United States
Environmental Protection
Agency
Office of A:r Quality
Planning and Standards
Research Triangle Park NC 27711
EPA-450/5-83-006R
November 1 984
Fir'dl Report
Air
Review and
Evaluation of the
Evidence for
Cancer Associated
with Air Pollution
-------
REVIEW AND EVALUATION OF
THE EVIDENCE FOR CANCER
ASSOCIATED WITH AIR POLLUTION
Final Report
Prepared for:
U.S. Environmental Protection Agency
Pollutant Assessment Branch
Office of Air Quality Planning and Standards
Under:
Contract No. 68-02-3396
Prepared by:
Clement Associates, Inc.
1515 Wilson Boulevard
Arlington, Virginia 22209
I.C.T. Nisbet, Ph.D.
M.A. Schneiderman, Ph.D.
N.J. Karch, Ph.D.2
D.M. Siegel, Ph.D.
Current Address: 6503 East Halbert Road
Bethesda, Maryland 20034
2
Current Address: Karch & Associates
7713 14th Street, N.W.
Washington, D.C. 20004
U.S. Environmental Protection Agencv
November 7, 1984 Region V, ' •'^•-?ry
230 So-Jt i f.':- '••>" ~ lijet
Chicago, W.nois 60b04
-------
DISCLAIMER
This report has been reviewed by the Office of Air Quality
Planning and Standards/ the United States Environmental Protection
Agency, and approved for publication as received from Clement
Associates, Inc. Approval does not signify that the contents
necessarily reflect the views and policy of the United States
Environmental Protection Agency, nor does the mention of trade
names or commercial products constitute endorsement or recommen-
dation for use.
__^
U,S. Environmental Protection Agency
-------
PREFACE
This report has been prepared for the Office of Air Quality
Planning and Standards (OAQPS), U.S. Environmental Protection
Agency. The first version of this report was prepared for OAQPS
in July 1981. The July 1981 report was revised to take account
of criticisms and suggestions generated during an extensive peer
review and to incorporate new material published during 1981
and 1982. A draft version of this report was submitted to OAQPS
on December 15, 1982. The December 1982 draft was further
revised in November 1983 to take account of comments gener-
ated during an internal EPA review, but no new material was added,
The November 1983 report, with minor revisions, was issued for
public comment in March 1984. This final report has been further
revised to respond to comments received from the public through
June 1984, but no material published after November 1982 has
been added. This report is intended to be a comprehensive
review of scientific data published through November 1982.
-------
-------
TABLE OF CONTENTS
Page
EXECUTIVE SUMMARY
CHAPTER I. INTRODUCTION
A. Nature of Cancer 1-4
B. Interaction Between Risk Factors 1-5
C. Nature of Air Pollution 1-9
D. Purpose and Scope of this Report 1-11
CHAPTER II. EPIDEMIOLOGIC EVIDENCE
A. Introduction II-l
B. Epidemiologic Considerations II-2
1. Case Reports II-4
2. Ecological Studies II-4
3. Cohort Studies II-6
4. Case-Control Studies II-7
5. Issues Arising in Studies of Cancer II-9
and Air Pollution
C. Source-Specific Studies 11-26
1. Arsenic 11-27
2. Asbestos 11-35
3. Vinyl Chloride 11-39
4. Petrochemical and Other Chemical Emissions 11-41
5. Steel Manufacturing 11-44
D. Migrant Studies 11-45
E. Urban-Rural and Other Geographic Studies 11-48
1. Introduction 11-48
2. Air Pollution as a Factor in Geographic 11-50
Variation in Cancer Rates
F. Summary 11-94
-------
TABLE OF CONTENTS
CHAPTER III. EXPERIMENTAL EVIDENCE AND MONITORING DATA
A. Introduction III-l
B. Experimental Evidence III-4
1. In Vivo Tests of Extracts of Air Pollution III-6
for Carcinogenicity
2. In Vivo Studies of the Irritant Effects III-l3
of Particulates
3. In Vivo Mutagenicity and Genotoxicity Testing III-l7
4. In Vitro Tests of Extracts of Air Pollution 111-21
C. Monitoring Data 111-31
D. Multimedia Exposure 111-34
E. Summary 111-36
CHAPTER IV. QUANTITATIVE ESTIMATES
A. Introduction IV-1
B. General Estimates IV-2
C. Estimates Based on the Analysis of IV-4
Epidemiologic Data
D. Summary IV-22
REFERENCES
-------
APPENDIXES
A. Table II-l: Urban-Rural and Other Geographic
Studies of Cancer
B. Table III-l: Concentrations of Suspected or Known
Carcinogenic Substances in the Air
C. Calculation of the Age-Adjusted Respiratory Cancer
Rates in Males and in the General Population
D. Calculation of the Risk of Lung Cancer to the
General Population as a Proportion of the Risk
to Males
E. Derivation of an Estimate of the Proportion of
Lung Cancers Associated with the Urban Environment
F. Time Trends in Lung Cancer Rates
G. Critique of Two Recent Reviews
H. Data on Smoking Habits in Northeastern England
-------
-------
LIST OF TABLES
Table 1-1:
Table 1-2:
Table II-l:
Table II-2:
Table II-3:
Table II-4:
Table I1-5:
Table II-6:
Table II-7:
Table I1-8:
Table I1-9:
Table 11-10:
Lung Cancer Death Rate by Smoking History
Estimates of Percentage Reduction in Lung
Cancer Mortality in Asbestos Workers by
Elimination of Exposure to Cigarettes and
to Asbestos
Urban-Rural and Other Geographic
Studies of Cancer
Urban-Rural County Ratios of U.S. Age-
Ad justed Cancer Mortality Rates, White
Population, 1950-1969
The Urban Factor in the Distribution of
Lung Cancer Mortality in the United States
Pa<
1-7
1-7
Appendix A
11-49
11-51
Age-Adjusted Lung Cancer Rates of 11-57
Individuals Who Had Never Smoked by Location
of Lifetime Residence
Urban-Rural Differences in Lung Cancer 11-59
Mortality Rates in Nonsmokers
Estimates of the Percentage of Current, 11-63
Regular Cigarette Smokers, Adults Aged
20 Years and Over, According to Family
Income, Selected Occupation Groups, and
Marital Status, United States, 1976
Estimated Relative Risks of Lung Cancer 11-66
Mortality Expected from Differences in the
Prevalence of Smoking in 1955 Between Urban
and Rural Populations
Cumulative Percentage of Persons Becoming 11-69
Regular Cigarette Smokers Prior to Age
Specified, By Sex and Age, for Urban,
Rural Nonfarm, and Rural Farm Populations
Differences in Smoking Habits Between 11-70
White Male Residents of Two Areas of
Allegheny County, Pennsylvania
Relative Risk of Mortality from Lung Cancer 11-76
Standardized for Age, Smoking Classifica-
tion, and Age at Starting to Smoke, 1963-1972
-------
Table 11-11: Lung Cancer Mortality in Male British
Doctors, Standardized for Smoking and
Age, Stratified by Location of Residence
Table III-l;
Table II1-2:
Table IV-1:
Table E-l:
Table E-2:
Table E-3:
Table F-l:
Table P-2:
Table H-l:
Table H-2:
Table H-3:
Table H-4:
Concentrations of Suspected or Known
Carcinogenic Substances in the Air
Estimated Human Exposure to PAH from
Various Ambient Sources
Estimates of Lung Cancer Deaths
Associated with Various BaP Levels
Lung Cancer Death Among Men by Place of
Residence and Occupational Exposures—
Smoking Adjusted—1959-1965
Relative Risks in Men of Lung Cancer
Mortality (Adjusted for Age and Smoking)
by Residence and Occupational Category
Attributable Risks of Lung Cancer
Mortality (Adjusted for Age and Smoking),
U.S. Males (25-State Study) Due to
Urban Factor as an Indicator of Air
Pollution
Cigarette Smoking per Adult and Lung
Cancer Mortality in Males, England and
Wales, United States
Smoking History: U.S. Males
Distribution of Age at Starting to Smoke
by Area and Sex in the Living Population,
1973
Distribution of Age at Starting to Smoke
by Area and Sex in the Living Population,
1973
Distribution of Depth of Inhalation by
District and Sex in the Living Population,
1973
Proportion of Manufactured-Cigarette
Smokers Who Smoke Filter Cigarettes—By
Area, Sex, and Period for Which Smoking
Habits Reported
11-81
Appendix B
111-37
IV-9
E-3
E-4
E-5
F-6
F-ll
H-2
H-3
H-5
H-6
-------
EXECUTIVE SUMMARY
This report is a comprehensive summary and compilation
of scientific evidence related to the hypothesis that cancer
rates in human populations are associated with individuals'
exposure to pollutants present in the ambient air. Critical
comments on the strengths and weaknesses of the studies are
presented, and general methodological problems in the conduct
and interpretation of the studies are discussed. However,
at the request of the U.S. Environmental Protection Agency,
no overall judgments about the weight of the entire body of
scientific evidence are proffered.
Chapter I of this report is an introduction, which defines
its purpose and scope. Scientific evidence on the association
between air pollution and cancer is of three main types: epidemi-
ological studies of factors associated with patterns and trends
in cancer rates; experimental studies of the carcinogenicity
and mutagenicity of substances and mixtures emitted into or
extracted from the ambient air; and monitoring studies of the
presence in the air of substances known to be carcinogenic.
