, there
is evidence of dependence between having a headache earlier in
the day and having a cold. The computed x2 for COLD and BREATH
is x2 = 1.58, which is not large enough to reject the hypothesis
of independence at a = 0.10. Shortness of breath and having a •
cold appear to be independent for the asthma panel. The com-
puted x for COLD and COUGH is x = 5.27, which is enough to
reject the hypothesis of independence at a = 0.05, but not at
a = 0.01.
In summary, there is evidence of dependence between having
a cold and throat discomfort, headache, nausea, other discom-
fort, headache earlier, or cough. There is not evidence of de-
pendence between having a cold and eye discomfort, chest discom-
fort, or shortness of breath. One implication for further
analysis is that the method should control for the presence of a
cold when explaining the presence of the qualitative variables
-61-
-------
by the air pollution variables. Otherwise, the investigator may
falsely attribute the discomfort symptoms to some measure of air
pollution.
Contingency Tables Between FEV, Q and Air Pollution Variables
A series of contingency tables were constructed between
MAXFEV and the air pollution variables OZONE, CO, N02, and NO.
MAXFEV was divided into two classes: 0^ MAXFEV < 300 and 300 <
MAXFEV. OZONE was divided into two classes: O^OZONE<0.1
and 0.1< OZONE. Variable CO was divided into two classes:
0.01
-------
The regression equation may be interpreted as the probability
that a symptom will be present given information about the value
of the explanatory variable, OZONE, CO, N02, NO, HUMID, or TEMP.
The slope of the regression line, 3, measures the effect on the
probability given a unit change in the explanatory variable.
Examination of the probability of the error term indicates
.certain properties of the model. . The. error has zero mean,, but
the variance"of the error is not constant for all observations.
The presence of heteroscedasticity results in a loss of effi-
ciency, but does not in itself result in biased or inconsistent
parameter estimates. However, the standard tests of signifi-
cance must be interpreted with caution.
Simple linear probability regressions using eye discomfort
as the dependent variable are in Table 15. It is noted that
low R.2 values are not unusual for regressions using a dichoto-
mous dependent variable. There appears to be a very definite
relationship between the probability of eye discomfort and the
air pollution variables except for NO.
Note that the regressions using HUMID and TEMP as the 'ex-
planatory variables have estimated slope coefficients which are
equal in magnitude, but opposite in sign. This was expected
given the collinearity indicated from the' correlation analysis .
This pattern is repeated for all the dependent variables. . The
variable TEMP appears to be better in most cases, from the .stand-
point of explanatory power as indicated by the R-squares and the
level of significance.
The linear probability regression for the remaining dis-
comfort symptom variables are not so impressive. Only those
which were statistically significant at a =0.05 or better are
presented in Table 16.
None of the regressions for NAUSEA, OTHER and COUGH are
significant. The variable PHLEGM was not tested because of the
small number of observations. Moreover, the variable HEADACHE
EARLIER was excluded because there seemed to be no direct cause
relationship between that variable and the levels of the air
pollution at times the asthma panelists reported for daily sur-
veillance.
In summary, only two of the discomfort symptoms variables
show any statistical dependence on the air pollution variables.
These symptoms are eye discomfort and headache. With EYES as
the dependent variable, the significant regressions are:
Prob (EYES) = 0.183 + 1.725(OZONE) R2 = 0.063
Prob (EYES) = 0.111 + 5.940(CO) R2 = 0.026
-63-
-------
TABLE 15. SIMPLE LINEAR PROBABILITY REGRESSIONS WITH
EYE DISCOMFORT AS THE DEPENDENT VARIABLE
EXPLANATORY: OZONE**
CORRELATION (R) = 0.25120
R SQUARED = 0.06310
SIGNIFICANCE = 0.00001
STD ERR OF EST = 0.45898
INTERCEPT (a) = 0.18330
SLOPE (b) = 1.72504
PLOTTED VALUES = 379
EXPLANATORY: CO**
CORRELATION (R) = 0.16185
R SQUARED = 0.02620
SIGNIFICANCE = 0.00159
STD ERR OF EST = '0.46824
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
0.11092
5.94007
378
EXPLANATORY: N02 **
CORRELATION (R) = 0.14593
R SQUARED = 0.02130
SIGNIFICANCE = 0.00588
STD ERR OF EST = 0.46833
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
0.23978
5.94007
355
EXPLANATORY: NO
CORRELATION (R) = 0.01196
R SQUARED = 0.00014
SIGNIFICANCE = 0.82276
STD ERR OF EST = 0.47305
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
0.32399
0.73372
353
EXPLANATORY: HUMID **
CORRELATION (R) = -0.20567
R SQUARED = 0.04230
SIGNIFICANCE = 0.00006
STD ERR OF EST = 0.46405
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
1.00531
•0.01074
379
EXPLANATORY: TEMP**
CORRELATION (R) = 0.21398
R SQUARED = 0.04579
SIGNIFICANCE = 0.00003
STD ERR OF EST = 0.46321
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
-0.45804
0.01072
379
**Signifleant at a = 0.05
-64-
-------
TABLE 16. SIGNIFICANT LINEAR PROBABILITY REGRESSIONS
FOR THE ASTHMA PANEL
DEPENDENT: THROAT
EXPLANATORY: TEMP
CORRELATION (R) = 0.10691
R SQUARED = 0.01143
SIGNIFICANCE = 0.03749
STD ERR OF EST = 0.46675
INTERCEPT (a) = -0.06905
SLOPE (b) = 0.00530
PLOTTED VALUES = 379
DEPENDENT: CHEST
EXPLANATORY: NO
CORRELATION (R) = 0.11174
R SQUARED = 0.01249
SIGNIFICANCE = 0.03586
STD ERR OF EST = 0.48867
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
0.30238
7.12290
353
DEPENDENT: HEADACHE
EXPLANATORY: OZONE
CORRELATION (R) = 0.12964
R SQUARED = 0.01681
SIGNIFICANCE = 0.01153
STD ERR OF EST = 0.41116
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
0.14930
0.77853
379
DEPENDENT: HEADACHE
EXPLANATORY: CO
CORRELATION (R) = 0.17759
R SQUARED = 0.03154
SIGNIFICANCE . = 0.00052
STD ERR OF EST = 0.40846
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
0.00102
5.70141
378
DEPENDENT: HEADACHE
EXPLANATORY: HUMID
CORRELATION (R) = -0.15254
R SQUARED = 0.02327
SIGNIFICANCE = 0.00291
STD ERR OF EST = 0.40981
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
0.65196
-0.00696
379
DEPENDENT: HEADACHE
EXPLANATORY: TEMP
CORRELATION (R) = 0.19772
R SQUARED = 0.03909
SIGNIFICANCE = 0.00011
STD ERR OF EST = 0.40648
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
-0.42399
0.00866
379
DEPENDENT: BREATH
EXPLANATORY: OZONE
CORRELATION (R) = 0.10600
R SQUARED = 0.01124
SIGNIFICANCE = 0.03915
STD ERR OF EST = 0.46376
INTERCEPT (a)
SLOPE (b)
PLOTTED VALUES
0.25252
0.71598
379
-65-
-------
Prob(EYES) = 0.240 + 1.187(N02) R = 0.021
Prob(EYES) = 1.005 - 0.011(HUMID) R2 = 0.042
Prob(EYES) = -0.458 + 0.011(TEMP) R2 = 0.046
The significant regressions with HEADACHE as the dependent vari-
ables are:
Prob(HEADACHE) = 0.652 - 0.007(HUMID) R2 = 0.023
Prob(HEADACHE) = -0.424 + 0.009(TEMP) R2 = 0.039
Thus, eye discomfort is seemingly dependent on the presence of
ozone, carbon monoxide, nitrogen dioxide, humidity, or tempera-
ture. The dependency of the headache is on humidity or tempera-
ture. These results are not surprising considering the outcome
of the correlation analysis presented earlier.
SELECTED MULTIVARIATE LINEAR PROBABILITY REGRESSIONS
The form of a multivariate linear probability model is
•V
k
9± = ProbO. = 1) = 30 +£>iX - i = 1.....N
The interpretation is the same as the simple linear probability
model. The only difference is that there are several explana-
tory variables. The predicted value of the dependent variable
can be interpreted as the probability that the qualitative char-
acteristic, discomfort, is present given values of the explana-
tory variables. Only two of the discomfort symptoms, EYES and
HEADACHE, were chosen for this analysis, because they showed
sensitivity to the air pollution variables in the correlation
analysis.
The correlation analysis indicated that OZONE and CO were
the most explanatory of the air pollution variables. Variable
TEMP was included as an environmental control, and HUMID was not
included in order to avoid multi-collinearity. Binary variable
COLD was included because the chi-square tests indicated depen-
dence between having a cold and reporting certain discomfort
symptoms. Differences between individuals required the inclu-
sion of AGE as a control variable. The results are reported in
Table 17.
-66-
-------
TABLE 17. MULTIVARIATE LINEAR PROBABILITY FUNCTIONS FOR EYE
DISCOMFORT AND HEADACHE--ALL ASTHMA PANELISTS
(STANDARD ERRORS IN PARENTHESES) ; '
Prob(EYE) = -0.390 + 1.138(OZONE)** + 2.707(CO) + 0.006(TEMP)**
(0.445) (2.045) (0.003)
+ 0.297(COLD)** + 0.002(AGE)
(0.109) (0.003)
R2 = 0.09, N = 368, F = 7.36
Prob(HEADACHE) . = -0.717 - 0.261(OZONE) + 5.299(CO)** + 0.009(TEMP)**
(0.393) (1.809)
+ 0.290(COLD)** + 0.002(AGE)
(0.096) (0.003)
R2 = 0.08, N = 368, F = 6.24
**Signifleant at a = 0.05
-------
The F-statistic for the eye discomfort function is F = 7.36
which is significant at a = 0.01 indicating .a rejection of a
null hypothesis that all coefficients are simultaneously z-ero.
The R2 = 0.09 indicates that 9 percent of the variation of this
dependent variable is explained by regression. Air pollution
variable OZONE and environmental variable TEMP have a signifi-
cant direct relationship with eye discomfort. The coefficient
for variable CO is not significantly different from zero at
a = 0.05.
The F-statistic for the headache function is also signifi-
cant at a = 0.01. The R^ = 0.08 indicates 8 percent of the var-
iation in this qualitative variable is explained by regression.
Air pollution variable OZONE is not significantly related to the
dependent variable HEADACHE. The variables CO and TEMP exhibit
a significant positive relationship in predicting the probabil-
ity of the headache discomfort symptom.
In both regressions, COLD has a positive coefficient which
is statistically significant. Also in both regressions, the co-
efficient of the variable AGE is not significantly different from
from zero.
Two linear probability specifications were estimated for an
asthma panel subset not having a cold. Variable AGE was dropped
because it was not significantly different from zero in the pre-
vious specifications. Variable N0£ was added as an explanatory
variable. The dependent variables were again EYES and HEADACHE.
The results are reported in Table 18.
The estimated coefficients were tested for significant at
a = 0.05. The estimated coefficient for OZONE is significantly
positive with EYES as the dependent variable, but the OZONE co-
efficient is not significant with HEADACHE as the dependent var-
iable. Variable CO is not significant with EYES as the depen-
dent variable; however, CO is significant with HEADACHE as the
dependent variable. Variable N02 is not significant in either
specification. TEMP is significant only with HEADACHE as the
dependent variable.
Comparison of the regression results between all asthma
panelists and the subset not having a cold leads to the conclu-
sion that the results are similar. Both sets of regressions
have estimated coefficients which are similar in magnitude,
have the same sign, and remain either significant or not signif-
icant. That is, the slope coefficients did not change drasti-
cally when the sample was restricted to asthma panelists not
having a cold. Of course, this conclusion might change if a
non-linear specification for the probability function were
utilized.
-68-
-------
TABLE 18. MULTIVARIATE LINEAR PROBABILITY FUNCTIONS
FOR EYE DISCOMFORT AND HEADACHE--ASTHMA PANELISTS
WITHOUT A COLD
(STANDARD ERRORS IN PARENTHESES)
Prob(EYES)
= -0.249 + 1.040(OZONE)** + 2.523(CO)
(0.452) (2.462)
+ 0.670(N02)
(0.503)
R2 = 0.08, N = 335, F > 6.83
+ 0.004(TEMP)
(0.003)
Prob(HEADACHE) = -0.574 - 0.299(OZONE) + 5.413(CO)**
(0.399) (2.170)
- 0.547(N02)
(0.401)
R2 = 0.05, N = 335, F = 4.47
**Signifleant at a = 0.05
+ 0.009(TEMP)**
(0.004)
-69-
-------
BETWEEN GROUP DIFFERENCES
Two possibilities which could affect usefulness of the
asthma panel results are considered in the following material:
1. That there were differences in sensitivity to
air pollution between asthma panelists having
and those not having a cold.
2. That there were differences between one or
more of the sub-panels which would not permit
certain data to be aggregated.
Sensitivity Analysis for Asthma Panelists Having and Not Having
a Cold^
The asthma panel was divided into two mutually exclusive
subsets, those panelists who reported having a cold and those
who reported not having a cold. This was done to see if there
was any difference in the sensitivity to air pollution between
these two groups.
The subset of panelists having a cold was examined first.
Pearson correlation coefficients were computed between each of
the air pollution variables and the discomfort symptoms vari-
ables along with MAXFEV.
Two primary differences were found between these correla-
tions and the correlations for the asthma panel as a whole.
First, the correlations are generally larger than before. How-
ever, most are not statistically significant. If the panelists
having a cold are generally more responsive to air pollution,
the data are not strongly supportive. Second, the correlation
coefficients between COUGH and the air pollution variables
OZONE, CO, and NOo are now quite large in magnitude and sta-
tistically significant at the a = 0.05 level. But the correla-
tions are negative.
Non-parametric correlations, the Spearman and Kendall coef-
ficients, were also computed. The non-parametric correlations
support the Pearson correlations. Again, the coefficients are
generally larger than those computed for the asthma panel as a
whole, and the cough variable-is significantly negatively corre-
lated with OZONE, CO, and N02.
To examine the MAXFEV performance of the subset having a
cold, scattergrams were plotted and linear regressions for
MAXFEV with the air pollution variables as explanatory were
computed. None of the regressions .were statistically signifi-
cant. The scattergrams did not exhibit any obvious non-linear
relationship. Therefore, the resulting statistics are not
-70-
-------
reported here. The implication of the analysis is that MAXFEV
shows no significant systematic relationship, with the air pollu-
tion variables for the subset having a cold.
Simple linear probability regressions were run with the
discomfort symptoms as the dependent variables. Only those re-
gressions having a level of significance of a = 0.10 or greater
are reported. Those regressions significant only with HUMID or
TEMP are not reported.
The only discomfort symptoms that are significantly related
to the air pollution variables for this subset were eye discom-
fort and headache. The estimates are:
Prob(EYES)
Prob(EYES)
Prob(HEADACHE)
Prob(HEADACHE)
Prob(HEADACHE)
0.255 + 4.684(OZONE) FT = 0.17
-0.031 + 8.904(N02) R2 =0.21
0.093 + 5.664(OZONE) R2 = 0.25
-0.416 + 24.838(CO) R2 = 0.45
-0.457 + 14.015(N02)
IT = 0.54
Comparison with the linear probability regressions for the
asthma panel as a whole leads to the observation that the coef-
ficients of determination are much higher—a larger percentage
of the dependent variable is explained for the restricted
sample. Moreover, the estimated slope coefficients are much
larger in magnitude. The implication is that the panelists
having a cold are more sensitive to OZONE and N0£ in experienc-
ing eye discomfort and are more sensitive to OZONE, CO, and N0£
in experiencing a headache than the whole panel.
The linear probability specification was applied separately
to the panelists who did not report a cold. The regressions,
with EYES as the dependent variable, significant at a = 0.05 or
better are:
Prob(EYES)
Prob(EYES)
Prob(EYES)
Prob(EYES)
Prob(EYES)
0.168 + 1.751(OZONE) R = 0.067
0.069 + 6.719(CO)
0.229 + 1.167(N02)
-1.010 + O.Oll(HUMID)
-0.470 + 0.011(TEMP)
R^ = 0.034
R2 = 0.022
R2 = 0.046
R2 = 0.046
-71-
-------
These estimates are almost identical to those computed for the
whole panel.
•s
The significant estimates with HEADACHE as the dependent
variable are:
Prob(HEADACHE) = 0.139 + 0.759(OZONE) R2 = 0.017
Prob(HEADACHE) = 0.016 + 5.Oil(CO) R2 = 0.025
Prob(HEADACHE) = 0.670 - 0.007(HUMID) R2 = 0.028
Prob(HEADACHE) = 0.451 + 0.009(TEMP) R2 = 0.042
The regressions with explanatory variables HUMID and TEMP are
almost identical to those estimated for the whole panel. How-
ever, the regressions with OZONE and CO as the explanatory vari-
ables are now significant at a = 0.05, but were not significant
for the whole panel.
The remaining significant linear probability regressions
for the subset not having a cold are:
Prob(CHEST) = 0.292 + 7.415(NO) R2 =0.014
Prob(THROAT) = 0.209 + 6.529(NO) R2 = 0.012
Prob(THROAT) = 0.662 - 0.006(HUMID) R2 = 0.022
Prob(THROAT) = 0.236 + 0.797(OZONE) R2 = 0.014
None of these specifications were significant for the whole
panel but they are for the subset not having a cold. The vari-
able NO becomes a significant explanatory, variable in predicting
the probability of experiencing chest or throat discomfort when
the panelist does not have a cold. Variable OZONE is a signifi-
cant predictor of the probability of shortness of breath for
those in the subset not having a cold. In general, the fore-
going analysis emphasizes the need to control for the presence
of a cold.
Differences Between Sub-Panels
The asthma panel was divided into four sub-panels for daily
surveillance. These groups were tested for four successive two-
week periods. Differences between the sub-panels is a potential
source of variation which must be examined. The following
-72-
-------
paragraphs present the results of difference in means tests and
two-way analysis of variance (ANOVA) for the four groups. The
difference in MAXFEV between sub-panels was examined. Table 19
gives the sample size, sample mean, and sample standard devia-
tion of MAXFEV for the four groups.
TABLE 19. SAMPLE MEANS AND STANDARD DEVIATIONS
FOR MAXFEV OF THE FOUR ASTHMA SUB-PANELS
(NO RESTRICTIONS ON TOTAL SAMPLE)
Sub-Panel
1
2
3
. 4
N
126
112
91 .
50
X
302.47
281.19
298.81
230.56
S
61.9
85.7
107.0
68.6
An F-test of sample variances was performed between pairs
of all four sub-panels. This was necessary to choose the appro-
priate t-test to test for differences in means between all four
groups. The null hypothesis to be tested first is
H : a? = a2, with the alternative H»: a? ^ a?, where
o i j A. i j
i, j = 1,...,4 with i i= j. From the sample variances F is com-
puted as F = larger S2/smaller S2. If the probability for F is
greater than some chosen level of significance (a), Ho is ac-
cepted. If the null hypothesis is rejected, then an approxima-
tion to t must be made which is based on the separate variance
estimate.
The results of the F-tests are given in Table 20. The
level of significance chosen was a = 0.10. A two-tailed test of
significance was applied. The results indicate a rejection of
the hypothesis of equal variance between all of the sub-panels
except one. Therefore, there is evidence that the population
variances of MAXFEV between sub-panels are not equal when there
are no restrictions placed on the selection of observations
other than the sub-panel classification.
Given populations with unequal variances, an approximation
to the t for the difference in sample means should be applied.
The results are given in Table 21. The null hypothesis is
H0: V»i = Uj for all i ? j, and the alternative hypothesis is
-73-
-------
TABLE 20. DIFFERENCE IN VARIANCES TEST ON ASTHMA SUB-PANELS
(NO RESTRICTIONS ON TOTAL SAMPLE)
0
0
0
0
0
0
H F
o
= 0 1.91
1 2
= 0 2.99
1 3
= 0 1.23
i i*
= 0 1.56
2 3
= 0 1.56
2 i»
=0 2.44
3 4
a/2
0.00
0.00
0.18
0.02
0.04
0.00
Significant
at a = 0.10
yes
yes
no
yes
Yes
yes
Decision
on H0
reject
reject
do not reject
reject
reject
reject
TABLE 21. DIFFERENCE IN
SUB -PANELS
MAXFEV MEANS TEST
ON ASTHMA
USING SEPARATE VARIANCE ESTIMATES
y
y
y
y
y
y
H t
o
, = y 2.17
1 2
= y 0.32
1 3
= y 6.45
1 4
= y -1.27
2 3
= y 4.01
2 4
= y 4.60
3 4
a/2
0.015
0.376.
0.000
0.102
0.000
0.000
Significant
at a = 0.05
no
no
yes
no
yes
yes
Decision
onHo
do not reject
do not reject
reject
do not reject
reject
reject
-74-
-------
HA: Vi £ y-i • The level of significance chosen was a = 0.05
(0.025 in each trial) using the two-tailed test.
The results in the difference in means test are mixed.
While mean MAXFEV for Sub-Panel 4 is significantly different
from the mean MAXFEVs of Sub-Panels 1, 2, and 3, the mean
MAXFEVs between Sub-Panels 1, 2, and 3 are not significantly dif-
ferent from each other. This indicated difference for Sub-Panel
4 may be due to the small sample size of 50, about half the size
of each of the other groups. Sub-Panel 4 represented 13.2 per-
cent of the total asthma panel.
Differences Between Sub-Panels with MAXFEV Adjusted for Age and
Height
A potential source of between group differences is age and
height of each panelist. This is particularly true if a given
sub-panel has a small sample size as does Sub-Panel 4. The hy-
pothesis to be tested is that the between group differences in
MAXFEV are due to age and height.
To test this hypothesis, the difference between the
recorded MAXFEV and predicted FEVn Q was computed from a linear
equation with age and height as the explanatory variables.
Furthermore, the sample was restricted to panelists not having
a cold. The estimated equation is:
Predicted FEV1>0 = 13.902 - 3.203(AGE) + 5.911(HEIGHT)
(0.484) (1.271)
2
The coefficient of determination for this equation was R =
0.20. The standard error of each coefficient is in parentheses
below the coefficient. Twenty percent of the variation in
MAXFEV is accounted for by age and height. Each of the esti-
mated slope coefficients is significantly different from zero at
a = 0.10 level of significance.
Again, a test of a hypothesis of no differences between
variances of the MAXFEV difference variable between groups is
appropriate. The results are given in Table 22. Only the vari-
ance of Sub-Panel 4 appears to be different at a = 0.10 level
of significance.
A pooled variance estimate was used to compute the t-
statistic to test for differences in sub-panel means of MAXFEV
difference variable, since the hypothesis of unequal variances
could not be rejected between all sub-panels. The results are
given in Table 23. Only the hypothesis that P! = y,, and y2 = y,,
can be rejected. Again, Sub-Panel 4 seems to be the only sub-
panel which is different from the rest. And much of the unex-
plained variation apart from age and height contained in
-75-
-------
TABLE.22. DIFFERENCE IN VARIANCES TEST ON ASTHMA SUB-
PANELS: DIFFERENCE IN MAXFEV FOR PANELISTS NOT HAVING A COLD
0
0
0
0
0
0
Ho F
= 0 1.03
1 2
=0 1.25
1 3
=0 2.20
= 0 1.22
2 3
= 0 2.25
2 If
= a 2.74
3 "t
a/2
0.45
0.13
0.01
0.17
0.01
0.00
Significant
at a = 0.10
no
no
yes
no
yes
yes
Decision
onHQ
do not reject
do not reject
reject
do not reject
reject
reject
TABLE 23. DIFFERENCE IN
DIFFERENCE IN
MEANS TEST ON ASTHMA
SUB-PANELS :
MAXFEV FOR PANELISTS NOT HAVING A COLD
y
y
y
y
y
y
Ho t
= ^2 0.34
1.47
! = y,, 3.32
= V3 1.10
2 = ^^ 2'9?
3 ~ <*
a/2
0.366
0.072
0.001
0.137
0.002
0.039
Significant
at a = 0.05
no
no
yes
no
yes
no
Decision
on HQ
do not reject
do not reject
reject
do not reject
reject
do not reject
-76-
-------
Sub-Panel 4 could easily be caused by differences in other
variables, such as humidity or temperature, which are not pan-
elist characteristics.
Analysis of Variance Between Sub-Panels
The model applied was a fixed effects model in a two-way
analysis. It was assumed that the observed MAXFEV is a sum of
systematic effects associated with experimental treatments, plus
random error.
A two-way ANOVA with MAXFEV as the dependent variable was
performed for the panelists who reported not having a cold.
Again, the main concern was whether or not the sub-panels which
were tested on different weeks are significantly different from
each other. Therefore, in every ANOVA performed, the sub-panel
was used as a treatment (explanatory) variable. A total of six
ANOVA were performed. The first three used OZONE, CO, and TEMP,
respectively, as the other treatment variable. The second
three used the same air pollution variables; however, Sub-Panel
4 was excluded from the analysis. The reason for the exclusion
was that the difference in means tests performed above indicated
that Sub-Panel 4 may be distinct. The hypothesis to be tested
is that Sub-Panel 4 is the distinct group.
OZONE 'was divided into three ranges which were chosen to
equal plus or minus one standard deviation from the mean. Like-
wise, CO and TEMP were divided the same way. The reason for
categorizing these explanatory variables was that extreme vari-
ation in any of them was expected to affect MAXFEV. The results
of the ANOVA with MAXFEV as the dependent variable and with sub-
panels and either OZONE, CO, or TEMP as treatment variables
showed that for the interaction between OZONE and the. sub-panels,
the F-ratio is not significant at a = 0.05, but it is significant
at a = 0.10. This indicates that there may have been some inter-
action between OZONE and the sub-panels, but" the evidence is not
strong. The other effect of primary interest is that of the sub-
panel variable itself. The sub-panel variable has an F-ratio
that is significant at a = 0.01 at the same time, the OZONE
variable has an F-ratio that is not significant. The interac-
tion variable for CO and the sub-panels is not significant.
Again, the sub-panel variable is significant, while the CO vari-
able is not. A third ANOVA indicates no significant interac-
tion between the sub-panels and TEMP in explaining the variation
in MAXFEV. The sub-panel variable by itself has an F-ratio
which is significant, while the F-ratio for TEMP is not. All
three ANOVA indicate that there is a significant difference be-
tween sub-panels in explaining the variation in MAXFEV which has
not been adjusted for differences in age and height of the study
subjects.
-77-
-------
The second set of three ANOVA excludes Sub-Panel 4 from the
analysis. The implications of the results are quite different.
In all three cases the interaction term has F-ratios which are
not significant. The F-ratios for the variables OZONE, CO, and
TEMP are also not significant. The primary difference between
these ANOVA which exclude Sub-Panel 4 and the first set of ANOVA
lies in the F-ratios for the group variable. The F-ratio for
sub-panel is not significant at a = 0.05 for all three ANOVA.
However, the F-ratio for the sub-panel variable is significant
at a = 0.10 in two of the ANOVA, one using OZONE and the other
using TEMP. The implication is that the between group variation
evident in the first set of ANOVA is absent in the second set of
ANOVA where Sub-Panel 4 is excluded.
The implications of the second set of ANOVA above is con-
sistent with the previous tests for differences in MAXFEV be-
tween groups. Sub-Panel 4 appears to be distinct from the other
three. Based on these results, Sub-Panel 4 is dropped from
further analysis except where between group differences can be
identified and controlled!.
ANOVA with MAXFEV Adjusted for Age and Height
The six ANOVA above were replicated with the MAXFEV differ-
ence variable. The MAXFEV difference variable was actual MAXFEV
minus predicted FEV^ Q where predicted FEVi Q was computed from
variables. The interaction term in all cases was not statisti-
cally significant.
The group effect term is statistically significant by the
F-ratio for all three ANOVA which included all sub-panels. But
the group effect term is not statistically significant for the
three ANOVA when Sub-Panel 4 is excluded. The implication is
that Sub-Panel 4 remains distinct from the other three sub-
panels after attempting to control for the age and height dif-
ferences .
Summary of the Between Group Analysis
There is no logical nor theoretical reason why any of the
sub-panels should be distinct or different from each other. The
temporal difference in the sub-panels should not, in itself,
create differences between the groups. However, differences be-
tween the sub-panels in their measured reactions to the air pol-
lution variables could exist. This is because of the possible
differences in the environment at the time of measurement and of
differences in the composition of the sub-panels due to small
sample or self-selection bias.
-78-
-------
The reason, then, for this analysis was to determine if
there is any difference between the sub-samples which cannot be
controlled using appropriate statistical methods. If there is
no difference that cannot be controlled statistically, then the
sub-panels may be analyzed simultaneously as a pooled sample r
representative of the population from which the sample was
drawn.
Different methods indicate that there is a difference be-
tween the four groups. These same methods indicate that Sub-
Panels 1, 2, and 3 are not different from each other. And the
methods identify Sub-Panel 4 as a distinct group. When MAXFEV
was adjusted for age and height, the conclusions remained essen-
tially unaltered. This does not mean that there is no logical
reason for the difference. For example, examination of the pro-
portions of panelists reporting discomfort symptoms reveals a
noticeable decrease in those proportions on the Sub-Panel 4 test
dates. Moreover, variation in the level of carbon monoxide on
those same test dates is noticeably absent. In addition, the
level of relative humidity was higher and the temperature was
lower on the test dates for Sub-Panel 4.
The reason why Sub-Panel 4 seems distinct remains uniden-
tified. A conservative approach, therefore, is to exclude Sub-
Panel 4 from further analysis as a possible source of contamina-
tion. The evidence is strong in favor of treating the other
three sub-panels as representative of the population from which
they were drawn.
DISCOMFORT SYMPTOMS ANALYZED BY DATE OF MEASUREMENT
The asthma panel was interviewed on 40 different dates. The
sample was divided into four sub-panels. The first sub-panel
was tested on ten consecutive days, excluding the weekend. The
second, third, and fourth sub-panels were tested over successive
ten-day periods, also excluding weekends. The dates were re-
corded as dates 1 through 40.
On each date, the proportion of the panelists reporting
discomfort with respect to each of the qualitative symptoms was
calculated. The number of subjects tested on each date ranged
from 4 to 13. For dates 1 through 30, i.e., during daily sur-
veillance of the first three sub-panels, the lowest number of
panelists tested was eight.
The average and maximum levels of OZONE, CO, N0£, NO,
HUMID, and TEMP were computed for each date also. The average
and maximum levels of air pollution variables were then related
to the proportion of panelists reporting each discomfort symptom
on each date.
-79-
-------
Analysis of Eye Discomfort Using Proportions of Panelists
Reporting
The .LOGIT specification consolidates observations of the
dependent and explanatory variables as follows. Let fg denote
the proportion of panelists reporting a discomfort symptom for
date g = 1,...,G. Let Xg denote the average level of an air
pollution variable on date g. Then the LOGIT is:
ln[f /(I - f )] =. 30 + 3-,X ; g = 1.....G,
O O V J- £>
with G denoting the number of classes. This grouping technique
tends to reduce the degrees of freedom, and the detail in the
original data will be muted. Aggregation and regression toward
the mean will increase the coefficient of determination. And,
the LOGIT specification exhibits heteroscedasticity due to the
unequal size of the sub-panels. However, the functional form is
logically consistent with the probability interpretation. The
alternative is maximum likelihood estimation of the logistic
function directly from the micro data, a much more expensive
procedure.
A linear probability specification using the proportions of
panelists reporting is:
fg = aQ + a^; g = 1, . . . ,G.
This linear specification of the probability (proportion) is in-
consistent with the probability interpretation, but it may lead
to close approximations within the range of the original data.
Extrapolation beyond the range of the data is particularly dan-
gerous when using the linear form.
The qualitative variable, EYES, was chosen for analysis
since it seemed more responsive than the others. The regres-
sions using the linear form with the proportion of panelists re-
porting as the dependent variable and average of OZONE, CO, N02,
and TEMP as the dependent variables with Sub-Panel 4 excluded
are:
2
e = 0.213 + 1.758(OZONE)*** R =0.22
6 = 0.300 + 2.014(CO) R2 = 0.06
6 = 0.105 + 3.300(N02)*** R2 = 0.49
6 = 0.132 + 0.003(TEMP) R2 = 0.02
where 6 denotes the estimated probability of eye discomfort and
*** indicates significance at a = 0.01. Both OZONE and N02 are
-80-
-------
significant predictors of eye discomfort, while CO and TEMP are
not.
The LOGIT regressions with Sub-Panel 4 excluded are:
-1.360 + 8.695(OZONE)*** R2 = 0.23
-0.954 + 10.478(CO)** R2 =0.06
-1.867 + 15.970(N02)*** R2 =0.70
1 , , - -1.002 + 0.020 (TEMP) R2 =0.03
Here, *** indicates significance at a = 0.01 and ** indicates
significance at a = 0.05. In both the linear case and the
LOGIT, these tests of significance should be interpreted with
caution because of the heteroscedasticity casued by unequal
size-panels. That influence should be small, however, since
almost all groups contained nine or ten responses (a similar
magnitude).
Scattergrams of the proportion of panelists reporting, f ,
were plotted against the average values of the air pollution °
variables OZONE, CO, N02, and TEMP. Superimposed on each were
the linear probability estimate of the proportion and the LOGIT
estimate of the proportion. The LOGIT probabilities were plot^
ted by computing the predicted 6 for several values of the
explanatory variable using the transformation
6 = I/ j 1 + exp [ - (30 + 3{X)]['
where 30 and 3, were taken from the results of the LOGIT regres-
sion.
Comparison of the two plotted curves for all four cases in-
dicated, that the linear specification provides a close approxi-
mation to the probabilities predicted by the LOGIT, but as the
plots are extrapolated beyond the original range of the dates,
the plots diverge. In the case with N02 as the explanatory var-
iable, the linear function gives a probability greater than one
when N02 reaches 0.27 ppm. At the same time, the LOGIT plot is
asymptotic to the 0 and 1 limits. Therefore, the LOGIT would pro-
vide more realistic predictions at the extremes of the explana-
tory variable.
-81-
-------
Multiple LOGIT Regression Using EYES and HEADACHE
Multi-LOGIT regression specifies the probability (or pro-
portion) of an event occurring (panelists experiencing eye dis-
comfort or headache) as a function of several explanatory vari-
ables simultaneously. Multi-LOGIT regression is the multiple
variable extension of simple LOGIT regression which uses a
single explanatory variable. The results reported in Table 24
and Table 25 use the LOGIT estimate of 8/(l - 9) where the pro-
portion of panelists reporting eye discomfort or headache were
substituted as estimates of 6, the probability of experiencing
the discomfort symptom.
The LOGIT results using the proportion of panelists report-
ing eye discomfort are reported in Table 24. Five variations in
the LOGIT model were estimated. The first two were estimated
for all sub-panels simultaneously. The last three were esti-
mated while excluding Sub-Panel 4. The last regression in-
cluded dummy variables for the sub-panels: Sub-Panel (Group) 1
is the base; G2 = 1 if the observation was for Sub-Panel (Group)
2; G3 = 1 if the observation was for Sub-Panel (Group) 3. The
first, third, and fifth regressions used the weighted average of
the hourly averages of OZONE, CO, N0£, and TEMP as explanatory
variables. (See Appendix I for an explanation of how the
weighted averages were derived.) The second and fourth regres-
sions used the daily maximum of the hourly averages of the same
air pollution variables as explanatory. . .
2
In Table 24, Equation 1 (shown as Column 1) has an R =
0.41 and only average temperature (AVETEMP) has a coefficient
significantly different from zero. Equation 2 has an R2 = 0.40
and only maximum ozone (MAXOZ) has a. coefficient significantly
different from zero. The Durbin-Watson (D.W.) test for serial
correlation is indeterminate for Equation 1, but indicates pos-
itive serial correlation for Equation 2.
When the Sub-Panel 4 is excluded the results change. Equa-
tion 3 has an R2 = 0.73, and the Durbin-Watson statistic is
still indeterminate in its indication of serial correlation.
Average nitrogen dioxide (AVEN02) now has a significantly posi-
tive coefficient, but none of the other coefficients are signif-
icantly different from zero. Equation 4, using the maximums of
the air pollution variables, has an R2 = 0.28, the lowest of all
the LOGIT regressions using eye discomfort. None of the coef-
ficients in Equation 4 are statistically significant at a =
0.10.
Equation 5 was estimated using the averages of the air
pollution variables, excluded Sub-Panel 4, and utilized dummy
variables for the remaining groups. The coefficients of deter-
mination for Equation 5 was R2 = 0.76. Furthermore, the Durbin-
Watson coefficient leads to acceptance of a null hypothesis of
-82-
-------
TABLE 24. MULTIPLE LOGIT PROBABILITY REGRESSIONS WITH PROPORTION OF
EYE DISCOMFORT AS THE DEPENDENT VARIABLE
(STANDARD ERRORS IN PARENTHESES)
Variable
AVEOZ
AVECO
AVEN02
AVETEMP
MAXOZ
MAXCO
MAXN02
MAXTEMP
G2
G3
CONSTANT
R2
F
N
D.W.