The existence and strength of the hypothesized association
between air pollution and cancer have been subject to extensive
scientific debate. One general problem is that a relatively
small effect of air pollution is difficult to establish conclu-
sively in the presence of larger (and variable) effects of
cigarette smoking and other factors (e.g., diet and alcohol).
-------
Another is that most cancers have multiple causes/ and there
are conceptual and methodological difficulties in attributing
cancers to more than one causative agent in the presence of
interactions. A third problem is that air pollution is complex
and variable in constitution and is difficult to characterize
adequately from existing types of monitoring data.
Chapter II summarizes epidemiologic studies of cancers
in the human population and their relation to air pollution
and other factors. Section II.B introduces the four principal
types of epidemiologic studies and discusses issues that arise
in applying them to the cancer/air pollution problem. Although
there is evidence that air pollutants may be associated with
cancers at a number of anatomic sites, only lung cancers have
been studied in sufficient detail for critical analysis. Air
pollution is a complex mixture of agents/ and most available
measurements are of conventional pollutants, which are unlikely
to"be carcinogenic in themselves; furthermore/ the use of a
single component, such as benzo(a)pyrene, as a surrogate measure
of the carcinogenic potential of polluted air may not be entirely
satisfactory. Significant exposure to some air pollutants
occurs in indoor environments/ where monitoring data are scanty.
The long latent periods for human cancers mean that current
cancers should be associated with exposures in past decades/
when some pollutants were present at higher levels and others
at lower levels. The most pervasive difficulty encountered
in the conduct and interpretation of epidemiologic studies
ii
-------
is the control of confounding factors, especially cigarette
smoking. Other problems that arise include the interpretation
of sex and racial differences in patterns of cancer mortality,
the insensitivity of many studies, and the selection of appro-
priate comparison populations.
Section II.C summarizes source-specific or "neighborhood"
studies. A number of studies have reported apparent elevations
in cancer rates in the vicinity of industrial facilities of various
types. Some of these studies were of the large-scale "ecologic"
type, whose results are usually regarded as no more than suggestive,
Most other studies in this category had substantial limitations,
including problems in identifying appropriate control populations;
in controlling for smoking, occupation, and demographic factors;
and in verifying exposure. The more persuasive evidence of this
kind is the finding of rare types of cancer characteristic of
exposure to vinyl chloride and asbestos near putative sources
of these materials. However, there are negative studies in
each of these cases.
Section II.D summarizes several studies that suggest that
migrants from one country to another with higher (or lower)
air pollution levels continue to experience cancer rates charac-
teristic of their native countries. However, the rigor of the
statistical comparisons of cancer rates is questionable, and
the differences were not related to specific data on exposure
to air pollution.
iii
-------
Section II.E summarizes urban-rural and other geographic
studies. Table II-l (Appendix A) tabulates 48 epidemiologic
studies (reported in 43 papers) of cancers of the lung and
other sites in human populations. In 28 of these studies,
a statistical association was reported between cancer rates
and one or more (direct or indirect) measures of air pollution,
and most of the rest reported excess frequencies of cancer
in urban areas relative to rural areas. Only seven or eight
studies reported finding no association between cancer rates
and either urban location or measures of air pollution. How-
ever, all the studies were subject to various limitations,
which complicate their interpretation.
The most pervasive and difficult problem in these studies
is control for confounding effects, of which cigarette smoking
is the most important. Ten studies of lung cancer rates in
nonsmokers have shown rather consistent urban-rural differen-
tials in males, but not in females. However, all but one of
these studies were limited by small sample size, and none was
controlled for occupational exposures. In a number of studies,
urban-rural differentials and statistical associations between
cancer rates and air pollution remained significant after attempts
were made to control for the effects of smoking, using data on
smoking habits in cancer victims or population groups. However,
the completeness of the control for smoking in these studies
is disputed. Some scientists have argued that differences
in aspects of smoking such as age at starting to smoke and
iv
-------
depth of inhalation cannot be controlled for. However, actual
data on these aspects of smoking do not confirm that they would
contribute significantly to urban-rural differentials.
Only a few studies have been controlled for the effects
of occupational exposures. One study that was so controlled
revealed significant urban-rural differentials in both occupa-
tionally exposed and unexposed groups/ after controlling for
smoking (see Appendix E). Other studies have suggested inter-
actions between effects of occupation and air pollution.
Chapter III compiles and summarizes experimental evidence
and monitoring data. A substantial number of studies have
shown that extracts of airborne materials from polluted air
and materials emitted from motor vehicle engines and stationary
sources are frequently carcinogenic and mutagenic when tested
in experimental bioassay systems. Results of in vivo tests
have included the induction of skin cancers, lymphomas, fibro-
sarcomas, liver tumors, and lung tumors in mice; lung tumors
in rats and hamsters; and chromosome damage and sister chromatid
exchange in hamsters. Respiratory irritants present in polluted
air may also enhance the effects of other carcinogenic agents.
Results of in vitro tests have included the induction of point
mutations in bacteria and Drosophila melanogaster, malignant
transformation of mammalian cells in culture, and sister chro-
matid exchange and DNA fractionation in cultured mammalian
cells, including human cells. Positive results in these in vitro
-------
tests are generally correlated with the potential for carcino-
genicity.
Table III-l (in Appendix B) lists more than 35 chemicals
that have been detected in ambient air and that are known or
suspected to be carcinogenic in humans or in experimental ani-
mals. Where comparative data are available, concentrations
of these chemicals tend to be higher in urban areas than in
rural areas/ and higher still in industrial emissions. There
is evidence of significant multimedia exposure to several pol-
lutants after their release into ambient air.
Chapter IV summarizes attempts to estimate the possible
magnitude of the association between lung cancer rates and air
pollution levels. For this purpose, the index of air pollution
most commonly used is the average atmospheric concentration
of benzo(a)pyrene (BaP). Use of this index, however, causes
difficulties because average levels of BaP in the United States
have declined considerably since 1958 and probably were higher
still prior to 1958. However, it is not clear that overall
hazards posed by air pollution would have declined, since levels
of other potential carcinogens have probably increased since
1940. BaP is thus not a stable index of the carcinogenicity
of polluted air, and estimates made at one time period cannot
be applied directly to others; for example, estimates based
on the study of lung cancers in the past cannot be used directly
to predict future effects of current pollution.
vi
-------
Recognizing this problem/ Table IV-1 tabulates 13 esti-
mates of the quantitative relationship between lung cancer
rates and air pollution levels as indexed by BaP concentrations.
Estimated slopes (regression coefficients) of this relation-
ship range from 0.1 to 5.0 x 10 lung cancer deaths per year
per ng/m BaP. Some of these figures should probably be adjusted
downwards by factors of 2 to 4 to take account of the likely
reduction in BaP levels since the 1930s and 1940s when most
effective exposures took place. The estimates derived from
studies in the general population (0.8-5.0 x 10" ) are signifi-
cantly higher than those derived from studies of workers exposed
to products of incomplete combustion (0.11-0.8 x 10" ). This
difference suggests that incomplete combustion products are
associated with only part of the excess lung cancer rates ob-
served in urban areas. Most of the studies were based on lung
cancer mortality data from the 1960s. The results are con-
sistent with the hypothesis that at that time, factors respon-
sible for the urban excess in lung cancer were associated with
about 19% of lung cancers in urban areas of the United States.
In the one study in which both cigarette smoking and potential
industrial exposure could be taken into account, this estimate
was about 23%. These quantitative estimates can be derived
without resolving the issue of whether the unexplained urban
excess of lung cancer can or cannot be attributed confidently
to air pollution, which depends on interpretation of data sum-
marized in Chapter II.
vii
-------
Several Appendixes to this report deal with technical
issues or tabulate information used in the text. Appendix E
presents a calculation of the relationship between lung cancer
rates and location of residence, after controlling for age,
smoking/ and occupational exposure. Appendix F discusses time
trends in lung cancer incidence and mortality, including results
from three recent cohort analyses that support the hypothesis
that changes in smoking habits cannot account for all features
or trends in the U.S. and the U.K. Appendix G presents a critique
of two recent reviews of the subject that concluded that the
association between air pollution and cancer rates was incon-
clusive or weak.
viii
-------
I. INTRODUCTION
The air contains a wide variety of hazardous substances,
exposure to which may be associated with a broad range of adverse
human health effects. Relatively high-level short-term exposures
to some types of air pollution may result in acute sickness,
alteration of important physiological functions, or impairment
of performance. Prolonged exposure to lower levels may result
in cancer or other chronic diseases, shortening of life, or
impairment of growth or development.
During the past several years, the relationship between
air pollution and cancer has received considerable attention.
We have come to recognize a number of air pollutants as known
or suspected carcinogens. Some of these are widespread and
derive from a variety of sources (e.g., benzene, arsenic, as-
bestos, and certain polycyclic aromatic hydrocarbons), while
others are limited to a few types of sources (e.g., certain -
chlorinated solvents or arsenic and other smelter emissions).