(1)
All
Sub-Panels
11.74
(8.67)
7.76
(20.96)
5.90
(12.86)
0 . 12*
(0.05)
-12.32
0.41
6.00
40
1.19***
(2)
All
Sub-Panels
11.53*
(6.66)
-38.16
(34.07)
2.24
(3,02)
0.11*
(0.04)
-9.95
0.41
5.41
40
1 . 11****
(3)
Three
Sub-Panels
3.03
(2.89)
-6.79
(7.05)
16.71*
(4.50)
-0.01
(0.02)
-1.04
0.73
7 .'04
30
1.53***
(4)
Three
Sub-Panels
2.73
(2.72)
15.23
(14.97)
1.71
(1.18)
-0.01
(0.02)
-0.84
0.28
2.53
30
1.61***
(5)
Three
Sub-Panels
6.35*
(3.43)
-7.66
(7.40)
14.31*
(4.66)
-0.01
(0.02)
0.24
(0.33)
0.64
(0.37)
-1.25
0.76
5.38
30
1.86**
*Signifleant for two-tailed test with a = 0.10
**Accept null hypothesis of no serial correlations at a = 0.05
***Durbin-Watson test for serial correlation indeterminate at a = 0.05
****positive serial correlation at a = 0.05
-83-
-------
TABLE 25. MULTIPLE LOGIT PROBABILITY REGRESSIONS WITH PROPORTION
OF HEADACHE AS THE DEPENDENT VARIABLE
(STANDARD ERRORS IN PARENTHESES)
Variable
AVEOZ
AVECO
AVEN02
AVETEMP
MAXOZ
MAXCO
MAXN02
MAXTEMP
G2
G3
CONSTANT
R2
F
N
D.W.
(1)
All
Sub-Panels
0.14
(8.02)
23.24
(19.38)
-0.05
(11.89).
0.16*
(0.04)
-__
-14.83
0.41
6.11
40
1.86*
(2)
All
Sub -Pane Is
•
8.03
(6.04)
-5.39
(30.89)
-0.25
(2.74)
0.11*
(0.04)
-11.84
0.43
6.62
40
1.85*
(3)
Three
Sub-Panels
-0.39
(3.26)
1.30
(7.90)
5.00
(4.97)
0.04*
(0.02)
-4.70
0.25
1.90
28
1.14**
(4)
Three
Sub-Panels
3.15
(2.32)
3.15
(13.00)
-0.72
(1.00)
0.03
(0.02)
-4.26
0.32
2.65
28
1.27***
(5)
Three
Sub -Panels
1.83
(2.96)
-7.78
(6.16)
6.50*
(3.91)
0.02
(0.02)
-1.05*
(0.28)
0.23
(0.31)
-2.62
0.63
5.85
28
2 . 31***
*Signifleant for two-tailed test with a = 0.10
**Accept null hypothesis of no serial correlations at a = 0.05
***Durbin-Watson test for serial correlation indeterminate at a = 0.05
-84-
-------
no serial correlation at a = 0.05. Serial correlation is not a
problem when the group dummy variables are included. AVEOZ and
AVEN02 are positive in sign and statistically significant^
AVECO and AVETEMP do not have coefficients significantly differ-
ent from zero.
The multi-LOGIT regressions using the proportions of panel-
ists reporting headache in Table 25 follow the same pattern.
Equations 1 and 2 using all groups have R2 = 0.41 and R2 = 0.43,
respectively. Only the temperature variables have significant
coefficients. The test for serial correlation leads to accep-
tance of the hypothesis of no serial correlation: Equation 3
has R2 = 0.25 with AVETEMP having the only statistically signif-
icant coefficient. None of the coefficients in Equation 4 are
significant. Both of the regressions which exclude Sub-Panel 4,
are significant. Both of the regressions which exclude Sub-
Panel 4, Equations 3 and 4, have Durbin-Watson statistics in the
indeterminate range.
Again, Equation 5 yields the best results for the LOGIT
transformation of the proportion of panelists reporting head-
ache. The coefficient of determination is R2 = 0.63. The equa-
tion excluded Sub-Panel 4 and utilized dummy variables for the
remaining groups. The only pollution variable having a signifi-
cant coefficient is AVEN02. The test for serial correlation is
indeterminate.
Summary of the Analysis by Date
The multi-LOGIT approach to estimating the probability of a
qualitative dependent variable is a useful extension of the
simple LOGIT which is consistent with the probability interpre-
tation. There was no indication that serial correlation was a
serious problem when Sub-Panel 4 was excluded or the group dummy
variables were used to control for between group differences.
The results which explained the greatest variance are shown in
Equation 5. Equation 5 using eye discomfort indicates a posi-
tive influence of OZONE and N02- For the headache symptom,
Equation 5 indicates a positive influence of N0£.
MULTIVARIATE ANALYSIS OF MAXFEV
Two multivariate regression models were used to explain
variations in MAXFEV, while controlling for differences between
.individuals. Following this, multivariate regressions were at-
tempted with the air pollution variables lagged. The extent to
which variations in MAXFEV were explained by differences between
individuals and differences in levels of air pollution are pro-
vided in this concluding discussion of the asthma panel.
-85-
-------
The Hasselblad Model
A specific model developed by Dr. Victor H. Hasselblad for
analysis on a St. Louis data set similar to that of the present
study makes use of dummy variables which control for differences
between individuals. The form of model is:
n-1 s
FEV. . =3 + g P . +3 X . + e..
11 o ^-* m mi L^^r. ri 11
J m=l J r=n J J
for i = l,...,n and j = l,...,k. The variable P . is defined
by: . m3
P.=lifim+l
mj
where m - l,...,n - 1. Since this variable controls for dif-
ferences between individuals, there is no need to include AGE,
HEIGHT, or COLD. The variable Xij is the inverse of the day of
the study; i.e., X]i = 1/k. The other Xrj are air pollution
variables which may be included separately or simultaneously.
The £j_* represents the random error term.
Translation of this notation into specific variables for
the asthma panel results in 40 Pmj variables. These are named
PI to P40 since the dates of the study were numbered 1 to 40,
the variable X]^ was named DINV which stands for day inverse.
DINV = 1/k where k = 1, ... ,40. The other explanatory variables
were OZONE, CO, N0£ , and NO. Five regressions were performed.
The first four used the air pollution variables separately; the
last one used the air pollution variables simultaneously. The
results are reported in Tables 26A through 26E.
2
All of the regressions have high R values. In every case,
more than 80 percent of the variation in MAXFEV was explained by
regression. The total F indicates rejection of a hypothesis
that all coefficients are zero at the same time. The Durbin-
Watson test of serial correlation indicates acceptance of a null
hypothesis of no serial correlation in all five regressions.
The data variable, DINV, is not statistically different
from zero at a level of significance of a = 0.05. The air pol-
lution variables are never statistically significant from zero
in any of the regressions. Many of the dummy variables for the
individuals, P •• are statistically significant. Therefore,
-86-
-------
TABLE 26A. HASSELBLAD REGRESSION ON MAXFEV
(IN LITERS x 100) WITH OZONE
R2 = 0.81,
Variable
PI
P2
P3
P4
P5
P6
P7
P8
P9
P10
Pll
P12
P13
P14
P15
P16
P17
P18
P19
P20
P21
P22
P23
P24
P25
P26
P27
P28
P29
P30
P31
P32
P33
P34
P35
P36
P37
P38
P39
P40
DINV
OZONE
CONSTANT
N = 379, Total
6
-15.78957
-49.64007
40.86285
-11.35473
-8.84465
2.98894
-4.44504 '
-13.28890
-15.15549
-0.69572
1.56220
-6.79830
-5.60478
1.89075
-1.17128
-2.57055
-0.75333
-1.55023
-1.95164 .
2.91307
-1.14707
-2.52392
-6.14406
1.10657
6.28333
1.67245
1.91744
0.37725
-0.95427
3.90674
0.09962
0.08798
0.35763
1.27750
2.10680
-3.03290
-1.00258
-3.44107
-1.76191
0.39697
20.94211
-24.04256
280.05108
F = 34.43, D.W. =
Standard Error
10.89973
5.34495
3.85492
2.74903
2.32977
1.83538
1.65966
1.39633
1.25960
1.12138
1.09826
0.94663
0.90126
0.81799
0.77348
0.73848
0.71803
0.64221
0.62277
0.58772
0.56549
0.52646
0.56551
0.48796
0.47256
0.47418
0.44963
0.90221
0.44662
0.38572
0.37734
0.36622
0.35001
0.53058
0.33686
0.32231
0.32961
0.30391
0.29838
0.29685
13.15471
38.51230
1.83
F=t2
2.099
86.254
112.364
17.061
14.412
2.652
7.173
90.574
144.769
0.385
2.023
51.575
38.674
5.343
2.293
12.116
1.101
5.827
9.821
24.568
4.115
22.983
118.038
5.143
176.790
12.440
18.186
0.175
4.565
102.587
0.070
0.058
1.044
5.797
39.115
88.545
9.252
128.206
34.869
1.738
2.534
0.390
-87-
-------
TABLE 26B. HASSELBLAD REGRESSION ON MAXFEV
(IN LITERS x 100) WITH CO
R2 = 0.81,
Variable
PI
P2
P3
P4
P5
P6
P7
P8
P9
P10
Pll
P12
P13
P14
P15
P16
P17
P18
P19
P20
P21 .
P22
P23
P24
P25
P26
P27
P28
P29
P30
P31
P32
P33
P34
P35
P36
P37
P38
P39
P40
DINV
CO
CONSTANT. .
N = 378, Total
B
-16.03449
-50.06063
40.68993
-11.62245
-8.50794
3.03131
-4.23387
-13.36425
-15.20223
-0.66490
1.68398
-6.86772
-5.50511
1.84561
-1.26273
-2.70575
: -0.86611
-1.61008
-2.06079
2.82119
-1.26300
-2.56860
-6.09700
-1.05978
6.42707
: 1.59025
1.93044
0.35290:
-0.98857
3.91339
0.10365
0.05318
0.37547
1.28370
2.07060
-3.06900 '
a. 02337
-3.44565
-1.79014
0.35730
21.17709
-49.47883
279.61475
F = 34.27, D.W. =
Standard Error .
10.88504
5.31065
3.84776
2.71202
2.24869
1.83306
1.61488
1.38981
1.26372
1.12360
1.07569
0.93934
0.88259
0.82005
0.75725
0.72109
0.70523
0.63643
0.60823
0.57750 .
0.55570
0.52305
0.56039
0.48106
0.48707
0.47066
0.44871
0.90286
0.44357
0.38497
0.37704
0.36210
0.35079
0.53027
0.33163
0.32197
0.32898
0.30416
0.29692
0.29455
13.27700
166.74573
1.83
F.= t2
2.170
88.858
111.830
18.366
14.315
2.735
6.874
92.465
144.716
0.350
2.451
53.454
38.906
5.065
2.781
14.080
1.508
6.400
11.480
23.865
5.166
24.116
118.372
4.853
174.121
11.416
18.509
0.153
4.967
103.337
0.076
0.022
1.146
5.860
38.984
90.269
9.677
128.334
36.350
1.471
2.544
0.088
-88-
-------
TABLE 26C. HASSELBLAD REGRESSION ON MAXFEV
(IN LITERS x 100) WITH N02
R2 = 0.90,
Variable
PI
P2
P3
P4
P5
P6
P7
P8
P9
P10
Pll
P12
P13
P14
P15
P16
P17
P18
P19
P20
P21
P22
P23
P24
P25
P26
P27
P28
P29
P30
P31
P32
P33
P34
P35
P36
P37
P38
P39
P40
DINV
N02
CONSTANT . .
N = 355, Total
B
-6.03457
-48.30260
41.57889
-11.11454
-8.10130
3.58742
-4.04938
-13.18349
• -15.08360
-0.57895
1.76683
-6.66454
-5.45445
1.89159
-1.24942
-2.66190
-0.82924
-1.58942
-2.02859
2.85027
-1.22332
-2.55197
-6.08941
1.16107
6.45446
1.61365
1.92868
0.36735
-1.02457
3.78322
0.11789
0.00412
0.27157
1.28877
1.96061
-3.04024
-1.00794
-3.44117
-1.77467
0.37584
20.67062
1.13692
. 278.07138 .
F =. 31.23, D.W..
. Standard Error
12.48188
5.73542
4.03225
2.82193
2.31121
2.06781
1.65494
1.66631
1.29482
1.20447
1.12277
1.00869
0.90251
0.83532
0.77355
0.73332
0.71759
0.64948
0.61858
0.58730
0.55963
0.53359
0.57143
0.54584
0.49633
0.47591
0.45772
0.91864
0.51909
. 0.41442
0.38364
0.38846
0.45564
0.54205
0.43366
0.32784
0.33586
0.30957
0.30313
. 0.29956
13.42492
36.64907
= L.82
F = t2
0.234
70.927
106.329
15.513
12.287
3.010
5.987
62.596
135.703
0.231
2.476
43.651
36.525
5.128
2.609
13.176
1.335
5.989
10.755
23.554
4.778
22.873
113.560
4.525
169.114
11.448
17.755
0.160
3.896
83.338
0.094
0.000
0.355
5.653
20.441
86.000
9.006
123.563
34.276
1.574
2.371
0.001
-89-
-------
TABLE 26D. HASSELRLAD REGRESSION ON MAXFEV
(IN LITERS x 100) WITH NO
R2 - 0.80,
Variable
PI
P2
P3
P4
P5
P6
P7
P8
P9
P10
Pll
P12
P13
P14
P15
P16
P17
P18
P19
P20
P21
P22
P23
P24
P25
P26
P27
P28
P29
P30
P31
P32
P33
P34
P35
P36
P37
P38
P39
P30
DINV
NOX
CONSTANT
N = 353,. Total
0
-6.01690 •
-47.87112
41.87309
-10.74706
-8.34287
3.47027
-4.18557
-13.18164
-15.09640
-0.63051
1.67274
-6.62364
-5.51711
1.86695
-1.11378
-2.64092
-0.81075
• -1.50242 .
-1.99574
2.85973
-0.91934
.-2.46238
-6.15167
1.16900
6.26174
1.81273
1.98004
0.33477
-0.93198
3.75552
0.08415
0.13316
0.25651
1.25820
2.05933
-2.95713
-1.01671 •
-3.44263
-1.72127
0.36323
17.09786
426.68439
272.74778
F = 30.07., D.W. =
Standard Error
12.52263
5.76182
4.05034
2,84165
2.32496
2.19622
1.66543
1.67306
1.29801
1.20919
1.10471
1.01282
0.90432
0.83833
0.78220
0.73465
0.71941
0.65451
0.62024
0.58864
0.62424
0.53871
0.57497
0.54641
0.52835
0.49488
0.45984
0.92185
0.48356
0.43792
0.38589
0.39061
0.45730
0.54248
0-.40153
0.33373
0.33630
0.31051
0.30558
0.30001
13.64908
299.63223
1.82
F = t2
0.231
69.028
106.878
14.303
12.877
2.497
6.316
62.075
135.267
0.272
2.293
42.769
37.220
4.959
2.028
12.923
1.270
5.269
10.354
23.602
2.169
20.893
114.472
4.577
140.455
13.417
18.541
0.132
3.715
73.545
0.048
0.116
0.315
5.379
26.304
78.513
9.140
122.922
31.727
1.466
1.569
2.028
-90-
-------
TABLE 26E. HASSELBLAD REGRESSION ON MAXFEV
(IN LITERS x 100) WITH ALL AIR POLLUTION
VARIABLES MEASURED DURING DAILY
SURVEILLANCE OF..THE ASTHMA PANEL
$2^= 0.80, N
Variable
PI
P2
P3
P4
P5
P6
P7
P8
P9
P10
Pll
P12
P13
PU
P15
P16
P17
P18
P19
P20
P21
P22
P23
P24
P25
P26
P27
P28
P29
P30
P31
P32
P33
P34
P35
P36
P37
P38
P39
P40
DINV
OZONE
CO
N02
NOX
CONSTANT
= 348, Total.
B
-6.03019
-47.75923
41.89505
-10.67156
-8.52124
3.44262
-4.29493
-13.14384
-15.11353
-0.63022
1.52866
-6.57314
-5.59551
1.87400
-1.07335
-2.60215
-0.77691
-1.48500
-1.96360
2.88701
-0.90436
-2.45571
-6.16779
1.20496
6.89114
1.79579
1.95623
0.33042
-1.00745
3.75571
0.06279
0.08442
0.24113
1.26254
2.00120 .
-2.96381
-1.02273
-3.44270
-1.72659
0.36648
16.93315
-11.68595
6.00126
-12.73782
407.04484
274.66639
F = 27.30,. D.W.
Standard Error
12.59058
5.85841
4.08670
2.90616
2.46456
2.21316
1.74278
1.70061
1.31553
1.22409
1.17651
1.03991
0.94089
0.85549
0.80849
0.81278
0.77644
0.67041
0.67914
0.63551'
0.68120
0.54946
0.58746
0.56186
0.62121
0.52532
0.46663
0.93211
0.52564
0.44239
0.38938
0.42043
0.. 46416
0.54750
0.44435
0.33894
0.34159
0.31295
0,31209
0.31735
14.15700
57.19537
269.56597
45.02380
323.01965
= 1.85
F=t2
0.229
66.459
105.095
13.484
11.954
2.420
6.073
59.736
131.987
0.265
1.688
39.953
35.367
4.799
1.763
10.250
1.001
4.907
8.360
20.637
1.763
19.975
110.232
4.599
123.057
11.686
17.575
0.126
3.673
72.073
0.026
0.040
0.270
5.318
20.283
76.462
8.964
121.016
30.607
1.334
1.431
0.042
0.000
0.080
1.588
-91-
-------
the conclusion is that most of the variation in MAXFEV is ex-
plained by differences between the individuals in the asthma
panel rather than the levels of air pollution measured during
daily surveillance.
Standard Linear Model
An alternative linear model utilizes a dummy variable for
the sub-panels tested on different dates. Sub-Panel 1, tested
on dates 1 to 10, was used .as the base group. Variable G2 = 1
if the observation was for Sub-Panel 2; G3 = 1 if the observa-
tion was for Sub-Panel 3; G4 = 1 if the observation was for Sub-
Panel 4. AGE and HEIGHT were also included as explanatory vari-
ables. A dummy variable to control for the panelist reporting a
cold was included as an explanatory variable. HUMID and TEMP
were included. The only air pollution measure included was CO,
since prior analysis indicated this variable as potentially sig-
nificant. The results are reported in Table 27.
TABLE 27. MULTIVARIATE REGRESSION OF MAXFEV WITH
SUB-PANEL (GROUP) VARIABLES
R2 = 0.25, N = 367, Total F = 13.27
Variable
AGE
HEIGHT
G2
G3
G4
COLD
CO
HUMID
TEMP
CONSTANT
3
-3.62**
4.69**
0.69
-8.52
. -48.56**
-54.50
31.35
-0.15
-0.10
128.653
Standard Error
0.52
1.34
10.33
10.84
14.78
17.12
315.92
0.73
0.77
F- t2
47.77
12.26
0.01
0.62
10.79
10.13
0.01
0.04
0.02
**Signifleant at a = 0.05
-92-
-------
The only variables having estimated coefficients .which are
statistically different from zero are those associated with mea-
suring differences between individuals, AGE, HEIGHT, COLD,, and
G4. Sub-Panel 4 is again significant as an explanatory variable
capturing between group differences that were not captured by
AGE, HEIGHT, and COLD.
None of the air pollution variables had estimated coeffi-
cients that were statistically different from zero. The conclu-
sion is again that most of the variation in MAXFEV was explained
by differences between the individuals in the asthma panel.
Lagged Pollution Variables
In an attempt to see if there was any cumulative or resid-
ual effects of air pollution on the pulmonary function variable,
the air pollution variables were lagged one day. The results of
the multivariate regression are shown in Table 28. The variable
names are different to indicate that they are lagged. AGE and
HEIGHT are the only variables that are statistically signifi-
cant. None of the lagged air pollution variables contributes to
the reduction in the explained variation in MAXFEV.
TABLE 28. MULTIVARIATE REGRESSION OF MAXFEV
WITH LAGGED AIR POLLUTION VARIABLES
R2 = 0.204, N = 290, F = 7.988
Variable
AGE
HEIGHT
COLD
OZONE- 1
CO-1
N02-1
NO-1
HUMID- 1
TEMP-1
3
-3.46
5.63
-11.70
64.71 '
-185.51
17.65
1,014.77
-0.11
-0.63
Standard Error
. 0.57
1.45
22.59
126.59
394.52
25.61
1,354.55
1.29
1.26
F-.t2
36.73
15.07
0.27
0.26
0.22
0.47
0.56
0.07
0.25
-93-
-------
Summary of the Multivariate Analysis
Both the Hasselblad model and the alternate linear model
include variables to control for differences in panelist char-
acteristics. The Hasselblad model utilizes a dummy variable
technique which assigns a value to separate variables for each
individual in the study. The alternative model uses AGE and
HEIGHT, as well as dummy variables for having a cold and for the
four sub-panels.
The primary conclusion is that the variables which capture
individual differences explain a large portion of the variance
in the dependent variable MAXFEV. There is very little varia-
tion left to be explained by the air pollution variables.
Therefore, it is concluded that MAXFEV was not statistically
sensitive to the levels of air pollution monitored during daily
surveillance of the asthma panel.
OVERALL SUMMARY OF THE ASTHMA PANEL DATA ANALYSIS
The levels of air pollution in the study area as monitored
at the Los Angeles County Air Pollution Control District station
in Azusa were light to moderate during the 40 days of asthma
panel symptom surveillance. The average values of the aero-
metric variables at the times asthma panelists reported their
symptoms (i.e., the weighted averages of these variables) are
presented in Table 9C. The weighted averages of responses to
daily symptom interviewing and to pulmonary function testing are
shown in Tables 9A and 9B. The symptom and aerometric data are
compared graphically in Appendix I.
The analysis of the symptom data showed that eye irritation
was significantly correlated with all of the aerometric vari-
ables (Tables 10, 12, and 13). The discomfort symptom headache
was significantly correlated with ozone, carbon monoxide, rela-
tive humidity, and temperature. There were few significant cor-
relations between the other discomfort symptoms and the aero- .
metric variables. Ozone was significantly correlated with the
most discomfort symptoms. Ozone was positively correlated with
reports of eye irritation, headache, shortness of breath, and
production of phlegm.
The results regarding the measurement of maximum FEVl.0,
whether adjusted for age and height, adjusted by the ratio of
maximum FEV]_ o/maximum FVC, or unadjusted maximum FEV^_Q showed
no linear association between pulmonary function and the aero-
metric variables. Most of the variation in maximum FEVi.O was
explained by differences in the age and height of the panelists.
The lack of association between pulmonary function and the aero-
metric variables may have been due to the generally low concen-
trations of air pollutants in the study area during the 40 days
-94-
-------
of symptom surveillance. An additional explanation is that
the asthma panelists tended to live fairly quiet lives by
avoiding strenuous activities and staying indoors during"the
hours of the day when air pollution levels could be expected
to be high.
-95-
-------
SECTION 8
DATA ANALYSIS—OUTDOOR WORKER PANEL
STATISTICAL DESCRIPTION OF THE OUTDOOR WORKER PANEL
This panel was intended to be composed entirely of healthy
outdoor workers. However, the comprehensive clinical examina-
tions and interviews revealed that 7 of the 95 workers had symp-
toms of bronchitis and two others had active cases of asthma.
Since the employers apparently considered all of the workers to
be healthy enough to perform their assigned jobs, data were col-
lected from all 95 workers. However, the analysis covers only
the 85 workers for whom there was sufficient data and no indica-
tion of preexisting lung disease.
The outdoor worker panel was composed of both smokers and
nonsmokers in the 21 to 50 age bracket. Slightly less than half
were smokers. Like the asthma panel, the outdoor worker panel
was divided into four sub-panels which were tested on four suc-
cessive two-week periods. The composition of the sub-panels is
shown in Table 29.
Description of the Variables
The discomfort symptom variables were recorded on the form
shown in Appendix F. The variables were dichotomous and were
coded so that "Yes" responses were equal to unity and "No" re-
sponses were equal to zero. A statistical profile of the dis-
comfort variables is given in Table 30A. The variables were the
same as those measured during daily surveillance of the asthma
panel. Variable COLD was used to measure whether or not the
panelist had a bad cold and MEDICINE was used to measure whether
or not the panelist had taken any medicine on the day of
testing.
A profile of the smoking variables is shown in Table SOB.
Variable SMOKE was given a value of unity for each day the
panelist had smoked during the two-week testing period and a
value of 'Zero for each day the panelist had not smoked during
the same period. CIGARETTES indicate the number of cigarettes
smoked each day.
Two pulmonary function variables were used in the analysis
of the outdoor worker panel. Variable MAXFEV measured the
-96-
-------
TABLE 29. COMPOSITION OF THE OUTDOOR WORKER .SUB-PANELS : :
Characteristics
Subjects enrolled in panel
(all male)
Current cigarette smokers
Subjects 21 to 30 years old
Subjects 31 to 40 years old
Subjects 41 to 50 years old
Subjects with 12 or more
years of school completed
Race other than white
Sub-
Panel 1
18
8
5
3
10
17
0 '
Sub-
Panel 2
24
14
4
8
12
19
8
Sub-
Panel 3
26
9
2
15
9
24
3
Sub-
Panel 4
18
8
7
4
7
15 -
5
Total
86
39
18
30
38
75
16
-------
TABLE 30A. STATISTICAL PROFILE OF DISCOMFORT SYMPTOM VARIABLES MEASURED DURING OUTDOOR WORKER PANEL SURVEILLANCE
v£>
00
I
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
EYES = E
- 0.174
= 0.144
- 0.977
= 0.379
- 1.726
THROAT -
- 0.132
= 0.115
= 2.714
- 0.339
- 2.172
CHEST =
- 0.149
- 0.127
- 1.890
= 0,356
= 1.973
HEADACHE
= 0.060
• 0.057
= 11.728
- 0.238
- 3.707
NAUSEA =
= 0.014
= 0.014
- 64.243
- 0.119.
- 8.143
OTHER -
- 0.257
- 0.191
- -0.760
= 0.437
- 1.114
ye discomfort now
RANGE
MINIMUM
MAXIMUM
VALID DBS
MISSING OBS =
Throat discomfort now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS -
Chest discomfort now
RANGE
MINIMUM
MAXIMUM
'' VALID OBS . =
MISSING OBS =
» Headache now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS -
Nausea now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
Other discomfort now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS "
1.000
0.000
1.000
899
0
1.000
0.000
1.000
899
0
1.000
0.000
•1.000
899
0
.1.000
0.000
1.000
899
0
1.000
0.000
1.000
899
0
1.000
0.000
1.000
899
0
VARIABLE :
MEAN
VARIANCE
KURTOSIS
' STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
. STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
HEADACHE
= 0.088
= 0.080
= 6.487
= 0.283
= 2.915
BREATH =
•= 0.088
= 0.080
- 6.487
= 0.283
= 2.915
COUGH =
= 0.356
= 0.230
- -1.636
- 0.479
= 0.602
PHELGM =
= 0.719
= 0.203
= -1.988
= 0.450
= -0.980
COLD = B
= 0.106
= 0.095
= 4.590
= 0.308
•» 2.568
MEDICINE
• 0.166
- 0.138
= 1.237
= 0.372
= 1.800
EARLIER = Headache f
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
Shortness of breath
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
Cough today
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
Phlegm today
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
ad cold today
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
- Medicine today
RANGE
MINIMUM
MAXIMUM
'VALID OBS
MISSING OBS
:arller today
- 1.000
" 0.000
- 1.000
= 899
= 0
today
= 1.000
- 0.000
- 1.000
- 899
- 0
- 1.000
- 0.000
- 1.000
- 899
= 0
- 1.000
- 0.000
•=• 1.000
• 317
= 3
1.000
0.000
1.000
899
0
- 1.000
• 0.000
- 1.000
- 899
- 0
-------
TABLE SOB. STATISTICAL PROFILE OF SMOKING
VARIABLES MEASURED DURING OUTDOOR WORKER
PANEL SURVEILLANCE :
VARIABLE: SMOKE = Smoke today
MEAN = 0.399 RANGE = 1.000
VARIANCE = 0.240 MINIMUM = 0.000
KURTOSIS = -1.830 ' MAXIMUM = 1.000
STD DEV = 0.490 VALID OBS = 899
SKEWNESS = 0.412 MISSING OBS = 0
VARIABLE: CIGARETTES = Number of cigarettes smoked
today
MEAN = 6.063 RANGE = 50.000
VARIANCE = 84.948 MINIMUM = 0.000
KURTOSIS = 2.876 MAXIMUM = 50.000
STD DEV = 9.217 VALID OBS = 899
SKEWNESS = 1.668 MISSING OBS = 0
-99-
-------
maximum FEV]_o score of three maneuvers attained each day by
each panelist. MAXFVC was the largest of two FVCs, one being
the maximum FVC score recorded at the clinical examination pre-
ceeding the two-week testing period, the other being the maximum
FVC score recorded at the clinical examination following the
testing period. The AGE and HEIGHT of each panelist were re-
corded as age in years and standing height in inches. These
data were used to adjust the MAXFEV and MAXFVC scores prior to
analysis. A profile of the MAXFEV, MAXFVC, AGE, and HEIGHT
variables is given in Table 30C. MAXFEV and MAXFVC are shown in
hundred liters (liters x 100).
The air pollution and weather variables used in this anal-
ysis are the same as those used in the asthma panel analysis.
A profile of these variables is given in Table SOD. The air
pollution variables were obtained as hourly averages measured in
ppm (CO in ppm/100). Relative humidity was estimated by hour
and expressed in percent. A discussion on how the estimates
were made is given in the paragraphs in Section 7 which describe
the variables used in the asthma panel analysis. Temperature
was obtained as hourly averages measured in degrees Fahrenheit.
There were no missing observations for any of the discom-
fort symptom variables, and none were missing for COLD, MEDI-
CINE, SMOKE, CIGARETTES, AGE, or HEIGHT. There were a large
number of unrecorded FEV scores, but a sufficient number of ob-
servations were recorded to maintain statistical credibility.
Very few FVC scores were missing.. There were a few missing ob-
servations for the-air pollution variables, but they were not
large in number. There were no missing observations for HUMID
and TEMP.
Although most of the members of the outdoor worker panel
were tested over fewer hours than members of the asthma panel,
weighted averages of exposure to air pollutant, humidity, and
temperature levels were used in the analysis of the outdoor
worker panel also. Once again, the hour at which each panelist
reported for testing was noted and the aerometric values for
that hour were included in the weighted average for that day.
The proportions of outdoor worker panelists who reported
discomfort symptoms are shown in Appendix J for each of the 54
days when symptom data were collected. The weighted averages of
the air pollutants, relative humidity, and temperature are also
given in Appendix J.for each of the same days. The charts are
provided to facilitate comparisons between proportions of panel-
ists who reported symptoms and the weighted averages of air
pollution.
-100-
-------
TABLE 30C. STATISTICAL PROFILE OF PULMONARY FUNCTION
MEASURED DURING OUTDOOR WORKER PANEL
SURVEILLANCE
VARIABLE: MAXFEV = Maximum FEVX^0 (in liters x 100)
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
376.627
5,671.790
-0.118
75.311
0.094
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
= 370.000
= 190.000
= 560.000
= 762
= 137
VARIABLE: MAXFVC = Maximum FVC (in liters x 100)t
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
= 7
570.351
,334.533
1.586
85.642
0.828
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
= 490.000
= 305.000
= 795.000
= 894
— C
VARIABLE: AGE = Age of panelist (in years)
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
38.616
60.616
-1.085
7.786
-0.331
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
25
25
50
86
0
VARIABLE: HEIGHT = Height in panelist (in inches)
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
68.440
5.165
0.626
2.273
0.262
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
13
63
76
84
2
tLargest of maximum FVC scores attained from clinical
examination, and during daily surveillance.
-101-
-------
TABLE 300. STATISTICAL PROFILE OF AIR POLLUTION
VARIABLES MEASURED DURING OUTDOOR WORKER
PANEL SURVEILLANCE
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
V
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
OZONE = Estimate
0.
0.
0.
0.
1
CO =
= 0.
0.
0.
0.
0.
N02 =
0.
0.
= -j
o!
i.
NO =
0.
0.
= 16.
0.
3.
HUMID
= 59.
= 89.
= -1.
= 9.
0.
TEMP
= 73.
= 114.
= -0.
= 10.
0.
080
004
627
065
173
for oxidant (in
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
ppm)
0.300
0.010
0.310
882
17
Carbon monoxide (in ppm/100)
036
000
309
Oil
544
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
0.060
. 0.010
0.070
893
6
.Nitrogen dioxide (in ppm)
077
002
122
041
971
Nitric oxide
020
000
607
022
757
= Relative
793
677
133
470
418 ,
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
(in ppm)
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
0.280
0.010
0.290
848
51
0.150
0.010
0.160
876
23
humidity (in percent)
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
25.000
50.000
75.000
899
0
= Temperature (in degrees
Fahrenheit)
112
389
635
695
332
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
43.000
53.000
96.000
899
0
-102-
-------
CORRELATION ANALYSIS
The results of performing Pearson and non-parametric 'corre-
lation analyses on the outdoor worker panel data are reported on
the following pages.
Pearson Correlation
The Pearson correlation coefficients for the outdoor worker
panel are shown in Table 31. The discomfort symptoms which are
significantly related to the air pollution variables at the a =
0.01 and a = 0.05 levels of significance are EYES, THROAT,
CHEST, BREATH, COUGH, AND PHLEGM. EYES, the most significant
discomfort symptom with respect to air pollution, is significant
at a = 0.01 when correlated with OZONE, NO, HUMID, and TEMP.
OZONE and TEMP are positively related to EYES, while NO and
HUMID are negatively related with less magnitude. CHEST is pos-
itively correlated with OZONE and TEMP at a =0.01 and a = 0.05,
respectively. MAXFEV is not significantly correlated with any
air pollution variable, but the sign of the relationship is fre-
quently negative.
COUGH and PHLEGM are positively correlated with OZONE, at
the a = 0.01 and a = 0.10 levels of significance, respectively.
PHLEGM is inversely related to CO and N0£ at a = 0.01. The fre-
quency of significant relationships between BREATH, COUGH, and
PHLEGM and the air pollution variables, and the fact that the
outdoor worker panel consists of both smokers and nonsmokers,
suggests that an analysis of differences between these two
groups would -be appropriate to determine if lung discomforts
are due solely to the air pollutants or to a combination of air
pollutants and individual characteristics such as smoking.
Non-Parametric Correlations
The Spearman correlation coefficients reported in Table 32
and the Kendall correlation coefficients reported in Table 33
for the outdoor worker panel confirm the relationship indicated
by the Pearson correlation coefficients for most pairs. Several
discomfort symptoms are correlated with the air pollution vari-
ables at the a = 0.01 and ct= 0.05 levels of significance:
EYES, THROAT, CHEST, BREATH, COUGH, and PHLEGM. EYES is the
symptom which is most significantly related with OZONE, NO, HUMID,
and TEMP, having non-parametric coefficients of the same magni-
tude as the Pearson coefficients.
CHEST is positively related to OZONE and TEMP, as shown in
the Pearson results, but the non-parametric correlations are
more significant. Moreover, the Kendall correlation between
CHEST and NO is now significantly negative. BREATH is signifi-
cantly related to OZONE, NO, and TEMP at the a = 0.01 level.
-103-
-------
•TABLE 31. PEARSON CORRELATION COEFFICIENTS FOR .THE OUTDOOR WORKER PANEL
o
•P-
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
MAXFEV
MAXFVC
OZONE
0 . 38***
0.04
0 . 16***
0.01
-O.Q2
0.04
0.03
0 . 19***
0 . 14***
0.09*
-0.02
-0.08**
CO
0.07**
0.01
0.03
0.05
-0.02
-0.02
-0.01
0.04
-0.04
-0.16***
-0.03
-0.03
N02
0.06*
0.04
-0.00
0.03
-0.00
-0.01
-0.02
-0.02
-0.08**
-0 . 13***
-0.00
0.02
NO
-0.16***
-0.02
-0.04
-0.02
• 0.02
-0.06*
-0.05
-0.06*
-0.04
•*0 . 08
-0.00
0.02
. . HUMID
-0.12***
-0.08**
-0.03
-0.00
-0.02
-0.05
-0.02
0.01
0.01
-0.01
0.01
0.01
. . TEMP
0.26***
0.03
0.07**
-0.00
-0.02
0.03
0.03
0 . 09***
0.06*
0.07
0.02
-0.05
-'Significant at a = 0.10
**Signifleant at a = 0.05
***Significant at a = 0.01
-------
TABLE 32. SPEARMAN CORRELATION COEFFICIENTS' FOR .THE OUTDOOR WORKER PANEL
o
Ul
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
OZONE
0 . 36***
0.03
0.15***
-0.01
-0.03
0.02
0.00
0 .17***
0 . 14***
0.11**
CO
0.07**
0.00
0.17
0.05*
-0.02
-0.04
-0.00
0.04
-0.06**
-0.17***
. . N02 .