The evidence that cancer risks may be associated with air pol-
lution or specific pollutants in air is of three main types:
• Data from epidemiologic studies, which include des-
criptive studies of trends in cancer by time, place,
or affected group (e.g., sex, age, race); ecological
studies, which relate group differences in exposure
to group differences in the frequency of cancers; and
case-control or cohort studies, depending on whether
the initial basis for study is a group of people with
cancer (cases) or a group exposed to air pollution
or another risk factor (cohort)
• Data from laboratory studies, which include a range
of in vitro studies (e.g., studies of the mutagenicity
1-1
-------
in cell cultures of substances identified in ambient
air), and long-term carcinogenesis bioassays in animals
of specific pollutants, complex mixtures of pollutants,
or concentrates of air samples
e Data from monitoring studies, which involve measurements
of individual pollutants in air and which are designed
to demonstrate the presence of specific substances
or mixtures, many of which may have been found to be
cancer causing in epidemiologic or laboratory studies.
Some^have interpreted this evidence as showing that cancer
risks are associated with air pollution, while others have argued
that the evidence does not support such an association. Although
several surveys of the problem have appeared in recent years
(see Appendix E), no comprehensive review of the scientific
evidence has yet been published. This report is intended to
provide a compilation and evaluation of this evidence. Although
we do not proffer an overall judgment as to the weight of evidence
that air pollution (or specific pollutants) is associated with
increased cancer risk, we point out the strengths, weaknesses,
and biases of individual studies and discuss a number of general
problems in conducting and interpreting studies of this problem.
At the request of EPA, this review covers all potential airborne
contaminants except radioactive substances.
Much of the debate on this question has focused on urban-
rural differences in cancer incidence or mortality, i.e., the
observation of excess mortality from cancer at certain anatomic
sites in urban compared to rural counties in the United States.
Elevated cancer risks in urban areas, whether attributable
to air pollution, cigarette smoking, occupational exposure,
or other factors, are cause for concern among public health
1-2
-------
officials because over three-fourths of the population of the
United States now lives in areas defined by the U.S. Census
Bureau as urban. Furthermore, rural air in certain parts of
the country may also contain carcinogenic pollutants/ in which
case urban risks calculated from urban-rural differences would
tend to underestimate the role of air pollution, if carcinogenic
4.
air pollutants are in fact a cause of these differences.
In the debate on the relationship between air pollution
and cancer in the United States, urban-rural differences have
been interpreted by a number of scientists as evidence for
an association. This has been supported by monitoring data
that demonstrate the presence in air of substances previously
shown in epidemiologic studies (usually of work place risks)
or animal studies to be carcinogenic. Also, when other risk
factors have been controlled for, urban-rural differences have
been used to compute estimates of the magnitude of the risks
posed by urban air pollution.
Other scientists have argued against the conclusion that
an association exists because (1) the evidence for increased
cancer risks from urban air pollution is not consistent, in
that some investigators have failed to find a. correlation between
lung cancer and measured levels of pollution; (2) urban lung
cancer rates have not declined although air pollution, as meas-
ured by the level of benzo(a)pyrene (BaP), has declined; and
(3) in some studies urban-rural differences have been observed
only for men. These scientists have cited differing patterns
1-3
-------
of cigarette smoking, workers' industrial exposure, or both
as alternative explanations for the urban-rural differences.
Several scientists have argued that, in the presence of large
and variable effects of cigarette smoking, it is impractical
or impossible to detect smaller effects of air pollution, and
that existing studies that appear to indicate such effects
are inconclusive.
A. Nature of Cancer
Most experts now recognize cancer as a multicausal, multi-
stage set of diseases (OSHA 1980). Cancer is a complex group
of diseases that characteristically progress through a number
of stages, each of which may be initiated or accelerated by
a number of different intrinsic and extrinsic risk factors.
Each factor may act at one or more stages, and different factors
may interact in an additive or a synergistic (multiplicative)
way. Furthermore, because of the frequently long latency period
between initial exposure and manifestation of cancer, typically
20-30 years or more for many carcinogens, numerous opportunities
exist for multiple exposures to potentially carcinogenic agents.
It follows from the complexity of cancer causation and develop-
ment that most cancers would have multiple "causes," and it
would be simplistic to assign to any cancer or type of cancer
a single causative agent.
The multistage, multicausal nature of cancer greatly compli-
cates the task of identifying whether complex mixtures of sub-
stances, such as air pollution, cigarette smoke, and certain
1-4
-------
work place exposures, are associated with increased cancer
risks. It offers, however, various opportunities for prevention,
particularly when there is an interaction between risk factors.
B. Interaction Between Risk Factors
It is reasonable to expect that there will be interactions
among cigarette smoking, air pollution, and other complex risk
factors. First, many of the substances identified as carcinogens
in cigarette smoke are also found often as pollutants in air
or as constituents of emissions in the work place. Second,
synergistic interactions lead to a combined risk that is greater
than the sum of the risks from each, in which case reduction
in exposure to either factor is likely to be accompanied by a
greater than proportionate reduction in risks. When two factors
interact synergistically, each factor is not a confounding
factor of the other, but an effect modifier (Rothman 1975).
Synergism in the induction of lung cancer is known to occur
in humans with a number of agents, e.g., between cigarette
smoke and asbestos, and between cigarette smoke and radionuclides
(Selikoff and Hammond 1975). In view of this, it is simplistic
to attribute all lung cancers in which smoking is involved
to cigarette smoking only.
Walker (1981) recently proposed a method for estimating
the proportion of disease attributable to the combined effect
of two factors. This method first identifies the etiologic
fraction of disease due to the simultaneous action of both
factors among exposed persons. This fraction is an estimate
1-5
-------
of the extent to which disease may depend on exposure to both
factors together. An interaction index is then calculated,
which is the proportion of disease attributable specifically
to the interaction between two factors rather than to the disease
expected from each acting alone.
As an illustration, if Walker's method is applied to EnterlineV
(1979) smoking, asbestos, and lung cancer data (see Table 1-1),
the etiologic fraction is 97%, i.e., the proportion of lung
cancer among smoking asbestos workers attributable to smoking,
asbestos, and their interaction, is 97%. (The remaining 3%
is attributable to other, unidentified, factors.) Of the 97% at-
tributable to smoking and asbestos, the proportion due specifically
to interaction is 73%; the remaining 27% is expected from the
effect of smoking and asbestos acting alone.
Another way of looking at interactions is to determine
the proportion of cancers that could be prevented by eliminating
either factor. This method attributes the interaction between
factors to the factor being eliminated. This is illustrated in
Table 1-2 (OTA 1981), based on the data of Lloyd (1979), which
are similar to those of Enterline (1979).
The potential for interaction among cigarette smoking,
air pollution, and other factors such as occupational exposure
requires careful evaluation. In such complex circumstances,
attributing all possible disease to cigarette smoking whenever
1-6
-------
TABLE 1-1
LUNG CANCER DEATH RATE BY SMOKING HISTORY
(Rates per 100,000 per Year)a
Cigarette
Smoking
Yes
No
Asbestos
Insulators
362.0 .
40.4
U.S.
Males
74.4
9.2
Relative
Risk
4.9
4.4
alf the combined effect of smoking and asbestos
changes with age, the age distribution in the popu-
lation to which these data are standardized will
affect the calculations of the etiologic fraction
and the interaction index.
SOURCE: Enterline (1979), Table 2
TABLE 1-2
ESTIMATES OF PERCENTAGE REDUCTION IN LUNG CANCER
MORTALITY IN ASBESTOS WORKERS BY ELIMINATION
OF EXPOSURE TO CIGARETTES AND TO ASBESTOS
Percentage
Reduction
from Current
Status Rate
Current 0.0
Eliminate smoking only 88.5
Eliminate asbestos only 79.6
Eliminate smoking and asbestos 97.8
SOURCE: OTA (1981), Table 11, p. 68
1-7
-------
cigarette smoking is a factor may lead to overestimation of
the role of smoking and an underestimation of the importance
of the other factors present. The implication for cancer pre-
vention is that interference with any (or all) identified risk
factors is likely to reduce disease incidence.
Synergistic effects between various substances, such as
BaP and N-nitroso compounds, both of which are often present
in ambient air, have also been demonstrated in animal experi-
ments. In one such experiment, Montesano et al. (1974) instilled
intratracheally into hamsters BaP adsorbed on ferric oxide
particles. This was followed by repeated injections of diethyl-
nitrosamine. BaP or diethylnitrosamine alone produced few
malignant tumors, but the two in combination produced a 35% inci-
dence of tumors, which appeared within a shortened latency
period. In a similar experiment, Kaufman and Madison (1974)
found that either N-nitroso-N-methylurea or BaP plus ferric
oxide induced tumors with a latency of about 50 weeks after
intratracheal instillation. When both substances were admin-
istered together adsorbed on ferric oxide, they caused a higher
tumor incidence with a latency of 20-35 weeks. In another
study, McGandy et al. (1974) examined the interaction of BaP
adsorbed on ferric oxide and polonium-210, a carcinogenic radio-
isotope. These substances were administered intratracheally
in hamsters either simultaneously or sequentially. In both
cases, the number of lung tumors observed was more than twice
1-8
-------
the number expected from the effects of each substance acting
alone.
C. Nature of Air Pollution
Polluted air is a complex and highly variable mixture
of substances. In many studies reviewed in this report, the
term air pollution is considered synonymous with the air in
areas with concentrations of heavy industry. Yet, since the
days of the dial-painters/ carcinogenic hazards have been known
to exist in a number of light and service industries; because
substantial strides have been made in the last two decades
in reducing emissions from a variety of types of heavy industry,
some of the most hazardous emissions may be from small, older
operations that are not classified as heavy industry.
Data have been collected on a number of common, widespread
pollutants, but the measurement of many pollutants is difficult
and expensive. In many areas, only a fraction of the pollutant
mixtures may be measured or even known. What is measured may
not easily be generalized to other areas. Also, data that
have been collected rarely cover the extended periods of time
necessary for cancer to develop. Current levels of pollutants,
often used as an indicator of past exposures, may not be represen-
tative of past exposures.
Even when the definition of air pollution is tied more
closely to measured levels of specific pollutants, the results
of a study can be substantially affected by the location, fre-
quency, and extent of measurements. Pollution levels tend
1-9
-------
to drop off as distance from the source increases/ and models
of dispersion and movement are sensitive to a number of assump-
tions about such factors as meteorological conditions and trans-
formations of pollutants. If peak levels of a pollutant induce
proportionately more damage than lower levels, the method of
averaging over time as well as over distance can be important.
Thus, because of the complexity of cancer induction and
the difficulty in knowing with any accuracy the exposure levels
to a pollutant, the task of assessing whether and under what
circumstances pollutants in ambient air may be associated with
increased cancer risk is a complicated one. Air pollutants
may act in several ways in the induction or promotion of cancer.
First, substances emitted into ambient air may act alone to
increase population cancer risks. This appears to be the case,
for example, with vinyl chloride. Exposure to this substance
in the work place and perhaps in communities surrounding certain
industrial plants increases the risk of developing angiosarcoma
of the liver and possibly brain cancer. Second, ambient air
pollutants may interact synergistically with other factors.
The interactions between smoking and asbestos or radionuclides
are prime examples of this. Third, substances present in the
ambient air may also promote or otherwise enhance the carcino-
genic effects of particular agents. The phenomenon of promotion
or cocarcinogenesis among chemical agents has been studied
in experiments with animal tissues (Sivak 1979). These exper-
iments show that the effect of some carcinogens may be enhanced
1-10
-------
by other substances often present in polluted air (i.e., fine
particulates and such respiratory irritants as sulfur dioxide).
Chemical carcinogens present as pollutants in air at low concen-
trations might be expected to have only slight effect by them-
selves but to have much greater effects when present in combin-
ation with these promoters or cocarcinogens. There is also
the possibility that substances in the air may act antagonistic-
ally, reducing the effectiveness of chemical carcinogens.
This might be the case when carcinogenic pollutants are adsorbed
to large, nonrespirable particulates.
D- Purpose and Scope of this Report
The purpose of this report is to review in a systematic
way the evidence for cancer risks associated with air pollution.
First, we review the epidemiologic literature on cancer risks
associated with pollutants in ambient air, excluding radiation.
The evidence has been divided into four major categories: source-
specific studies, urban-rural comparisons, migrant studies,
and time trend analyses. Second, we review the experimental
and analytical data indicating that ambient air may contain
a wide variety of carcinogenic or mutagenic substances. A third
section of this report reviews studies in which the possible
magnitude of the association between air pollution and cancer
rates has been estimated in quantitative terms. Summaries
at the end of each section give an overall characterization
of the extent of each type of scientific evidence and of the
strengths and weaknesses of this evidence. However, in accor-
1-11
-------
dance with the request of the U.S. Environmental Protection
Agency that this report present an objective summary and unbiased
review of the available data, no overall judgments about the
weight of the entire body of scientific evidence are proffered.
1-12
-------
II. EPIDEMIOLOGIC EVIDENCE
A. Introduction
This chapter reviews the epidemiologic evidence for the
proposition that ambient air pollutants contribute (either alone
or in combination with other factors) to cancer rates observed
in human populations. For purposes of this review/ the chapter
has been divided into four major sections (Sections B-E):
• Epidemiologic considerations and issues
• Source-specific studies
• Migrant studies
• Urban-rural contrasts and other geographic studies/
including attempts to correct or control for the con-
tribution of other factors
Temporal trends in cancer rates are discussed in Appendix F,
with a review of attempts to interpret them in terms of temporal
changes in air pollution and in human exposure to other causative
factors.
In the first section of this chapter (Section B), four
major types of epidemiologic studies that can be used to investi-
gate the association of air pollution with cancer frequencies
are described. The strengths and weaknesses of each type of
study are considered, and some specific problems that arise
when they are applied to the air pollution/cancer problem are
discussed.
In the second section, source-specific studies, i.e.,
studies that examine the relationship between air pollution
II-l
-------
from a particular industrial source and cancer rates in nearby
communities, are reviewed. These include studies on the risks
of cancer in communities surrounding several types of industrial
facilities/ such as smelters, asbestos factories, vinyl chloride
manufacturing plants, and petroleum refineries. The strengths
and weaknesses of each study are reviewed, including considera-
tion of inconsistent data.
In the third section, studies of migrants from areas of
high pollution to areas of low pollution (or vice versa) are
reviewed. In the fourth section, urban-rural and other geographic
comparisons are reviewed. In these studies cancer rates in
urban (and/or industrial) areas are compared with those in
rural (and/or nonindustrial) areas. The major difficulty with
these studies is the problem of confounding, i.e., the presence
of differences between urban and rural areas in such factors
as smoking and occupation can obscure the effects of air pollu-
tion itself. In this section we review attempts to isolate
or control for the confounding factors and thus estimate the
effects of air pollution, alone and in combination, in accounting
for the elevated rates of cancer in urban areas.
Recent trends in cancer mortality and incidence are reviewed
in Appendix F.
B. Epidemiologic Considerations
Properly designed and controlled epidemiologic studies
can provide direct evidence that human exposure to a particular
substance or pollutant is associated with a risk of disease.
II-2
-------
Such studies, however, are unfortunately vulnerable to many
biases, leading to a wide range of limitations and uncertainties.
Because of these limitations, the findings of a single study are
rarely accepted as conclusive. Epidemiologic findings carry
more weight when the results of independent studies conducted
under different circumstances support each other. The results
of epidemiologic studies may draw strength from, or may be
challenged by, the results of other epidemiologic studies,
as well as other types of scientific evidence.
Epidemiologic studies .have been classified into four main
types:
• Case reports
• Ecological or "descriptive" studies
• Cohort studies
• Case-control studies
The latter two types of study, which are also called "analytic"
studies, carry more weight than the first two types because
they are better controlled and usually reflect the consequences
of exposure for specific individuals. Ecological and descriptive
studies usually generate evidence of the circumstantial type
and help to generate hypotheses about associations. Where
the circumstantial evidence is very strong, they and certain
case reports can lead to relatively firm conclusions. However,
in most cases it is necessary to test the hypotheses generated
by these studies using the more rigorous methodology of cohort
or case-control studies.
II-3
-------
1. Case Reports
Case reports take the form of reporting illness or death
in one or several individuals—with the illness putatively
associated with an exposure of an unusual type or a set of
common exposures. Case reports often serve as the starting
point in implicating specific exposures as possible causative
factors. The hypotheses generated from these reports generally
need to be tested systematically in controlled studies before
they are regarded as conclusive. In some instances, when the
effect is both pronounced and specific, such observations may
provide strong evidence for an association between a substance
and the outcome observed.
2. Ecological Studies
Ecological studies relate group differences in exposure
to group differences in the frequency of disease. The groups
typically comprise residents of geographic areas, such as dis-
tricts, cities, or counties. Data on geographic differences
in cancer frequencies among these groups are related statisti-
cally to data on differences in exposure to chemicals or other
possible causative factors. Other ecological studies report
trends in disease over time or by demographic characteristics
(sex, race, income, etc.) and attempt to associate them with
specific trends or differences in exposure. These studies
generally use data that are readily available and thus may serve
for preliminary examination of an hypothesis or for generating
II-4
-------
other hypotheses. Such studies often provide a basis for deci-
sions on whether to initiate more intensive studies and, more
rarely, for definitive conclusions about associations.
Ecological and other descriptive studies are sensitive
to misclassifications and the inappropriate handling of confound-
ing factors. If sufficiently important, these may lead to
underestimates/ overestimates, or even reversals in the direction
of the relationship between exposure and outcome at the individ-
ual level (Robinson 1950, Greenberg 1979). Results of these
studies,, therefore, are usually considered tentative until
confirmed by other evidence. In evaluating the descriptive
and ecological studies bearing on the relationship between
air pollution and cancer, the degree and manner in which poten-
tial confounding factors, such as age, sex, race, cigarette
smoking, and occupation, are taken into account influence the
outcome.
Statistical sensitivity (the probability of detecting
a true association when it exists) is an important concern
in epidemiologic studies. Ecological studies usually are insen-
sitive—or have a high noise-to-signal ratio. For example,
sensitivity may be lost by considering all residents in a certain
geographic area as "exposed." All residents are rarely equally
exposed. If only a proportion of residents is actually exposed
and at risk, the risk estimated in such a study will be diluted
and may not even be detectable. Migration between geographic
areas can also reduce sensitivity. As people migrate between
areas, the distinction between exposed and unexposed is gradually
II-5
-------
lost. As a result, the ability of geographic studies to reveal
an effect is likely to be reduced substantially if migration
is not taken into account. The longer the cancer latency period,
the larger this dilution effect is likely to be. It has been
estimated that when migration has taken place over a 30-year
period (roughly the latent period of the disease of concern),
40-50% of the actual excess risk will not be detected (Polissar
1980).
3. Cohort Studies
Cohort studies (and the case-control studies discussed
below) measure the association between the risk of disease
in individuals and their individual exposures to etiological
factors. In cohort studies, a population of individuals is
defined at the start of the study as being exposed, or "at
risk," and is then followed over time in order to observe the
incidence and timing of disease. A control population closely
similar to the exposed population except for the exposure is
established at the same time and followed in the same way.