0.06**
0.04
0.02
0.02
.0.00
-0.01
-0.01
0.02
-0.07
-0 . 07*
. . . NO. .
-0.20***
-0.03
-0.10***
0.02
0.01
-0.04
-0.02
-0.11***
-0.10***
-0.19***
. . HUMID .
-0 . 12***
-0.08***
-0.03
• -0.00
-0.02
-0.05*
-0.02
0.01
0.02
0.00
. . TEMP
0.28***
0.04
0.08***
-0.01
-0.03
0.03
0.02
0 . 10***
0.07**
0.07*
''Significant at a = 0.10
**Significant at a = 0.05
***Significant at a = 0.01
-------
TABLE 33. KENDALL CORRELATION COEFFICIENTS FOR THE .OUTDOOR WORKER PANEL :
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
M HEADACHE
g EARLIER
i
BREATH
COUGH
PHLEGM
OZONE
0.30***
0.02
0 . 12***
-0.01
-0.03
0.02
0.00
0'. 14***
0.12***
0.09**
CO
0 . 06**
0.00
0.02
0 . 04*
-0.01
-0.03
-0.00
0.04
-0.05**
-0.16***
N02
0.05**
0.03
0.02
0.02
0.00
-0.01
-0.01
0.02
-0.06**
-0.06*
NO
-0.19***
-0.03
-0.09 ***
0.02
0.01
-0.04
-0.01
-0.10***
-0.10***
-0.18***
HUMID
-0.11***
-0.07***
-0.03
-O-.OO
-0.02
-0.05*
-0.02
0.01
0.02
0.00
TEMP
0 . 23***
0.04
0.07***
-0.00
-0.02
0.03
0.02
-0.08***
0.06**
0.06*
''Significant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
BREATH is positively correlated with OZONE and TEMP, but in-
versely correlated with NO. COUGH is directly related to
OZONE and TEMP at the a = 0.01 and a = 0.05 levels of signif-
icance, respectively. COUGH is negatively related to CO and
NO at a = 0.05 and a = 0.01, respectively. PHLEGM is positive-
ly correlated with OZONE and TEMP at significance levels of
a = 0.05 and a = 0.10, but PHLEGM is negatively correlated
with CO and NO at a = 0.01.
NAUSEA and HEADACHE EARLIER do not appear to be associated
with any of the air pollution variables. This is true whether
one examines the Pearson or non-parametric correlation coeffi-
cients .
The similarity of correlation between the lung variables,
BREATH, COUGH and PHLEGM, and the air pollution variables in
both the Pearson and non-parametric correlations suggests reli-
ability of the results. The number of significant correlations
for this panel is strikingly larger than the number of signifi-
cant correlations for the asthma, bronchitis, and athlete
panels. In fact, one must conclude from the correlation analy-
sis that the panel of healthy outdoor workers was more sensitive
to air pollution than were members of the other three panels.
The correlations between the lung variables and the air pollu-
tion variables are again examined in the following pages, but
this time separately for the smokers and nonsmokers.
Pearson Correlation for Smokers
Pearson correlation coefficients for the smokers in the
outdoor worker panel are summarized in Table 34. EYES, THROAT,
CHEST, HEADACHE, NAUSEA, OTHER, BREATH, COUGH, PHELGM, and
MAXFEV are correlated with the air pollution data at the a =
0.01 or a = 0.05 level of significance. Thus, there are more.
significant correlations for smokers than for the panel as a
whole.
EYES is significantly correlated at a = 0.01 with OZONE,
NO, HUMID, and TEMP. There is a positive relation between EYES
and OZONE and TEMP, and there is a smaller, negative relation
between EYES and NO and HUMID. THROAT is positively related to
OZONE and CO at a = 0.05. CHEST is significantly related to
OZONE at a = 0.01. HEADACHE, which was not significant when
analyzing the entire panel, is positively correlated with OZONE
and CO at a = 0.05. NAUSEA is another discomfort symptom which
was not significant when all workers were analyzed, but for
smokers alone it is significantly correlated with HUMID at a =
0.05. The sign of this relationship is negative as it was for
the entire panel.
BREATH, COUGH and PHLEGM are again correlated with air
pollution variables. BREATH is positively related to OZONE,
CO, and TEMP at d = 0.01, a = 0.05, and a = 0.01, respectively.
-107-
-------
TABLE 34. PEARSON CORRELATION COEFFICIENTS FOR SMOKERS IN
• : • : : : THE OUTDOOR WORKER PANEL .:.::. • : : : : : :
o
00
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
MAXFEV
MAXFVC
OZONE . .
0 . 40***
0 . 11**
0 . 18***
0.13**
-0.04
0.11**
0.11*
0.31***
0 . 18***
0.04
0 . 14**
-0.08
CO
0.08
0 . 11**
0.04
0 . 13**
-0.00
0.04
0.02
0.11**
-0.03
-0.14**
-0.04
-0.02
N02 . .
0.04 '
0.04
. 0.03
-0.02
0.02
0.02
-0.09*
0.03
-0.13**
-0.04
-0.12**
. 0.00
... NO
-0.18***
-0.00
-0.06
-0.02
0.05
-0.07
-0.08
-0.08
-0.06
-0.07
-0.11**
-0.03
. . . HUMID . . .
-0.17***
-0.04
-0.00
0.02
-0.12**
-0.04
0.01
-0.02
0.07
-0.01
0.02
0.01
. . . TEMP
0.36***
0.05
0.09*
0.05
0.01
-0.00
0.07
0 . 19***
0.09*
0.02
0.11**
-0.09*
'-Significant at a = 0.10
**Significant at a = 0.05
***Significant at a = 0.01
-------
COUGH is positively related to OZONE at a = 0.10, and negatively
related to N02 at a = 0.05. PHLEGM is negatively correlated with
CO at a = 0.05. Comparing the Pearson correlations for smpkers
to those for the entire panel, it is evident that there are more
significant correlations for smokers, and the sign of those cor-
relations is the same for most pairs.
Spearman and Kendall Correlations for Smokers
The Spearman and Kendall correlation coefficients are shown
in Tables 35 and 36. Note that the Spearman and Kendall results
strongly agree with the Pearson results for most pairs. BREATH
is one discomfort symptom which has a higher number of signifi-
cant correlations in the non-parametric results. More reliabil-
ity should be placed on the non-parametric correlation analysis,
since the data for the discomfort symptoms do not represent a
continuous, cardinal form of measurement.
Pearson Correlation Coefficients for Nonsmokers
Pearson correlation coefficients for nonsmokers in the out-
door worker panel are shown in Table 37. Variables EYES,
THROAT, CHEST, COUGH, PHLEGM and MAXFEV are significantly corre-
lated with the air pollution variables at the a = 0.01 or .a =
0.05 levels of significance. The total number of significant
correlations is smaller for the nonsmokers than those for the
.entire panel. This may be an indication that smoking is another
•variable to be considered when analyzing sensitive or insensi-
tive subsets.
EYES is positively correlated with OZONE and TEMP at the
a = 0.01 level of significance. EYES is negatively correlated
with NO at a = 0.01 and with HUMID at a = 0.05. THROAT is neg-
atively related to HUMID and COUGH is positively related to
OZONE at the a = 0.01 level of significance. HEADACHE, NAUSEA,
OTHER, and HEADACHE EARLIER, are not significantly related to
the air pollution variables at a = 0.01 or a = 0.05.
The number of significant correlations for the lung vari-
ables BREATH, COUGH, and PHLEGM is much lower than those for
smokers and the panel as a whole. BREATH is not significantly
paired with any air pollution variable at the a = 0.01 or a =
0.05 level of significance. COUGH and PHLEGM are positively
correlated with OZONE at a = 0.01. The absence of significant
correlations between the lung variables and the rest of the air
pollution variables supports the conclusion that smoking is a
significant variable in the analysis of discomfort symptoms.
MAXFEV is negatively correlated with OZONE at the a = 0.01
level of significance. The direction of this relationship is
consistent with the rest of the panels, as would be expected.
However, this is the only group in any of the panels studied
-109-
-------
TABLE 35. SPEARMAN CORRELATION COEFFICIENTS FOR SMOKERS IN
THE OUTDOOR WORKER PANEL .... . . . . : . : .
o
i
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
OZONE
0.39***
0.10**
0..17***
0 . 11**
-0.04
0 . 10**
0.09**
0.27***
0 . 20***
-0.03
CO
0.07*
0.09*
0.03
0.13*
-0.01
0.04
0.02
0.11**
-0.05
-0.15**
. . N02
0.04
0.06
.0.04
0.01
0.01
0.04
-0.08*
0.07*
-0.12**
-0.02
. . NO ...
-0.24***
-0.05
-0.09**
-0.04**
0.02
-0.05
-0.09**
-0.13***
-0.14***
-0.12**
. . . HUMID. .
-0.16***
-0.03
0.00
0.00
-0.12**
-0.03
0.00
-0.02
0.07*
,0.01
TEMP
0 . 38***
0.06
0.09**
0 . 04**
-0.01
-0.01
0.05
0.18***
0.10**
0.00
^'Significant at a = 0.10
**Significant at a = 0.05
***Significant at a =0.01
-------
TABLE 3.6. KENDALL CORRELATION COEFFICIENTS FOR SMOKERS IN
. THE OUTDOOR WORKER PANEL .... ..::.:..:
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
OZONE
0. 33***
0 . 08**
0 . 14***
0.09**
-0.03
0.08**
0.08**
0.22***
0.17***
-0.02
CO
0.06*
0.08*
0.03
0 . 11***
-0.01
0.03
0.02
0 . 10**
-0.05
-0.13**
. N02
0.04 '
0.05
. 0.03
0.01
0.01
0.03
-0.07*
0.06*
-0.10**
-0.02
NO
-0. 23***
-0.05
-0.08**
-0.04
0.02
-0.05
-0.09**
-0.13***
-0. 14***
-0.11**
. . HUMID .
-0.15***
-0.03
0.00
0.00
-0.11**
-0.03
0.00
-0.02
0.07*
0.01
. . TEMP
0.31***
0.05
0 . 07**
0.03
-0.01
-0.01
0.04
0.15***
0.08**
0.00
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
TABLE 37. PEARSON CORRELATION COEFFICIENTS FOR NONSMOKERS IN
. . THE OUTDOOR WORKER PANEL . . : : . : ::..:.
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
MAXFEV
MAXFVC
OZONE
0 . 37***
-0.02
0. 15***
-0.08*
-0.01
0.01
-0.01
0.08*
0. 12***
0 . 19***
-0. 13***
-0.08 .
. . CO.
0.08*
-0.06
0.05
0.00
-0.02
-0.06
-0.03
0.00
0.00
-0.09
-0.05
,-0.05
. . N02 .
0.08*
0.05
.0.01
0.07
-0.01
-0.02
0.02
-0.05
-0.01
-0.11
0.03
.. 0.01 .
NO. ...
-0.14***
-0.30
-0.00
-0.01
-0.01
-0.04
-0.04
-0.02
0.02
-0.02
0.05
-0.00
. . . HUMID
-0.10**
-0 . 12***
-0.07
-0.01
. 0 . 08*
-0.07
-0.04
0.04
-0.04
-0.05
0.01
0.03
. TEMP
0.20***
0.02
0.06
-0.04
-0.05
0.04
0 .00
0.01
0.05
0.07
-0.03
-0.04
*Signifleant at a =0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
where the MAXFEV was significantly related to any of the air
pollution variables. And the signs of the estimated coeffi-
cients are in the direction expected. Healthy nonsmoking 'out-.
door workers may be more responsive to acute photochemical air
pollution than.any of the other groups studied.
Spearman and Kendall Correlations for Nonsmokers
The Spearman correlation coefficients, Table 38, and
Kendall correlation coefficients, Table 39, for nonsmokers rein-
force the Pearson results. There are fewer significant correla-
tions for nonsmokers than for the entire panel, especially for
the lung variables BREATH, COUGH, and PHLEGM.
Variable EYES has the largest number of significant corre-
lations with the air pollution variables. The direction and
magnitude of the non-parametric correlations are smiliar to the
Pearson correlations. This is true for THROAT also. Variable
CHEST has more significant pairs in the non-parametric results.
CHEST is positively correlated with OZONE at a = 0.01 and TEMP
at a = 0.05, and it is negatively correlated with HUMID and NO
at a = 0.05. The larger number of significant correlations for
CHEST indicates that the non-parametric analysis may capture
associations which the Pearson analysis missed. HEADACHE,
NAUSEA, and OTHER also have significant correlations which were
not indicated in the Pearson results. The signs are consistent
for both Pearson and non-parametric results. HEADACHE EARLIER
is not significantly related to any pollution variable included
in the study.
The lung variables, BREATH, COUGH, and PHLEGM, have few
significant non-parametric correlations with the air pollution
data. BREATH is not significantly related to any air pollution
variables at a = 0.01 or a = 0.05. COUGH and PHLEGM are posi-
tively related to OZONE at a = 0.01. PHLEGM is also signifi-
cantly correlated with NO at a = 0.05.
Summary of the Correlation Analysis
Both the Pearson and the non-parametric correlation coeffi-
cients for smokers, nonsmokers, and for the panel as a whole indi-
cate that EYES is the most frequently significant of the dis-
comfort symptoms correlated with the air pollution variables.
The lung variables BREATH, COUGH, and PHLEGM have fewer signifi-
cant correlations with the air pollution variables for non-
smokers than for smokers or for the panel as a whole. This sug-
gests that the nonsmokers in the outdoor worker panel were less
sensitive to air pollution. However, when MAXFEV is considered,
the reverse seems to have been true. MAXFEV is negatively cor-
related with OZONE at a = 0.01 only for the nonsmokers. In
order to identify further differences which may be attributable
-113-
-------
TABLE 38. SPEARMAN CORRELATION COEFFICIENTS FOR NONSMOKERS IN
: THE OUTDOOR WORKER PANEL ...
•P-
i
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
OZONE
0. 33***
-0.04
0 . 14***
-0.10
-0.02
-0.03 ,
-0.05
0.07*
0 . 10***
0 . 18***
CO
0 . 08**
-0.05
0.04
-0.01
-0.01
-0.08**
-0.02
-0.00
-0.02
-0.11*
N02. . . .
0.07**
0.03
0.03
0.04
0.00
-0.04
0.03
-0.02
-0.01
-0.04
. NO .
-0.17***
-0.01
-0.09**
0.07*
0.00
-0.02
0.03
-0.06*
rO.02
-0.13**
. HUMID
-0.09**
-0. 12***
-0.08**
-0.00
0.07**
-0.06*
-0.03
0.04
-0.04
-0.05
TEMP
0.21***
0.03
0.08**
-0.04
-0.05
0.06*
-0.00
0.02
0.06*
0.08
*Signifleant at a = 0.10
**Significant at a = 0.05
***Signifleant at a = 0.01
-------
TABLE 39. KENDALL CORRELATION COEFFICIENTS FOR NONSMOKERS IN
Ul
i
...::::.::::: THE: OUTDOOR. WORKER: PANEL. :::.:....:.::.::..:::::::
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
0
-0
0
-0
-0
-0
-0
0
0
.0
OZONE
. 28***
.03
. 12***
.08**
.02
.03
.05
.06*
. 09***
. 15***
0
-0
0
-0
-0
-0
-0
-0
-0
-0
CO
.07**
.05
.03
.01
.01
.07**
.02
.00
.02
. 10*
0
0
0
0
0
-0
0
-0
-0
-0
N02
.06**
.02
.02
.03
.00
.03
.03
.02
.00
.03
-0
-0
-0
0
0
-0
0
-0
-0
-0
NO . . --
.16***
.01
.08**
.06*
.00
.02
.03
. 06*
.02
.12**
-0
-0
-0
-0
0
-0
-0
0
-0
-0
HUMID
.09**
.12***
.07
.00
. 07**
..06*
.03
.04
.04
.04
TEMP
0 . 18***
0.03
. 0 . 06**
-0.03
-0.04
0.05*
-0.00
0.02
-0.05*
0.07
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
to smoking, smokers and nonsmokers are tested separately in the
remaining analyses of data collected from this panel.
MEASURES OF ASSOCIATION BETWEEN VARIABLES
Three measures of association are applied to the discomfort
symptoms, smoking, pulmonary function, and air pollution vari-
ables in the following pages.
Simple Linear Regressions with FEV-j^ Q and FEV1 Q/FVC
Single variable regressions were performed with MAXFEV as
the dependent variable. None of the regressions using MAXFEV
and data collected from the whole panel resulted in statisti-
cally significant results. However, significant results were
obtained when the outdoor worker panel was divided into smokers
and nonsmokers . The estimated equations are reported in
Tables 40 and 41.
Variables OZONE, N02, NO, and TEMP are significantly related
to MAXFEV for smokers at the a = 0.05 level of significance.
However, the sign of the coefficient for OZONE is positive.
Variables CO and HUMIDITY are not significant for smokers. For
nonsmokers, only OZONE is statistically significant at a = 0.01,
and the sign of the coefficient is now negative. None of the
other air pollution variables are statistically significant for
nonsmokers .
Regression equations were also estimated using maximum
FEVi . 0 divided by maximum FVC as the dependent variable. The
FVC score used for a panelist was the largest of two FVC scores
obtained by the panelist, one recorded at the clinical examina-
tion given before the two weeks of daily surveillance and the
other recorded at the clinical examination following daily sur-
veillance. This MAXFVC variable was treated as a constant for a
given individual. It does not vary with the air pollution
values since it was not measured at the same time as MAXFEV and
the air pollution variables. The transformation maximum FEVi.Q/
maximum FVC measures the proportion of MAXFEV with respect to
MAXFVC. This transformation is intended to control for individ-
ual differences; therefore, when the transformation is used, it
is not necessary (or appropriate) to control for age, height, or
other characteristics of an individual that could influence
MAXFEV. The regression results are reported in Tables 42 and
43.
Table 42 presents the results for smokers. N0£ is the only
air pollution variable that is not statistically significant.
The results for nonsmokers, reported in Table 43, are comparable
except that OZONE and N02 are not statistically significant. In
-116-
-------
TABLE 40. SIMPLE REGRESSIONS WITH MAXFEV (IN LITERS
x 100) AS DEPENDENT VARIABLE FOR SMOKERS
MAXFEV = 341.192 + 151.352(OZONE)** R2 = 0.020
MAXFEV = 362.554 - 262.267(CO) R2 = 0.002
MAXFEV = 371.185 - 237.880(N02)** R2 = 0.015
MAXFEV = 362.791 - 586.121(NO)** R2 =0.013
MAXFEV = 342.620 + 0.190(HUMID) R2 = 0.001'
MAXFEV = 293.137 + 0.821(TEMP)** R2 = 0.013
**Signifleant at a =.0.05
TABLE 41. SIMPLE REGRESSIONS WITH MAXFEV (IN LITERS x
- 100) AS DEPENDENT VARIABLE FOR .NONSMOKERS- .
MAXFEV
MAXFEV
MAXFEV
MAXFEV
MAXFEV
MAXFEV
= 404
= 404
= 388
= 387
= 387
= 409
.462
.059
.221
.640
.930
.055
- 149. 976 (OZONE)***
- 332.504(00)
+ 49.460(N02)
+ 291. 371 (NO)
+ 0.073 (HUMID)
- 0.227 (TEMP)
R
R
R
R
R
R
2
2
2
2
2
2
= 0
= 0
= 0
= 0
= 0
= 0
.017
.003
.001
.003
.000
.001
***Signifleant at a = 0.01
-117-
-------
TABLE 42. SIMPLE REGRESSIONS WITH FEV]_ o/FVC AS
DEPENDENT VARIABLE FOR SMOKERS'
FEV1 o/FVC = 0.631 + 0.688(OZONE)*** R2 = 0.024
FEVj^ o/FVC = 0.816 - 3.966 (CO)*** R2 = 0.021
FEV1 o/FVC = 0.712 - 0.520(N02) R2 = 0.004
FEVX o/FVC = 0.757 - 4.440(NO)*** R2 = 0.096
FEV-, O/FVC = 0.931 - 0.004 (HUMID)** R2 = 0.016
•L •
FEV, o/FVC = 0.201 + 0 .007 (TEMP)*** R2 = 0.047
**Signifleant at a = 0.05
***Signifleant at a = 0.01
TABLE 43. SIMPLE REGRESSIONS WITH'FEVi 0/pVC AS
DEPENDENT VARIABLE FOR NONSMOKERS
Q/FVC = 0.670 + 0.288(OZONE) R2 = 0.004
0/FVC =0.815 - 3.744(CO)*** R2 = 0.019
^ o/FVC = 0.653 + 0.169(N02) R2 = 0.001
o/FVC = 0.772 - 4.815(NO)*** R2 = 0.116
o/FVC = 1.047 - 0.006 (HUMID)*** R2 = 0.036
^ o/FVC = 0.179 + 0.007(TEMP)*** R2 = 0.057
***Signifleant at a = 0.01
-118-
-------
summary, the single variable regressions indicate that CO, NO,
and HUMIDITY are inversely related to the variable maximum
FEVl.Q/maximum FVC for both smokers and nonsmokers. Variable
OZONE is positively related to maximum FEV]_ Q/maximum FVC, but
the estimated coefficient is not statistically significant for
nonsmokers.
Contingency Tables for FEV, « and Air Pollution Variables
Contingency tables were constructed and x^ tests were ap-
plied between MAXFEV and the air pollution variables. Three
categories for all variables were constructed with the transi-
tion points being one standard deviation above and below the
mean with the exception of the NO variable. The two categories
for NO were constructed with the division occurring at the mean.
The contingency tables and computed X^ values are reported in
Tables 44A through 44D.
The hypothesis of independence for a (3x3) table requires
four degrees of freedom. The computed x^ must exceed 9.49 at
a = 0.05 if the hypothesis of independence is to be rejected.
For the (3x2) table there are two degrees of freedom, and the
computed x must exceed 5.99 at a = 0.05 in order to reject the
hypothesis of independence. As shown in the table, all of the
computed x values are large enough to reject the null hypoth-
esis. This lends support to the hypothesis of dependence or as-
sociation between MAXFEV and the pollution variables OZONE, CO,
N02, and NO for outdoor worker panel.
Linear Probability Model
A linear probability formation was estimated using the di-
chotomous discomfort variables as dependent variables. A dif-
ferent air pollution variable was used as the explanatory vari-
able in each regression. The specifications were estimated for
the entire outdoor worker panel. The regressions which were
significant at a = 0.10 or better are shown in Table 45. Then,
the specifications were estimated for the smokers; the signifi-
cant regressions for smokers are shown in Table 46. Finally,'
the specifications were estimated for the nonsmokers on the
panel; the significant regressions are shown in Table 47.
For the outdoor worker panel as a whole, EYES is signifi-
cantly correlated with OZONE, CO, N02, NO, HUMID, and TEMP.
Variables NO and HUMID have estimated coefficients of.negative
sign. EYES is directly related to the other variables. THROAT
is significantly related only to HUMID. CHEST is significantly
related to OZONE and TEMP. OTHER is inversely related to NO.
BREATH is directly related to OZONE and TEMP and inversely re-
lated to NO. COUGH is directly related to OZONE and TEMP and
inversely related to N02- None of the other discomfort
-119-
-------
TABLE 44A. CONTINGENCY TABLE FOR MAXFEV (IN LITERS x 100)
AND OZONE FOR THE OUTDOOR WORKER PANEL
MAXFEV
< 302
Count
Row 7»
Column 7o
Total 7,
302 * MAXFEV * 452 .
MAXFEV
Total
Total
Count
Row 7o
Column 7o
Total 7o
> 452
Count
Row 7o
Column 7o
Total 70
Count
Percent
OZONE < .015
.54
21.2
52.9
6.0
39
7.3
38.2
4.3
9
8.1
8.8
1.0
102
11.3
.015 * OZONE s .145
t
156
. 61.2
24.1
17.4
412
77.3
63.6
45.8
80
72.1
12.3
8.9
648
72.1
.145 < OZONE
. 45
17.6
30.2
5.0
82
15.4
55.0
9.1
22
19.8
14.8
2.4
149
16.6
Row Total
255
28.4
533
59.3
111
12.3
ro
o
37.93
-------
TABLE 44B. CONTINGENCY TABLE FOR MAXFEV (IN LITERS"
X 100) AND CO. FOR THE OUTDOOR WORKER PANEL
MAXFEV < 302
Count
Row 7o
Column 7<>
Total 7o
302 s MAXFEV s 452
Count
Row 70
Column 7o
Total 7o
MAXFEV > 452
Count
Row 70
Column %
Total 7o
Total Count
Total Percent
CO < .025
16
6.3
12.1
1.8
91
18.2
73.5
10.8
19
17.1
14.4
2.1
132
14.7 .
.025 ^ CO s .047
177
69.4
29.5
19.7
352
66.0
58.6
39.2
72
64.9
12.0
8.0
601
66.9
.047 < CO
62
24.3
37.3
6.9
84
15.8
50.6
9.3
20
18.0
12.0
2.2
166
18.5
Row Total
255
28.4
533
59.3
111
12.3
X = 24.43
-------
-TABLE 44.C;, -CONTINGENCY TABLE FOR MAXFEV (IN LITERS
100) AND N02 FOR THE OUTDOOR WORKER PANEL
MAXFEV < 302
Count
Row 7o
Column 7o
Total 70
302 =s MAXFEV ^ 452
Count
Row 7o
Column 70
Total 70
MAXFEV > 452
Count
>Row 7o
Column %
Total 7o
Total Count
Total Percent
N02 < .036
16
6.3
13.8
1.8
91
17.1
78.4
10.1
9
8.1
7.8
1.0
116
12.9
.036 s N02 s .118
214
83.9
30.8
23.8
370
73.2
56.2
43.4
90
81.1
13.0
10.0
694
77.2
.118 < N02
25
9.8
28.1
2.8
52
9.8
58.4
5.8
12
10.8
13.5
1.3
89
9.9
Row Total
255
28.4
533
59.3
111
12.3
ro
CO
20'. 78
-------
TABLE 44D. CONTINGENCY TABLE FOR MAXFEV (IN LITERS x
100.) AND NO FOR OUTDOOR WORKER PANEL
to
MAXFEV < 302
Count
Row 7o
Column 7o
Total 7o .
302 *
MAXFEV z 452
Count
Row 7o .
Column 70
Total 7o
MAXFEV > 452
Total
Total
Count
Row 70
Column 70
Total 7o
Count
Percent
NO < .02
r
130
51.0
22.0
. 14.5
387
72.6
65.5
43.0
74
66.7
12.5
8.2
591
65.7
NO * .02
125
49.0
40.6
13.9
146
27.4
47.4
16.2
37
33.3
12.0
4.1
308
34.3
Row Total
255
28.4
533
59.3
111
12.3
= 35.87
-------
TABLE 45. SIMPLE LINEAR PROBABILITY REGRESSIONS
FOR THE ENTIRE OUTDOOR WORKER PANEL (N = 882)
-
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
(EYES)
(EYES)
(EYES)
(EYES)
(EYES)
(EYES)
(THROAT)
(CHEST)
(CHEST)
(OTHER)
(BREATH)
(BREATH)
(BREATH)
(COUGH)
(COUGH)
(COUGH)
(PHLEGM)
(PHLEGM)
(PHLEGM)
= -0
0
0
0
0
= -0
0
0
= -0
=• 0
0
0
. = -0
0
0
0
0
0
0
.003
.086
.139
.230
.468
.496
.299
.077
.031
.277
.020
.104
.093
.271
.437
.150
.482
.786
.688
+ 2.
+ 2.
+ 0.
- 2.
- 0.
+ 0.
- 0.
+ 0.
+ 0.
- 1.
+ 0.
- 0.
+ 0.
+ 1.
- 0.
+ 0.
+ 0.
- 6.
- 1.
241 (OZONE)***
45 6 (CO)**
571(N02)*
7 39 (NO)***
00 5 (HUMID)***
00 9 (TEMP)***
003 (HUMID)**
87 2 (OZONE)***
002 (TEMP)**
126 (NO)*
840 (OZONE)***
742 (NO) *
00 2 (TEMP)***
008 (OZONE)***
945(N02)**
00 3 (TEMP)**
641 (OZONE)*
9 17 (CO)***
673(N02)***
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
R2
R2
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
.147
.005
.004
.025
.015
.062
.006
.026
.005
.003
.038
.003
.009
.019
.006
.004
.008
.026
.018
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-124-
-------
TABLE 46. SIMPLE LINEAR PROBABILITY REGRESSIONS
FOR THE SMOKERS (N = 354.)
Prob(EYES)
Prob(EYES)
Prob(EYES)
Prob(EYES)
Prob( THROAT)
Prob( THROAT)
Prob(CHEST)
Prob( CHEST)
Prob(HEADACHE)
Prob(HEADACHE)
Prob(NAUSEA)
Prob (OTHER)
Prob(BREATH)
Prob(BREATH)
Prob(BREATH)
Prob(COUGH)
Prob(COUGH)
Prob(COUGH)
Prob(PHLEGM)
0.007 + 2.458(OZONE)*** R^ = 0.159
0.268 - 3.424(NO)*** R2 = 0.032
0.652 - 0.007(HUMID)*** R2 = 0.027
-0.839 + 0.014(TEMP)*** R2 = 0.126
0.131 + 0.617(OZONE)** R2 = 0.011
0.053 + 3.788(CO)*** R2 = 0.011
0.139 + 1.181(OZONE)*** R2 = 0.033
-0.041 + 0.004(TEMP)* R2 = 0.008
0.033 + 0.500(OZONE)** R2 = 0.016
-0.031 + 2.969(CO)** R2 = 0.016
0,134 - 0.002(HUMID)** R2 = 0.013
0.236 + 0.744(OZONE)** R2 = 0.011
0.013 + 1.651(OZONE)*** R2 = 0.093
0.019 + 3.709(CO)** R2 = 0.013
0.341 + 0.007(TEMP)*** R2 =0.036
0.483 + 1.356(OZONE)*** R2 = 0:033
0.733 - 1.803(N02)** R2 = 0.017
0.266 + O.OOS(TEMP)* R2 = 0.009
0.953 - 5.733(CO)** R2 = 0.021
*Significant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-125-
-------
TABLE 47. SIMPLE LINEAR PROBABILITY REGRESSIONS
FOR THE NONSMOKERS (N = 528)
*
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
(EYES)
(EYES)
(EYES)
(EYES)
(EYES)
(EYES)
(THROAT)
(CHEST)
(HEADACHE)
(NAUSEA)
(BREATH)
(COUGH)
(PHELGM)
= -0
= 0
0
0
0
= -0
= 0
0
0
= -0
0
0
0
.009
.059
.112
.203
.368
.309
.313
.038
.076
.036
.026
.133
.167
+ 2
+ 2
+ 0
- 2
- 0
+ 0
- 0
+ 0
- 0
+ 0
+ 0
+ 0
+ 1
.087 (OZONE)
.577
.661
(CO)*
(N02)*
***
R2
R
2
R2
.264 (NO)***
.004 (HUMID)
**
.006 (TEMP)***
.004 (HUMID)***
.640
.289
.001
.274
.710
.227
(OZONE)
(OZONE)
(HUMID)
(OZONE)
(OZONE)
(OZONE)
***
* v
*
*
***
***
R
R
R
R
R
R
R
R
R
R
2
2
2
2
2
2
2
2
2
2
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
.139
.007
.006
.020
.009
.038
.014
.021
.007
.006
.007
.014
.036
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-126-
-------
symptom variables are significantly related to any of the air pol-
lution or weather variables for trie panel as a whole.
s
EYES is not significantly related to CO or N02 for the
smokers in the panel. However, the relationship of eye discom-
fort with OZONE, NO, HUMID, and TEMP appears to be stronger for
smokers than for nonsmokers since the magnitude of the estimated
slope coefficients is larger. THROAT is significantly related
to OZONE and CO for the smokers. CHEST is directly related to
OZONE and TEMP as with the whole panel. HEADACHE is directly
related to OZONE and CO. OTHER is significantly related to
OZONE. BREATH is strongly related to OZONE, CO, and TEMP.
COUGH is directly related to OZONE and TEMP and inversely re-
lated to N02- PHLEGM is inversely related to CO for the
smokers.
There are fewer significant linear probability specifica-
tions for the nonsmokers in the panel. However, EYES is again
significantly related to all of the air pollution variables for
the nonsmokers. THROAT remains inversely related to HUMID and
not significantly related to the other air pollution and weather
variables. CHEST is no longer significantly related to TEMP,
but chest discomfort remains directly related to OZONE. HEAD-
ACHE is now inversely related to OZONE, but the level of statis-
tical significance is not high. NAUSEA is now directly related
to HUMID, but the relation is not strong. BREATH is no longer
significantly related to NO or TEMP, but remains directly re-
lated to OZONE. Similarly, COUGH remains directly related to
OZONE, but is no longer significantly related to N0£ or TEMP.
PHLEGM is related to OZONE, but not to CO or N0£ for the non-
smokers .
SELECTED MULTIVARIATE LINEAR PROBABILITY REGRESSIONS
The form of the multivariate linear probability model is
k
6
. = Prob(Y = 1) = 3 +Xii • i = 1..-..N.
This model was introduced and discussed in the analysis of the
asthma panel data. The predicted value of the dependent vari-
able, the qualitative discomfort symptom, can be interpreted as
the probability that the discomfort symptom is present given
values of the explanatory variable. Six of the discomfort
symptom variables were used as dependent variables: EYES,
THROAT, CHEST, BREATH, and PHLEGM. They were chosen because
they were most frequently related to at least some of the air
-127-
-------
pollution variables. All of the air pollution variables were
used as explanatory variables. The data were restricted to
those outdoor workers who reported not having a cold and not
taking medication. The results are reported in Table 48.
One of the regressions, that for THROAT, resulted in a
total F which indicated that none of the estimated coefficients
were significantly different from zero. All of the other re-
gressions resulted in at least one explanatory variable whose
coefficient was statistically significant at a = 0.05 or better.
The EYES regression resulted in OZONE as the only statis-
tically significant explanatory variable. None of the other air
pollution variables were significant in reducing the variation
in eye discomfort. The estimated equation indicates that when
OZONE increases by 0.10, the probability of experiencing eye dis-
comfort increases by more than 0.20, i.e., by more than 20 per-
cent.
CHEST and BREATH are significantly related to OZONE and
HUMID in the multivariate regressions. The estimated coeffi-
cients are positive. When the level of OZONE increases by 0.10,
the estimated probability of experiencing chest discomfort and
shortness of breath increases by about 0.12 or 12 percent.
COUGH is significantly and positively related to OZONE, NO, and
HUMID, and COUGH is inversely related to CO. When OZONE in-
creases by 0.10, the probability of having a cough increases by
almost 0.03 or 3 percent. An increase in NO of 0.01 is esti-
mated to increase the probability of having a cough by almost
0.01 or 1 percent. An increase in CO of 0.10 is estimated to
cause a reduction in the probability of having a cough by:about
0.10 or 10 percent. PHLEGM is significantly related to OZONE and
CO only. The estimate indicated a 0,10 increase in OZONE is esti-
mated to cause an increase in the probability of having phlegm by
about 0.15 or 15 percent, when CO increases by 0.01, the probabil-
ity of having phlegm is reduced by about 0.14 or 14 percent.
In general, the subjects in the outdoor worker panel show
more sensitivity in reporting discomfort symptoms than demon-
strated by the asthma, athlete, or bronchitis panels. This may
be caused by a more direct exposure to the environment or to the
greater physical stress of working prior to being tested.