After a long enough time, incidence of disease in the two popu-
lations is compared.
The cohort approach is often used when the exposure under
study is common. For example, with such risk factors as smoking
or air pollution, large cohorts can be readily identified.
However, when the number of exposed individuals is small, the
combination of a small cohort and a relatively uncommon outcome
II-6
-------
(i.e., some specific cancer) can considerably reduce the statis-
tical power of a study, and small-to-moderate associations
generally will not be detectable. Schlesselman (1974) has
shown that the sample size necessary to detect a twofold increase
in lung cancer among exposed individuals (with a statistical
confidence level of 95% that false positive results will not
be accepted and a statistical power of 80% that true associa-
tions will be detected) would require over 24,000 persons in
both the study and comparison populations. Such large sample
requirements often make it important that the power of a study,
particularly one with "negative" findings, be carefully eluci-
dated. Cohort studies are also subject to biases and confounding
factors, unless detailed information about the characteristics
and exposures of the cohort and control group is collected.
These problems are especially important in retrospective cohort
studies, i.e., studies in which a cohort is identified as it
existed at some prior time and its subsequent disease history
is compiled.
4. Case-Control Studies
Case-control (or case-referent) studies work in the opposite
direction from cohort studies (hence they are sometimes called
"trohoc" studies, which is cohort spelled backwards). Cases
(and appropriate controls) are identified, and an attempt is
made to discover the extent of prior exposure in both groups.
Case-control studies can usually be done much more quickly
(and much more cheaply) than cohort studies, particularly where
the disease (outcome) is rare. For relatively rare conditions,
II-7
-------
they are able to provide estimates of relative risk for exposed
vs. unexposed persons. They usually cannot provide estimates of
absolute risk or the magnitude of risk that follows from a given
exposure, although methods are being developed for estimation
of exposure-specific rates (Schlesselman 1982). Case-control
studies suffer from recall bias—i.e./ people are asked to
recollect exposures after the fact, and persons with a disease
may probe their memories more deeply or more imaginatively
in order to provide (for themselves) an explanation of their
illness. These studies are also subject to distortion as a
result of confounding and are very sensitive (especially in
their risk estimates) to the choice of appropriate controls.
A schematic for both case-control and cohort studies is
given below:
Exposure
Present
Absent
Total
Disease
Present Absent
a b
c d
nl n2
Total
ml
m2
N
In the cohort study one defines at the outset the popula-
tions m, and m-- After a suitable period of time, an observation
is made of a and c (b and d fall out automatically, by subtrac-
tion). The question is then asked:
II-8
-------
i.e./ is the proportion of cases among the exposed greater
than among the nonexposed?
In a case-control study, the comparison is usually made
of ^ (the "odds" that disease occurred in previously exposed
persons) with £ (the "odds" that disease occurred in previously
unexposed persons). The resulting "odds ratio," tr/g-
is an estimate of the relative risk to an exposed person.
It does not matter that n, could be all persons with the disease
(in a given hospital, say) and n2 a sample of all persons without
the disease. If the n2 persons are appropriately chosen, the
computation ?j— yields an unbiased result (Siemiatycki et al . 1981,
Schlesselman 1982).
5- Issues Arising in Studies of Cancer and Air Pollution
In succeeding sections, we review a number of epidemiolo-
gic studies in which the association between cancer and air
pollution has been investigated. The results of 46 of these
studies are summarized in tabular form in Appendix A (Table II-l).
Most of these studies have been of the descriptive or ecological
type, but there have been several major prospective cohort
studies (e.g., Hammond and Horn 1958, Hammond and Garfinkel 1980)
and several large case-control studies in which large samples
of lung cancer cases were compared to unmatched control popula-
tions (e.g., Haenszel et al . 1962; Dean et al . 1977, 1978). Many
of the studies were not designed specifically (or exclusively)
to investigate air pollution, and some merely provide evidence
on urban-rural differences in cancer frequency.
II-9
-------
Eight general problems arise frequently in the interpre-
tation of these studies and will be discussed summarily at
the outset.
a. Sites of action
Although some of the descriptive studies analyze data
on cancers at a number of sites, most of the detailed studies
are limited to lung cancers. The rationale for this focus
(where stated) is that the lung is the primary site of contact
with carcinogenic agents that may be inhaled from the ambient
air, that lung cancer is the primary effect of cigarette smoking,
that air pollution has components and characteristics in common
with cigarette smoke, and that some evidence exists to suggest
that air pollution may act to augment the effects of cigarette
smoking (see infra). Although all of these points have some
validity, there are several reasons to suspect that air pollution
may also act at sites other than the lung. First, air pollutants
(like cigarette smoke and other airborne carcinogens) come
into direct contact with other organs, including the upper
respiratory tract, the gastrointestinal tract, and the skin.
Second, cigarette smoking is associated with elevated cancer
rates at sites other than the lung, including the mouth, pharynx,
larynx, esophagus, pancreas, kidney, and bladder; indeed, for
every excess lung cancer in cigarette smokers, there is between
0.5 and 1.0 excess cancer at other sites (Doll and Peto 1981,
Wilson et al. 1980). Third, although the air pollutants that
result from incomplete combustion include components that are
11-10
-------
found in cigarette smoke, ambient air also contains many other
inorganic and organic carcinogens (see Chapter III below).
Some of these are known to cause cancer in humans at sites
other than the lung, including the skin, pleura, peritoneum,
hematopoietic system, central nervous system, liver, and bladder
(Althouse et al. 1980). Indeed, source-specific studies have
yielded some evidence for excess frequency of cancers in the
central nervous system, pleura, peritoneum, liver, lung, nasal
cavity, skin, and breast in residents living in the neighborhood
of industrial sources (for a review, see Section II-C below).
Fourth, there is a marked urban excess of cancer at a number
of anatomic sites, including sites not known to be affected
by cigarette smoking or other identified urban factors (see
Section II-E below). Finally, if air pollution acts to enhance
the effect of cigarette smoking, it might well be conjectured
that this enhancement takes place at sites other than the lung.
In principle, it would be desirable for these reasons
to review and analyze studies of cancer frequencies at all
sites where an association with air pollution might reasonably
be hypothesized. In practice, data to support such an analysis
are scanty and inadequate. Descriptive studies that suggest
excess cancers at other sites rarely control for smoking, and
there is not enough quantitative information on the effects
of smoking at other sites to attempt to subtract out its effects.
Accordingly, this review follows others in focusing on lung
cancer.
11-11
-------
Wilson et al. (1980) suggested that since cigarette smoking
causes about one cancer at other sites for each lung cancer, it
would be reasonable to assume that the same would hold for air
pollution. Hence, they estimated the total number of cancers
caused by air pollution by doubling his estimate for lung cancers.
Although this assumption is probably more reasonable than ignoring
other sites altogether, it is questionable for at least three
reasons. First, more precise analysis of cancers attributable
to cigarette smokng indicates that the ratio of excess cancer
at other sites to excess cancers in the lung is between 0.5:1
and 0.7:1 rather than 1:1 (Doll and Peto 1981, Tables 10 and 11).
Second, the dose-response relationships for airborne carcinogens
at different sites may differ, so that the ratio for excess
cancers at other sites to excess cancers of the lung observed
in cigarette smokers may be too high (or too low) for persons
exposed to lower concentrations of the same carcinogens. Third,
as pointed out earlier, ambient air contains a wider variety
of carcinogens than cigarette smoke, and many of them act at
sites other than the lung. Hence, Wilson et al.'s assumption
may understate the likely risks at other sites. However, the
epidemiologic data needed to investigate this hypothesis are
very scarce.
b. Nature and measurement of air pollution
"Air pollution" is a complex and variable mixture of agents
that exist in many chemical and physical forms. No single
measure of air pollution can suffice to characterize fully
11-12
-------
its potential to increase cancer risks. Unfortunately, most
of the quantitative measures of air pollution levels that are
available—particularly for the periods in the past when the
exposures likely to be most significant in causing current
cancers occurred—have been of conventional pollutants, such
as CO, SO2* hydrocarbons, NOX/ ozone, etc., which are unlikely
to be carcinogenic in themselves. These measures serve at
best as indirect measures of fossil fuel combustion or industrial
activity and may or may not be well correlated with ambient
levels of carcinogens. Other conventionally measured pollutants,
such as total suspended particulate matter or "smoke," include
products of incomplete combustion and are probably better cor-
related with at least one class of airborne carcinogens. How-
ever, neither these nor other available measures of air pollution
have any direct relation to emissions or ambient concentrations
of many of the inorganic carcinogens or industrial organic
chemicals listed in Appendix B (Table III-l).
Estimating air pollution exposure involves (l) selecting
an appropriate indicator of the carcinogenic potential of air
pollution and (2) estimating the levels of exposure to that
indicator. Ideally, one could then combine the contributions
of each pollutant known or suspected to be related to lung
cancer (see Appendix B, Table III-l). This would require a
detailed historical inventory of the substances present in
the urban atmosphere and their relative carcinogenic activity.
11-13
-------
Such information is not available. In its place, several indi-
cators of carcinogenic potential have been suggested. For
example, benzo(a)pyrene (BaP), a product of fossil fuel combus-
tion, has been used as a surrogate by several investigators.
The early choice of benzo(a)pyrene appeared to be reason-
able in that BaP has been found to be carcinogenic and is rela-
tively easy to measure. However, similar levels of BaP may
occur with wide variations in the levels of other carcinogenic
air pollutants. It has been shown that the relative quantities
of polynuclear aromatic hydrocarbons (PAHs) emitted from different
sources are not in a constant relationship to each other or
to that of BaP (Friberg and Cederlof 1978, Wilson et al. 1980).
The use of BaP as a quantitative predictor of risk is discussed
further in Chapter IV.
More recent work (Walker et al. 1982) suggests that it
may be possible to correlate health effects (lung cancer mortal-
ity) with the presence of mutagenic airborne materials. The
short-term mutagenesis tests, such as the Ames test, could
be used to evaluate the mutagenic potency of air samples.
This approach needs considerable development before it will
become practical.
There are also problems associated with attempts to monitor
exposure of the population to air pollutants. Monitoring is
often done from a single sampling station in a community, and
measurements are used to characterize the levels of various
pollutants in the surrounding census tract, city, or county.
Any extrapolation from monitoring data involves some error,
11-14
-------
but when data from a few stations are used for a large area
involving a diffuse population, the likelihood of substantial
error is greater.
To remedy this would require detailed data on environmental
release and behavior in relation to the size and characteristics
of the exposed populations. The work of Greenberg (1979) indi-
cates that the use of more refined estimates of exposure increases
the strength of the association between industrial air pollution
and lung cancer mortality. He found that total suspended particu-
late emissions, when corrected for land area and wind direction,
showed a much higher correlation with lung cancer mortality
than did the uncorrected emission figures.
The lack of information on cumulative exposure of individ-
uals to air pollution is also a problem. This is particularly
important with respect to cancer, in that incidence and mortality
are in general proportional to cumulative exposure for many
carcinogens (Schneiderman and Brown 1978). Only in situations
where a single measurement of the indicator substance is propor-
tional to the cumulative exposure to that material will the
estimated relationship reflect the true effects of air pollution.
Over the last 10 years, levels of many air pollutants have
been declining (CEQ 1980). If this decline has been uniform
throughout the country, then estimates based on current cancer
mortality (affected by past air pollution levels) would over-
estimate the role of air pollution. If, on the other hand,
air quality was improving in some areas while declining in
11-15
-------
others (or improving at different rates), the full effect of
air pollution would be underestimated.
c. Outdoor and indoor air pollution
Although the term air pollution usually connotes pollu-
tion of outdoor air, it has recently been recognized that human
exposure to many airborne pollutants is often greater indoors,
even in nonoccupational settings. Although systematic measure-
ments of indoor air pollution are scanty, it appears that ambient
concentrations are generally greater outdoors than indoors for
pollutants that .are emitted into or produced in the ambient air
(e.g., SO~, photochemical oxidants, and industrial chemicals),
but they are generally greater indoors for pollutants that
are released or concentrated indoors (e.g., cigarette smoke,
wood smoke, radon, formaldehyde, asbestos, and components of
consumer products) (for a recent review, see NRC 198la). Since
most people (other than outdoor workers) spend much more time
indoors than outdoors (Szalai 1972), indoor exposures are poten-
tially very significant. Two studies that indicated excess
frequencies of lung cancer in nonsmoking wives of smoking hus-
bands (Hirayama 1981, Trichopoulos et al. 1981; but see Garfinkel
1981 for conflicting data) suggest that indoor exposure, at
least to components of cigarette smoke, may be sufficiently
high to lead to measurable increases in cancer risk. In addi-
tion, two descriptive studies (Bean et al. 1982, Edling et al.
11-16
-------
1982) and a case-control study (Axelson et al. 1981) have sug-
gested an association between lung cancer and indoor exposure
to radon and its daughters.
In the absence of systematic monitoring or epidemiologic
studies of indoor exposure, it is only possible to speculate
about its likely contribution to the results of the epidemiolo-
gic studies reviewed in this section. For pollutants that
are generated outdoors, concentrations are frequently lower
indoors. For example, Wilson et al. (1980) estimated that
average BaP levels indoors would be only about 40% of those
outdoors, so that risks posed by BaP to the average person
would only be about 60% of those calculated on the basis of
outdoor levels. Hence, it seems reasonable to assume that
for these pollutants differences in exposure between polluted
and unpolluted areas would be reduced in magnitude in proportion
to the time spent indoors. For pollutants that are generated
indoors, it seems reasonable to assume that indoor concentrations
would be relatively independent of the degree of urbanization
and of the degree of industrialization, although they might
vary from one part of the country to another according to differ-
ences in climate and building characteristics. For both reasons,
we expect that indoor exposures would be more likely to dilute
than to enhance the effects of outdoor air pollution in leading
to geographic and urban-rural differences in air pollution.
However, direct study of this issue is needed to confirm this
expectation. One limited exception to this generalization
11-17
-------
is the indoor exposure of nonsmokers to cigarette smoke: To
the extent that smoking is (or was) more prevalent in urban
areas, urban nonsmokers might be at correspondingly greater
risk.
d. Latency period and trends in exposure
A complicating factor in studies of the association between
air pollution and cancer—as in all epidemiologic studies of
factors associated with cancer—is the long latency period
that usually elapses between exposure to carcinogenic agents
and the clinical manifestation of the resulting effect.
For most carcinogenic agents/ the minimum latent period
before excess cancers can be observed is 20-30 years and for
agents such as asbestos the effective latent period may be
45 years or more. This means that associations have to be
estimated between present cancers and exposures far in the
past. Unfortunately, systematic measurements of exposure to •
air pollutants were limited in extent and reliability in the
period when they were likely to have been most significant
in causing current cancers—the 1930s, 1940s, and 1950s.
A particular problem with air pollution is that its compo-
sition and distribution, as well as its intensity, have changed
since this critical period of interest. One major recorded
change is the reduction in concentrations of particulates,
smoke, and S0« in cities, which has resulted from the reduction
in the use of coal for space heating and the location of fossil-
fuel-fired power plants in rural areas (CEQ 1980).
11-18
-------
While this has resulted in a reduction in measured levels
of BaP, the primary indicator of incomplete combustion, it has
also led to a general reduction in urban-rural differentials.
Since the 1940s there has also been a massive increase in the
production of synthetic organic chemicals/ including volatile
carcinogenic compounds that can now be found in ambient air
(Davis and Magee 1979). However, this has been accompanied
by a general improvement in industrial hygiene, housekeeping,
and pollution control, and by substantial efforts to reduce
the emissions of agents known to be carcinogenic, such as as-
bestos and vinyl chloride. The consequence of all these changes
is that reductions in ambient levels of some carcinogenic agents
have been offset by increases in others, so that it is not
possible to determine even the direction of trends in the overall
risks likely to be posed by ambient air. However, it appears
likely that the early control of combustion sources means that
BaP is now less useful as a surrogate measure of the potential
carcinogenicity of ambient air, since its reduction has been
accompanied by the introduction of other (and more uniformly
distributed) pollutants.
e. Sex and racial differences
Most of the studies reviewed in this report have been
limited to (or focused upon) lung cancer in white males. In
principle, useful information could be derived from sex and
racial differences in cancer frequencies and patterns. For
example, lung cancer rates in black males are higher than those
11-19
-------
in white males, although the former smoke less; this suggests
that black males are either inherently more susceptible or are
exposed more to other carcinogenic agents. Also, urban-rural
differences in lung cancer rates are smaller in white females
than in white males, even when they are crudely matched for
smoking habits; this has been used to argue that the unexplained
differences must be due to occupational exposures in the males.
However, females also have substantial exposure to potential
carcinogens in the work place, and it has not been shown that
the difference in their exposure is sufficient to explain the
differences in their patterns of lung cancer. Another explana-
tion of this difference is that females spend more time indoors
in nonoccupational settings (Szalai 1972), so that they would
be less exposed to urban-rural differentials in outdoor air
pollution. A third possibility is that females are intrinsi-
cally less susceptible than males to carcinogens in the urban
environment, because of hormonal or other factors. Although
we comment on these and other features of some of the studies
under review, in general the analyses of data on blacks and
females have not been sufficiently rigorous to yield the precise
information that could be derived from them.
f. Confounding and effect modification
The most pervasive difficulty encountered in the conduct
and interpretation of epidemiologic studies reviewed here is
the control of confounding (Rothman and Boice 1982, Schlesselman
1982). In the present context, "confounding" refers to the
11-20
-------
influence of an extraneous variable that may wholly or partially
account for an observed effect of air pollution or may mask
a true association between air pollution and lung cancer.
A "confounding variable" is an extraneous variable that satisfies
both of two conditions (Schlesselman 1982):
1. It is a risk factor for lung cancer.
2. It is associated with exposure to air pollution, but
it is not a consequence of that exposure.
An obvious example of a confounding variable in epidemiologic
studies of lung cancer and exposure to air pollution is age. The
risk of lung cancer increases with age/ and sizeable differences
in the age distribution between "exposed" and "unexposed" groups
(or between cases and controls) could result in a spurious
association if the "exposed" group contained older individuals •
than the "unexposed" group. Similarly, if the "unexposed"
group contains older individuals than the "exposed" group,
an association may be masked. For these reasons, epidemiologic
studies of air pollution and lung cancer generally control for
age differences, either by stratifying data according to age
or by standardizing them to a reference population with a spe-
cific age distribution. Other risk factors for lung cancer
that may be confounding variables are cigarette smoking and
occupational exposures to certain chemical or physical agents.