Eye discomfort was again the symptom that was reported most
consistently. But the outdoor workers also exhibited signifi-
cant response to the air pollution variables in reporting throat
discomfort, chest discomfort, shortness of breath, cough, and
phlegm. When multivariate linear probability specifications
were estimated, throat discomfort was not significantly related
to any of the air pollution variables. Eye discomfort, chest
discomfort, and shortness of breath were significantly related
-128-
-------
TABLE 48. MULTIVART.ATE LINEAR PROBABILITY FUNCTIONS FOR EYE. THROAT,
AND CHEST DISCOMFORT, SHORTNESS OK BREATH, COUCH, AND PHLEGM--ALL
OUTDOOR WORKER PANELISTS
(STANDRAD ERRORS IN PARENTHESES)
Prob(EYES) = -0.548 H- 2.185 (OZONE)** - 1.390 (CO)
(0.532) (3.584)
- 0.037(NO) + 0.0025(HUMID)
(2.396) (0.0037)
R2 - 0.18, N = 281, F = 9.98
+ 0.235(N02)
(0.862)
+ 0.0064(TEMP)
(0.0039)
+ 0.986(N02)
(0.791)
- 0.0028(TEMP)
(0.0036)
Prob(THROAT) t 0.356 + 0.738(OZONE) - 3.684(CO)
(0.488) (3.290)
- 2.209(NO) + 0.0003(HUMID)
(2.200) (0.0034)
R2 = 0.02, N = 281, F = 0.75
Prob(CHEST) = -0.962 + 1.252(OZONE)** - 1.414(CO) + 0.598(N02)
(0.493) (3.320) (0.799)
+ 1.866(NO) + 0.0080(HUMID)** + 0.0068(TEMP)
(2.220) (0.0035) (0.0036)
R2 = 0.08, N = 281, F = 3.76
Prob(BREATH) = -0.630 + 1.186(OZONE)** - 1.899(CO) + 0.850(N02)
(0.422) (2.841) (0.683)
- 1.278(NO) + 0.0069(HUMID)** + 0.0033(TEMP)
(1.899) (0.0030) (0.0031)
R2 = 0.08, N = 281, F = 4.01
Prob(COUGH) = -0.078 + 2.552(OZONE)** - 9.748(CO)** - 0.519(N02)
(0.617) (4.158) (0.999)
+ 7.675(NO) + 0.0092(HUMID)** + 0.0037(TEMP)
(2.780) (0.0043) (0.0046)
R2 = 0.12, N = 281, F - 5.99
Prob(PHLEGM) - 0.164 + 1.497(OZONE)** - 13.941(CO)**
(0.687) (4.630)
+ 5.511(NO)
(3.096)
R2 - 0.07, N - 281, F • 3.37
+ 0.0061(HUMID)
(0.0048)
H 0.264(N02)
(1.114)
I- 0.0031(TEMP)
(0.0051)
**Signifleant at a «• 0.05
-129-
-------
to only one air pollution variable, OZONE. However, cough was
directly related to both OZONE and NO and inversely related to
CO. Phlegm was directly related to OZONE and inversely related
to CO in the multiple regression. Relationships between the
discomfort symptoms and the air pollution variables were tested
again utilizing a more appropriate sigmoid probability function
in the analysis of the proportions of respondents reporting each
discomfort symptom by date. The results of the tests are dis-
cussed later in Section 8.
BETWEEN GROUP DIFFERENCES
The outdoor worker panel was divided into four sub-panels
prior to the daily surveillance. The absolute frequency and
relative frequency of observations for each of the four sub-
panels are given in Table 49. Each sub-panel was tested during
different two-week periods. Differences between groups is a
potential source of variation which was examined. As in the
analysis of the asthma panel, the procedure was to check for
small sample bias or self-selection bias; i.e., to determine if
there were unmeasured and unexpected differences between the
groups that cannot be controlled using appropriate statistical
methods. Under this procedure, if any differences encountered
can be controlled, then the four sub-panels can be pooled and
analyzed together.
TABLE 49. ' BREAKDOWN OF THE FOUR OUTDOOR WORKER
SUB-PANELS BY ABSOLUTE AND RELATIVE FREQUENCY
OF OBSERVATIONS
Sub-
Panel
1
2
3
4
Absolute
Frequency
209
275
226
189
899
Relative
Frequency (7o)
23.2%
30.6
25.1
21.0
100 . 0%
Difference in Means Tests
The following paragraphs present the results of difference
in means tests for the four sub-panels. Three variables were
-130-
-------
tested. The first variable tested was MAXFEV which was unad-
justed for age, height, or other explanatory variables. How-
ever, the sample size was reduced by examining only those panel-
ists who reported not having a cold and not taking medication.
In addition, missing observations for the MAXFEV variable neces-
sitated omission of those observations.
The second variable tested was the difference between
actual recorded FEV^ Q and predicted FEV]__o where
Predicted FEVL Q = -214.60 - 4.089(AGE)
+11.110(HEIGHT) - 32.715(SMOKE)
and where the coefficient of determination for predicted FEVj^o
was R2 = 0.33. The coefficients of all three of the explanatory
variables for this equation were significant at a = 0.01. The
FEV]_o difference variable was then compiled as
DIFFEV = MAXFEV - Predicted FEV;L Q
Variable DIFFEV controls for age, height, and whether or not the
panelist was a smoker.
The third variable tested was maximum FEV^Q divided by
maximum FVC. This variable, FEV]_ g/FVC, measured the proportion
of MAXFEV with respect to MAXFEV since MAXFVC should vary di-
rectly with age, height, and other unmeasured physical charac-
teristics .
For all three of these variables, MAXFEV, DIFFEV, and
FEV]_ o/FVC, a F-test of sample variances was performed between
pairs. This was necessary to choose the appropriate t-test to
examine the differences in means between all four sub-panels.
Between Group Differences in MAXFEV
The sample means and sample standard deviations of MAXFEV
for the four groups are given in Table 50. The results of the
F-tests are reported in Table 51. The level of significance
chosen was a = 0.10. A two-tailed test was applied and the
results indicate a rejection of the null hypothesis of equal
variance of MAXFEV between Sub-Panel 1 and Sub-Panel 2, Sub-
Panel 1 and Sub-Panel 3, Sub-Panel 2 and Sub-Panel 4, and Sub-
Panel 3 and Sub-Panel 4. The hypothesis of equal variance of
MAXFEV cannot be rejected at a = 0.10 between Sub-Panel 1 and
Sub-Panel 4 nor between Sub-Panel 2 and Sub-Panel 3. Since the
results tend to favor the hypothesis of unequal variance, the
conservative strategy of using separate variance estimates was
applied in the performance of the t-tests.
-131-
-------
TABLE 50. SAMPLE MEANS AND STANDARD DEVIATIONS
FOR MAXFEV OF THE FOUR OUTDOOR WORKER SUB-PANELS
Sub-
Panel
1
2
3
4
N
131
138
179
126
X
382.14
373.19
396.70
341.67
S
96.11
60.00
53.88
86.78
TABLE 51. DIFFERENCES IN MAXFEV VARIANCES TEST ON
OUTDOOR WORKER SUB-PANELS
Ho
a = a
1 2
a = a
I 3
a = a
i •*
a = a
2 3
a = a
2 <*
a = a
3 "t
F
2.57
3.18
1.23
1.24
2.09
2.59
Significant
a/2 at a = 0.10
0
0
0
0
0
0
.00 yes
.00 yes
.12 no
.09 no
.00 - Yes
.00 yes
Decision
on HQ
reject
reject
do not reject
do not reject
reject
reject
The results of the t-tests on the mean MAXFEV between
groups are reported in Table 52. The results indicate that the
null hypothesis yr = y and the null v^ = y3 cannot be rejected
at a = 0.05. However, the hypothesized equality between the
means of the remaining groups is rejected at a = 0.05. Some of
the results are contradictory e.g., yx = y2 and \il = y3, which
-132-
-------
TABLE 52. DIFFERENCE IN MAXFEV MEANS TEST ON OUTDOOR WORKER
SUB-PANELS USING SEPARATE VARIANCE ESTIMATES
Significant
HQ t a/2 at a = 0.05
y = y 0.91 0.182 no
I 2
y = y -1.56 0.060 no
1 3
y = y 3.55 0.000 yes
i "»
y = y -3.62 0.000 yes
2 3
y = y 3.40 0.001 yes
2 U
y = y 6.31 0.000 yes
3 >*
Decision
onHo
do not reject
do not reject
reject
reject
reject
reject
implies that y2 = y3> but this latter hypothesis is rejected.
Nevertheless, this preliminary investigation does indicate a
difference between Sub-Panel 4 and the other three sub-panels.
Between-Group Difference in DIFFEV
The above analysis was replicated for the FEV]_ Q difference
variable, DIFFEV. The general hypothesis tested was that the
between-group differences in MAXFEV are due to age, height, and
smoking. The F-tests results of the hypothesis of no difference
in the variance of DIFFEV between each group are reported in
Table 53. The hypothesis that ax = a and the hypothesis that .
o3 = On cannot be rejected at a = 0.10. However, the other
pairs are rejected. A conservative strategy again leads to the
utilization of separate variance estimates in the performance of
the t-tests of differences between the group means. The results
of the t-tests for DIFFEV are reported in Table 54. Again, the
results are mixed. It is not possible to reject the hypothesis
that ya = y2> vl = y,, and yz = yu. But the hypothesis that
y = y3, y2 = y3, and y3 = y4 are rejected at the a = 0.05 level
of significance. These tests support an argument that Sub-Panel
3 is distinct from the other three groups, while Sub-Panel 1,
Sub-Panel 2, and Sub-Panel 4 are indistinguishable once age,
height, and smoking are used to control for between-group dif-
ferences in MAXFEV.
-133-
-------
TABLE 53. DIFFERENCES IN DIFFEV VARIANCES TEST ON OUTDOOR
WORKER SUB-PANELS
Ho
a = a
1 2
a = a
1 3
a = a
1 4
a = a
2 3
a = a
2 4
a = a
3 4
F
1.18
1.73
1.92
1.47
1.62
1.11
a/2
0.136
0.001
0.001
0.004
0.002
0.260
Significant
at a = 0.10
no
yes
yes
yes
yes
no
Decision
onHQ
do not reject
reject
reject
reject
reject
do not reject
TABLE
54. DIFFERENCE IN
WORKER
DIFFEV MEANS TEST
SUB -PANELS
ON OUTDOOR
H
o
», -",
*,-",
y = y
1 4
y = y
2 3
y = y
2 4
y = y
3 it
t
0.47
-2.39
-0.40
-3.19
-0.98
2.30
a/2
0.314
0.018.
0.343
0.001
0.165
0.011
Significant
at a = 0.05
no
yes
no
yes
no
yes
Decision
on H
o
do not reject
reject
do not reject
reject
do not reject
reject
-134-
-------
Between-Group Difference in FEV-^ Q
The difference in variances and difference in means tests
were replicated for the FEV, n/FVC variable. The results are
reported in Table 55 and Table 56. As with the FEV]_Q differ-
ence variable, the hypotheses that Oj = o2 and a3 = a^ cannot be
rejected, but the statistical inference is that the other pairs
of variances are not equal. Therefore, the separate variance
estimates were utilized in the t-tests of differences between
means of the groups for the variable
The t-tests do not infer rejection of the hypotheses that
PI = y2 and Vi = Vn- The remaining pairs of means are rejected,
being equal at a = 0.05. However, the hypothesis that y2 = y^
cannot be rejected at the a = 0.01 level of significance.
Therefore, the results tend to support the conclusion that Sub-
Panel 3 is the only sub-panel that is statistically different
from the other three sub-panels. But the evidence is not strong
based on the results of the DIFFEV or FEV]_ Q/FVC analyses.
Difference Between Nonsmokers and Smokers
Difference in variances and means tests were also performed
between nonsmokers and smokers within the outdoor worker panel .
Again, the sample was restricted to panelists not having a cold
and not taking medication. The results are reported in Table
57. The variances of MAXFEV between smokers and nonsmokers can-
not be rejected as being significantly different from each other
at a level of significance of a = 0.10, but the means of MAXFEV
between smokers and nonsmokers are different at a level of sig-
nificance greater than a = 0.05. For the variable MAXFVC, the
statistical inference is that both variances and means are dif-
ferent from smokers and nonsmokers. The same conclusion holds
for the variable YEVi^Q/FVC. The FEV]_ Q difference variable,
DIFFEV, however, does 'not have statistically different means for
smokers and nonsmokers at the a = 0.05 level. This result is
reasonable since DIFFEV was created using a dependent variable
which controlled for smoking..
In summary, the comparison of MAXFEV, MAXFVC, and FEV^ Q/
FVC between smokers and nonsmokers indicates that there is a
significant difference. Smokers have lower recorded MAXFEV,
MAXFVC, and YEVi^Q/FVC scores than nonsmokers. This further
demonstrates the* importance of treating the two groups separ-
ately or of controlling for the influence of smoking in the
analyses of the outdoor worker panel.
-135-
-------
TABLE 55. DIFFERENCE IN FEVi Q/FVC VARIANCE TEST ON
OUTDOOR WORKER SUB-PANELS
Ho
a = a
I 2
a = a
1 3
a = a
a = a~
2 3
a = a
2 U
a = a
3 <*
F
1.08
1.35
1.45
1.46
1.57
1.07
a/2
0.302
0.023
0.012
0.005
0.003
0.326
Significant
at a = 0.10
no
yes
yes
yes
yes
no
Decision
onHQ
do not reject
reject
reject
reject
reject
do not reject
TABLE
56.
DIFFERENCE IN FEVx Q/VVC MEANS
OUTDOOR WORKER SUB-PANELS
TEST ON
Ho
y = y
1 2
y = y
1 3
yi = \
y = y
2 3
y = y
2 "»
y3 = y.
t
1.31
3.04
-0.78
-4.55
-2.17
2.36
a/2
0.096
0.002.
0.218
0.000
0.016
0.010
Significant
at a = 0.05
no
yes
no
yes
yes
yes
Decision
onHQ
do not reject
reject
do not reject
reject
reject
reject
-136-
-------
to
TABLE 57. DIFFERENCE IN VARIANCES AND MEANS TESTS BETWEEN NONSMOKERS (NS)
AND SMOKERS (S) ON OUTDOOR WORKER SUB-PANELS
Variable
MAXFEV
(In Liters
x 100)
MAXFVC
(In Liters
x 100)
DIFFEV
(In Liters
x 100)
FEV/FVC
Sub -Panel
(N)
NS(348)
S(226)
NS(420)
S(267)
NS(348)
S(226)
NS(348)
S(226)
X
389.74
353.94
473.39
457.69
0.0002
0.0002
0.829
0.781
S
75.70
73.61
89.61
78.32
60.10
66.89
0.11
0.13
Significant Significant
F at a = 0.10 t at a = 0.05
1.06 no 5.60 yes
1.31 yes 2.71 yes
1.24 yes -0.00 no
1.46 yes 4.44 yes
-------
DISCOMFORT SYMPTOMS ANALYZED BY DATE OF MEASUREMENT
Daily surveillance of the outdoor worker panel was con-
ducted on 54 different dates. On each date, the proportion of
the panelists reporting each discomfort symptom was calculated.
The average levels of OZONE, CO, N02, NO, HUMID, and TEMP were
computed for each date also. The average levels of air pollu-
tion were then related to the proportion of panelists reporting
each discomfort symptom on each date. This was done using two
techniques, simple linear regressions, and selected multivariate
regressions.
Simple Linear Regressions^
Let fg denote the proportion of panelists reporting a dis-
comfort symptom for date g = 1,...,G. Let Xg denote the average
level of an air pollution variable on date g. Then a linear
probability specification using the proportions and averages is:
fg = «0 + ce^g ; g = 1.....G.
In applying this specification, each of the discomfort propor-
tions were used as the dependent variable and each of the air
pollution variables were used as the explanatory variable. The
simple linear regressions which are significant at a = 0.10 or
better are shown in Table 58.
EYES is significantly related to OZONE, CO, NO, HUMID, and
TEMP. THROAT is significantly related to OZONE, N02, and HUMID.
CHEST is significantly related to OZONE, NO, and TEMP. HEADACHE
is significantly related only to CO. BREATH is significantly
related to OZONE, NO, and TEMP. COUGH is significantly related
to OZONE and TEMP. PHLEGM is significantly related only to
OZONE, but the relation is not strong at a = 0.10. Therefore,
PHLEGM is omitted from further analysis.
Single variable LOGIT regressions were estimated for EYES,
THROAT, CHEST, HEADACHE, BREATH, and COUGH. The significant re-
gressions are shown in Table 59. The LOGIT transformation is:
; g = 1.....G
In ( JL__L
' """^g
where Xp is the mean of the explanatory variable for date g and
fg is the proportion of panelists reporting symptoms on date g.
The specification is sigmoid in shape; hence, it is consistent
with a probability specification.
-138-
-------
TABLE.58. SIGNIFICANT SIMPLE REGRESSIONS OF DISCOMFORT
SYMPTOM PROPORTIONS
Proportion(EYES)
Proportion (EYES)
Proportion(EYES)
Proportion (EYES)
Proportion (EYES)
Proportion (THROAT)
Proportion(THROAT)
Proportion (THROAT)
Proper tion(CHEST)
Proper tion(CHEST)
Proportion(CHEST)
Proportion (HEADACHE)
Proportion (BREATH)
Proportion(BREATH)
Proportion (BREATH)
Proportion(COUGH)
Proportion (COUGH)
Proportion (PHLEGM)
= -0
= -0
0
0
= -0
0
0
0
0
0
= -0
0
= -0
0
= -0
0
0
0
.043
.018
.241
.459
.533
.103
.104
.319
.066
,179
.075
.011
.009
.124
.156
.260
.026
,385
+ 2
+ 5
- 3
- 0
+ 0
+ 0
+ 0
- 0
+ 1
- 1
+ 0
+ 1
+ 1
- 2
+ 0
+ 0
+ 0
+ 0
.7 28 (OZONE)***
.925
.346
.005
.010
.360
.370
.003
.006
.777
.003
(CO)**
(NO)*
(HUMID)*
(TEMP)***
(OZONE)**
(N02)*
(HUMID)**
(OZONE)***
(NO)*
(TEMP)**
.496 (CO)**
.182
.067
.003
(OZONE)***
(NO)*
(TEMP)**
.864 (OZONE)**
.004 (TEMP)**
.960
(OZONE)*
R2
R2
R2
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
= 0
.584
.097
.041
.036
.201
.056
.037
.084
.233
.034
.058
.060
.352
.050
.075
.096
.060
.035
*Significant at a = 0.10
**Significant at a = 0.05
***Significant at a = 0.01
-139-
-------
TABLE 59. SIGNIFICANT LOOIT REGRESSIONS OF DISCOMFORT
SYMPTOM PROBABILITIES ^
ln 1 - Prob(EYES) = -5-475 + 30.491(OZONE)*** R2 = 0.471
ln {l - Prob(EYES) i = ~5'756 + 81-857(00)*** R2 = 0.119
In
1 - Prob(EYES) = 2.364 + -0.092(HUMID)*** R2 = 0.123
Prob(EYES)
1 - Prob(EYES)
= -13.214 + 0.139(TEMP)*** R2 = 0.266
1 - Prob(BREATH) } = -5-482 + 16.322(OZONE)*** R2 = 0.150
In I. Prob(BREATH)
jl - Prob(BREATH)
= -9.329 + 0.070(TEMP)** R2 = 0.076
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-140-
-------
None of the LOGIT regressions for THROAT, CHEST, HEADACHE,
or COUGH were statistically significant. EYES remains statisti-
cally significantly related to OZONE, CO, HUMID, and TEMP,- but
it is no longer significantly related to NO. BREATH remains
significantly related to OZONE and TEMP, but is no longer sig-
nificantly related to NO.
Selected Multivariate Regressions
Seven of the discomfort symptom variables were chosen for
rnultivariate regression analysis. The dependent variable was
the proportion of panelists reporting a discomfort symptom or
the LOGIT transformation of that proportion. The explanatory
variables were the average levels of each of the air pollution
variables on the date of testing.
Initially, seven linear probability specifications were es-
timated. The four specifications using EYES, BREATH, CHEST, and
COUGH demonstrated some statistical significance. The results
are reported in Table 60. Three regression estimates, those for
THROAT, HEADACHE, and PHLEGM, are not reported since none of the
explanatory variables had estimated coefficients statistically
different from zero. The variable measuring OZONE was statisti-
cally significant in the regression for EYES, BREATH, and CHEST.
Secondly, the LOGIT transformation of each of the propor-
tions--EYES, BREATH, CHEST, and COUGH--were regressed on the air
pollution variables. The results are reported in Table 61. In
the first regression using the LOGIT of EYES, OZONE, and TEMP
were statistically significant. In the second regression using
BREATH, OZONE, and N02 were significant. In the third using
CHEST, none of the explanatory variables were significant. In
the fourth regression, only CO contributed significantly in ex-
plaining COUGH.
Since the LOGIT transformation is subject to heteroscedas-
ticity due to unique size groups in computing the proportions, a
correction technique was applied. The transformation is to
utilize
J-l
in place of
k
jig ; g
-141-
-------
TABLE 60. MULTIVARIATE REGRESSIONS OF PROPORTIONS OF
DISCOMFORT SYMPTOMS
(STANDARD ERRORS IN PARENTHESES) ,
Explanatory
Variable
CONSTANT
OZONE
CO
N02
NO
HUMID
TEMP
R2
F
D.W.
Dependent Variable
EYES
-0.273
2 . 494**
(0.522)
1.809
(3.000)
-0.098
(0.604) .
1.228
(2.198)
-0.0003
(0.0025)
0.0026
(0.0030)
0.602
11.832
1.785
BREATH
-0.130
1 . 341**
(0.354)
0.023
(2.035)
-0.477
(0.410)
0.828
(1.491)
0.0018
(0.0017)
0.0003
(0.0020)
0.412
5.488
1.278
CHEST
-0.036
1.219**
(0.416)
-1.573
(2.393)
0.042
(0.482)
0.705
(1.754)
0.0018
(0.0020)
0.0003
(0.0024)
0.257
2.714
1.408
COUGH
-0.062
0.094
(0.591)
-1.675
(3.397)
-0.352
(0.684)
1.226
(2.490)
0.0029
(0.0028)
0.0028
(0.0034)
0.160
1.491
0.981
**Significant at a = 0.05
Note: N = 54
-142-
-------
TABLE 61. MULTIVARIATE REGRESSIONS OF LOGIT OF
DISCOMFORT SYMPTOMS
(STANDARD ERRORS IN PARENTHESES)
Explanatory
Variable
CONSTANT
OZONE
CO
N02
NO
HUMID
R2
F
D.W.
Dependent Variable
EYES
-12.804
17.111**
(6.724)
61.013
(38.679)
6.679
(7.793).
-23.337
(28.343)
-0.0359**
(0.0385)
0.572
10.488
1.802
BREATH
-10.968
16.831**
(8.120)
17.443
(46.709)
-21. 322**
(9 '.228)
32.110
(34.228)
0.0499
(0.0464)
0.307
3.471
1.700
CHEST
-3.823
10.952
(8.553)
-4.995
(49.198)
-0.608
(36.051)
6.876
(36.051)
-0.0072
(0.0489)
0.059
0.496
1.250
COUGH
-6.119
-3.261
(4.895)
59.666**
(28.157)
-9.430
(20.633)
-0.869
(20.633)
0.0462
(0.0280)
0.147
1.346
1.390
**Signifleant at a = 0.05
Note: N = 54
-143-
-------
where £„ is the proportion in group g and ng is the number of
cases in group g used to compute the proportion. This correc-
tion helps the small sample properties of the estimation pro-
cess, but has no effect on the large sample properties. The
correction yields a variance of the error which is normally dis-
tributed with zero mean. Therefore, application of the t-tests
do not rely on the asymptotic properties of the estimators.
The results of the LOGIT regressions corrected for hetero-
scedasticity are reported in Table 62. The regressions using
CHEST and COUGH do not yield any estimated relationships with
the air pollution variables which are statistically significant.
In the two remaining regressions, that for EYES and BREATH, only
OZONE demonstrates a statistically significant relationship.
Summary of the Analysis by Date
Single linear probability regressions of each discomfort
symptom with each of the air pollution variables resulted in
several statistically significant associations. The proportion
of panelists reporting eye discomfort was significantly corre-
lated with OZONE, CO, NO, HUMID, and TEMP which were measures
of the average levels of the respective air pollution and weather
data on the date of testing. Throat discomfort was significant-
ly correlated with OZONE, N02, and HUMID. Chest discomfort was
significantly correlated with OZONE, NO, and TEMP. The propor-
tion of panelists reporting headache was correlated significant-
ly only with CO, while PHLEGM showed weak statistical correla-
tion with OZONE.. Shortness of breath was correlated signifi-
cantly with OZONE, NO, and TEMP. The proportion of panelists
reporting cough was correlated significantly with.OZONE and
TEMP. Simple LOGIT transformation of the proportions resulted
in significant correlations only between eye discomfort and the
variables OZONE, CO, HUMID, and TEMP; and between shortness of
breath and OZONE and TEMP.
Multivariate regression was then applied to the proportion
of the discomfort symptoms and the logit transformation of the
proportions. A correction for heteroscedasticity was applied
to the LOGIT specification. Only eye discomfort and shortness
of breath generated consistently significant estimates. And
only OZONE remained statistically significant as an explanatory
variable. It is possible, however, that multicollinearity be-
tween the explanatory variables may account for the seeming lack
of influence between the discomfort symptoms and the air pollu-
tion variables. But no collinearity was readily apparent.
MULTIVARIATE ANALYSIS OF MAXFEV
As in the analysis of the asthma panel, two multivariate
regression models were used to explain variations in MAXFEV,
while controlling for differences between individuals. An
-144-
-------
TABLE 62. MULTIVARIATE LOGIT REGRESSIONS CORRECTED
FOR HETEROSCEDASTICITY: OUTDOOR WORKERS
(STANDARD ERRORS IN PARENTHESES)
Explanatory
Variable
CONSTANT
OZONE
CO
N02
NO
HUMID
TEMP
R2
F
D.W.
Dependent Variable
EYES
-0.401
1.652**
(0.390)
2.695
(2.245)
-0.121
(0.452)
0.999
(1.645)
0.0006
(0.0019)
0.0029
(0.0022)
0.587
11.125
2.097
BREATH
-0.294
0.726**
(0.264)
1.063
(1.519)
-0.511
(0.306)
0.705
(1.114)
0.0020
(0.0013)
0.0017
(0.0015)
0.377
4.732
1.576
CHEST
-0.107
0.516
(0.286)
0.246
(1.644)
-0.029
(0.331)
-0.021
(1.205)
0.0012
(1.2046)
0.0006
(0.0016)
0.166
1.558
1.190
COUGH
-0.076
0.458
(0.394)
-0.303
(2.265)
-0.151
(0.456)
-0.015
(1.660)
0.0013
(0.0019)
0.0019
(0.0022)
0.121
1.081
1.272
**Significant at a = 0.05
Note: N = 54
-145-
-------
extension of this analysis in which the air pollution variables
are lagged concludes the discussion of the outdoor worker panel.
\
The Hasselblad Model
The Hasselblad model utilizes dummy variables to control
for differences between individuals. The dependent variable is
MAXFEV. The form of the model is described in Section 7. Since
the model controls for differences between individuals, it is
not appropriate to include variables such as age, height, the
presence of a cold, or smoking.
It was not possible to use the entire outdoor worker panel
in this analysis because the number of explanatory variables was
too large for the regression program used. The maximum number
of individuals possible for inclusion was 60. Therefore, the
Hasselblad specification used the 494 observations on the first
60 individuals in the panel. This resulted in 59 Pmj variables
named PI to P59. Variable Xij was named DINV which stands for
day-inverse, so DINV = 1/k. The other explanatory variables
were OZONE, CO, N02, and NO.
The regression is shown in Table 63. The estimated coef-
ficients for the dummy variables Pmj are not reported as they
only control for differences between individuals in order to
isolate the effects of air pollution. It should be noted that
many of the dummy variables are statistically significant. The
R.2 = 0.95 indicates that 95 percent of the variation in the de-
pendent variable has been explained by regression. The total F
indicates rejection of a hypothesis that all coefficients are
simultaneously zero. Apart from the dummy variables, none of
the other explanatory variables are statistically significant.
The primary conclusion must be that the variables which
capture individual differences explain a large portion of the
variance in the dependent variable and that there is very little
variation left to be explained by the air pollution variables.
Therefore, MAXFEV was not statistically sensitive to the levels
of air pollution measured during daily surveillance of the out-
door worker panel.
Standard Multivariate Regression of MAXFEV and FEV^ Q/FVC
The first four multivariate regression specifications esti-
mated for the outdoor worker panel utilized MAXFEV and FEViQ/
FVC as dependent variables. Each dependent variable was used
separately for smokers and nonsmokers. The results are reported
in Table 64. When MAXFEV was the dependent variable, AGE and
HEIGHT were included as explanatory variables to control for in-
dividual differences. Use of YEVi^fFVC as the dependent
-146-
-------
TABLE 63. HASSELBLAD REGRESSION FOR MAXFEV WITH AIR
POLLUTION VARIABLES (USING DATA FROM 60 INDIVIDUALS IN
THE OUTDOOR WORKER PANEL) .
R2 = 0.95, N = 494, Total F = 129.7, D.W. = 1.87
MAXFEV = - 1,694.02
59
I—* m mj
m=l
- 40.057(DAYINV)
- 10.229(OZONE)
- 195.790(CO)
+ 35.815(N02)
- 56.303(NO)
Note: The estimated coefficients for the dummy variables
Pmj are not reported as they only control for dif-
ferences between individuals so that the effects of
air pollution variables may be isolated.
-147-
-------
TABLE 64. MULTIVARIATE REGRESSIONS FOR MAXFEV AND FEV]^ Q/FVC—SMOKERS
AND NONSMOKERS IN OUTDOOR WORKER PANEL
(STANDARD ERRORS IN PARENTHESES)
Variable
CONSTANT
OZONE
CO
N02
NO
HUMID
TEMP
AGE
HEIGHT
R2
F
N
MAXFEV,
Smokers
35.5A404
21.15038
(101.49903)
858.46059
(703.66308)
-139.71547
(177.44272)
-477.10936
(422.59835)
1.95632**
(0.69204)
2.05934**
(0.75571)
-5.08527**
(0.68980)
3.40367
(1.993.34)
0.301
11.178
217
MAXFEV,
Nonsmokers
-596.47103
-43.66274
(83.34288)
-675.54668
(585.18125)
233.20581
(140.25433)
-630.01413
(439.15076)
0.98538
( (0.62679)
0.47845
(0.57293)
-3.58388**
(0.44972)
15.31423**
(1.43680)
0.406
25.945
312
FEVX o/FVC,
Smokers
-0.00791
0.17436
(0.19724)
1.61645
(1.38860)
-0.38937
(0.35109)
0.72760
(0.81840)
0.00475**
(0.00137)
0.00609**
(0.00150)
— — —
0.142
5.794
217
FEV-L o/FVC,
Nonsmokers
0.58291
-0.33708**
(0.15155)
-0.75733
(1.06001)
0.49401*
(0.25440)
-1.61697**
(0.79784)
0.00212*
(0.00113)
0.00215**
(0.00103)
— — —
0.041
2.188
312
*Signifleant at a = 0.10
**Signifleant at a = 0.05
-148-
-------
variable does not require the inclusion of control variables for
individuals since the transformation accomplishes that objec-
tive .
For the regressions using MAXFEV as the dependent variable,
none of the air pollution variables have coefficients statisti-
cally different from zero. The only coefficients statistically
significant involve HUMID, TEMP, AGE, or HEIGHT.
For the regressions using FEV]_Q/FVC °f smokers as the de-
pendent variable, only HUMID and TEMP contribute significantly
to the reduction in overall variance. However, the results
using FEVi.Q/FVC of nonsmokers indicate that OZONE, N02, NO,
HUMID, and'TEMP are statistically significant. But in the
latter use, the overall measure of the goodness of fit,
R = 0.04, is small indicating that the association is not
strong.
These four regression specifications were reestimated after
Sub-Panel 3 was excluded from the data. This was done since
previous analysis indicated that Sub-Panel 3 was potentially
different from the others. The results reported in Table 65 do
not indicate any stronger associations than before. In the
MAXFEV regressions, only HUMID, AGE, and HEIGHT are statisti-
cally significant. In the FEV^o/FVC regressions, only HUMID
and TEMP are significant for smokers while none of the explana-
tory variables are significant for nonsmokers.
Alternative linear specifications were estimated which
utilized a dummy variable for sub-panels tested on different
dates. As with the asthma panel, there were four sub-panels in
the outdoor worker panel. Sub-Panel. (Group) 1 was used as the
base group; G2 = 1 if the observation was for Sub-Panel (Group)
2; G3 = 1 if the observation was for Sub-Panel (Group) 3; G4 = 1
if the observation was for Sub-Panel (Group) 4. The results are
reported in Table 66.
Apart from the dummy variables, the only statistically sig-
nificant variables explaining MAXFEV from smokers or nonsmokers "
are TEMP, AGE, or HEIGHT. Also, for the regressions predicting
FEV]__o/FVC for smokers, none of the air pollution variables are
statistically significant. However, for the FEV]_tQ/FVC regres-
sion for nonsmokers, CO is statistically significant. None of
the other explanatory variables are significant, and the implied
inverse correlation between FEVi.Q/FVC and CO may be spurious.
Lagged Explanatory Variables
The average level of each of the air pollution variables
was computed for each date. These variables were then lagged
one day and used as explanatory variables with MAXFEV as the
-149-
-------
TABLE 65. MULTIVARIATE REGRESSIONS FOR MAXFEV AND FEVi Q/FVC--SMOKERS
AND NONSMOKERS WITH SUB-PANEL 3 EXCLUDED'
(STANDARD ERRORS IN PARENTHESES) ,
Variable
CONSTANT
OZONE
CO
N02 ..
NO
HUMID
TEMP
AGE
HEIGHT
R2
F
N
MAXFEV,
Smokers
210.72203
110.85588
(117.66221)
326.03352
(784.31580)
-240.98382
(201.59030)
-202.14476
(466.18577)
1.72680**
(0.74427)
1.31751
(0.94992)
-5.78183**
(0.75212)
2.34305
(2.11581)
0.351
11.814
184
MAXFEV,
Nonsmokers
-268.38956
58.43143
(123.20335)
-813.61658
(799.60494)
50.50919
(180.47025)
-88.68855
(571.31320)
0.54677
(0.81066)
0.04526
(0.85769)
-4.41017**
(0.85769)
11.89836**
(0.56207)
0.426
18.109
204
FEV1.0/FVC>
Smokers
0.11536
0.34284
(0.23431)
0.51066
(1.58060)
-0.61154
(0.91546)
1.55049
(0.91546)
0.00394**
(0.00151)
0.00531**
(0.00192)
-
0.149
5.175
184
FEVX Q/FVC,
Nonsmokers
0.85057
-0.08167
(0.20626)
-2.03921
(1.33654)
0.36832
(0.95562)
-0.52144
(0.95562)
0.00044
(0.00315)
-0.00014
(0.00143)
---
0.028
0.955
204
**Signifleant at a = 0.05
-150-
-------
TABLE 66. MULTIVARIATE REGRESSIONS FOR MAXFEV AND FEV1 0/FVC—SMOKERS
AND NONSMOKERS WITH DUMMY VARIABLE INCLUDED FOR bUB-PANELS
(STANDARD ERRORS IN PARENTHESES) ,
Variable
CONSTANT
OZONE
CO
N02
HUMID
TEMP
AGE
HEIGHT
G2
G3
G4
R2
F
N
MAXFEV,
Smokers
479.22652
-25.22124
(91.82008)
-29.93116
(654.48034)
-18.29724
(150.55112)
-0.95098
(0.69681)
-1.60883**
(0.81854)
-5.32284**
(0.60779)
4.19671**
(1.70719)
-21.66984
(11.74915)
19.93225
(13.31532)
-113.58543**
(17.14194)
0.476
18.716
217
MAXFEV,
Nonsmokers
-424.99038
-1.89032
(92.17671)
-975.73363
(641.04083)
182.93001
(127.81478)
0.01643
(0.72906)
-0.38070
(0.73550)
-3.63778**
(0.44951)
14.64094**
(1.45461)
3.40644
(12.61516)
9.46227
(11.00845)
-17.96280
(15.93409)
0.413
21.139
312
FEVli0/FVC,
Smokers
0.92505
-0.09735
(0.18476)
0.70104
(1.31966)
0.04675
(0.30356)
-0.00062
(0.00141)
-0.00093
(0.00165)
.
-0.04844**
(0.02352)
0.00277
(0.02639)
-0.23143
(0.03453)
0.331
12.866
217
FEV1<0/FVC,
Nonsmokers
0.76546
-0.00526
(0.16546)
-2.32312**
(1.14889)
0.39861
(0.22901)
0.00075
(0.00131)
0.00053
(0.00132)
0.01572
(0.02264)
0.06848
(0.01972)
0.01777
(0.02847)
0.076
3.111
312
**Signifleant at a = 0.05
-151-
-------
dependent variable. The results are reported in Table 67. Only
AGE, HEIGHT, and COLD are statistically significant. None of
the lagged air pollution variables are statistically signrfi- .
cant. Moreover, the analysis did not suggest that lags of
greater magnitude or weighted averages of the lagged values
would yield more significant results. The data "do not support a
hypothesis that lagged or cumulative exposure to air pollution
significantly affected MAXFEV.
TABLE 67. MULTIVARIATE REGRESSIONS OF MAXFEV WITH
ONE-DAY LAGGED AIR POLLUTION VARIABLES
o
Variable 3 Standard Error F = t
CONSTANT -311.325
OZONE -1
CO-1
N02-1
NO-1
HUMID- 1
TEMP-1
AGE
HEIGHT
COLD
-44.596
-442.751
60.849
157.758
-0.689
-0.548
-4.069
13.903
-15.978
0.207
382.181
88.598
276.817
0.416
0.434
0.313
1.030
7.538
0.587
1.342
0.472
0.325
2.744
1.592
168.592
182.183
4.493
R2 = 0.394
N = 589
F = 41.844
In addition to the average value of each of the air pol-
lution variables being lagged one day, the averages were lagged
two and three days. The lagged variables were then used as ex-
planatory variables along with AGE, HEIGHT, and COLD. The de-
pendent variable was MAXFEV. Again, the only variables statisti-
-152-
-------
cally significant in these regressions were AGE and HEIGHT.