Confounding can be controlled by separating the effect of
air pollution from the effect of confounding factors (Rothman
and Boice 1982). Three strategies can be used to accomplish
11-21
-------
this: (1) strict matching of "exposed" and "unexposed" individ-
uals or of cases and controls/ (2) stratification according
to levels or categories of the confounding factor/ or (3) multi-
variate mathematical modeling. Strict matching is rarely pos-
sible, especially when large studies are undertaken, and it
is employed only for certain case-control studies. With strati-
fication, the comparison of "exposed" with "unexposed" groups
(or of cases with controls) occurs within the various categories
of the confounding factor. In each stratum, the confounding
factor is set within a limited range so that the comparison
will not be significantly confounded. When confounding is
controlled by stratification, an overall measure of the effect
of exposure can be obtained by taking a weighted average of
the stratum-specific estimates. There are two basic ways of
combining such data (Rothman and Boice 1982): pooling and
standardization. An assumption underlying pooling is that
differences among stratum-specific groups are due to sampling
error. Standardization does not require such an assumption.
Stratification is often preferred to multivariate analysis
because it permits closer examination of the data by the inves-
tigator and it is easier for readers to interpret (Rothman
and Boice 1982). Multivariate analysis, on the other hand,
reduces the investigator's "feel" for the data; it involves
a set of mathematical assumptions about dose-response and related
relationships that can rarely be tested and verified; and its
results are often difficult to interpret in direct epidemiologic
terms.
11-22
-------
A further complication in the control of confounding is
the potential for interaction between a confounding variable
(such as cigarette smoking) and a study variable (such as a
measure of air pollution). If the effects of air pollution
were enhanced in the presence of smoking, smoking would be
an "effect modifier" for air pollution (and vice versa).
Effect modifiers are not true confounding variables/ and treating
them as such could bias the estimate of effect and hence the
conclusion about the nature and strength of an association.
In situations in which there may be several confounding
factors, stratification may not be practical and multivariate
analysis may be the preferred way to control several factors
simultaneously. In addition, multivariate analysis may include
various interaction terms in the event that some factors modify
the effects of the exposure under study. The multivariate
model can give an estimate of the importance of the interaction.
Thus, in the presence of interactions, multivariate analysis
may constitute a more rigorous tool than stratification, but
the results of such an analysis must be interpreted with care.
Most of the studies reviewed below employed stratification
and standardization to control for confounding, but no study
fully considered all potential confounding factors. Further-
more, a general limitation in these studies is the failure to
consider interactions between study and confounding variables
in a rigorous way.
11-23
-------
g. Study sensitivity
Several factors operate to reduce the sensitivity of many
studies. Migration tends to blunt distinctions. Small studies
are notoriously insensitive. For example/ Winklestein et al.
(1967), Dean (1966), and others, made computations on the basis
of a small number of cases (often less than five). Conclusions
based on such small numbers must be viewed with caution in
that the variability among cases can be large and a few cases
can substantially affect an apparent association. As with
the failure to control for potential confounding factors, this
could result in either an increase or a decrease in the observed
associations. Dean (1966) reported that in inner Belfast the
age-standardized lung cancer mortality rate for male nonsmokers
was 36 per 100,000 men. This conclusion was based on six cases.
The upper and lower 95% confidence limits on this estimate
(Table A-5 in Lilienfeld et al. 1967) are 78.5 and 13.2, respec-
tively. For male non-smokers residing in the "environs of
Belfast," a lung cancer mortality rate of 16 per 100,000 men
was calculated on the basis of one observed case. Upper and
lower 95% confidence limits on this estimate are 89.1 and 0.4.
h. Comparison populations
Rural populations are often used as "control" or comparison
populations. Rural residents are not without exposure to environ-
mental hazards, such as farm chemicals and pesticides. Indeed,
as pollution has become more widespread, the distinctions between
11-24
-------
exposed and unexposed populations have become blurred. Higginson
and Muir (1979, p. 1992) noted this complicating factor:
Often people assume that industrial and urban
environments are more heavily contaminated by such
agents as chemical carcinogens, mutagens, and pro-
moters, and that comparison with nonindustrial areas
should provide measure of their effect. However,
these comparisons are complicated by widespread
pollution by such chemicals as pesticides and herbi-
cides occurring in modern agricultural societies
as well as by behavioral and dietary variables.
Shabad (1980) recently made the same point/ noting the many
sources of atmospheric benzo(a)pyrene and its ubiquitous nature
in the environment. A recent analysis of cancer mortality
data led Greenberg et al. (1980) to hypothesize that factors
leading to environmentally induced cancer are diffusing and
are in turn leading to higher cancer mortality rates in parts
of the United States other than the historically high rate
areas of the Northeast and Great Lakes states. Blot and Fraumeni
(1982) have reported on the recent great increase in lung cancer
rates in both rural and urban areas of the southeastern United
States. The rates in the Southeast now exceed those in the
Northeast. Whether this is due to the rapid industrialization
of the Southeast during and following World War II (and the
possible concomitant increase in pollution) or to cigarette
smoking differentials (if there are any) is not at all clear.
It is thus unlikely that present urban-rural ratios provide
a full statement of urban excess relative to a pristine environ-
ment. Future urban-rural differences may be even smaller.
11-25
-------
C. Source-Specific Studies
The air in communities surrounding industrial point sources
has often been found to contain carcinogenic substances. Prom
this it has been anticipated that residents of such communities
would be at increased risk of developing cancer. The issue
discussed in this section is whether this risk is sufficiently
large to be significant and measurable.
This local type of pollution (point source, source-specific,
or neighborhood pollution) has been distinguished from pollution
of the general ambient air derived from diverse sources. For
example, Hammond and Garfinkel (1980, p. 207) stated:
General air pollution should be distinguished
from "neighborhood pollution" of fumes or particulate
matter from a factory or similar source. The effects
of this type of exposure may certainly increase
the risk of cancer in people living across the street
from a factory from which chemical or mineral conta-
minations are discharged. But the effects of such
risks for people living within several miles of
such factories has not yet been clearly delineated.
Many carcinogenic substances have been identified through
studies of work place exposure; of the 36 compounds or processes
that have been linked more or less strongly to cancer in humans,
23 are chemicals or processes identified in the work place
(Althouse et al. 1980). The impact of such substances may
be restricted entirely to the work place or may extend to the
surrounding communities. Community or neighborhood studies
are usually undertaken to see if they give results that are
consistent with worker studies. Attention has been drawn speci-
fically to studies of this kind that have reported associations
11-26
-------
of excess cancer with community exposure to arsenic, asbestos,
and vinyl chloride (EDF/NRDC 1980).
Ambient community exposure levels are likely to be consider-
ably lower than worker exposures, and the risks to individual
persons are expected to be correspondingly lower. However,
the differences in ambient concentrations are offset by several
other factors. Ambient exposure may occur over a longer period
of time (i.e., be of greater duration) than work place exposure.
The age at first neighborhood exposure may be considerably
lower than at first work place exposure. The population at
risk may be larger for ambient pollution than for work place
exposure, and may include more highly susceptible individuals.
Therefore, exposure levels that may have resulted in only a
few cancers among a small worker population could theoretically
lead to a substantial number of cancers among the larger (and
more diverse) populations exposed to ambient"pollution. However,
any such effects would be more difficult to detect in the general
population because of their low expected frequency and the
difficulty in controlling for other factors.
1. Arsenic
Several studies have shown that workers exposed to high
levels of inorganic arsenic are at an increased risk of develop-
ing lung cancer (Lee and Fraumeni 1969, Pinto et al. 1977,
Ott et al. 1974). Because of these findings, several investi-
gators have studied the risks to residents of communities in
which smelting and refining industries are located. To date,
11-27
-------
the evidence for an association between cancer and community
exposure to arsenic is mixed, witli some studies providing evi-
dence for increased cancer risks and others not. Blot and
Fraumeni (1975), Newman et al. (1976), and Pershagen et al. (1977)
have reported that residents in counties in which smelters
are located are at increased risk of developing cancer. Matanoski
et al. (1981) have reported that lung cancer rates are signifi-
cantly higher in areas near an arsenical insecticide plant.
However, similar increased risks were not found by Greaves
et al. (1981), Lyon et al. (1977), and Perry et al. (1978).
Blot and Fraumeni (1975) studied the distribution of lung
cancer mortality in 71 U.S. counties with primary smelting
and refining industries. Using the data compiled by Mason
et al. (1975), cancer mortality rates (for the period 1950-1969)
were calculated for the white population in each county. Data
on the possible confounding factors of population density,
percentage urban, percentage nonwhite, percentage foreign born,
median number of years of schooling, median income, and geographic
region were obtained from the 1960 census statistics.
A general linear, multiple regression model with adjust-
ments for confounding was used to test for differences in cancer
mortality between the smelting/refining counties and the remain-
ing U.S. counties. It was found that lung cancer mortality,
corrected for demographic variables, was significantly higher
among both males (17%, p<0.0l) and females (15%, p<0.05) residing
11-28
-------
in the 36 counties with copper, lead, or zinc smelting or refin-
ing operations than in counties without these operations.
This excess was found in all counties, independent of population
size, but the magnitude of the excess was lower in the more
populated, urban areas. The authors concluded that these "findings
suggest the influence of community air pollution from industrial
emissions containing inorganic arsenic."