TEMP was statistically significant in the three-day lag speci-
fication. But none of the air pollution variables were signifi-
cant in either the two- or three-day specifications.
Summary of the Multivariate Analysis
Alternative linear regression specifications estimated for
data obtained from the outdoor worker panel used MAXFEV and
FEV1.0/FVC as dependent variables. Separate regressions were
used'to control for differences between individuals and sub-
panels. The Hasselblad model included a dummy variable for each
individual tested. The more standard linear specification
either included AGE and HEIGHT variables or applied the FEV^ Q/
FVC transformation to control for differences between individ-
uals. A dummy variable technique was also applied to each of
the four sub-panels. Finally, the air pollution variables were
averaged and lagged to investigate the possibility of lagged
effects of air pollution on the pulmonary function results.
The conclusion is that differences in MAXFEV are explained
quite well by AGE and HEIGHT and other explanatory variables
which controlled for differences between individuals. But
MAXFEV measured for the outdoor worker panel was not statisti-
cally sensitive to air pollution variables in the multivariate
specifications.
OVERALL SUMMARY OF THE OUTDOOR WORKER PANEL DATA ANALYSIS
The levels of air pollution in the study area.were light to
moderate during the 54 days of outdoor worker testing.
Typical seasonal changes were observed in the concentrations of
the air pollutants. Maximum hourly averages of ozone gradually
declined from around 0.30 ppm to 0.05 ppm, while concentrations
of nitrogen dioxide increased from 0.01 ppm to around 0.30 ppm.
The average values of these and the other aerometric variables
at the times outdoor worker panelists reported their symptoms
(i.e., the weighted averages of these variables) are presented
in Table 30D. The weighted averages of responses to daily
symptom interviewing and to pulmonary function testing are shown
in Tables 30A and 30C. The symptom and aerometric data are com-
pared graphically in Appendix J.
The data from the outdoor worker panel provided the most
significant results. It is believed that this happened because
the panel members were exposed to concentrations of air pollu-
tants while at work out of doors.
-153-
-------
The analysis of the data involved several statistical
tests. Significant results were obtained from most of them.
For example, correlation analyses were performed on the da'ta of
the outdoor worker panel. Both parametric and non-parametric
correlations were computed and tested for statistical signifi-
cance. Eye discomfort, chest discomfort, cough, and phlegm
were positively correlated with ozone. Eye discomfort was also
positively correlated with all of the other aerometric vari-
ables.
The correlation analysis was replicated separately for
smokers and nonsmokers of the outdoor worker panel. There were
more significant correlations for smokers than for the outdoor
worker panel as a whole. Headache was not significant for the
entire panel but was positively correlated with ozone for
smokers. The total number of significant correlations for non-
smokers was smaller than for the entire panel. This was par-
ticularly true for the discomfort symptoms—shortness of breath,
cough, and phlegm. However, MAXFEV was negatively correlated
with ozone and statistically significant only for the non-
smokers . The nonsmoking outdoor workers seemed to be more re-
sponsive in terms of decreased lung function to the air pollu-
tion variables than any other group examined in this study.
These differences between smokers and nonsmokers in the outdoor
workers panel indicated that separate treatment of the two
groups was appropriate.
Single variable linear regressions were estimated with
MAXFEV as the dependent variable. Most of the air pollution
variables were significantly related to MAXFEV for smokers.
However, the sign of the coefficient for ozone was positive.
For nonsmokers, only ozone was statistically significant and the
sign of the coefficient was negative.
Contingency tables were constructed and chi-square tests
of association were made between MAXFEV and the air pollution
variables. These tests supported the hypothesis of dependence
between MAXFEV and the air pollution variables for the outdoor
worker panel.
Linear probability functions were estimated using the di-
chotomous discomfort variables as dependent and the air pollu-
tion variables as explanatory. The relationship of eye discom-
fort with ozone was stronger for smokers than for nonsmokers.
Eye discomfort, throat discomfort, chest discomfort, shortness
of breath, and cough were all directly related to ozone.
There were fewer significant linear probability regressions
for nonsmokers. However, eye discomfort was significantly re-
lated to all the air pollution variables for the nonsmokers.
-154-
-------
Multivariate linear probability regressions were estimated
for the whole outdoor worker panel. The data were restricted to
those outdoor workers who reported not having a cold and not
taking medication. The results showed reports of eye discom-
fort, chest discomfort, shortness of breath, cough, and phlegm
to be significantly related to concentrations of the air pollu-
tants .
Discomfort symptoms were analyzed by computing proportions
of the symptoms and the means of the air pollution variables for
each of the 54 days of surveillance. Single variable and multi-
variate regressions were performed using both the linear proba-
bility and LOGIT probability specifications. The simple linear
probability regressions resulted in several statistically sig-
nificant associations. Eye discomfort, throat discomfort,
shortness of breath,'and cough were significantly related to
ozone. Multivariate regressions resulted in only eye discomfort
and shortness of breath having consistently significant esti-
mates after a correction for heteroscedasticity was applied.
And, only ozone remained statistically significant as an explana-
tory variable.
-155-
-------
SECTION 9
DATA ANALYSIS--BRONCHITIS PANEL
STATISTICAL DESCRIPTION OF THE BRONCHITIS PANEL
This panel, like the asthma panel, was composed entirely of
females. All of the.bronchitics selected for participation
smoked cigarettes; all were in the 35 to 55 age bracket. The
bronchitis panel was tested on 12 days over a period of seven
weeks. Although 54 females were selected for participation,
only 38 attended testing sessions regularly enough for suffi-
cient test results to be entered into the data file for analy-
sis. Table 68 summarizes the composition of the bronchitis
panel.
TABLE 68. COMPOSITION OF THE
BRONCHITIS PANEL
Total
Characteristics Panel
Subjects enrolled in panel
(all female) 38
Current cigarette smokers 38
Subjects 31 to 40 years old 14
Subjects 41 to 50 years old 14
Subjects over 50 years old 10
Subjects with 12 or more
years of school completed 28
Race other than white 2
The comprehensive clinical examinations administered before
and after the testing period and the clinical interviews held
after the testing period confirmed that all but one of the
-156-
-------
panelists had symptoms of bronchitis. Even this single panelist
exhibited symptoms of bronchitis during the testing period. The
types of data obtained from each bronchitis panelist are listed
in Section 4 of this report. All data were included in the
analysis except for heart function measures and the results of
the hematological and nasal smear evaluations. The latter data
were eliminated for the following reasons.
Heart function was evaluated by measuring blood pressure,
heart rate, and 12-lead electrocardiograms. The electrocardio-
grams were converted to a digital format that yielded 44
descriptive codes or numerial quantities. All the measures of
heart function stayed essentially constant throughout the test-
ing period for all panelists, so they were not included in the
numerical data analysis.
A complete hematological analysis was obtained on venous
blood samples which were taken on four or five selected testing
days. Immunoglobulins were also determined. The measures of
MCV, MCH, and MCHC remained nearly constant for all panelists
and therefore were not used in the statistical analysis. The
counts of basophiles and monocytes were so low that analysis
against air pollution variables was not possible. Differential
counts were converted to absolute counts for the statistical
analysis. There were few significant correlations of any sort,
and there was no indication of any systematic association be-
tween air pollution levels and any of the blood components.
Nasal smears were taken from each panelist and submitted
for a semiquantitative determination of eosinophiles. The re-
sults were categorized as "negative," "few," "moderate," or
"many." Most of the readings were "negative," so no attempt was
made to analyze the results by any formal method.
Description of the Variables
The discomfort symptom variables were measured with the
form shown in Appendix F. The same coding was.used on the re-
sponses as in the other panels: unity for "Yes," zero for "No."
A statistical profile of the discomfort variables is given in
Table 69A. The variables were the same as those measured of the
asthma panel.
Two pulmonary function variables were used in the analysis
of the bronchitis panel. MAXFVC was the best of five maneuvers
attained each study day by each panelist. Variable MAXFEV was
the highest FEVi.Q taken from the volume-time tracings of the
FVC maneuvers. The AGE and HEIGHT of each panelist were re-
corded as age in years and standing, height in inches. These
data were used to adjust the MAXFVC and MAXFEV scores prior to
-157-
-------
TABLE 69A. STATISTICAL PROFILE OF DISCOMFORT SYMPTOMS REPORTED BY BRONCHITIS PANELISTS ON 11 TESTING DAYS
00
I
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWHESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
EYES » Eye discomfort now
= 0.478
= 0.250
- -1.989
- 0.500
«• 0.089
THROAT -
= 0.459
- 0.249
- -1.969
= 0.499
- 0.167
CHEST -
- 0.547
- 0.248
- -1.962
- 0.498
- -0.189
HEADACHE
- 0.343
- 0.226
• -1.556
- 0.475
- 0.665
NAUSEA =
= 0.091
- 0.083
- 6.095
- 0.238
- ' 2.849
OTHER -
• 0.149
- 0.127
- 1.893
- 0.357
- 1.975
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
Throat discomfort now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS -
Chest discomfort now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS -
•" Headache now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS -
Nausea now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS -
Other discomfort now
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS -
1.000
0.000
1.000
362
0
1.000 .
0.000
1.000
362
0
1.000
0.000
1.000
362
0
1.000
0.000
1.000
362
0
1.000
0.000
1.000
362
0
1.000
0.000
1.000
362
0
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :'
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
HEADACHE EARLIER = Headache earlier today
- 0.348
= 0.228
- -1.589
- 0.477
- 0.640
BREATH
• 0.392
• 0.239
= -1.802
" 0.489
- 0.443
COUGH -
• 0.856
= 0.123
= 2.143
- 0.351
*> -2.038
PHLEGM
= 0.884
= 0.103
" 3.792
- 0.320
= -2.410
COLD =
- 0.075
= 0.069
= 8.520
- 0.263
= 3.247
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
" Shortness of breath
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
Cough today
RANGE'
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
«• Phlegm today
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
Bad cold today
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
- 1.000
- 0.000
- 1.000
- 362
- 0
today
- 1.000
.= 0.000
- 1.000
- 362
- 0
- 1.000
" 0.000
- 1.000
- 362
- 0
- 1.000
- 0.000
= 1.000
- 311
- 0
= 1.000
- 0.000
° 1.000
- 362
- 0
MEDICINE - Medicine today
- 0.328
- 0.221
- -1.457
- 0.470
- 0.736
RANGE
• ' MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
- 1.000
- 0.000
- 1.000 •
- 360
- 2
-------
analysis. A profile of the pulmonary function variables is
given in Table 69B. MAXFVC and MAXFEV are shown in hundred
liters (liters x 100).
The air pollution and weather variables used in this analy-
sis are the same as those used in the asthma panel analysis. A
profile of these variables is given in Table 69C. The air pol-
lution variables were obtained as hourly averages measured in
ppm (CO im ppm/100). Relative humidity was estimated by hour
and expressed in percent. A discussion on how the estimates
were made is given in the paragraphs of Section 7 which describe
the variables used in the asthma panel analysis. Temperature
was obtained as hourly averages measured in degrees Fahrenehit.
There were no missing observations for any of the discom-
fort symptom variables, except for phlegm. The highest number
of missing observations were those for MAXFVC and FEV^.O' with
110 or 30 percent missing. Still, a sufficient number were
available for use in data analysis. Among the air pollution
variables, NC-2 and NO had a sizeable number of missing observa-
tions, but there were enough data points to include both of
these pollutants in the analysis.
Weighted averages of exposure to air pollutant, humidity,
and temperature levels were used in the analysis of the bron-
chitis panel. As in the analysis of the asthma panel, the hour
at which each panelist reported for testing was noted and the
aerometric values for that hour were included in- the weighted
average for that day.
The proportions of bronchitis panelists who reported dis-
comfort symptoms are shown in Appendix K for each of the 12
days when symptom data were collected. The weighted averages of
the air pollutants, relative humidity, and temperature are also
given in Appendix K for each of the same days. The charts are
provided to facilitate comparisons between proportions of panel-
ists who reported symptoms and the weighted averages of air
pollution.
CORRELATION ANALYSIS
Pearson and non-parametric correlation analyses were ap-
plied to the data collected from the bronchitis panel. The re-
sults are presented below.
Pearson Correlation
The Pearson correlation coefficients for the bronchitis
panel are reported in Table 70. The significant levels are in-
dicated at a = 0.01, a = 0.05, and a = 0.10. The pulmonary
-159-
-------
TABLE 69B. STATISTICAL PROFILE OF PULMONARY FUNCTION
VARIABLES AND AGE AND HEIGHT OF BRONCHITIS
PANELISTS MEASURED ON 12 TESTING DAYS
VARIABLE: MAXFVC = Maximum FVC (In liters x 100)
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
266.623
5,648.419
-0.614
75.156
-0.218
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
323.000
I 94.000
417.000
252
~ 110
VARIABLE: MAXFEV = Maximum FEV1>0 (in liters x 100)
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
212.266
4,619.893
-0.238
67.970
-0.280
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS =
290.000
54.000
344.000
252
110
VARIABLE: AGE = Age of panelist (in years)
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
44.622
44.377
-1.478
6.662
-0.050
RANGE = 20
MINIMUM =35
MAXIMUM = 55
VALID OBS = 43
MISSING OBS = 2
VARIABLE: HEIGHT = Height of panelist (in inches)
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
63.156
7.680
0.270
2.771
0.375
RANGE = 13
MINIMUM = 56
MAXIMUM = 69
VALID OBS =43
MISSING OBS =2
-160-
-------
' TABLE 69C. STATISTICAL PROFILE OF AIR POLLUTION
VARIABLES MEASURED ON 11 BRONCHITIS PANEL TESTING DAYS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
OZONE = Estimate
0.082
0.004
0.714
0.067
= 1.154
for oxidant (in ppm)
RANGE = 0.260
MINIMUM = 0.010
MAXIMUM = 0.270
VALID OBS = 360
MISSING OBS = 2
CO = Carbon monoxide (in ppm/100)
0.036
= 0.000
0.344
= . 0.013
0.686
RANGE = 0.060
MINIMUM = 0.020
MAXIMUM = 0.080
VALID OBS = 361
MISSING OBS = 1
N02 = Nitrogen dioxide (in ppm)
0.081
0.002
5.795
0.045
2.069
NO = Nitric oxide
0.015
0.000
8.400
0.009
2.665
HUMID = Relative
= 59.712
= 78.256
= -0.901
8.846
= 0.399
RANGE = 0.250
MINIMUM = 0.020
MAXIMUM = 0.270
VALID OBS = 333
MISSING OBS = 29
(in ppm)
RANGE = 0.050
MINIMUM = 0.010
MAXIMUM = 0.060
VALID OBS = 328
MISSING OBS = 34
humidity (in percent)
RANGE = 25.000
MINIMUM = 50.000
MAXIMUM = 75.000
VALID OBS = 361
MISSING OBS = 1
TEMP = Temperature (in degrees Fahrenheit)
= 77.393
= 95.823
= -1.001
9.789
= 0.134
RANGE = 37.000
• MINIMUM - 58.000
MAXIMUM = 95.000
VALID OBS = 361
MISSING OBS = 1
-161-
-------
TABLE: 70:.:: PEARSON. CORRELATION COEFFICIENTS: FOR THE BRONCHITIS PANEL
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
MAXFVC
MAXFEV
OZONE
0.15***
0.01
-0.04
-0.13**
0.04
0.00
-0.15***
-0.04
-0.02
-0.10**
-0.01
-0.00
CO
0.08
-0.03
0.05
-0.02
0.06
0.01
0.03
0.08
0.02
0.02
0.03
0.07
... N02 .
0.09
-0.00
0 . 12**
0.04
0.04
0.02
0.09
0.03***
0.02
-0.00
0.04
0.07 .
. . . .NO
0.03
0.08
0 . 14**
0.04
-0.04
0.01
0.11**
0 . 22***
0.13**
0.01
-0.04
0.02
HUMID
-0.07
0.07
0.08
0.10**
0.05
-0.08
0.10*
-0.07
-0.03
0.05
-0.04
-0.03
TEMP
0.05
-0.03
-0.07
-0.16***
-0.05
0.07
-0.13**
0.03
-0.01
-0.08
0.00
-0.03
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
function variables, MAXFVC and MAXFEV, are not significantly
correlated with any of the air pollution variables for the bron-
chitis panel.
Examination of the correlations between the qualitative
discomfort variables and the air pollution variables reveals
several significant correlations, Variables EYES is positively
correlated with OZONE, at a = 0,01, but EYES does not have
significant Pearson correlation coefficients with respect to any
of the other air pollution variables, CHEST has significant
positive correlations with NO? and NO. HEADACHE variable is
negatively correlated with OZONE at a level of significance of
a = 0.05, and HEADACHE is also significantly correlated with
HUMID and TEMP. BREATH, the variable indicating shortness of
breath, is positively correlated with NOo and NO at a = 0.01,
COUGH is only significantly correlated with NO. Finally, PHLEGM
is negatively correlated with OZONE at a = 0.05. None of the
other correlations were significant at a level better than a =
0.10. These results need to be compared to the more appropriate
non-parametric correlations.
Non-Parametric Correlations
The Spearman and Kendall correlation coefficients are re-
ported in Tables 71 and 72, respectively. The computed Spear-
man and Kendall correlations reinforce each other, but they are
different in some respects to the Pearson correlations. EYES
is now significantly correlated with OZONE, CO, and N02 at a =
0.10 or better. Variables THROAT and CHEST are now significantly
correlated with HUMID; however, the level of significance is not
strong. CHEST is correlated with NOo and NO at a greater level
of significance, a = 0.01, than is estimated by the Pearson
correlations.
Variable NAUSEA is again not significantly correlated with
any of the air pollution variables. However, OTHER is now sig-
nificantly correlated with HUMID and TEMP. BREATH is now sig-
nificantly correlated with HUMID as well as with N02 and NO.
COUGH remains significantly correlated with NO, but it is sig-
nificantly correlated with TEMP at a = 0.05.
Based on the results of the correlation analysis, further
analysis of EYES, CHEST, HEADACHE, BREATH, and COUGH seemed war-
ranted and was undertaken as described below. But due to the
lack of statistically significant relationships of THROAT, NAUSEA,
OTHER and PHLEGM with the air pollution variables, these dis-
comfort symptom variables were not included in further analysis
of the bronchitis panel. For this panel, the results for HEAD-
ACHE and HEADACHE EARLIER were almost identical. This cast
doubt on the usefulness of HEADACHE EARLIER as a different
measure of the effects of air pollution; thus, HEADACHE EARLIER
was also dropped from further analysis,
-163-
-------
TABLE 71. SPEARMAN CORRELATION COEFFICIENTS: FOR THE BRONCHITIS: PANEL .
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
. OZONE
O.«08*
-0.02
-0.06
-0.13***
0.02
0.00
-0.17***
-0.03
-0.01
-0.05
CQ
0.09**
-0.03
0.04
-0.02
0.06
-0.00
0.04
0.07
0.02 .
0.02
. . . . N02 .
0.09*
0.04
0 . 13***
0.03
0.03
-0.00
0.11**
0.19***
0.03
-0.01
. ... NO ...
0.04
0.06
0 . 14***
0.05
-0.00
0.05
0.12**
0.27***
0.12**
-0.14
. HUMID
-0.07
0.07*
0.07*
0 . 10**
0.04
-0.09**
0.09**
-0.08*
-0.03
0.05
TEMP
0.06
-0.02
-0.06
-0.16***
-0.05
0.07*
-0.13***
0.03
-0.01
-0.08***
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
TABLE 72. KENDALL: CORRELATION COEFFICIENTS. FOR THE BRONCHITIS. PANEL
EYES
THROAT
CHEST
HEADACHE
NAUSEA
OTHER
HEADACHE
EARLIER
BREATH
COUGH
PHLEGM
. OZONE
0.0?*
-0.02
-0.05
-0.11***
0.02
0.00
-0.14***
-0.03
-0.01
-0.04
CO .
0 . 08**
-0.02
0.04
-0.02
0.06
-0.00
0.03
0.06
0.01
0.02
. . . N02 .
0.08*
0.03
0.11***
0.02
0.02
-0.00
0.09**
0 . 16***
0.03
-0.01
. . NO
0.04
0.06
0 . 14***
0.05
-0.00
0.05
0.12**
0 . 26***
0 . 12**
-0.03
. HUMID
-0.06
0.07*
0.07*
0,09**
0.03
-0.09** .
0.09**
-0.08*
-0.03
0.05
. . TEMP
0.05
-0.01
-0.05
-0 . 13***
-0.04
0.06*
_0 . 11***
0.03
-0.01
-0.07*
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
MEASURES OF ASSOCIATION BETWEEN VARIABLES
In parallel with the analyses of the asthma and outdoor
worker panels, attempts were made to identify and evaluate'as-
sociations between variables in the bronchitis panel. Several
methods were applied. The results are explained on the follow-
ing pages.
Simple Linear Regressions of FVC and FEV-. Q
Simple linear regressions were computed with MAXFVC or
MAXFEV as the dependent variable. None-produced statistically
significant results. Inspection of scattergrams revealed no
non-linear relationship for MAXFVC or MAXFEV with the air pollu-
tion variables for unrestricted bronchitis data.
Contingency Tables Between FEV-, Q and. the Air. Pollution Variables
Contingency tables were constructed using MAXFEV as the row
variable and the air pollution variables as the column variables.
MAXFEV was divided into the same classes as the asthma panel.
The classes for the air pollution variables were also the same
as for the asthma panel (refer to Section 7"> . All of the tables
are (2x2) and have critical values of the x ~ 6-63 at a = 0.01,
X2 = 3.84 at a = 0.05, and x2 = 2.70 at a = 0.10. Computed x
values greater than the chosen critical value lead to rejection
of the hypothesis of independence between the variable classes.
?
None of the computed x values are significant at a = 0.10.
Thus, the hypothesis of independence between MAXFEV and the air
pollution variables OZONE, CO., NO2, and NO is not rejected. The
bronchitis panel showed no strong sensitivity to the levels of
air pollution measured during the tests.
Linear Probability Model
Linear probability functions were estimated for the bron-
chitis panel for each of the discomfort symptom variables. The
explanatory variable for each was a different measure of air
pollution."" The regressions which were significant at a = 0.05,
or better, are reported in Table 73. At first glance, the co-
efficients of determination may seem small, but given a binary
dependent variable, it is impossible to obtain R-2 = 1 when fit-
ting a straight line to the data.
-166-
-------
•TABLE 73. SIMPLE LINEAR PROBABILITY REGRESSIONS
FOR THE BRONCHITIS PANEL
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
Prob
(EYES)
(CHEST)
(CHEST)
(HEADACHE)
(HEADACHE)
(HEADACHE)
(BREATH)
(BREATH)
(COUGH)
= 0.
= 0.
= 0.
= 0.
= 0.
= 0.
= 0.
= 0.
= 0.
086
455
446
421
Oil
933
205
233
787
+ 1.
+ 1.
+ 7.
- 0.
+ 0.
- 0.
+ 2.
+ 12
+ 4.
114
273
718
927
006
008
500
(OZONE)***
(N02)***
(NO)**
(OZONE)**
(HUMID) **
(TEMP)***
(N02)***
.106 (NO)***
942 (NO)***
R2
R
R
R
2
2
2
R2
R
R
R
2
2
2
R2
= 0.
= 0.
= 0.
= 0.
= 0.
= 0.
= 0.
= 0.
= 0.
022
013
019
017
Oil
025
053
050
016
**Signifleant at a = 0.05
***Signifleant at a = 0.01
EYES is significantly related to OZONE. Given an increase
in OZONE of 0.10 will increase the probability of experiencing
eye discomfort by approximately 0.10 or 10 percent. CHEST is
significantly related to N02 and NO. The proportion of bron-
chitics experiencing chest discomfort will increase by approxi-
mately 0.13 or 13 percent given an increase in N02 of 0.10.
That proportion is predicted to increase about 0.08 or 8 percent
for a 0.01 increase in NO.
HEADACHE is significantly related to OZONE, HUMID, and
TEMP. However, the level of significance is not strong for
OZONE and HUMID. And the sign of the coefficient attached to
OZONE is suspect. BREATH is significantly related to N02 and
NO. The percentage of respondents experiencing shortness of
breath is predicted to increase by 0.25 or 25 percent given a
0.10 increase in N02- That percentage will increase by 0.12 or
12 percent given a 0.01 increase in NO. Finally, COUGH is sig-
nificantly related to NO. A 0.05 or 5 percent increase in the
percentage of bronchitics having a cough is predicted for each
0.01 increase in NO.
-167-
-------
SELECTED. MULTIVARIATE LINEAR PROBABILITY REGRESSIONS
The previous section showed five discomfort symptom vari-
ables which demonstrated significant statistical association
with some of the air pollution variables. Those discomfort
symptom variables are EYES, CHEST, HEADACHE, BREATH, and COUGH.
Multivariate linear probability specifications were estimated
for each using the qualitative discomfort symptom as the depen-
dent variables with OZONE, CO, N02, HUMID, TEMP, and COLD as ex
planatory variables. The potential explanatory variable NO was
excluded because it was expected to be collinear with N02- The
results are reported in Table 74 except for the regression with
COUGH as the dependent variable. That regression has no statis
tically significant coefficients for any of the explanatory
variables.
For EYES as the dependent variable, only N0£ has a statis-
tically significant coefficient. This result may be contrasted
to the single linear probability regression where OZONE has the
only coefficient that was statistically significant. It appears
that when HUMID, TEMP, and COLD are included as control vari-
ables in the multivariate specification, N0£ explains more of
the variation in eye discomfort than OZONE. The simple linear
probability regressions for CHEST show N02 and NO as significant
explanatory variables. Likewise, the multivariate regressions
show N02 as a significant explanatory variable, but NO is not
included. The simple regressions for HEADACHE show OZONE,
HUMID, and TEMP are no longer significant in the presence of the
other variables. CO and N02 are statistically significant ex-
planatory variables; however, the sign of the CO coefficient is
negative. The multivariate specification for BREATH resulted in
CO and N02 having statistically significant coefficients, but
again the CO coefficient is negative.
DISCOMFORT SYMPTOMS ANALYZED BY DATE OF MEASUREMENT
The bronchitis panel was tested on 11 different dates. On
each date, the proportion of the panelists reporting each dis-
comfort symptom was computed. On each date, the average level
of each air pollution variable was also computed. The average
levels of air pollution were then related to the proportion of
panelists reporting each discomfort symptom on each date. This
was done using two regression techniques.
Simple Linear Regressions
The LOGIT specification was applied to selected discomfort
symptom proportions. The discomfort variables chosen were EYES,
CHEST, HEADACHE, BREATH, and COUGH. The explanatory variables
were OZONE, CO, N02 and NO. The linear probability specifica-
tion was also estimated. The form of these specifications is
-168-
-------
TABLE 74. MULTIVARIATE LINEAR PROBABILITY REGRESSIONS FOR
THE BRONCHITIS PANEL
(STANDARD ERRORS IN PARENTHESES)
Explanatory
Variable
CONSTANT
OZONE
CO
N02
HUMID
TEMP
COLD
R2
F
N
Dependent Variable
EYES
0.82523
1.84982
(0.10597)
-6.30112
(4.33739)
2.16677***
(1.09490)
-0.00197
(0.00528)
-0.00452
(0.00550)
-0.15350
(0.10597)
0.045
2.163
332
CHEST
0.39688
0.31416
(0.57754)
-3.50222
(4,22136)
2.11782***
(1.05382)
0.00461
(0.00508)
-0.00246
(0.00541)
0.05614
(0.10606)
0.027
1.492
332
HEADACHE
2.02789
0.02662
(0.54309)
-9.03733***
(3.96958)
1.91167**
(0.99096)
-0.00571
(0.00478)
-0.01507
(0.00509)
0.07346
(0.09973)
0.058
3.353
332
BREATH
0.89568
0.54533
(0.55941)
-8.41667***
(4.08887)
4.11930***
(1.02014)
-0.00398
(0.00492)
-0.00417
(0.00524)
0.00938
(0.10273)
0.066
3.848
332
*Significant at a = 0.10
**Significant at a = 0.05
-169-
-------
discussed in connection with the asthma panel (Section 7) and
outdoor worker panel (Section 8). The significant regressions
are reported in Table 75. None of the regressions for HEADACHE
and COUGH are significant at a = 0.10 or better, so they are not
reported.
EYES is significantly related to OZONE in both the LOGIT
and linear probability regressions. None of the other air pol-
lution variables showed significance in explaining eye discom-
fort. CHEST is significantly related to N02 in both the linear
probability and LOGIT regressions. An increase in N0£ is pre-
dicted to cause an increase in the probability of chest discom-
fort. BREATH has significant predictors, N0£ and NO, for both
the linear probability and LOGIT regressions. An increase in
N02 and NO is predicted to cause an increase in the proportion
of bronchitics experiencing shortness of breath.
Selected Multivariate Regressions
The statistically significant single variable regressions
using the proportions of panelists reporting discomfort symptoms
involved the EYES, CHEST, and BREATH variables. Therefore,
these three variables were used to estimate multivariate proba-
bility regression. Linear and LOGIT specifications were esti-
mated with explanatory variables average ozone (AVEOZ), average
carbon monoxide (AVECO), average nitrogen dioxide (AVEN02),
average :temperature (AVETEMP), and average humidity (AVEHUM).
Average nitric oxide was excluded to avoid collinearity with
AVEN02. The results of the estimations for the linear specifi-
cation are reported in Table 76, and the results for the LOGIT
specification are reported in Table 77.
The total F rejects a hypothesis that all estimated coeffi-
cients are simultaneously zero at a = 0.05 for all regressions.
The R2 values are all high, indicating that a large percentage
of the variation in the respective dependent variables is ex-
plained. Explanatory variables AVEOZ, AVECO., and AVEN02 have
estimated coefficients significantly different from zero at a =
0.10. AVETEMP is statistically significant only once and AVEHUM
is not statistically significant at all, but they are retained
to avoid biasing the other estimated coefficients.
One phenomenon which should be noted is how the estimated
coefficients change in size and significance when compared to
the single explanatory variable estimates. OZONE was a signifi-
cant explanatory variable when used alone only in explaining
eye discomfort, but AVEOZ is significant in explaining eye dis-
comfort, chest discomfort, and shortness of breath in the multi-
variate regressions. Even more striking, CO was not significant
in any of the single variable estimates, but AVECO, like AVEOZ,
is significant in all of the multivariate estimates. Further-
more, AVEN02 remains significant in explaining chest discomfort
-170-
-------
TABLE 75. SIGNIFICANT SIMPLE AND LOGIT REGRESSIONS OF
_ DISCOMFORT SYMPTOM PROPORTIONS . _ .
Prob(EYES) = 0.30564 + 2. 09268 (OZONE)*** R2 = 0.552
In - Prob(EYES) — / _0< 32137 + 8 . 840 78 (OZONE)*** R2 = 0.556
1 - Prob(EYES) ..
In
Prob (CHEST) = 0.43526 + 1. 29806 (N02)*** R = 0.287
(CHEST) — ( _0 27245 + 5.44977(N02)** R2 = 0.291
—
1 - Prob (CHEST) )
Prob (BREATH) = .0.20190 + 2 . 50020 (N02)** R = 0.412
In (BREATH) • = _li30189 + io . 89243 (N02)** R2 = 0.416
( 1 - Prob (BREATH) )
Prob (BREATH) = 0.17715 + 15 . 17725 (NO)* R = 0.258
In
Prob(BREATH)
1 - Prob(BREATH)
+ 66. 13254(NO)*
R = 0.261
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-171-
-------
TABLE 76. MULTIVARIATE LINEAR PROBABILITY REGRESSIONS WITH PROPORTIONS
OF EYE DISCOMFORT, CHEST DISCOMFORT, AND SHORTNESS OF BREATH
AS DEPENDENT VARIABLES
' : : : : : . . (STANDARD ERRORS: IN PARENTHESES) : ::.::.::: . . :
Prob(EYES) = 1.432 + 3.630(AVEOZ)** - 16.519(AVECO)** + 3.821(AVEN02)**
(0.605) (4.721) (0.844)
- O.IO(AVETEMP)** - O.OOS(AVEHUM)
(0.005) (0..005)
R2 = 0.90, F. = 11.01, D.W. = 2.57, N =12
Prob(CHEST) = 1.188 + 1.902(AVEOZ)** - 15.221(AVECO)** + 3.922(AVEN02)**
(0.596) (4.651) (0.831)
- 0.009(AVETEM) + 0.002(AVEHUM)
(0.005) (0.005)
R2 = 0.84, F = 6.34, D.W. = 1.80, N = 12
Prob(BREATH) = 1.987 + 2.704(AVEOZ)** - 19.267(AVECO)** + 5.328(AVEN02)**
(1.234) (9.622) (1.720)
- 0.012(AVETEMP) - 0.Oil(AVEHUM)
(0.011) (0.010)
R2 = 0.86, F = 3.36, D.W. - 2.19, N = 12
**Signifleant at a = 0.05
-------
TABLE 77. MULTIVARIATE LQGIT PROBABILITY REGRESSIONS WITH PROPORTIONS OF EYE
DISCOMFORT, CHEST DISCOMFORT, AND SHORTNESS OF BREATH AS DEPENDENT VARIABLES
: ' . : ' : . . : : : . . ;(STANDARD ERRORS IN PARENTHESES)' : . : : : . . : .
In J Prob(EYES) 1 = 4.022 + 15.340(AVEOZ)** - 69. 391(AVECO)** + 15.905(AVEN02)**
( 1 - Prob(EYES) ) (2.607) (20.327) (3.634)
- 0.043(AVETEMP) - 0.014(AVEHUM)
(0.023) (0.021)
R2 = 0.90, F .= 10.47, D.W.. .= 2.58, N = 12
i Prob(CHEST) )
1 Prob(CHEST) ( = 3'02A + 7.952(AVEOZ)** - 64.259(AVECO)** + 16.498(AVEN02)**
(2.464) (19.216) . (3.436)
- 0.038(AVETEMP) + 0.008(AVEHUM)
(0.021) (0.020)
R2 ='0.84, F =6.49, D.W. = 1.81, N = 12
, Prob(BREATH) (
- Prob(BREATH)) = 5>33° + H.372(AVEOZ)** - 80.713(AVECO) + 22.987(N02)**
(5.358) (41.784) (7.471)
- 0.042(AVETEMP) - 0.040(AVEHUM)
(0.046) (0.043)
R2 = 0.86, F = 3.33, D.W. = 2.15, N = 12
**Signifleant at a = 0.05
-------
and shortness of breath, but it is not significant in explaining
eye discomfort in the multivariate regression. The sign of the
coefficients remains consistent, but the coefficients change in
magnitude indicating that the variables probably interact in
their effect on the discomfort symptoms.
OVERALL SUMMARY OF THE BRONCHITIS PANEL DATA ANALYSIS
Air pollution levels were light to moderate on the 12 days
members of the bronchitis panel were tested. The weighted
values of the air pollution and weather variables are presented
in Table 69C. The weighted averages of responses to symptom
interviewing and pulmonary function testing are shown in Tables
69A and 69B. The symptom and aerometric data are compared
graphically in Appendix K.
Correlation analysis did not demonstrate any statistically
significant association between MAXFVC or MAXFEV and'the air
pollution variables for the bronchitis panel. Scattergrams and
contingency tables upheld this finding. The results were not so
sparse for the discomfort symptoms, however,
Parametric and non-parametric correlation coefficients were
statistically significant for eye discomfort with OZONE, CO, and
N02- Chest discomfort and shortness of breath were signifi-
cantly correlated with N02 and NO. Finally, having a cough was
correlated with NO. Simple linear probability estimates using
the dichotomous discomfort symptom variables estimated the re-
lationships of OZONE on eye discomfort, N02 and NO on chest dis-
comfort, OZONE, HUMID, and TEMP on headache, N02 and NO on
shortness of breath, and NO on cough. All of the above esti-
mates were significant at a = 0.05 or better.
The discomfort symptoms proportions by date were computed
and the average levels of the air pollution variables were also
computed. The proportions were used to estimate linear proba-
bility and LOGIT regressions. Comparisons of the single explan-
atory variable estimates to the multiple explanatory variable
estimates demonstrated several differences. Statistical theory
supports the multivariate as the more precise technique. Logic
indicates the sigmoid LOGIT specification is consistent with
the probability interpretation. The LOGIT estimates showed a
statistically significant relationship of EYES, CHEST, and
BREATH with OZONE. There was also a statistically significant
relationship of the same discomfort symptom variables with CO.
The LOGIT estimates showed an inverse relationship of EYES,
CHEST, and BREATH with CO. These results were reinforced by the
linear probability regressions.
-174-
-------
SECTION 10
DATA ANALYSIS—ATHLETE PANEL
STATISTICAL DESCRIPTION OF THE ATHLETE PANEL
Members of a cross-country team at Citrus College in Azusa,
California were recruited for the athlete panel. Seventeen
athletes agreed to participate. All were 18 or 19 years of age,
physically conditioned for distance running, and apparently free
of disease. The qualifications for participation did not cover
smoking; however, none of the panelists were smokers. Table 78
summarizes the composition of this panel.
TABLE 78. COMPOSITION OF THE
ATHLETE PANEL
Total
Characteristics Panel
Subjects enrolled in panel 17
Current cigarette smokers 0
Subjects under 21 years old 17
Subjects with 12 or more
years of school completed 17
Race other than White 5
The comprehensive clinical examinations administered before
and after the testing period and the clinical interviews held
after the testing period confirmed that all of the athletes were
in good health. The athlete panel was tested on 11 days over a
period of seven weeks. The schedule of testing is given in Ap-
pendix L. Attendance ranged from 7 to 11 subjects, with an
average daily attendance of 8. The types of data obtained from
each athlete panelist are listed in Section 4 of this report.
All data were included in the analysis except for heart function
measures, the results of the hematological and nasal smear eval-
uations, the nausea symptom information, and the information on
-175-
-------
medication. The reasons for not including the heart, blood, and
nasal smear data are the same as those given for the bronchitis
panel. None of the athlete panelists suffered symptoms of-
nausea and none took medication on any of the testing days.
Description of the Variables
The discomfort symptom variables were measured with the
form shown in Appendix F. The same coding was used on the re-
sponses as in the other panels: unity for "Yes," zero for "No."
The main difference between the athlete panel and the other pan-
els was that discomfort symptom information was taken before and
after the athletes subjected themselves to stress, i.e., ran a
distance of at least two miles. A statistical profile of the
discomfort variables is given in Table 79A.
The discomfort symptoms reported before running are indi-
cated by placing a "1" following the variable name. For
example, EYES1 is the variable name for eye discomfort reported
before running. The discomfort symptoms reported after running
are indicated by placing a "2" following the variable name.
Thus, EYES2 denotes eye discomfort reported after running.
The proportions of the discomfort symptoms reported were
extremely small for all of the discomfort variables listed. For
example, only 8 percent of all responses were "Yes" to the
presence of eye discomfort before running, and only 9 percent
were "Yes" to eye discomfort after running. The pattern of
little or no discomfort before stress and little or no change
after stress is consistent for all of the discomfort variables.
Two pulmonary function variables were used in the analysis
of the athlete panel. Each was measured before and after run-
ning each testing day. A profile of these variables is given in
Table 79B. Maximum FVC and maximum FEVi.Q are listed as BFVC
and BFEV for the measurements made before running. Maximum FVC
and maximum FEV^o are listed as AFVC and AFEV for the measure-
ments made after running. BFVC and AFVC were each taken from
the best of five maneuvers. BFEV and AFEV were the highest
FEV]_o taken from the volume-time tracings of the FVC maneuvers.
The AGE and HEIGHT of each panelist were recorded as age in
years and standing height in inches. These data were used to
adjust the BFVC, BFEV, AFVC, and AFEV scores prior to analysis.
A profile of the distances run between reporting discomfort
symptoms and taking pulmonary function tests is shown in
Table 79C.
The air pollution and weather variables used in this analy-
sis are the same as those used in the asthma panel analysis.
The profile of these variables is given in Table 79D. The air
pollution variables were obtained as hourly averages measured in
-176-
-------
TABLE 79A. STATISTICAL PROFILE OK DISCOMFORT SYMPTOMS REPORTED RY ATHLETE PANELISTS ON 11 TESTING DAYS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
KEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
HEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
HEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
KEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
HEAN
VARIANCE
KURTOSIS
STD DF.V
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
EYES1 • Eye discomfort before running
0.080
0.075
7.638
0.274
3.121
THROAT1 - Throat
0.034
0.034
24.350
0.184
5.162
CHEST1 -
0.011
0.011
- 83.000
0.107
- 9.272
HEADACHE1
0.057
• 0.055
- 12.641
- 0.234
• 3.847
OTHER1 -
- 0.092
0.084
- 6.081
- 0.291
2.857
COUGH1 -
- 0.069
0.065
- 9.720
0.255
• 3.442
PHLEGM1 -
0.000
0.000
0.000
0.000
0.000
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
discomfort before
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
Chest discomfort before
RANGE
. MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
„
1.000
0.000
1. 000
87
0
running
1.000
0.000
1.00
87
0
running
-
1.000
0.000
1.000
87
0
» Headache discomfort before
running
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
Other discomfort before
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
m
1.000
0.000
1.000
87
0
running
K
1.000
0.000
1.000
87
0
Cough before running
Phlegm
HEADACHE EARLIER
0.046
0.044
17.028
0.211
4.386
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
before running
RANGE
MINIMUM
' MAXIMUM
VALID OBS
MISSING OBS
n
-
• Headache earlier
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
1.000
0.000
1.000
87
0
0.000
0.000
0.000
6
0
today
1.000
0.000
1.000
87
0
VARIABLE:
MEAN
VAIUANCE
KUKTOS1S
STD DK'/
SKEWNESS
VARIABLE:
HEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
KEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE:
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
EYF.S2 - Eye Discomfort after running
0.092
0.084
6.081
0.291
2.857
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
- 1.000
- 0.000
- 1.000
- 87
- 0
THROAT2 - Throat discomfort after
0.115
0.103
3.909
0.321
2.443
C1IEST2 - Chest
—
0.092
0.084
6.081
0.291
2.857
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
discomfort after
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
HEADACHE2 - Headache discomfort
running
0.069
0.065
9.720
0.255
3.442
OTHER2 - Other
=
.0.161
0.137
1.457
0.370
1.867
COUGH2 = Cough
—
0.069
0.065
9.720
0.255
3.442
PHLECM2 - Phlegi
a
0.167
0.167
6.000
0.408
2.449
COLD - Bad cold
m
m
0.023
0.023
39.006
0.151
6.440
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
discomfort after
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
after running
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
en after running
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
today
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
-
running
1.000
0.000
1.000
37
0
running
e
1.000
0.000
1.000
87
0
after
!
1.000
0.000
1.000
87
0
running
i
—
m
•
1.000
0.000
1.000
87
0
1.000
0.000
1.000
87
0
1.000
0.000
1.000
6
0
1.000
0.000
1.000
87
0
BREATH - Shortness of breath today
- 0.046
- 0.044
- 17.028
- 0.211
- 4.386
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
m
1.000
0.000
1.000
87
0
-177-
-------
TABLE 79B. STATISTICAL PROFILE OF PULMONARY FUNCTION
VARIABLES AND AGE AND HEIGHT OF ATHLETE PANELISTS
BEFORE AND AFTER RUNNING ON 11 DAYS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV.
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
RANGE
MINIMUM
MAXIMUM
VARIABLE :
RANGE
MINIMUM
MAXIMUM
BFVC = Maximum FVC before running
(in liters x 100)
494.253 RANGE
= 4,140.726 MINIMUM
-0.612 MAXIMUM
64.348 VALID DBS
-0.321 MISSING DBS =
BFEV = Maximum FEV^ Q before running
(in liters x'lOO)
416.069 RANGE
= 3,511.111 MINIMUM
0.116 MAXIMUM
59.255 VALID OBS
-0.136 MISSING OBS =
AFVC = Maximum FVC after running
(in liters x 100)
489.250 RANGE
= 3,958.744 MINIMUM
-0.342 MAXIMUM-
62.919 VALID OBS
0.173 MISSING OBS =
AFEV = Maximum FEV^ Q after running
(in liters x'lOO)
410.738 . .RANGE
= 3,253.786 MINIMUM
0.716 MAXIMUM
57.042 VALID OBS
0.273 MISSING OBS =
AGE = Age of panelist (in years)
= 1 ' VALID OBS
= 18 MISSING OBS =
= 19
292.000
324.000
616.000
87
0
284.000
280.000
564.000
87
0
311.000
355.000
666.000
84
3
296.000
280.000
576.000
84
3
17
0
HEIGHT = Height of panelist (in inches)
= 13 VALID OBS
= 64 MISSING OBS =
= 77
17
0
-178-
-------
TABLE 79C. STATISTICAL PROFILE OF DISTANCES RUN BY
ATHLETE PANELISTS ON 11 TESTING DAYS
VARIABLE: DIST = Distance run (in miles)
MEAN
VARIANCE =
KURTOSIS =
STD DEV
SKEWNESS =
2.092
0.364
39.006
0.603
6.440
RANGE
MINIMUM
MAXIMUM
VALID DBS
MISSING DBS =
4.000
2.000
6.000
87
0
TABLE 79D. STATISTICAL PROFILE OF AIR POLLUTION
VARIABLES MEASURED ON 11 ATHLETE PANEL TESITNG DAYS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIABLE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
VARIABLE :
MEAN
VARIANCE
KURTOSIS
STD DEV
SKEWNESS
OZONE
0.
0.
0.
= 0.
1.
CO =
= 0.
0.
= -0.
0.
0.
N02 =
= 0.
0.
= -0.
0.
0.
NO =
= 0.
0.
2.
0.
1.
HUMID
= 61.
= 47.
0.
6.
0.
TEMP
= 73.
= 80.
0.
8.
0.
= Estimate for oxidant (in ppm)
107
005
416
071
187
RANGE
MINIMUM
MAXIMUM
VALID DBS
MSSING OBS
cs
1=
=:
0.260
0.020
0.280
86
1
Carbon monoxide (in ppm/100)
038
000
784
012
558
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
s=
s=
0.040
0.020
0.060
86
1
Nitrogen dioxide (in ppm)
076
001
572
036
614
Nitric oxide
013
000
349
006
912
= Relative
570
095
403
863
127
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
(in ppm)
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
humidity (in
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
=
=
=
=
0.130
0.020
0.150
78
9
0.020
0.010
0.030
78
9
percent)
=5
=5
=
25.000
50.000
75.000
86
1
= Temperature (in degrees
Fahrenheit)
733
575
912
976
271
RANGE
MINIMUM
MAXIMUM
VALID OBS
MISSING OBS
=
S
43.000
52.000
95.000
86
1
-179-
-------
ppm (CO in ppm/100). Relative humidity was estimated by hour
and expressed in percent. A discussion on how the estimates
were made is given in the paragraphs of Section 7 which describe
the variables used in the asthma panel analysis. Temperature
was obtained as hourly averages measured in degrees Fahrenheit.
Weighted averages of exposure to air pollutant, humidity,
and temperature levels were used in the analysis of the athlete
panel. As in the analysis of the asthma panel, the hour at
which each panelist reported for testing was noted and the
aerometric values for that hour were included in the weighted
average for that day.
CORRELATION ANALYSIS
As in the other panels, Pearson and non-parametric correla-
tion analyses were applied to the data collected from this
panel. The results are highlighted in the following tables.
Pearson Correlation
The Pearson correlation coefficients for the athlete panel
are reported in Tables 80 and 81. Table 80 reports the coeffi-
cients for the air pollution variables with the discomfort
symptoms and pulmonary function results before running. Table
81 reports the coefficients after running. Before running, the
only variables which have significant correlations are: EYES1
with OZONE and CO,. THROATl with HUMID and TEMP, and OTHERl with
OZONE. None of the other pairs have significant correlations
better than a = 0.10. After running, the only variables with
significant correlations are: EYES2 with OZ.ONE, HEADACHE2 with
OZONE, and OTHER2 with TEMP. None of the other pairs have sig-
nificant correlations.
Non-Parametric Correlations
The non-parametric correlation coefficients, Spearman's Rs
and Kendall's T, were computed between the qualitative discom-
fort symptoms and the air monitoring data. The Spearman and
Kendall coefficient variables before running are reported in
Table 82 and 83, respectively. The results tend to confirm the
Pearson correlations. Looking at Table 82, only EYES1 with
OZONE, EYES1 with CO, and OTHERl with OZONE are significant at
a level of a = 0.05 or better before running. In Table 83,
examination of the air pollution variables with the discomfort
symptom variables results in non-parametric correlations signif-
icant at a = 0.05 or better for EYES1 with OZONE, EYES1 with CO
THROATl with HUMID, and OTHERl with OZONE.
-180-
-------
TABLE 80. PEARSON CORRELATION COEFFICIENTS FOR THE ATHLETE PANEL BEFORE RUNNING
1
oo
M
I
EYE SI
THROAT1
CHEST1
HEADACHE1
OTHER1
BREATH1
COUGH1
PHLEGM1
HEADACHE
EARLIER
BFVC
BFEV
OZONE
0.52***
-0.05
-0.01
0.15
0 . 28***
0.03
-0.01 .
0.01
0.01
0.01
0.02
CO
0.25**
-0.13
-0.08
0.11
0.14
-0.02
0.12
-0.12
0.12
0.01
-0.06
N02
-0.07
-0.07
-0.08
-0.04
0.01
-0.08
0.05
-0.05
0.01
0.01
-0.04
NO
-0.15
-0.10
-0.06
-0.04
-0.08
-0.02
0.22
-0.22
-0.02
0.07
-0.01
' HUMID
0.10
-0.21*
0.01
0.11
-0.12
-0.08
-0.07
0.07
0.12
-0.03
-0.03
TEMP
0.02
0.20*
0.09
-0.14
0.09
0.06
0.05
-0.05
-0.17
-0.10
-0.06
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
TABLE 81. PEARSON CORRELATION COEFFICIENTS FOR THE ATHLETE PANEL AFTER RUNNING
1
M
OO
NJ
1
EYES2
THROAT2
CHEST2
HEADACHE2
OTHER2
COUGH2
PHLEGM2
AFVC
AFEV
OZONE
0.45***
0.14
0.07
0.23**
-0.06
0.10
-0.10 '
-0.06
-0.03
CO
0.18
0.08
0.01
0.15
-0.18
0.15
-0.11
0.03
0.01
N02
-0.04
t
-0.07
0.01
'-0.07
-0.10
-0.02
0.06
0.02
0.00
NO
-0.15
-0.04
0.04
0.02
-0.01
0.02
-0.03
0.11
0.10
HUMID
0.10
0.10
0.17
0.02
0.15
0.10
-0.09
-0.03
-0.03
TEMP
0.09
-0.05
-0.01
-0.08
-0.19*
-0.01
-0.07
-0.07
-0.02
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
TABLE 82. SPEARMAN CORRELATION COEFFICIENTS FOR THE ATHLETE PANEL BEFORE RUNNING
EYES1
THROAT1
CHEST1
HEADACHE1
OTHER1
BREATH1
M COUGH1
oo
V PHLEGM1
HEADACHE
EARLIER
OZONE
0 . 40***
-0.00
0.05
0.09
0.18**
-0.05
-0.07 '
0.07
-0.04
CO
0.25***
-0.14*
-0.08
0.11
0.12
-0.04
0.11
-0.11
0.13
N02
-0.00
-0.09
-0.09
-0.03
0.04
-0.08 .
0.04
-0.04
0.01
NO
-0.16*
-0.10
-0.06
-0.02
-0.06
0.01
0.15
-0.15*
0.01
HUMID
0.10
-0.22**
0.01
0.11
-0.12
-0.08
-0.07
0.07
0.12
TEMP
0.04
0 . 20**
0.13
-0.13
0.16*
0.10
0.09
-0.09
-0.16*
— Significant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
TABLE 83. KENDALL CORRELATION COEFFICIENTS FOR THE ATHLETE PANEL BEFORE RUNNING
EYES1
THROAT1
CHEST1
HEADACHE1
OTHER1
BREATH1
^ COUGH1
oo
f PHLEGM1
HEADACHE
EARLIER
. OZONE
0.34***
-0.01
0.04
0.08
0.15**
-0.04
-0.06 '
0.06
-0.04
CO
0.23**
-0.13*
-0.07
0.10 •
0.11
-0.03
0.10
-0.10
0.11 .
N02
-0.01
-0.07
-0.08
-0.03
0.03
-0.07
0.03
-0.03
0,01
NO
-0.16*
-0.10
-0.06
-0.02
-0.06
0.01
0.15
-0.15
0.01
HUMID
0.10
-0.21**
0.01
0.11
-0.11
-0.08
-0.07
0.07
0.12
TEMP
0.03
0.17
0.11
-0.11
0.13*
0.09
0.07
-0.07
-0.13*
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
The non-parametric correlations for the discomfort symptoms
present after running are reported in Tables 84 and 85 . The
correlation coefficients which are significant at a = 0.05 or
better are those for EYES 2 with OZONE and CO, OTHER2 with CO,
and PHLEGM2 with TEMP. The only two non-parametric correlations
which are significant both before and after running are EYES1
and EYES2 with OZONE and CO.
MEASURES OF ASSOCIATION BETWEEN VARIABLES
Simple linear regressions and linear probability analyses
were performed on the health data collected from the athlete
panel. In addition, contingency tables were constructed and
X2 tests were applied between the FVC and FEV^ Q variables and
the air pollution variables. The results are reported below.
Simple Linear Regressions of FVC and FEVy Q
FVC and FEV^ Q before and after running were regressed on
the air pollution data for OZONE, CO, N02 , NO, HUMID, and TEMP.
The results indicate no significant linear relationship between
either FVC or FEV-^ Q on the level of air pollution variables for
the athlete panel. ' Scattergrams were examined to see if there
were any discernable non-linear relationships between FVC or
FEV^ Q and the level of air pollution for this panel. None were
found .
Simple linear regression equations were computed with FVC
or FEV^ Q before and after running as the dependent variables .
The explanatory variables were the air pollution variables . The
test of the hypothesis of a significant linear relationship be-
tween the dependent and independent variable involved a two-
tailed test that R is significantly different from ze.ro. That
hypothesis could not be rejected at a = 0.10 or greater levels
of significance for any of the regressions.
Contingency Tables Between the Pulmonary Function and the Air
Pollution Variables '•
A series of contingency tables were constructed using FVC
and FEV^ Q before and after running with the air pollution vari-
ables. FVC (in liters x 100) was divided into two classes:
0
-------
TABLE 84. SPEARMAN CORRELATION COEFFICIENTS FOR THE ATHLETE PANEL AFTER RUNNING
oo
ON
EYES2
THROAT2
CHEST2
HEADACHE2
OTHER2
COUGH2
PHLEGM2
OZONE
0 . 40***
0.07
0.08
0.14
-0.05
0.09
-0.16* -
CO
0.18**
0.06
0.01
0.14*.
-0.19**
0.11
-0.08
N02
0.05
-0.12
0.00
-0.03
-0.12
-0.04 .
0.09
NO
-0.16*
-0.01
0.03
-0.02
-0.03
0.06
-0.08
HUMID
0.10
0.10
0.17*
0.03
0.15*
0.11
-0.10
TEMP
0.07
-0.02
-0.03
-0.04
-0.17*
0.07
-0.18**
*Signifleant at a = 0.10
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
TABLE 85. KENDALL CORRELATION COEFFICIENTS FOR THE ATHLETE PANEL AFTER RUNNING
1
I-1
00
">J
1
EYES2
THROAT2
CHEST2
HEADACHE2
OTHER2
COUGH2
PHLEGM2
*Signific<
OZONE
0 . 34***
0.06
0.06
0.12
-0.04
0.08
-0.13* -
ant at a =
CO
0.17**
0.05
0.01
0.13 .
-0.18**
0.10
-0.07
0.10
N02
0.04
-0.10
0.00
-0.02
-0.10
-0.04 ,
0.07
NO
-0.16*
-0.01
0.03
-0.02
-0.03
0.06
-0.08
HUMID
0.10
0.09
0.17*
0.03
0.15*
0.10
-0.10
TEMP
0.06
-0.02
-0.02
-0.03
-0.14*
0.06
-0.15**
**Signifleant at a = 0.05
***Signifleant at a = 0.01
-------
The results for FVC before and after running with the air
pollution variables OZONE, CO, N02, and NO were calculated.
None of the computed x^ values were significant .at a = 0.10
or better. The results for FEV^ Q before and after running
with the air pollution variables'were also calculated. Again,
none of the x values were significant. It is not possible to
reject the hypothesis of independence between the FVC or
FEV]^ Q variables and the air pollution variables either before
or after running. The indication is that the athlete panelists
were relatively insensitive to the levels of air pollution
measured during the survey.
A separate set of contingency tables was constructed be-
tween the variable COLD and the discomfort symptoms. The null
hypothesis was that reporting a discomfort symptom is indepen-
dent of having a cold. The hypothesis of independence between
THROAT1 and COLD is rejected at a level of significance of a =
0.10. The hypothesis is rejected at a = 0.01 for CHEST1 with
COLD. None of the other discomfort symptoms before running
show any association with having a cold. After running, none
of the discomfort symptoms show any association with having a
cold; thus, the hypothesis of independence between the discom-
fort symptoms and having a cold cannot be rejected.
Linear Probability Model
Linear probability functions were estimated using the qual-
itative discomfort symptoms for the athlete panel as the depen-
dent variables. Regressions were performed for the dependent
variables both before and after running. The functions which
were significant at a = 0.10 or better are:
Prob(EYESl) = -0.136 + 2.040(OZONE) R = 0.274
Prob(EYESl) = -0.137 +
Prob(THROATl) = 0.384 .'-
5.679(CO)
0.006(HUMID)
0.063
0.044
Prob(THROATl) = -0.264 + 0.004(TEMP) R = 0.039
:Prob(EYES2) = -0.108 + 1.882(OZONE) R2 = 0.207
Prob(OTHER2) = 0.758 - 0:008(TEMP) R2 = 0.038
-188-
-------
None of the other discomfort symptoms showed any association
with the air pollution variables at a = 0.10 or better.
The estimated linear probability functions with eye discom-
fort, both before and after running with OZONE as the explana-
tory variable, have high R.2 values for regressions with a
dichotomous dependent variable. OZONE explains 27 percent of
the variance in EYES1. OZONE also explains almost 21 percent of
the variance in EYES2. CO explains 6 percent of the variance in
EYES1. The only other discomfort symptoms which show signifi-
cant relationships in the linear probability specification are
THROAT1 with. HUMID, THROATl with TEMP, and OTHER2 with TEMP.
OTHER2 is inversely related to TEMP with almost 4 percent of the
variance explained.
PAIRED DIFFERENCE TEST OF MEANS OF PULMONARY FUNCTION VARIABLES
An analysis of paired differences was possible for the
athlete panel, where the same individuals were measured before
and after the planned exercise of running a distance of at least
two miles. Both FVC and FEV]_ _ Q we^e measured before and after
running for each athlete. The purpose of pairing was to reduce
extraneous influences on the variable (either FVC or FEy )
being measured. Pairing reduced subject-to -subject variability.
The paired, difference variable was formed by computing the
difference between mean BFVC and AFVC scores. Also, the paired
difference between mean BFEV and AFEV scores was computed. The
null hypothesis in each case was that the mean of the difference
is equal to zero, Ho : yj = 0 . The alternative hypothesis was
that H^: yd >0; i.e., FVC and FEV]_ Q scores were expected to
decline as a result of running. If the null hypothesis is re-
jected, then the data supports the alternative hypothesis of a
decline in the measured variable. Since the alternative hypoth-
esis was directional, a one- tailed test was appropriate.
The results of the paired difference test for FVC and
FEVi o f°r tne athletes are reported in Table 86. Neither of
the t-values are large enough to reject the null hypothesis at a
level of significance of a = 0.05. The implication is that no
significant decrease in either FVC or FEV^Q occurred as a re-
sult of running.
OVERALL SUMMARY OF THE ATHLETE PANEL DATA ANALYSIS
Very few significant results were obtained from the athlete
panel. One explanation for this outcome is that the ambient
concentrations of air pollution on the 11 days when the panel was
tested were not high enough to cause measurable biochemical and
-189-
-------
TABLE 86. ' PAIRED DIFFERENCEJ'-!TEST OF MEAN FVC AND MEAN FEV
(BOTH IN LITERS x 100) BEORE AND AFTER-RUNNING
Standard (Difference)
Variable Mean Deviation Mean
i
VO
O
1
BFVC
AFVC
BFEV
AFEV
Note:
492.3690
489.2500
414.5000
410.7381
Degrees of freedom
64.538
62.919 3'1190
59.275
3.7619
57.042
= 83
Standard t 1-Tail
Deviation Value Probability
25.723 1.11 0.135
22.490 . 1.53 0.065
-------
physiologic responses in the panel members. An equally plaus-
ible explanation, however, is that the data collection was
centered on the wrong variables and/or was conducted under the
wrong circumstances.
It was probably a mistake to collect data during practice
instead of competition and to depend entirely on health data in-
stead of both health and performance data to assess the impact of
air pollution. The results of a study conducted in the Los
Angeles community of San Marino in the early to mid-1960's dem-
onstrated the importance of including performance variables.*
The study involved high school long distance runners. Relation-
ships were found between concentrations of photochemical air
pollution one hour before races and the performance of the run-
ners (i.e., group running times) clocked during the races.
In the present study, there was no assurance that the mem-
bers of the athlete panel achieved maximal exercise on any of
the testing days. Maximal exercise could be assumed if the
members of the panel had been running in competition against
teams from other schools. It cannot be assumed under non-
competitive, practice conditions. In addition, the runners'
times were not consistently recorded, so an attempt to relate
concentrations of air pollution to performance was not possible.
A few results were obtained which showed relationships be-
tween discomfort symptoms and the aerometric variables. Corre-
lation analysis indicated that there was a direct relationship
between eye discomfort, both before and after running, and
OZONE. That relationship was upheld by linear probability es-
timates. The linear probability functions for both EYES1 and
EYES2 with OZONE, were highly significant with coefficients of
determination above 20 percent. Correlation analysis also in-
dicated a direct relationship between eye discomfort before run-
ning and CO. Again, the relationship was upheld by the linear
probability estimate. None of the other variables (discomfort
symptoms, maximum FEVi Q, or maximum FVC) measured in connection
with the athlete panel'showed any significant relationship with
respect to the air pollution variables. Throat discomfort be-
fore running showed a statistically significant relationship
with weather variables HUMID and TEMP, and other discomfort re-
ported after running was estimated as being significantly in-
fluenced by TEMP.
*W.S. Wayne, P.F. Wehrle, and R.E. Carroll, "Oxidant Air
Pollution and Athlete Performance," Journal of the American
Medical Association, 199(12), March 20, 1967, 901-904.
-191-
-------
APPENDIX A
INTERVIEWER INSTRUCTIONS
-192-
-------
INTERVIEWER INSTRUCTIONS
BACKGROUND
The Four Panel Study is funded by the EPA and is similar in
some ways to previous studies we have conducted for the EPA and
very different in others. The purpose of the study is to mea-
sure how the environment affects health on days when pollution
is unusually high. The study is not designed to evaluate the
commulative effect of several years of exposure.
Four panels will be used in the study: athletes, asthmat-
ics , bronchitics, and healthy outdoor workers. CIC is managing
the study and will be assisted by the Lung Association of Los
Angeles County, the University, of California, Riverside, and
Rockwell International. Your task as interviewers is to recruit
people for the Asthma and Bronchitis panels. These interviews
for the Four Panel Study are designed to determine whether the
person qualifies for the panel (medical condition, age, etc.)
and whether he is willing and able to participate. All other
information about the panelist will be obtained in interviews
conducted by the Lung Association during the test periods.
The Four Panel Study differs from previous studies in
another way because it is a pilot study for testing new tech-
niques for gathering information about health effects. This
means that CIC is making out the schedule, designing the ques-
tionnaires, and writing instructions for interviewing and test-
ing. We will be the first group to do these panel studies, and
we may find that some instructions and procedures will not work
and will have to be changed. YOU SHOULD NOT MAKE CHANGES ON
YOUR OWN, BUT SHOULD TELL YOUR TEAM LEADER ABOUT PROBLEMS YOU
ENCOUNTER.
The study will be conducted in portions of Azusa, Covina,
and Glendora. A street map of the study area showing exactly
where we will be working will be given to you. We have tried to
choose an area that will have essentially the same environment;
this means that we do not go up into the hills, as the environ-
ment changes there.
We will be recruiting asthmatics and bronchitics at the
same time, so you will have to be familiar with two studies in-
stead of one. You will administer a short, simple questionnaire
-193-
-------
to those who have asthma or bronchitis for the purpose of allow-
ing us to judge whether they are truly qualified. At present,
we are assuming that we may have to ring every doorbell in the
entire study area to recruit enough asthmatics. You must try to
do such a good job of selling the study that every qualified
panelist will want to participate.
INTERVIEW ASSIGNMENTS
Each day your team leader will give you a set of Inter-
viewer Record sheets like the one attached to these instruc-
tions. Only the street name and city will be filled in; you
will fill in house numbers as you make the interviews. A record
should be made even if you ring the bell and no one answers.
An example of a completed Interviewer Record sheet is also
attached.
At the end of the day, check over your Interviewer Record
sheets to make sure that they are legible. Mark in some conspic-
uous way all cases where you have promised callbacks. Make
sure your team leader sees these. Give all Interviewer Records
to your team leader every day.
INITIAL CONTACT
The purpose of this initial contact is to determine as ef-
ficiently as possible whether any adult member of the household
has chronic bronchitis or asthma. This is not a statistical
survey, so we do not need to worry about leading the respondent
and biasing the results. The following is suggested:
"Hello. I'm from Copley International
Corporation. We are trying to locate adults with
chronic bronchitis or asthma who would be interested
in helping with a study to see how the environment af-
fects their health. This study is being sponsored by
the federal government. Does anyone in this house-
hold qualify? Do you know of anyone else who might
be interested?"
If the respondent is a potential panelist, proceed as directed
in the PANEL INTERVIEW section of these instructions. If the
respondent wants more information to decide whether some other
member of the household qualifies, proceed as follows:
• If asked, "What is asthma?," explain that the main
criterion is that he has episodes of shortness of
breath and wheezing in the chest, and that a doctor
has diagnosed it as asthma.
-194-
-------
• If asked, "What is chronic bronchitis?," explain
that this is a persistent cough that raises phlegm
from the chest that may or may not have been called
bronchitis by a doctor.
• If asked, "What do you mean by adult?," explain that
asthmatics must be 30-45 years old, while bronchitics
must be 35-55 years old.
If a prospective panelist is not at home, try to make an ap-
pointment for an interview at some time when the prospective
panelist is available—preferably the following day.
PANEL INTERVIEW
The first step in conducting the interview is to determine
whether the respondent has asthma or bronchitis. Select the in-
terview questionnaire that correspondsto the respondent's con-
dition. Find out whether the respondent is the proper age for
the panel. IF HE IS NOT IN THE RIGHT AGE BRACKET, TERMINATE THE
INTERVIEW AND DO NOT FILL OUT THE QUESTIONNAIRE.
The Asthma Panel Questionnaire is relatively straightfor-
ward. The answer to Question 1 should always be "Yes" in this
survey. Som respondents may not understand Question 4a, "Short-
ness of breath" so you may want to substitute one of the follow-
ing:
o "Difficulty taking in enough breath"
• "Difficulty breathing out the air that you take in"
After you have completed the first five questions, stop and ex-
plain the study to the respondent. Use the Memo to Asthma
Panelist for reference, and emphasize that he will receive a
chest X-ray and other lung tests and will be notified if any
medical problems are discovered. Also emphasize that his par-
ticipation will only be for two weeks (ten weekdays) plus two
days..
After you have explained the study, fill out the bottom
portion of the questionnaire. DO NOT encourage the respondent
to specify a defininte time of day for his appointment, such as
3:15 or 4:30, but try for general times, such as "afternoon" or
"between 4.and 6 p.m.," etc. Under "Comments" write down any
additional information that relates to the availability for
testing or the mechanics of getting him to the test location.
If a respondent refuses to participate, write down his reason
for refusing.
-195-
-------
For every respondent who agrees to participate, fill out a
Memo to Asthma Panelist and leave it with him. DO NOT LEAVE
THIS MEMO UNLESS THE PERSON AGREES TO BE A PANELIST. Before you
leave, tell the respondent that there is a possibility that he
might not be selected for the study. Usually this is because he
hasn't had enough attacks or because his asthma began when he
was a child instead of an adult. The memo explains the time
schedule for notification of panelists. BEFORE YOU LEAVE, ASK
THE RESPONDENT IF HE KNOWS ANYONE IN THE AREA WHO MIGHT BE A
CANDIDATE FOR EITHER OF THE PANELS.
The Bronchitis Panel Questionnaire is also relatively
straightforward.The biggest problem will probably be with
Question 1 because chronic bronchitis is not always diagnosed as
such by a doctor. You may wish to ask Questions 2, 3, and 4 to
help the respondent make up his mind. If he answers "Yes" to at
least two of these questions, and if he has had the condition
for several years, he is eligible for the study. At this point,
stop and explain the study, using the Memo to Bronchitis
Panelist as a reference. Do your very best low-key selling job
here.Emphasize that he will receive a full chest X-ray and
electrocardiogram, and will be notified if any medical problems
are found. Be sure to tell him that a small blood sample will
be taken at 4 of the 11.examinations. Don't overemphasize this,
but you may want to assure him that the Breathmobile technicians
take pride in making the procedure painless. If you have to,
you can tell the respondent that we will not repeat any part of
the examination that causes him excessive discomfort or dis-
tress. You may also want to explain that the dates for the
examinations cannot be set in advance because they will depend
on the weather. You should also remind the women that they will
have to disrobe partially for the electrocardiogram. You will
be supplied with a memo containing suggestions for those who may
not want to disrobe.
After you have explained the study, fill out the rest of
the quesionnarie which deals with scheduling and transportation.
This is self-explanatory. If the respondent refuses to partici-
pate, write down the reason.
For every respondent who agrees to participate, fill out a
Memo to Bronchitis Panelist and leave it with him. DO NOT LEAVE
THIS MEMO UNLESS THE PERSON AGREES TO BE A PANELIST. Before you
leave, tell the respondent that there is a possibility that he
might not be selected for the study. This could happen because
we are trying to recruit a panel with equal numbers of men and
women spread uniformly over the study area. The memo explains
the time schedule for notification of panelists. BEFORE YOU
LEAVE, ASK THE RESPONDENT IF HE KNOWS ANYONE IN THE AREA WHO
MIGHT BE A CANDIDATE FOR EITHER OF THE PANELS.
-196-
-------
MISCELLANEOUS
If you should interview an individual who seems genuinely
interested in being a panelist but wants to discuss it with his
doctor first, try to accommodate him. If you think if appropri-
ate, you may tell the individual that his doctor may call
Dr. Stanley Rokaw at the Lung Association of Los Angeles County,
telephone 213/483-3220, for more information. DR. ROKAW WILL
NOT TALK TO INDIVIDUAL RESPONDENTS--ONLY TO THEIR DOCTORS.
The first time an individual visits the Breathmobile, he
will be given additional information about the tests and will
have a chance to ask questions. When he is satisfied that he
understands the procedures, he will be asked to sign a consent
form. He can, of course, refuse to consent even at this late
date, but we hope this won't happen.
-197-
-------
INTERVIEWER RECORD
(street)
(city)
(interviewer)
House
Number
Date/Time
#1
V
•
#2
Contact?
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No . Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
Eligible
Panelist?
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
Asth - Bron -
matic chitic
A B
A B
A B
A B
A B
.A B
A B
A B
A B
A B
A B
A B
A B
A B
A B
Comments
(quest, completed, callback, refused, etc.)
-198-
-------
INTERVIEW Ell R ECORD
A
(street)
"(city)
(inter viewer)
House
Number
3rt
322.
Aft- 1
3Z2-
Apt ^
322
Afl'3
312
Date/Time"*''
#1
||5P
1 \*-°
ii
,130
1 1 *""
II
1 1 —
II
nt§'
#?.•*#
I
Cong^
/"^
No <£es-
N!o) Yes
No^Yes
No ($e£.
No ([Yes,
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yet:
No Yes
Eligible
Panelist?
J^Q) Yes
No Yes
No Yes
iNo ^ejy
ira^be.
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
No Yes
Asth- Bron-
imitic cliitic
A B
A B
A B
A (£)
(h) B
.A B
A B
A B
A B
A B
A B
A' B
A B
A S
A 'U
Comments
(quest, completed, callback, refusei:!, etc.)
^Y'^VMJ. VOO/K — nOme. on'Tci'^ —
(0 c.ompeTcJt - <9GGepT ed
VWiioanA, voork^-CaU 2-T9-1B2.
•^OY" e^finio^ 'i^et'v'lCvO
"!%^ms1'eil^arec^-
•$:$: "~rt\\s co\tviv\rl ^o "be u.Spcl '(V
We -iv^ -4o re contact -fKc^>c
V-^o voeyen-t hono£..
•someone, \olno tsvioiiX, ah>occt
-199-
-------
ASTHMA PANEL QUESTIONNAIRE
Start Time
Mr.
Mrs.
Miss (last)
Address
(number)
(first)
(street)
(city)
Telephone
(home) (work)
1. Have you ever had ASTHMA?
(IF "Yes" TO 1, ASK:) 'Has a doctor told you this?
(IF "Yes" TO 1, ASK:) Are you now taking medicine for this?
2. How old were you when your asthma first began?
3. About how many asthma attacks have you had within the past year?
4. When you have an asthma attack, do you usually have:
a. Shortness of breath?
b. Wheezing in the chest?
c. Fever?
d. Increased sputum or phlegm condition?
5. What provokes your asthma attacks?
Dusts or pollens [ J Emotions | [
Infection | | Don't know [ I
6. Do you smoke cigarettes?
Finish Time
Occupation
Work Location
Birthdate
Yes Q No | I
™\
Yes| 1
Yes|~|
YesQ
Yesf~l
YesH]
Noj |
NO r~i
NoQ
Nod
No| |
NA) 1
years
attacks
Yes f~~| No|~~|
Subject agrees to participate:
Times NOT available for testing: Sept. 9-20 f""|
Sept. 23-Oct. 4 [~~|
Preferred time of day for appointment:
Preferred transportation: Subject's car [~~~|
Comments:
Yes | | No[~~l
Oct. 7-18 ["""I
Oct. 21-Nov. 1 | |
Our car
-200-
-------
MEMORANDUM
TO: Volunteer for Asthma Panel
FROM: Copley International Corporation
12511 Brooklmrst Street, Garden Grove, California 92640
Telephone: 714/539-7751
SUBJECT: Study Plans, 1974
Sometime before October 9, 1974, you will be notified (by mail or telephone) whether
you have been chosen to be a member of the Asthma Panel. As a panelist you have
volunteered to do the following:
1. Report to a test location in Covina (address to be supplied later) at an
appointed lime on ten consecutive weekdays. Your appointment will be
at the same time each day, and transportation will be supplied, if
necessary.
2. Each day, blow into a machine that will measure how big a breath you
can take. Answer a few questions about the kinds of discomfort you
feel. The test will take about five minutes and will be administered by
Copley International Corporation personnel.
3. On two days, one before and one after your two-week test period, go
to the Breathmobile and blow into several machines that will do a more
thorough job of measuring your lung function. These tests will take
about 10 minutes. The Breathmobile will be located in the Covina
area. You will be contacted by telephone about specific times
for these appointments.
4. On one of these visits to the Breathmobile, complete a clinical inter-
view questionnaire that includes questions about your present symptoms,
past health, smoking habits, and exposures to dust or fumes. Have a
full chest X-ray taken during this visit.
You and your doctor will be notified if the tests indicate that you have any abnormal
condition that might require treatment. The effects of the environment on your condition
will not be known until we have studied many people. When the results are published,
we will try to see that you receive a copy. • . • . .
Thank you for your help.
You have indicated that you will not be available for testing during:
Sept. 9-20 [~~] Oct. 7-18
Sept. 23-Oct. 4 \~\ Oct..21-Nov. 1
You have indicated as a preferred
time of day for your appointment.
-201-
-------
BRONCHITIS PANEL QUESTIONNAIRE
Mr.
Mrs._
Miss
(last)
(first)
Address
(number)
(street)
Start Time
Finish Time
Occupation
.(city)
Telephone
(home) (work)
1 . Have you ever had CHRONIC BRONCHITIS?
(IF "Yes" TO 1, ASK:) Has a doctor told you that you NOW
have this?
(IF "Yes" TO 1, ASK:) Are you now taking medicine for this?
2 . Do you cough first thing in the morning (when you get up) on more
than 50 days in a year?
(IF "Yes" TO 2, ASK:) How many years have you coughed like
tills?
5. Do you smoke cigarettes?
Work Location
Birthdate
Yes [~~) No [""I
Yes [~] No[~~| NA [""]
Yes ~1 No | | NA fH
Yes |~~) No
3. Do you bring up any phlegm from your chest first tiling in the morning
(when you get up) on more tlian 50 days in a year? Yes
(IF "Yes" TO 3, ASK:) How many years have you brought up
phlegm like this?
4. Do you bring up any phlegm from your chest later in the day on more
than 50 days in a year? Yes
(IF "Yes" TO 4, ASK:) How many years have you brought up
phlegm like this?
Yes
years
No
years
No | |
years
No | |
Subject agrees to participate:
Preferred transportation:
Subject's car |~~|
Yes No
Our car \ [
Page 1
over.../
-202-
-------
Times available for testing:
M
Tu
W
Tli
F
12 1
5678
Is there any period during September and October when subject
expects to be away?
Comments:
Page 2
-203-
-------
MEMORANDUM
TO: Volunteer for Bronchitis Panel
FROM: Copley International Corporation
12511 Brookhurst Street, Garden Grove, California 92640
Telephone: 714/539-7751
SUBJECT: Study Plans, 1974
Sometime before September 6, 1974, you will be notified (by mail or telephone) whether
you have been chosen to be a. member of the Bronchitis Panel. As a panelist, you have
volunteered to do the following:
1. On eleven (11) occasions between September 3 and November 8, 1974,
receive a telephone call notifying you to keep an appointment at the
Breathmobile the following day. All appointments will be scheduled
*• oh weekdays between 12:00 noon and 7:50 p.m.
2. Present yourself at the Breathmobile at the appointed time (or be ready
for our vehicle to pick you up if we have agreed to do so). The Breath-
mobile will be located in the Covina area and you will be told of its
exact location later.
3. Allow the Breathmobile staff to perform a series of tests that will take
about 20 minutes and will require you to (1) blow into several machines
to see how well your lungs function; (2) have your pulse, blood pressure,
and an electrocardiogram taken; (3) have a swab taken of material from
your noseband (4) have a small blood sample taken on 4 of the 11 visits.
4. On one of your eleven visits to the Brcathmobile, complete a clinical
interview questionnaire that includes questions about your present
symptoms, past health, smoking habits, and exposures to dust or
fumes. Have a full chest X-ray taken during this visit.
. You and your doctor will be notified if the tests indicate that you have any abnormal
condition that might require treatment. The effects of the environment on your condi-
tion will not be known until we have studied many people. When die results are
published, we will try to see that you receive a copy.
Thank you for your help.
You have indicated that you are available for testing at the following times:
-204-
-------
REMINDER TO FEMALE BRONCHITIS PANELISTS
In preparation for the electrocardiogram (heart test) at
the Breathmobile, you will be asked to partially disrobe so that
your chest and legs are bare. You will be covered by a sheet or
gown during the test. If you would be offended or embarrassed
by disrobing in the presence of the Breathmobile's medical
personnel, we suggest that you wear a two-piece bathing suit
which you can leave on for the test. We also suggest that you
do not wear panty hose; if you do, you must remove them for the
test.
-205-
-------
APPENDIX B
INSTRUCTIONS TO PANELISTS
-206-
-------
B (HI f]
*aaauu*x**xtsa
COPLEY INTERNATIONAL CORPORATION
Economic Research • Corporate Planning • Systems Engineering • Management Services
August 28, 1974
Dear Panelist:
Thank you for agreeing to participate in our study of bronchitics in the Covina-
Azusa area. You have been chosen as a panelist on one of our upcoming panels
and we appreciate the time and effort which will be required on your part to help
us with this important research project.
On September 3, you will receive a telephone call telling you of your first appoint-
ment to be scheduled for September 4 or 5. Transportation arrangements, if
necessary, will also be made at that time. Later, as our interviewer told you,
you will be called and asked to come in for tests on 10 separate occasions. In
each case the testing will be done on a weekday and you will be called a day in
advance. All testing will be done at the Breathmobile which will be located at
Dalton Community Park (see enclosed map). Your first appointment at the
Breathmobile will take about 20 minutes, with the subsequent Visits taking approx-
imately 10 minutes each.
Please save this letter and post it as a reminder to you of your appointments.
After your telephone call on September 3, write your appointment times and dates
on the appropriate line below. You will keep the same time for all of your visits
unless other arrangements have been made. It is important that you be prompt,
as we will be testing many people and do not wish to keep anyone waiting. If one
of our drivers will be picking you up, please be ready 15 minutes before your
scheduled appointment time.
Once again thank you for your cooperation. We look forward to working with you
very soon.
Sincerely,
Katherine W. Wilson
Project Director
Initial Breathmobile appointment:
1974, at
a.m./p.m.
-207-
-------
BREATHMOBILE LOCATION
Footliill Freeway
Citrus Edge Street
Armsteacl Street
O
Cl)
g
(B
Gladstone Street
Dalton
Community
Park
Breathmobile located at:
Dalton Community Park
(formerly Barranca Park)
18867 E. Armsteacl
Azusa, California
N
-E
-208-
-------
APPENDIX C
ELECTROCARDIOGRAPH DATA FORM
-209-
-------
ELECTROCARDIOGRAPH ANALYSIS
LUNG ASSOCIATION
'OF
LOS ANGELES COUNTY
1 . D. //
Name
Date Time
Rate: Atrlal
Ventricular
Mechanism
Axis Deviation
P-R Interval
P -Waves Deflections • •
QRS Complexes
T-Waves
S-T Segments
Others
,
Conclusions:
•
_.
INTERPRETATION BY:
Time
-210-
-------
APPENDIX D
ELECTROCARDIOGRAPH VARIABLES
-211-
-------
• CARD 37 - ECG ' . ...
COLUMN(S) DESCRIPTION
5-6 "37" (Card Number)
7-10 1.0. Number
11-16 . • Date
17-20 Time
21 (Before Exercise 1
•••'•( ...-•'
(After Exercise 2
22-2^ ' Rate Atrial •
25-27 ' Rate Ventricular • •
28-29 •'-..•• Mechanism (Rhythm)
„.'..' ... 01 Sinus Rhythm (normal) . .'.'...
. ''•:.' 02 Supraventricular Tachycardia
03(Atrial Premature Contractions
• "'-. ' • ' ( •'••.:•••'•
. . . . . • .• (Nodal Premature Contractions
. . . 04 Atrial Flutter
• 05 Atrial Fibrillation
. '• .': . . . 06 Ventricular Premature Contractions
'-.... * 07 Ventricular Tachycardia
" "'• . .. 08 Idioyentricular Rhythm
: : . . 09 Sinus Arrhythmias
10 Nodal Pacemaker
11 Coronary Sinus . . .
30-33 . Axis Deviation t XXX °
3i»-35 P-R It.terval
P-Waves Deflections
Codes; l=Nor'mal, 2=Bifid, 3=Peaked,
'l=0ecrcascd, 5~lnverted, 6=Diphasic,
7=Widcned
-212-
-------
CARD 37 - ECG (Cont.)
COLUHN(S)
36 . -'
37
38
39 •
kO
DESCRIPTION
.IA .
Leads
1
'.2/
-3
AVR
AVL
AVF
V]
.V2
_
V3
V|»
50 .
51
52
53
5*»
55
V6 . .
QRS Duration
QRS Deviation
Codes ; l=Progress ion Normal, 2=rR ,
3=R/S rat.io^l, /4=Low Vo 1 ta ge ,
Q noted, 6=lnverted
Progression Delayed
I
2
3
AVR
AVL
AVF
-213-
-------
CARD 37 - EGG (Cont.)
COLUMN(S)
.56
57 .
58
59
60
61 '
62-63 . . :
DESCRIPTION
65
66
67.
68
69
70
71
72 .
73
7**
75
76
Leads (cont.)
V,
T-Wave Duration
T-Wave Position
Codes; l=Upright, 2=lsoel ectric, flat-low
amp] i tude, -3=lnverted
S-T Segment
Codes; l=Normal, Isoelectric, 2=Depressed,
3=Biphasic, A=Coved, 5=Elevated
Leads; . '.'"'•'.
1
2 .... ' .
3 • . ''•• • • • • '• • '•
AVR
AVL
AVF
VJ
V2 -
V6
•214-
-------
CARD 37 - ECG (Cent.)
COLUMM(S)
77-78
.DESCRIPTION
Conclusion
Codes; l=Norma]
2=Enlargement Atrial RT
3-Enlargement Atrial LT . '
4=Enlargement Ventricle RT
5=Enlargement Ventricle LT
6=Infa retion• • .
7=Injury
' 8=Dru& Effect
9=Pulmonary Disease
10=Non-Specific
ll=Rhythm Disturbance .
12=Premature Junctional Systole
l3=Low voltage to the limb leads
1^=BiVentricular Hypertrophy
15=Improper Lead Position
I6=lnferior Subendocardial Ischemia
-215-
-------
APPENDIX E
BLOOD SAMPLE FORM
-216-
-------
HEMATOLOGY
HEMATOLOGY REOUEST.£r°
DATE
[3] CBC H ADM O Hgb C Hct Q RBC Q WBC D DIFF.
CURRENT
DIAGNOSIS:
DATE
TIME SPEC OBTAINED
IF 999 RE 'DILUTE
.
.
.
.
.
.
TEST
WBC
X101
RBC
X10"
Hgb
gm
Hct
MCV
MCH
H/1.9
MCHC
NORMALS
M 'F
4.8 4.8
10.8 10.8
4.6 4.2
6.2 5.4
14 12
IS 16
42 37
52 47
82 TO
99
27TO
32
32TO
36
REMARKS
o
I
»
t*
DIFFERENTIAL:
NEUTROPHILES %
SEGMENTED %
BAND FORMS %
METAMYELOCYTES %
MYELOCYTES %
EOSINOPHILES %
BASOPHILES ' %
LYMPHOCYTES %
MONOCYTES %
REMARKS :
<;T ppAMri1; WD^PITAI
PLATELETS: •
NORMAL
INCREASED
DECREASED
RBC
NORMAL
ANISOCYTOSIS
HYPOCHROMIA
POIKILOCYTOSIS
POLYCHROMASIA
R. F. HUFNER. M D
LYNWOOD. CALIFORNIA
J N GAP8EKRY. M D
DIRtOORS Of LABORATORY
-217-
-------
APPENDIX F
DAILY SYMPTOM RECORD
-218-
-------
DAILY SYMPTOM RECORD
Identification No.
Mr.
Mrs.
Miss Last First
DATE OF TEST:
Mo. L
Hr. |
Day
Min.
ALL PANELISTS
1. Do you have any discomfort now?
2.
3.
4.
5.
6.
7.
Eyes?
Throat?
Chest?
Headache?
Nausea?
Other?
Have you had a headache earlier today?
Have you felt any unusual shortness of breath today?
Have you been coughing at all today?
4a. IF "Yes" TO 4, ASK: Has your coughing brought up any
phlegm from your chest?
Did you smoke cigarettes today?
5a. IF "Yes" TO 5, ASK: How many?
Do you have a bad cold today?
Have you taken any medicine today?
Yes
Yes Q]
YesD
Yes Qj
Yes Q
Yes Q
Yes Q]
Yes Q]
Yes Q]
Yes Q
Yes G
No
N°D
No Q]
No [~~|
No Q
N°D
No Q
NoQ-
NO [~~|NA| |
No [~~[
cigarettes
No Q
ASTHMA AND OUTDOOR WORKER PANELS
Temp.
FEV
1.0
FEV
1.0
FEV
1.0
MAX. FEV1>0
-219-
-------
ATHLETE PANEL - (After their exertion)
Time Trial: .Distance
Time of Day: Race ended
8. Do you have any discomfort now?
9. Have you coughed since the race?
phlegm from your chest?
Time
minutes, seconds
Final tests start
hour, minute
3W?
Eyes?
Throat?
Chest?
Headache?
Nausea?
Other?
ice?
d your coughing bring up any
hour,
Yes Q
Yes Q
Yes Q
Yes Q
Yes Q
Yes n
Yes ["I
minute
No £
No £
No [|
No Q
No £
No {[
No T
]
]
:
3
D
\
1 r-.
-220-
-------
APPENDIX G
CLINICAL INTERVIEW QUESTIONNAIRE
-221-
-------
Mr.
Mrs.
Miss
Birthdate
Age
Sex
Height
CLINICAL INTERVIEW QUESTIONNAIRE
Month Day Year
| | M
ft. inches
Interviewer No.
I.D. No.
Date of Test
Form Approved
OMB No. 158R0120
. Month Day Year
PREAMBLE: I am going to ask you some questions about your chest. Please
answer "Yes" or "No" whenever possible.
1. Do you usually cough first thing in the morning in winter?
(COUNT TWO OR MORE COUGHS UPON ARISING, OR WHEN
SUBJECT FIRST GOES OUT OF DOORS, OR WHEN SUBJECT
SMOKES THE FIRST CIGARETTE OF THE DAY IF HE/SHE
IS A SMOKER. DO NOT COUNT CLEARING OF THROAT.)
Yes No
CD D
(IF "Yes" TO QUESTION 1, PLEASE ASK QUESTION 2.)
2. Do you cough like this on most days (or nights) for as much as three Yes No NA
months each year? [ | | | | j
3. Do you usually cough during the day or night in winter? (DO NOT Yes No
COUNT AN OCCASIONAL COUGH.) [ | | )
(IF "Yes" TO QUESTION 3, PLEASE ASK QUESTION 4.)
4. Do you cough like this on most days (or nights) for as much as three
months each year?
Yes No NA
Dan
Page 1 .
-222-
-------
I.D.No.
5. Do you usually bring up phlegm (thick fluid) from your chest first thing
In the morning in winter? (COUNT PHLEGM WHETHER SWALLOWED
OR EXPELLED, UPON ARISING, OR WHEN SUBJECT FIRST GOES OUT
OF DOORS, OR WHEN SUBJECT SMOKES THE FIRST CIGARETTE OF
THE DAY IF HE/SHE IS A SMOKER. DO NOT COUNT PHLEGM FROM Yes No
THE NOSE.) | | ("I
(IF "Yes" TO QUESTION 5, PLEASE ASK QUESTION 6.)
6. Do you bring up phlegm like this on most days (or nights) for as much Yes No NA
as three months each year? | [ [ [ | |
7. Do you usually bring up phlegm from your chest during the day or Yes No
night in winter? ' | | | |
(IF "Yes" TO QUESTION 7, PLEASE ASK QUESTION'S.)
8. Do you bring up phlegm like, this on most days or nights for as much Yes No NA
as three months each year? | | | | | |
9. In the past three years, have you had a period of cough and phlegm
lasting for three weeks or more? (ALL PERSONS SHOULD ANSWER
THIS QUESTION. IF SUBJECT USUALLY HAS COUGH OR PHLEGM,
THE QUESTION REFERS TO PERIODS OF MORE THAN USUAL
COUGH OR PHLEGM.) No | |
Yes, 1 period | |
Yes, 2 or more periods | |
10. Are you troubled by shortness of breath when hurrying on level ground Yes No
or walking up a slight hill? . [ j | |
11. Do you get short of breath walking with other people of your own age Yes No
on level ground? , | | | |
Page 2
-223-
-------
I.D. No.
12. Does the weather affect your chest or your breathing? If yes,
specify how and what type of weather, e.g., fog, damp, heat, cold. No | |
Yes, chest | |
Yes. breathing [ [
Yes No
13. Do you usually have a stuffy nose or runny nose in the winter? | J [ |
Yes No
14. Do you have this in summer? [ | | ]
15. During the past three years, have you had any chest illness which has Yes No
kept you from your usual activities for as much as a week? | | | |
(IF "No" TO QUESTION 15, ASK QUESTION 18.)
(IF "Yes" TO QUESTION 15, ASK QUESTION 16.)
Yes No NA
16. Did you bring up more phlegm than usual in any of these illnesses? | ^ | | [ |
(IF "No" TO QUESTION 15, ASK QUESTION 18.)
(IF "Yes" TO QUESTION 15, ASK QUESTION 17.)
17. How many illnesses like this have you had in the past three years? 1 illness | |
2 or more illnesses [ |
NA
Present and Past Illnesses
18. Do you now have any serious illness? If yes, specify. Yes No
n n
Yes No
19. Has a doctor ever told you that you have emphysema? | | [ j
\
(IF "Yes" TO QUESTION 19, ASK QUESTION 20.)
9 Yes No NA
20. Are you now taking medicine for this? ' | | [ | ( [
Page 3
-224-
-------
I.D. No.
21. Have you ever had chronic bronchitis? Yes No
—; an
Page 4
-225-
\
(IF "Yes" TO QUESTION 21, ASK QUESTION 22.)
Yes No NA
22. Has a doctor told you that you now have this? | | | | | |
(IF "Yes" TO QUESTION 21, ASK QUESTION 23.) .
' Yes No NA
23. Are you now taking medicine for this? • | | | | | |
Yes No
24. Have you ever had asthma diagnosed by a doctor? | | | |
\
(IF "Yes" TO QUESTION 24, ASK QUESTION 25.)
Yes No NA
25. Has your asthma been active in die past two years? | | | | | [
Yes No
26. Did you ever have hay fever? | | [ |
\
(IF "Yes" TO QUESTION 26, ASK QUESTION 27.)
Yes No NA
27. Has your hay fever been active in the past two years? ) | | | | |
Yes No
28. Did you ever have chronic sinusitis? | | | |
\
(IF "Yes" TO QUESTION 28, ASK QUESTION 29.)
Yes No NA
29. Has your chronic sinusitis been active during the past two years? ' | | | | [ |
Yes No
30. Did you ever have any allergies? [ | | |
\
(IF "Yes" TO QUESTION 30, ASK QUESTION 31.).
Yes No NA
31. Has your allergy been active during the past two years? | [ [ | | |
-------
I.D. No.
32. Have you ever had treatment for tuberculosis or any other chronic Yes No TB
lung condition? 1 | | | [ |
(IF "Yes," NOTE CONDITION:)^
Smoking Habits
33. Have you ever smoked as many as five packs of cigarettes, that is, Yes No
as many as 100 cigarettes, during your entire life? • | ~| [ [
Yes No
34. Do you now smoke cigarettes? | | [ |
35. If you are a current or an ex-cigarette smoker, how many cigarettes
do (did) you smoke per day?
Less'than 1/2 pack per day (1-5 cigarettes per day) | |
About 1/2 pack per day (6-14 cigarettes per day) | |
About 1 pack per day (15-25 cigarettes per day) [ j
About 1-1/2 packs per day (26-35 cigarettes per day) | |
About 2 packs per day (35 or more cigarettes per day) | |
36. If you are a current or an ex-cigarette smoker, how old were you when _____
you first started smoking? . | | | Years
37. If you are an ex-cigarette smoker, how old were you when you last gave ^_^_^
up smoking? | | | Years
' Yes No Exsm.
38. Do you smoke a pipe? | | | ) | |
Date stopped:
If yes, how many pipefuls per day?
Yes No Exsm.
39. Do you smoke cigars? • | | | | | |
Date stopped:
If yes, how many per day?
Page 5
-226-
-------
I.D. No.
General Questions
40. At your job are you now or have you been frequently exposed to Yes No
irritating smoke, dust, or fumes? | | | |
(IF "Yes" TO QUESTION 40, ASK QUESTIONS 41, 42, AND 43.)
41. What kind of irritant were you exposed to? (For example: coal dust,
cutting oils, asbestos, mine dust, smelter fumes, raw cotton dust,
foundry dust) Specify:
42. How long were you exposed?
Less than 1 year | |
1 to 5 years [ |
6 to 10 years | |
More than 10 years | |
43. Have you been exposed to irritating smoke, dust, or fumes at your job Yes No NA
during the past year? • | | [ | | | ^—
44. Where were you born?
City State Country
45. How long have you lived at your present address? | | | Years
46. How long have you lived in the Los Angeles area? • | | | Years
47. Have you ever changed occupations because of a breathing (lung) Yes No
problem? ' |~~1 ~
48. Have you ever changed residence location because of a breathing Yes No
(lung) problem? | | ~
Page 6
-227-
-------
I.D. No.
49. Have any of your "blood" relatives ever had persistent asthma,
bronchitis, or emphysema?
50. How many rooms are there in your living quarters? (Do not count
bathrooms, porches, balconies, foyers, halls, or halfrooms.)
51. How many people live in your household?
52. What kind of stove is used for cooking in your home?
Yes No
n
| | | Rooms
| || People
Gas
Electric | |
Other
53. What educational level have you and/or the head of your household
completed?
Head of Household
Respondent (if applicable)
Elementary school
Part of high school
High school graduate
Part of college
'College graduate
Graduate school, including advanced and professional
degrees
Trade, technical, or business school beyond high school
54. What is your current marital status?
Single . | | Separated car divorced f_ |
Married ( | Widow or widower | |
CD
n
n .
n
n
a
a
a
. a
o
a
a
a
a
Other (specify) | ]
Page?
-228-
-------
I.D. No.
55. Are you now pregnant? (We cannot X-ray pregnant women on the Yes No NA
Breathmobile.) | | | II I
(INTERVIEWER: GIVE YOUR BEST-ESTIMATE OF THE SUBJECT'S
RACE/ETHNIC GROUP, BUT DO NOT QUESTION HIM DIRECTLY.)
American Indian | |
Black/Afro American | |
Spanish/Mexican/Puerto Rican-American | |
Oriental-American " | |
White/Caucasian-American | |
Other | |
(SPECIFY, INCLUDING MIXED)
Thank you for your cooperation.
Page 8
-229-
-------
APPENDIX H
AEROMETRIC DATA BASE
-230-
-------
AEROMETRIC DATA BASE
The original study protocol required that aerometric data
from the EPA air monitoring stations at Glendora (0841) and
Covina (0842) be used in the analysis of the health information.
During the study period, many of these data were missing be-
cause of various operational difficulties. This section of the
report describes the content of the EPA data base and justifies
the steps that were taken to fill gaps in the base.
OXIDANT
If oxidant values are tabulated for all study days for
which data are available from both EPA air monitoring stations
0841 and 0842 (with no more than four hourly averages missing),
it is immediately apparent (see Table H-l) that the two stations
did not always track each other. On some days, for example 257,
278, 283, and 288, the agreement was excellent. On many days
(303 through 311) the oxidant values were so low that they did
not constitute a good test for agreement between the two sta-
tions. On still other days (253, 255, 259, 273, 274, and 275)
the recorded oxidant values were different at the two stations
with lower values occurring at 0842.
It is helpful to know whether these differences represent
genuine area differences in oxidant levels or merely a local
distrubance or instrument malfunction at one of the monitoring
sites. A comparison of EPA air monitoring station readings with
those from nearby monitoring stations operated by the California
Air Resources Board (ARE) at Temple City and the Los Angeles
County Air Pollution Control District (APCD) at Azusa, help to
clarify the situation. Data from all four stations are summar-
ized in Table H-2 and relative locations of the stations are
shown in Figure H-l. The Azusa and Glendora (0841) stations are
both located in the foothills about four miles apart and, as
expected, tend to give similar oxidant values. Readings at
Glendora are slightly higher than those at Azusa. Readings at
Covina (0842) tend to be considerably different from those at
Azusa and Glendora, and on days 253, 255, and 259 the Temple
City and Covina readings are similar. However, on days 273,
274, and 275 the oxidant levels at Covina were lower than those
at any of the other stations which suggests a local disturbance
or instrument malfunction. In summary, there appear to be
-231-
-------
^s
"%•
PDTX^
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
TABLE H-l. OXIDANT VALUES (PPM) FOR EPA AIR MONITORING STATIONS 0841 AND 0842
253
41* 42**
.00
.02
.07 .03
.12 .10
.19 .21
.26 .31
.36 .37
.40 .26
.35 .22
.31 .26
.26 .21
.14 .09
.07
255
41* 42**
.01
'.02
.04 .03
.07 .06
.13 .09
.15
.21 .18
.22 .15
.18 .09
.14 .08
. 11 . 05
.06 .01
.01 .00
257
41* 42**
.00 .00
.02 .01-
.04
.06 .04
.08 .07
.11
.16 .17
V21 .23
.24 .22
.26 .28
.20 .17
.12 .08
.05 .05
259
41* 42**
.02 .00
.04 .00
.06 .04
.09 .07
.11
.19 .28
.29 ' .26
.22 .18
.19
.15 .11
.09 .07
.04 .05
.01 .00
273
41* 42**
.00 .00
.00 .00
.01 .00
..07
.09
.19 .18
.27
.39 .22
.45 .22
.36 .16
.22 .09
.11 .03
.05 .01
274
41* 42**
.00 .00
.02 .00
.03 .01
.06 .03
.10 .05
.14 . .08
.20 .10
.23 .12
.18 .09
.12 .05
.09 .03
.00
.00 .00
275
41* 42**
.00 .00
.00 .00
.00 .00
.02 .00
.04 .02
.09 .05
.13 .06
.11 .04
.11 .05
.09
.04
.00
.00
278
41* 42**
.00 .00
.00 .00
.03 .02
.04 .04
.07 ..06
.08 .08
.11
.13
.17 .16
.14 . .13
.10 .08
.06 .05
.03 .01
283
41* 42**
.02 .00
.02 .00
.03 .00
.04 .05
.07 .07
.09 .11
.12 .13
.15 .17
.18 .19
.18 .17
.14 .12
.09 .07
.06
288
41* 42**
.03
.03
.03
.05 .05
.06 .07
.07 .08
.08 .09
.09 ..10
.11 .11
.14 .14
.11 .08
.07 .02
.09 .00
*41
**42
Station OS41 in
Station 0842 in
Glendora (EPA)
Covina (EPA)
-------
UJ
U>
^&>
PDTN.
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
TABLE H-l (continued). OXIDANT
290
41* 42*-*
.02 .00
.02 .00
.02 .00
.04 .03
.08 .05
.10
.16
.18. .21
.27 .23
.27 .23
-.20 .18
.10 .08
.00
303
41* 42**
.00 .00
.01 .02
.03 .03
.03 .04
.04' .05
.05 .06
.05 .07
.06 .04
.06 .04
.04 .02
.03 .00
.00 .00
.00 .00
304
41* 42**
.00 .00
.01 .00
.02 .03
.04 .04
.06 .05
.'06 .03
.05 .01
.01
.02 .00
.00
.00
.00
.00 .00
VALUES (PPM) FOR EPA AIR MONITORING STATIONS 0841 AND 0842
305
41* 42**
.00 .00
.00 .00
.01 .02
.02 .03
.03 .03
.03 .04
.03 .04
.04 .03
.04 .02
.02 .00
.01 .00
.00 .00
.00 .00
306
41* 42**
.01 .00
.02 .00
.02 .02
.02 .02
.03 .02
.03 .03
.04 .04
.04 .04
.05 .03
.05 .01
.02
.02
.00
308
41* 42**
.01 .00
.01 .00
.03 .01
.04 .03
.06 .07
.08 .09
.08 .10
.10
.06
.02 .00
.00
.00 .00
.00 .00
309
41* 42**
.02 .00
.03 .00
.04 .01
.05 .03
.07 .08
.08 .09
.09 .11
.08 .12
.08 .08
.06 .01
.01 .00
.03 .00
.00 .00
310
41* 42**
.02 .00
.03 .00
.03 .02
.04 .04
.05 .05
.05
.06 .07
.07 .07
.07 .08
.07 .04
.06 .00
.06 .00
.04 .00
311
41* 42**
.04 .00
.02 .00
.02
.05
.06
.05 ' .06
.06 .07
.06 .07
.08 .06
.00
.00 .00
.01 .00
.02 .00
*41
**42
Station 0841 in
Station 0842 in
Glendora (EPA)
Covina (EPA)
-------
I
N>
CO
PDT\
0700
caoo
0900
1000
1100
12CO
1300
1400
1500
1500
1700
1300
1900
TABLE H-2. COMPARISON OF OXIDANT VALUES FROM FOUR NEIGHBORING STATIONS
253
41 42 A* T**
.00 -- .01 .01.
.02 -- .02 .03
.07 .03 .05 .06
.12 .10 .09 .12
.19 .21 .15 .22
.26 .31 .25 .31
.36 .37 .36 .37
.40 .26 .33 .26
.35 .22 .30 .25
.31 .26 .27 .20
.26 .21 .20 .17
.14 .09 .14 .11
.07 -- .08 .06
255
41' 42 A* T**
.01 -- .01 .01
.02 -- .02 .01
.04 .03 .04 .02
.07 .06 .06 .04
.13 .09 .09 .07
-- .15 .14 .11
.21 .18 .17 .14
.22 •• .15 .18 .13
.18 .09 .13 .08
.14 .08 .09 .06
.11 .05 .06 .04
.06 .01 .04 .02
.01 .00 .01 .01
259
41 42 A* T**
.02 .00 .01 .02
.04 .00 .01 .03
.06 ' .04' .02 .04
.09 .07 ' .05 .08
.11 -- .10 .13
.10 .29 .19 .22
.29 .26 .25 .38
.22 .18 .38 .36
.19 -- .39 .29
.15 .11 .29 .20
.09 .07 .19 .11
.05 .05 .10 .08
.01 .00 .06 .04
273
41 42 A* T**
.00 .00 .01 .02
.00 .00 .01 .03
.01 .00 .02 .04
.07 -- .05 .08
.09 -- .10 .13
.19 .18 .19 .22
.27 -- .25 .38
.39 .22 .38 .36
.45 .22 .39 .29
.36 .16 .29 .20
.22 .09 .19 .11
.11 .03 .10 .08
.05 .01 .06 .04
274
41 42 A* T**
.00 .00 .01 .01
.02 .00 .01 .02
.03 .01 .02 .03
.06 .03 .05 .05
.10 .05 .09 .09
.14 .08 .15 .11
.20 .10 .18 .14
.23 .12 .19 .14
.18 .09 .15 .11
.12 .05 .10 ' .09
.09 ' .03 .08 .05
-- .00 .02 .02
.00 .00 .01 .01
275
41 42 A* T**
.00 .00 .01 .01
.00 .00 .01 .01
.00 .00
.02 .00 .01
.04 .02 .01
.09 .05 .08 .09
.13 .06 .11 .03
.11 .04 .09 ..06
.11 .05 .09 .05
.09 -- .06 .03
.04 — .02 .02
-- .00 .01 .01
-- .00 .01 .01
*A - Azusa,
,**T - Temple
L.A.
City
Co. APCD values were
(California ARB)
multiplied by 1.28 to correct for different calibration procedure
-------
ce.
D
BLVD.
I
ro
to
Ln
I
SIERRA MADRE BLVD.
Sierra Madre
SIERRA MADRE BLVD.
Monrovia
BLVD. :.
XKQLORADO BLVD.
iBaldwin Park
AIR MONITORING STATIONS
(in kilometers)
Figure H-l. Air Monitoring Station Locations
-------
genuine area differences in oxidant levels between Glendora and
Covina on some days with higher levels occurring at Glendora.
In addition, there is evidence that suggests that some local ef-
fect at the Covina station caused unusually low oxidant values
on days 273, 274, and 275.
If complete data were available from the Azusa, Covina, and
Glendora stations, it would have been possible to prepare maps
showing oxidant levels in all parts of the study area on each of
the study days. Unfortunately, there were too many gaps in the
Covina and Glendora data to make this possible. Instead, the
oxidant values from the Azusa station were used in the data
analysis for this study. The Azusa oxidant levels usually lie
between the levels at the Glendora and Covina stations.
OXIDANT ADJUSTMENT FACTORS
Measurements of oxidant made by the Los Angeles County
APCD, the California ARB, and the EPA are not directly compar-
able because of differences in calibration methods. This situ-
ation was recognized early in 1974, and an ad hoc committee*
was appointed by the California ARB to determine which method was
most accurate and to relate the correct method to prior measure-
ments. The committee was also requested to recommend to the
California ARB a reliable procedure for field monitoring of
ozone. The final report of the committee was completed on Feb-
ruary 20, 1975. Simultaneous measurements of oxidant were made
by the three agencies on laboratory generated mixtures of ozone
and air. Most of the discrepancies in oxidant readings were
traced to differences in methods used for calibration of the
measuring instruments. Questions were raised concerning the ef-
fects of absolute humidity on readings, and the California ARB
carried out a special series of studies to investigate this ef-
fect. A draft report of the results was completed on May 14,
1975. The results show that the effects of humidity are mea-
surable but small and do not override the differences caused by
using different calibration methods.
The main purpose of the California ozone studies was to de-
termine which method was most nearly correct and to evaluate the.
California episode criteria levels in view of the findings. The
ad hoc committee found that the Federal Reference Method gave
results that were approximately 24 percent too high while the
*Dr. William B. DeMore, Chairman (Calif. Inst. of Technol-
ogy, Jet Propulsion Laboratory); Mr. J. Cyril Romanovsky,
Secretary (EPA); Mr. Milton Feldstein (Bay Area APCD);
Mr. Walter J. Hamming (Los Angeles County APCD, retired);
Dr. Peter K. Mueller (Environmental Research and Technology,
Inc.).
-236-
-------
Los Angeles County APCD method gave results that were 4 percent
too low. Adjustment factors were proposed for converting oxi-
dant levels from one agency's method to another's.
The California ozone studies provide information that can
be used not only to adjust each ozone reading to the "correct"
value but also to convert values measured by the Los Angeles
County PACD to equivalent values measured by the EPA. The Feb-
ruary 4, 1975, report finds that Los Angeles County APCD read-
ings should be multiplied by 1.28 to give equivalent EPA read-
ings, and the May 14, 1975, draft report indicates that this
number should be 1.25. Other studies which are currently in
progress may change the recommended adjustment factor again, but
in view of all the other uncertainties associated with oxidant
measurements, these changes are expected to be inconsequential.
The factor of 1.25 was used in this report for converting Los
Angeles County APCD oxidant values to equivalent EPA values.
OXIDES OF NITROGEN
The N0£ values for all study days for which data are avail-
able for both EPA air monitoring stations (with no more than
four hourly averages missing) are summarized in Table H-3..
Readings from the two stations sometimes agree closely (e.g.,
day 306) and sometimes differ considerable (e.g., day 309). In
instances where the two stations differ, the higher N0£ levels
are registered at the Covina station. A comparison of the EPA
air monitoring station readings with those from the Azusa and
Temple City stations (Table H-4) shows that the Azusa and Temple
City stations track reasonably well and give readings that are
similar to those from the Covina station. On days when the
Glendora readings differ, it appears that the Glendora station
registered lower N0£ values than are measured elsewhere in the
study area.
For purposes of data analysis, the N02 values from the
Azusa station were used. There were so many gaps in the EPA
data that they could not be used as a consistent data base. The
Azusa values are believed to be close to the EPA values for
Covina and slightly higher than the EPA values for Glendora on
days when the Covina and Glendora readings differ.
-237-
-------
00
I
^
0700
0800
0900
1000
1100
1200
1300
1400
1500
'1600
1700
1800
1900
TABLE H-3. N02 VALUES (PPM) FOR EPA AIR MONITORING STATIONS 0841 AND 0842
280
41 42
.06 .07
.08 .09
.09 .10
.11
.13 .16
.13 .13
.12'
.13
.12 .12
.11 .09
.09 .09
.10 .10
.10 .10
290
41 42
.02 .06
.02 .06
.03 .10
.07 .14
.07 .06
.04
.06 T-
.10 .10
.11 .08
.12 .11
.14 .15
.14 .20
.17
304
41 42
.02 .06
.04 .07'
.05 .03
.03 .02
.02 .02
.00 .04
..02 .05
.05
.05 .05
.06 .06
.07 .08
.07 .08
.07 .08
305
41 42
.04 .05
.04 .04
.03 .02
.03 .02
.02 .02
.02 .04
.04 .05
.05 .04
.05 .05
.06 .06
.06 .06
.06 .06
.06 .05
306
41 42
.00 .03
.01 .03
.02 .04
.02 .04
.02 .04 .
.03 .04
.03 .05
.04 .06
.04 .06
.05 .06
.07 .06
.06 .07
.08 .07
308
41 42
.02 .07
.04 .09
.04 .13
.06 .10
.06 .06
.04 .05
.04 .08
.08
.06 .07
.07 ,--
.07 .12
.07 .10
.07 .10
309
41 42
.03 .08
.04 .11
.04 .13
.05 .13
.05 .07
.04 .09
.03 .08
.04 .07
.04 .05
.06 .14
.10 .17
.06 .15
.06
310
41 42
.01 .05
.02 .06
.02 .07
.01 .05
.01 .02
.01.
.02
.02 .03
.04 .05
.06 .10
.06 .13
.04 .13
.05 .09
-------
I
N>
U>
VO
I
p
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
TABLE H-4. N02 VALUES (PPM) FOR NON-EPA STATIONS
280
A* T**
.06 .08
.08 .11
.09 .12
.12 .16
.15 .15
.12 .15
.16
.12 .17
.12 .13
•
.10 .10
.09 .10 '
'.09 .11
.03 .11
290
A* T*
.02 .06
.03 .09
.03 .13
.15 .15
.06 .12
.05 .11
.16
. 10 . 16
.12 .18
.11 .15
".14 .21
.17 .24
.14 .27
304
A* T**
.01 .05
.08 .05
.05 .06
.06 .05
.03 .02
.04 .04
.07
.09 .08
.09 .07
.08 .07
.09 .08
.10 .07
.10 .07
305
A*v ' f **
.05 .04
.06 .06
.06 .06
.06
.06
.04 .06
.08
.08 .07
.08 .06
.08 .05
.07 .05
.08 .06
.08 .05
306
A* T**
.01 .03
.01 .04
.04 .04
.04 .04
.04 .05
.04 .05
-- ...05
.04 .06
.05 .08
.08 .08
.08 .08
.09 .08
.09 .06
308
A* T**
.04 .05
.06 .05
.09 .09
.12 .12
.09 .10
.07 .08
.09
.11
.10 .08
.09 .07
.10 .10
.13 .11
.13 .11
309
A* T**
.06 .04
.06 .05
.09 .07
.10 .09
.11 .09
.09 .10
.11
.08 .14
.07 .10
.07 .09
.13 .10
.18 .13
.17 .13
310
A* T**
.02 .03
.02 .05
.02 .07
.05 .04
.04 .04
.04 .04
.04
.04 .06
.04 .07
.07 .08
.12 .11
.15 .12
.15 .12
*A = Azusa
**T •= Temple
City
-------
APPENDIX I
PROPORTION OF ASTHMA PANEL REPORTING
DISCOMFORT SYMPTOMS AND WEIGHTED AVERAGES OF
AIR POLLUTION LEVELS CHARTED BY DAY NUMBER
-240-
-------
ASTHMA PANEL DATA COLLECTION SCHEDULE
Discomfort symptoms and aerometric data were collected from
members of the asthma panel on the dates shown below. This was
during late summer and fall of 1974.
Day Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Date
September 9
10
11
12
13
16
17
18
19
20
23
24
25
26
27
30
October 1
2
3
4
Sub-Panel
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
Day Number
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Date
October 7
8
9
10
11
14
15
16
17
18
21
22
23
24
25
28
29
30
31
November 1
Sub-Panel
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
The portions of asthma panelists who reported discomfort symp-
toms are shown in Figures 1-1 through 1-9 for each of these 40
days.
In order to account for the effects on health associated
with exposure to air pollutants and to humidity and temperature
at the time the panelists were asked to complete Daily Symptom
Records (shown in Appendix F), weighted averages had to be
developed. In doing so the hour at which each panelist reported
for testing was noted and the aerometric values for that hour
were included in the weighted average for that day. The
weighted averages of exposure are shown in Figures I-10 through
1-15.
-241-
-------
SCATttRGKAM OH
•DOWN) EYES
2.00 6.00
10.00
It. 00
18.00
{ACROSS) DATE
,22.00 2.6,jO
34.00
3B.OO
i.ooooo 4
1
I
ro
•P-
O.UOOOO
0.00 4.00 8.00 12.00 16.00 20,00 24.00 - 28.00 . 32.00 36.00 40.00
0.00000
Figure 1-1. Proportion of asthma panel reporting eye discomfort by date.
-------
SCATttRGKAN
IOOMN) THROAT
2.00 6.00
10.00 14.00
JACKOSS) DATE
18.00 22.00 26.00
30.00
34. 0(
8.00
1.00000 4
1
1.00000
O.VOOOO 4
1
i
0.90000
1
I
0.80000 4
_D_»flOJUlfL
0.70000 4
1
0.70000
"0". 00000 4 i_»__4 4 * + 0.00000
0.00 4.00 8.00 12.00 16.00 20.00 24.00 28.00 32.00 36.00 40.00
Figure 1-2. .Proportion of asthma panel reporting throat discomfort by date.
-------
StAlItRSKAH Uh (DOWN) CHEST
2.00 6.00
10.00 It.00
18.00
UCKOSS) DATE
_22...00 26..PJL
30.00 3H.OO
3B.OO
1.00000 + + 1.00000
I - I_
I
TO
0.00000
0.00 t.QO 6.00 12.00
16.00
20.00 24.00 26.00 32.00
36.00 HO. 00
0.90000
0.00000
Figure 1-3. Proportion of asthma panel reporting chest discomfort by date.
-------
SCATTLR6KAM Ut-
I DOWN ) HEADACHE
2.00 6,00
10.00
.00
18.00
(ACROSS) DATE
22.00 26.00
30.00
3t.OO
38.00
1.00000
1.00000
0.90000 +
1
1
0.90000
I
1
o.eoooo «•
_o_._6joj>ao_
0.70000 »
i
0.70000
I
N>
a.ooooo +
.*
0.
8.00
12.00
16.00
20.00
24.00
28.00
82.00
36.00.
to.oo
00 t.OO
Figure 1-4. Proportion of asthma panel reporting headache by date,
0.60000
.00000
-------
I
ro
SCAITLR6KAN
1.00000
O.VOOOO
0.00000
0.70000
0.60000
0.50000
0.40000
O.iOOOO
0.20000
0.10000
o.uodoo
Ul- (DOWN) NAUSEA
2.00 6.00
1
1
1
1
1
i
1
1
1
1
1
i
1
1
I
1
I
1
i .
1
i
1
1
1
4
1
1
1
1
4
1
I
1
1
I
1
^ \
1 Ay
+ 4 \—4
(ACKOSS) DATE
10.00 14.00 18.00 22.00 26.00 30.00 31.00 38.00
I
I
X
I
4
' I
I
I
. . I
*
I
I
I
I
I
1
I
I
*
'i
I
I
1
I
I
I
I
I
X
I
I
A n
\ \ \ n
Y J \ 1 \ I \ 1 \ ' Nausea I
1.00000
0.90000
0.80000
0.70000
O.bOOOO
o.soono
0.40000
0.30000
0.20000
0.10000
t
0.00000
0.00
4.00
Figure 1-5. Proportion of asthma panel reporting nausea by date.
-------
SCATTtRGHAM
IUUWNJ
2.00
OTHER
6.00 10.00
DATE
».OQ 18.00 22.00 26.00 3JJ..JUL
36.00_
1.00000 +
_ I
1.00000
o.yoooo +
i
i
0.90000
I
to
• 00 4
00 8.00 12.00 16.00 20.00 24.00 28.00 32.00 36.00 40
+
+ .
.00
0.00000
Figure 1-6. Proportion of asthma panel reporting other discomfort by date.
-------
SCATTtRGKAN Ut-
(DOWN) BREATH
2.00 6.00
(ACROSS) DATE
2.00 26.00
SO. 00
34.00
38.00
1.00000
0.90000
0.90000.
1
I
0.80000+
o.sonnn
0.70000
o.TOono
I
to
J>
00
O.bOOOO
0.00000
0.00 4.00 8.00 12.00 16.00 20.00 24.00 28.00 32.00
36.00
•»•
40.00
O.faOOOO
0.00000
Figure 1-7. Proportion of asthma panel reporting shortness of breath by date.
-------
SCATItRGKAM Oh
(DOWN) COUGH
2.00 ' 6.00
10.00
It.00
18.00
(ACROSS) DATE
22.00 26,
30.00
3M.OO
3B.OO
1.00000
c »j^u o» uu ,-..„"• ".".__, * • "" *"_*L"" &.*_Oi-y go« u u_x_ au,a_u.u *?.?_.• ,U_U JQ« u u
A
1.00000
I
ro
0.90000
o.gonoo
0.10000
0.10000
0.00000 +
o.oo "».oo e.oo 12.00 le.oo 20.00 .. 21.00 28.00
52.00
36.
00 ">0.00.
Figure 1-8. Proportion of asthma panel reporting cough by date.
0.00000
-------
SCAITLRGKh* OK
(DOWN)
2.00
6.00 10. 00 It. 0 0
(ACHOSS) DATE
16.00 82.00 2 6.00 30.00 3H.QQ 38.00
I.00000
• +
•»
1
1.00000
Ul
o
0.90000
0.90000.
C.CO t.Ofl 8.00 12.00 16.00 20.00 24.00 28.00 32.00 36.00 40.00
Figure 1-9. Proportion of asthma panel reporting phlegm by date.
0.00000
-------
SCATTtRGMH OF
OZONE . (ACKOSS) DAY
_2_-Q_o 6-_nn in.on lit, no iB.nn 22_^D_Q gft.n
0.30
•»--—+-—-«•--- -+— -•».
»
0.30
1
1
___.!..
0.2^ +
1
1
O.Z7
0.21
0.21
to
M
I
0.18
P.
2U.Q. . ..28^.0J) 5Z^O.O 36..00.
Figure I-10. Average level of ozone by date: asthma panel.
0.16
o.oo
-------
SCATTtRGKBM OF JJOWN) CO
_ Z.OQ _ 6.00
10. 00 11.00 16.00
(ACKOSS) DAT
22.00 26.00
30. BO _ SH.QQ 38.00
0.20 +
1
0.20
0.1B
0.18
O.lb
0.16
o.m
I
NJ
Ln
NJ
I
0.12
0.10
I
1
0.12
0.10
O.OtJ
o.na
"0700-
o.oo
6.00 12.00 16.00 20.00 2<«.00 2B.OO 32.00 36.00. 40.00
Figure I-11.- Average level of carbon monoxide by date: asthma panel,
-------
StATTtRGHAM Uf-
(JOWN) N02
g.nn c.nn
(ACKOSS) DAY
g. nn 9A-J
0.30
0.30
1
1
1_
0.27 +
I
l_
0.27
0.21
Ul
U)
I
o.ia
/ /
/
I
+
1
0.12
0.00
o.oo. t.oo e.oo 12,00 le.oo 20^00 2i».oo 20.0.0 92.00 .ZS.QO ta»oo
Figure 1-12. Average level of nitrogen dioxide by date: asthma panel,
0.03
0.00
-------
SCATTLRGKAM Uf-
(JOHN) NO
2 , o o _ 6.00
lo.oo
m.oo
la.oo
(ACKOSS) UAT
22.00 _ 26.00
30.00
3».OQ
38.00
0.070
+ •
+
I
0.070
0.063
0.063
0.056
JLQ56_
0.049
I
ho
01
4^
I
0.049
0.007
0.007
U7000 <
0.00
-------
SCATTCRGKAM
90.00
(DOWN) HUMID . . . > (ACKOSS) OAT
.2.JU1 6.00 10.00 It.00 18.00 22.00 26.00 SO.00 St.Bfl 38.00
90.00
N>
Ln
^n
I
50.00
o.oo t.oo e.oo 12.00
16.00 20.00 Zt.OO 28.00
32.00 36.00 40.00
Figure 1-14. Average relative humidity by date: asthma panel.
86.00
-------
SCATTLRGKAM OF
< UOWNI
_2,_00
TEMP
6.00
10.00
m.oo
18.00
(ACKOSS) OAT
22.00 26.00
30.00
St.00
36.00
100.00
ioo.no
95.00
95.00
I
fO
Ln
50.00 +
00
"4.00
6.00
12.00
16.00
20.00
2<».00
26.00
32.00
56.00
1*0.00
Figure 1-15. Average temperature by date: asthma panel.
50.00
-------
APPENDIX J
PROPORTION OF OUTDOOR WORKER PANEL REPORTING
DISCOMFORT SYMPTOMS AND WEIGHTED AVERAGES OF
AIR POLLUTION LEVELS CHARTED BY DAY NUMBER
-257-
-------
OUTDOOR WORKER PANEL DATA COLLECTION SCHEDULE
Discomfort symptoms and aerometric data were collected from members
of the outdoor worker panel on the dates shown below. This was during
late summer and fall of 1974.
Day Number
Date
Sub-Panel Day Number
Date
**Clinical pre-surveillance tests.
***Clinical post-surveillance tests and symptom interviews,
Sub-Panel
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
September 11
12
13
14
16
17
18
19
20
21
23
24
25
26
27
28
30
October 1
2
3
4
5
7
8
9
10
11
12
1*, 2*
1
1
1
1
1
1*
1
1
1
1, 3**,
4**
1
1
1
f
1
2
2
2
2
2
2
2
2
2
2
2
2
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
• 54
October 15
16
17
18
19
21
22
23
24
25
28
29
30
31
November 1
2
4
5
6
7
8
9
11
12
13
14
3
3
3
3
3
3
3
3
3
3
4
4
4
1*** 2***
3***,' 4
4
4
4
4
4
1*** 2***
3***^ 4***
4
4
4
4
4
4
-258-
-------
The proportions of outdoor worker panelists who reported discom-
fort symptoms are shown in Figures J-l through J-9 for each of
these 54 days.
In order to account for the effects on health associated
with exposure to air pollutants and to humidity and temperature
at the time panelists were asked to complete Daily Symptom
Records (shown in Appendix F), weighted averages had to be de-
veloped. In doing so, the hour at which each panelist reported
for testing was noted and the aerometric values for that hour
were included in the weighted average for that day. The
weighted averages of exposure are shown in Figures J-10 through
J-15.
-259-
-------
sc«nt.HGt()00tu 3 ; j 1.00000
1 . I
1 I
0.90000
0.90000
O.dOOOO +
I
0.60000
I
NJ
ON
O
0.00000 4
.oo
o.ooono
6.00 12.00 16.r)0 24.00 30.00 36.00 "»2.00 16.00 54.00 60.00
t.c
Figure J-l. Proportion of outdoor worker panel reporting eye discomfort by date,
-------
SCAULKGHnM Uh
(DOWN) THhOAT (ACKOSS) DATE
3.00 9.00 15.00 21.00 27.00 33.00 39.00 MS.00 51.00 37.00
l.UOOOO
0.90000
0.00000
+
1
1
1
1
•»
i
1
I
1
4
I
I
I
I
I
I
I
I
•f
1.00000
0.90000
0.6QOOO
0.70000
0.70000
O.faOOCO
0.60000
I
N>
O.bOOCO
0.50000
O.nOOOO
0.40000
O.iOOOO
Throat
Discomfort
0.30000
0.20000 +
1
0.20000
0.10000
0.10000
0.00000
it i y v v/\j i i 4 *
12.00 18.00 24.00 30.00 36.00 42.00 46.00 54.00 60.00
lT~t~ 1 *"^T"I f~\f /™*1T+*/1/^/^'V- T.-TS\^~1r f^-v ^^«*^x^1 ^•xN*^*-*i*»4-'$'v*^. ^-L.v.^«M^I ««^^.^^.^.^_^
0.00000
0.00
6.00
Figure J- 2. Proportion of outdoor worker panel reporting throat irritation by date
-------
SCAI ILHGKhM lit-IUUWN) CHCSt (ACKOSii) DATE
3.00 9.00 15.00 21.00 27.00 33.00 29.00 15.00 51.00 57.00
1.00000
i.ooono
o.^nooo
o'.soonn
O.tfOOOO
o.bOono
0.70000
0.70000
tsi
I
O.bOOOO
o.uoooo
O.O'OOOO
0.00
Figure J-3,
6.00 It.CO 18.00 24.00 30.00 36.00 42.00 48.00 5H.OO 60.00
Proportion of outdoor worker panel reporting chest discomfort by date,
-------
t- (UOwfi) HEftO " (ACKOSS) DATE
3.UO V.OO 15.00 21.00 27.00 . 33.00 39.00 45.00 SI.00 ST.00
1.00000 +
» 1.00000
I
I
o.yoooo
0.90000
o.eoooo *
1
0.80000
0.70000 *
I
1
0.70000
1
1
o.bnooo •»
O.feOOOO
K)
a\
oo
i
O.bOOOO *
o.soono
o.^ooco
o.*«oooo
o.uoooo
0.30000
0.00000
0.00 6.00 12.00 18.00 24.00 30.00 36.00 42.00 48.00 54.00 60.00
Figure J-4. Proportion of outdoor worker panel reporting headache by date,
-------
ro
4>
SCAlTtKGKAM
1.00000
0.90000
o.eoooo
0. 70000
0.60000
O.bOOOO
0.40000
0.30000
O.iiOOOO
0.10000
0.00000
UK
.4
4
1
1
I
I
4
1
1
1
1
+
1
i
I
I
4
1
1
1
1
4
I
1
I
4
1
1
1
4-
1
1
1
1
4
1
1
i
I
4
1
i
I
1
»
1
1
i
.4
0.
(UOWN) NAuSCA (ACKOSS) DATE
3.00 9.00 15.00 21.00 27.00 33.00 39.00 45.00 SI. 00 57.00
I
I
I
I
*
' I
••..-.-•••• ' • I
I
:
*
i
i
i
i
4
I
I
I
I
I
I
I
I
1
I
I
I
. . 1
I
I
1
I
I
I
I
Jl !
• A fl '<
\...N\ AA..A- -^ /1A,,,,,,,_ !
00 6.00 12.00 18.00 24.00 30.00 36.00 42.00 48.00 54.00 60.00
i.ooono
0.90000
0.80000
0.70000
0.60000
0.50000
0.40000
0.30000
0.20000
o.iooon
o.ooono
Figure J-5. Proportion of outdoor worker panel reporting nausea by date.
-------
T/V" (DOin'J) OTHE.R
3.00 9.on 15.OC 31.00
S) DATE
27.00 33.00 39.00 "»5.00 51.00 57.00
l.OOOCO «
1
1
i.ooono
o.voooo +
0.90000
I
NJ
O.OCOOO
* 0.00000
.00 6.00 12.00 18.00
2<4.00 30.00 36.00 12.00 IB.00 5H.OO 60.00
Figure J-6. Proportion of outdoor worker panel reporting other discomfort by date,
-------
SCAT1C.RCHAM Oh (DOWN) SHOUT "'" " " (ACKOSS) DATE""
3.00 V.OO 15.OC 21.00 27.00 33.00 39.00 13.00 SI.00 37.00
1.00000 »
1
1
1.00000
1
1
0.90000 »
I
I
* 0.90000
0.00000 +
1
o.eoooo
I
to
o.uooou
Shortness
• • •» i nf Breath-:
0.00 6.00 1?.00 16.00
.on jo.oo 36.oo <»2.oo tB.oo st.oo 60.00
0.00000
Figure J-7. Proportion of outdoor worker panel reporting shortness of breath by date,
-------
SC«t1t.KGM«* Ot- tCOWl) COUGH
3.00 9.nn
15.00 21.00 27.00 33.00
)"DATE
39.00 15.00 51.00 57.00
I
ro
o.uoooo *
1.00000
o.ooono
0.00 6.00
Figure J-8.
12.00 16.00 24.00 30.00 36.00 42.00 46.00 54.00 60.00
Proportion of outdoor worker panel reporting cough by date,
-------
Uh (DOWN) HHLE.GM (ACKOSS) DATE
3.00 9.00 15.00 21.00 27.00 33.00 79.00 45.00 51.00 97.00
I
N>
cr>
oo
O.OCOOO
0.00000
U.OO 6.00 12.00 18.00 24.00 30.00 36.00 42.00 46.00 54.00 60.00
Figure J-9. Proportion of outdoor worker panel reporting phlegm by date
-------
SCOTTtRGKfl* OF
OZONE: OAT
.3.00— 9.00 15.00 21.00 27.00- 33.00 39. OC
.45.00 51.00 57.00-
I
N>
ON
VO
I
0.00
1.
00 6.00
Figure
12.
J-10.
00 18.00 24.00 30.00 36.QQ 42.00 .4^00 54.00 ^O.Ott
Average level of ozone by date: outdoor worker panel.
0.40
-------
SCATTlflGKAM Oh (JUWIJI CO
3.00 9..00
(ACKOSS)
33.00
51.00
_57.QQ_
I
to
0.07
0.01 4
1
__
0.00 4
.«---
1.00
6.00
12,00
18.00
24.00
30.00
36,00
42,00
48.00
54.00.
60.00
0.01
O.oo
Figure J-ll. Average level of carbon monoxide by date: outdoor worker panel.
-------
I
N>
StATTtRGHn* Oh
(JOWN) N02
.3.00 9.
.15.00 21.00 27.OC
(ACROSS) CAT
.33.00 -39-.00 45.00 -51.00 57
.00-
0.00
1.00
6.00
12.00
18.00
24.00
.30.00
36.00
42.00
43.00
J4.0.0
6.0.00
O.flO
Figure J-12. Average level of nitrogen dioxide by date: outdoor worker panel,
-------
UK
0.10 +
_ .1
I
(JOHN) NO
3.00
9 .00
27 . 00
(ACKOSS) OAT
33. 00 _ 39 . 00
57.00.
0.09
0.01
+'
I
0.10
0.09
0.01
0.00 +
• +-•-
1.00
Figure
6.. 00
J-13.
.U. 0.0
.+----+-_.-+_ ___<.. -..+
24. Ott 30.00
'42.00
.48,00
54.00
^0.00
0.00
Average level of nitric oxides by date: outdoor worker panel.
-------
SCATTLRGHftN Oh
(OUMNI HUMID
.3.00 9.00-
-15.00
-21,0
(ACKOSS) DAY
'27.00——33.00 39.00
7.00
I
NJ
•vj
<«0.00
1,00
6, 0.0
Figure J-14.
12,00
Average
<»0.00
IS^QQ. 2A,00. . iO.QQ . .Ji.M 42L.OO 4&-Qa . 5AJ3D ^0.00- . . .
level of relative humidity by date: outdoor worker panel,
-------
(JOWH) TtMP
-3.00 9.C
-15..00-
-21.00-
-27.00
(ACKOSS)
. 33.. 00
OftY
-39.00
-45,00
-51.00-
100.00
55.00 +
100.00
55.00
50.00
14---
1,00
6*00 * 112.00
18.00 24*00 30*00 36*. OQ * .42. QO . 48.00 34.00 .60.00
Figure J-15. Average temperature by date: outdoor worker panel.
5o.no
-------
APPENDIX K
PROPORTION OF BRONCHITIS PANEL REPORTING
DISCOMFORT SYMPTOMS AND WEIGHTED AVERAGES OF
AIR POLLUTION LEVELS CHARTED BY DAY NUMBER
-275-
-------
BRONCHITIS PANEL DATA COLLECTION SCHEDULE
Discomfort symptoms and aerometric data were collected from
members of the bronchitis panel on the dates shown below. This
was during late summer and fall of 1974.
Day Number Date
1 September 4
2 5
3 18
4 24
5 25
5 October 1
7 9
8 15
9 .16
10 18
11 22
12 25
The proportions of bronchitis panelists who reported discomfort
symptoms are shown in Figures K-l through K-9 for each of these
12 days.
In order to account for the effects on health associated
with exposure to air pollutants and to humidity and temperature
at the time panelists were asked to complete Daily Symtpom
Records (shown in Appendix F), weighted averages had to be de-
veloped. In doing so, the hour at which each panelist reported
for testing was noted and the aerometric values for that hour
were included in the weighted average for that day. The
weighted averages of exposure are shown in Figures K-10 through
K-15.
-276-
-------
SCATItRGKAH Ul- (OUtfN) EYES (ACROSS) DATE
1_._00 3 ..Q.Q 5.00 7.00 9.00 11.00 13juP..O 15.00 17.00 19.00
+ _.__+._.-4.__.. + .-.- + ..».+-. _.+....+.-..+_... 4.. —+_—.+.... + ..- - + -....f._. -4......«._...+_.„_+„__.+._..+ .
1.00000
0.90000
O.bOOOO
0.70000
O.bOOOO
0.50000
O."*0000
0. .40000
0.20000
0.10000
0.00000
•»
1
1
1
1
+
1
1
1
1
«
1
1
1
\
\
\^ A
1 "\ s \
i . ^ \
5 V
i \
1
+
i
1
1
1
4
i
1
1
1
+
1
1
I
I
+
0.00 2.00 t.OO 6.00
•f
I
X
I
I
«•
I
I
* I
I
»
I
I
I
I
+
I
I
I
* I
A l
\ !
/ \ '
\ '
\ / \ • !
\ / \ Eye Discomfort I
V i
I
I
+
I
I
I
I
4
I
I
1
I
*
6.00 10.00 12.00 It. 00 16.00 16.00 20.00
1.00000
0.90QOO
O.BOOOO
0.70000
o.boono
0,50000
O.<)0000
0. 300(10
0.20000
0.10000
0.00000
Figure K-l. Proportion of bronchitis panel reporting eye discomfort by date.
-------
StantKGKAM UK (DOWN) THROAT (ACKOSS) DATE
1.00 3.00 5.00 7.00 9.00 11. DO 13.00 15.00 17.00 19.00
1
fO
^1
00
1
1.00000
0.90000
O.BOOOO
0.70000
O.bOOOO
0.50000
O.<»0000
0.30000
0.20000
0.10000
0.00000
+ • H
1
1
I
1
•i
I
I
1
1
4
1
J
1
1 ]
+ ' " <
1 ]
1 1
1 A / \ i
** / \ / \ ^
1 / V \
* / N. R > Throat Discomfort
i \ / NV/
i V
i
i
+ . .
i
i
i
i
»
i
i
i
i
+ ' 4
>• 1.00000
0.90000
o.eoono
> 0.70000
[
t
[
I
<• 0.60000
I
t
t
[
K 0,50000
0.40000
0.30000
0.20000
0.10000
> 0.00000
o.oo
Figure K-2. Proportion of bronchitis panel reporting throat discomfort by date,
-------
SCATTtKGKAH Oh
(DOWN) CHEST
1.00 3.00
3.00
7.00
9.00
{ACROSS) DATE
11.00 13.00
IS.00
17.00
19.00
I
N5
^J
VO
I
0.30000
_P,20000
0.10000
0.30000
_0.20000L
0.10000
o.uoooo»
«+--.
0.00
2.00 4.00 .6.00 8.00 10.00 12.00 .1.4.00 .16.00 18.00 .
+
•---+•
20.. 00
0.00000
Figure K-3. Proportion of bronchitis panel reporting chest discomfort by date.
-------
SCATltKGKAM
I
to
00
Ul- (DUHfJ) HEAD
i.oo 3.on
5.00
7.00
9.00
(ACROSS) DATE
1.1«9.P_ 13...P.Q I?. 00 17.00
19.00
0.00000
0.00000
0.00
2.00
4.00 6.00 8.00 10.00 12.00 14.00 16.00 18,00 .
20.00
Figure K-4. Proportion of bronchitis panel reporting headache by date.
-------
I
fO
00
SCATTLRGKAM
1.00000
0.90000
0.00000
0.70000
O.bOOOO
0.50000
O.tOOOO
O.AOOOO
0.20000
0.10000
0.00000
Oh (DOWN) NAUSEA
1.00 3.00
4
I
I
1
1
4
1
1
i
1
4
1
1
1
1
4
1
i
1
4
I
I
I
i
4 ,
i
I
I
1
I
1
1
I
4
1
1
1
i
4
1
1
1
i v A
L V V
+ V
(ACROSS) DATE
5.00 7.00 9.00 11.00 13.00 15.00 17.00 19.00
4
I
I
I
I
4
' I
I
I
1
I
I
I
1
4
I
I
I
I
4
I
I
I
I
4
I
I
1
I
4
I
I
I
I
4
I
I
I
I
4
I
/\. /^ / x^ \
^f V V >, ., I
• — Nausea -
I
4
i.ooono
0.90000
o.eoono
0.70000
0.60000
0.50000
0. "40000
0.30000
0.20000
0.10000
,
0.00000
0.00 2.00 4.00 6.00 8.00 10,00 12.00 I1*.00 16.00 18,00 . 20.00
Figure K-5. Proportion of bronchitis panel reporting nausea by date.
-------
SCATTLRGKAM Ul- (OOwN) OTHER (ACROSS) DATE
_-lAPJL ? •. P 5 5.00 7.JO ? «J
CO
N>
1
1.00000
o.voooo
0.00000
0.70000
O.bOOOO
0.50000
O.BOOOO
0.10000
0.20000
0.10000
0.00000
1
1
/
i
1
1
1
1
1
1
1
i
1
1
1 .
4
1
1
1
1
1
1
i \ X*^
1 \ /\ s \
i \ / \ / \
i ^ v ' \ „„,
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
4
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
4
I
4
1.00000
0.90000
o.eoooo
0.70000
0.60000
0.50000
o.toono
0.30000
0.20000
0.10000
0.00000
Figure K-6. Proportion of bronchitis panel reporting other discomfort by date.
-------
SCATTLRGKAM UF
(DOWN) SHORT
1.00 _ 3.00
5.00
7.00
9.00
(ACROSS) DATE
11.00 _ 18.00
1.00000
o.voooo
19.00
17.00
19.00
I
NJ
00
0.00000
1.00000
0.90000
P., 80000
0.70000
0.60000
0.50000_
'
Shortness of Breach
O.HOO
0.300
.0.20000
o.iooon
o.ooooo
o.oo
Figure K-7,
2.00
4.00
6.00
e.oo
.10.00
12.00
m.oo
16.00
18.00
20..00
Proportion of bronchitis panel reporting shortness of breath by date
-------
SCATltKGKAM Uh (OOMN) COUGH (ACKOSS) DATE
1.00 3.00 5.00 7.00 9.00 11.00 13.00 15.00 17.00 19.00
I
N>
00
4>
I
1.00000
0.90000
0.00000
0.70000
O.bOOOO
0.60000
0.00000
0.30000
0.20000
0.10000
o.uoooo
4
I .*
i A -
'- X
: '\^
\
4
J
1
I
1
4
1
I
I
I
t,
1
1
I
I
4
1
I
i
1
4
1
1
1
1
4
1
1
1
1
4
1
1
1
1
4
O.CO 2.00 "4.00
4
I
I
I
I
f Cough +
\ / ^~^ *
V i
I
4
I
I
I
4
I
1
1
I
4
I
I
I
I
4
I
I
z
I
4
I
I
I
I
4
I
I
1
1
4
I
1
I
I
4
6.00 6.00 10.00 12.00 It. 00 16.00 18.00 . 20.00
1.00000
0.90000
0.80000_
0.70000
0.60000
0.50000
o.uoooo
0.30000
0.20000
0.10000
,
0.00000
Figure K-8. Proportion of bronchitis panel reporting cough by date,
-------
1
N>
00
Ul
1
SCAT1LRGKAM
1.00000
O.VOOOO
O.bOOOO
0.70000
O.bOOOO
O.bOOOO
0.10000
0.40000
0.20000
0.10000
0.00000
ui- PHLEGM
1.00 3.00 5, 00
4
^_y
i f
4 /
1 /
i 1
1 I
1 /
I /
1 /
1 i
1
1
1
1
1
* • . . .
1
I
1
1
4
1
1
1
1
4
1
1
1
1
4
1
1
1
1
4
1
1
1
1
4
0.00 ?.00 4.00
(ACKOSS) DATC
7.00 9.00 11.00 13.00 15.00 17.00 19.00
4
I
*. _-*^ / >• Phlegm I
I
I
X
4
I
X
I
I
4
X
X
I
I
4
I
I
X
I
4
X
X
I
I
4
I
I
I
I
4
I
I
I
I
X
I
I
X
4
I
I
X
I
4
6.00 6.00 10.00 12.00 lt.00 16, 00 18.00 20.00
1.00000
0.90000
_ 0.80000
0.70000
0.60000
0.50000
O.HOOOO
0.30000
0.20000
0.10000
f
0.00000
Figure K-9. Proportion of bronchitis panel reporting phlegm by date.
-------
SCATTE.RGKA*
(UOWN) OZONE
l.OQ 3.00
5.00
7.00
9.00
(ACKOSS) OAT
11.00 13.00 15.00 17.00 19.00
0.27
0.27
0.2"*
0.22
0.82
ro
oo
"OTST5
0.00 2.00 4.00 6.00 8.00 10.00 12.00 It.00 16.00 16.00
20.00
Figure K-10. Average level of ozone by date: bronchitis panel.
o«niT
-------
SCATTtRGKAM OF
-------
SCATTtRGNAM Ot- (DOWN) NOa
1-00 3.00 5.00
7.00
(ACKOSS) DAT
9.00 11.00 13.00 15.00 17.00 19.00
00
00
I
0.27
0.03
0.00 *
0
00 2.00 <«.00 6.00 8.00 10.00 12.00 1H.OO 16.00 IS.00
20.00
o.no
Figure K-12. Average level of nitrogen dioxide by date: bronchitis panel.
-------
SCATTtRGKBM
0.070
0.063
0.056
0.049
0.042
1
N> 0.035
00
0.028
0.021
0.014
0.007
0.000
Vt- (OOHfO NO . (ACROSS) OAT
1.00 3.00 5.00 7.00 9.00 11.00 13.00 15.00 17.00 19.00
1
1 ]
1
1
1
1
1
1
4
1
1
1
1
*
1
1
i
1
4 . t
i
I
1
1
1
1 f NO
i A /
| / \ /
i A A / \ /
! / \ /\ / \ ^ V
i
i
i
i '
i
+ 4
y 0.070
I
[
0.063
0.056
0.049
;
> 0.042
[
[
[
I
t 0.035
[
I
I
t
* 0.028
I
I
I
I
» 0.021
I
I
I
I
f 0.014
I
t
I
f 0.007
I
I
I
I
^ 0.00
0.00 2.00
Figure K-13,
.00
Average level of nitric oxides by date: bronchitis panel
-------
VO
O
I
SCATTLRGKAH Of
-------
SC«TTLRGK«M OF
(JOHN) TEKP
1.00 3.00 5.00
7.00
(ACKOSS) OAT
9.00 11.00 13.00 15.00 17.00 19.00
I
to
95.00
61.70
61.70
SB.00
u.
0.00
2.00
H.OO 6.00 8.00 10.00 12.00 14.00 16.00 18.00.
-4
20.00
5S.OO
Figure K-15. Average temperature by date: bronchitis panel.
-------
APPENDIX L
ATHLETE PANEL DATA COLLECTION SCHEDULE
-292-
-------
ATHLETE PANEL DATA COLLECTION SCHEDULE
Discomfort symptoms and aerometric data were collected from
members of the athlete panel on the dates shown below. This was
during late summer and fall of 1974.
Day Number Date
1 September 9
2 10
3 12
4 19
5 October 3
6 10
7 17
8 24
9 26
10 29
11 November 6
-293-
------- |