EPA (USEPA 1978) subsequently reanalyzed Blot and Fraumeni's
data after eliminating the four counties containing only refin-
eries. This recalculation did not substantially alter the
results.
Newman et al. (1976) studied the incidence and histologic
types of bronchogenic cancer occurring among residents of Butte
and Anaconda, two communities close to the Anaconda Copper
Company smelter in Montana. Using data from the Montana State
Register and the U.S. Census, incidence rates for lung cancer
during 1969-1971 among men and women residing in Butte and
Anaconda were calculated. These figures were compared to state-
wide incidence rates for all of Montana. It was found that
the incidence of cancer of the bronchus and lung was signifi-
cantly (p<0.01) elevated among men in both Anaconda and Butte,
and among Butte women (p<0.00l). Three respiratory cancer
cases were found among Anaconda women, an incidence greater
than the expectation but not statistically significant. When
Newman et al. (1976) calculated the incidence of respiratory
cancer among Anaconda women for a 10-year period of observation, "
11-29
-------
they found that the Anaconda rate of 2.9 cases/10 persons
was significantly higher (p<0.05) than the state rate of 1.4/10 .
However/ this study did not control for smoking habits or for
occupation, so it is not clear that the elevated rates were
attributable to exposure via the ambient air.
Histological slides were available for 143 cases of lung
cancer diagnosed between 1959 and 1972. These slides were
re-evaluated by a panel of pathologists, and information on
occupation, residence, and other factors was obtained for each
case. Information on smoking habits was also obtained, but
only for 41% of the cases. The distribution of histological
types among four groups (copper smelter workers, copper mine
workers, "other" men, and women of Butte) was examined. Newman
et al. (1976) reported a high percentage of poorly differentiated
epidermoid carcinomas among smelter workers. This finding
was consistent with similar reports of excess lung cancer of
this histological type among smelter workers (Lee and Fraumeni
1969) and patients receiving arsenic medication (Novey and
Martel 1969). Poorly differentiated epidermoid carcinomas
were also the predominant histological type in female residents.
Newman et al. concluded that arsenic must be strongly suspected
as the etiologic agent of excess cancer in both the smelter
workers (males) and in females in the general Butte and Anaconda
populations. However, well-differentiated epidermoid carcinomas
were the predominant type in male residents of Butte and in
miners, and Newman et al. suggested that these might have resulted
11-30
-------
from exposure to a specific type of friable sanding material
used on the city streets during the winter months.
Pershagen et al. (1977) studied the mortality from different
causes in an area surrounding the Ronnskarsverken smelter works
in northern Sweden. A reference population with similar degree
of urbanization, occupational profile, fraction of population
working, and geographic location was chosen. For these two popu-
lations, causes of death over a 14-year period (1961-1974) were
extracted from the National Registry on Causes of Death. The
age structure of each population was derived from the national
censuses of 1960, 1965, and 1970. The standard mortality ratio
(SMR) for lung cancer among males in the exposed population
surrounding the smelter works was significantly (p<0.0l) elevated
when compared to that of the reference population, but it was
not significantly higher than national rates. Closer examination
by Pershagen et al. of the 28 males with primary respiratory
cancer revealed that 15 had been employed at the Ronnskarsverken
smelter. Excluding these individuals, a nonoccupational SMR
of 173 was calculated, which although greater than 100, was
not significantly greater than national rates (p<0.05). However,
the reported difference (13 observed vs. 7.5 expected) is actually
statistically significant (Z = 2.01, p<0.05) when analyzed with
a one-tailed test. Female lung cancer rates in the Ronnskarsverken
area (relative risk = 1.08) were not significantly different
from the national or comparison population rates.
11-31
-------
Matanoski et al. (1981) studied cancer mortality among
residents of an area surrounding an arsenical insecticide plant
in Baltimore. A significant excess of lung cancers was observed
among males relative to a comparison population matched for
race, sex, age, and socioeconomic status. These comparisons
were based on 25 lung cancer deaths. The excess in lung cancer
remained when two cases of lung cancer death among plant employees
were removed. The remaining cases were distributed in an area
lying north and east of the plant. This area had the highest
levels of arsenic in the soil, which tends to confirm that
exposure occurred. No significant excess was found in females.
Lyon et al. (1977, 1978) investigated the incidence of
lung cancer in communities surrounding a copper smelter near
Salt Lake City. They identified all new cases of lung cancer
during 1969-1975; and all new cases of lymphoma were used as
a control. Using individuals' addresses at the time of death
or diagnosis, cases and controls were grouped according to
their positions in relation to the smelter. There were no
significant differences in the numbers of cancers between cases
and controls at any specific distances from the smelter.- The
observed numbers of cases within four zones classified by distance
from the smelter were all close to those expected. The authors
concluded that these findings were not consistent with previous
reports of increased rates of lung cancer among persons living
near smelters. Salt Lake County, in which the smelter was
located, was the only county in the study by Blot and Fraumeni
11-32
-------
(1975) which did not show an elevated lung cancer rate (Stellman
and Kabat 1978). This was possibly due to a low prevalence
of smoking in this predominantly Mormon area. However, neither
study could control for smoking habits (Lyon et al. 1977).
Greaves et al. (1981) studied the incidence of lung cancer
in ten communities surrounding nonferrous smelters. For the
majority of these counties, the SMRs for lung cancer exceeded
100 (the range was 46-246). The authors identified all lung
cancer cases (using as controls all cases of three other types
of cancer: breast, prostate, and colon) occurring between
1970 and 1977 within a 20 km radius of each smelter. Using
addresses for each reported case at the time of death or diag-
nosis, the distance of the residence from each smelter was
calculated. The authors concluded that there was no relationship
between distance from the smelter and the incidence of lung
cancer.
Rom et al. (1982) conducted a study of communities surround-
ing a nonferrous smelter at El Paso, Texas, using a methodology
similar to that of Greaves et al. (1981). Cases of lung cancer
(413 males, 162 females) reported between 1944 and 1973 within
20 km of the smelter were compared with controls (376 males
with prostate cancer, 114 females with breast cancer) after
classifying the address of last residence according to distance
from the smelter. No statistically significant differences
in the distribution of cases and controls were found. The
11-33
-------
authors stated that this conclusion was unaltered after control-
ling for race, type of surname, age, and sex.
Both the positive and negative studies cited above have
been criticized (Nelson 1977, Stellman and Rabat 1978, USEPA
1978), and the validity of the positive studies has been vigor-
ously debated in comments submitted by ASARCO (1980), Air Products
and Chemicals Corp. (1980), AIHC (1981), and by the same parties
and by Kennecott on drafts of this report. A major limitation
of both the positive and negative studies is the failure to
control for smoking habits. Only two studies (Pershagen et
al. 1977, Matanoski et al. 1981) controlled for occupational
exposure to arsenic, and only the study by Blot and Fraumeni
(1975) controlled for urbanization and population density.
Only the study by Matanoski et al. (1981) included even an
indirect measure of exposure; both the negative studies and
the other positive studies incorporated populations with a
wide range of potential exposures, including populations upwind
of the smelters. Other criticisms directed at the positive
studies include their use of national cancer rates rather than
state or local rates to calculate expected cancer frequencies
and possible diagnostic biases. Thus, both the positive and
the negative results are subject to considerable uncertainty.
The case-control studies of Greaves et al. (1981) and Rom et
al. (1982) are methodologically superior to the ecological
studies conducted by others, but the use of other types of
cancer as controls in these studies is subject to question.
11-34
-------
2. Asbestos
A large number of investigators have demonstrated that
occupational exposure to asbestos results in an increased risk
of lung cancer, pleural and peritoneal mesotheliomas, and gastro-
intestinal cancers (IARC 1977). The indestructibility of this
material, its wide use, and (at least in the past) large indus-
trial emissions make it reasonable to hypothesize that such
risks extended beyond the work place. This is a particularly
suitable example for study because two of the diseases associated
with asbestos exposure (pleural and peritoneal mesotheliomas)
are extremely rare in persons without exposure to asbestos,
so that they serve as markers for asbestos-induced disease.
Several studies have reported apparent clusters or excesses
of mesotheliomas in the vinicity of asbestos factories, mills,
or mines. Newhouse and Thompson (1966) studied a series of
83 patients at the London Hospital with a diagnosis of mesothe-
lioma in order to determine the extent (if any) of asbestos
exposure. Full occupational and residential histories were
obtained for 76 of these patients. Using 76 patients from
the same hospital suffering from other diseases as controls,
it was found that a significantly greater number of mesothelioma
patients (p<0.0l) with no evidence of occupational or domestic
exposure lived within one-half mile of an asbestos factory.
This study has been criticized (AIHC 1981) for the choice
of comparison groups. The controls, although matched for date
of birth and sex, differed from the mesothelioma cases in that
11-35
-------
all were admitted to the hospital during 1964 while the mesothe-
lioraa cases were admitted between 1917 and 1964. This could
be a source of bias because exposure conditions might have
changed considerably between 1917 and 1964. Such biases would
be expected to have reduced rather than increased the reported
association because the greatly increased use of asbestos would
have made general population exposure to asbestos more common
in 1964 than 1917, thus leading to greater potential for exposure
in the controls than in the cases. The authors stated that
there was no evidence that the controls were less likely than
the study group to have worked in contact with asbestos or
to have lived in close proximity to asbestos factories. However/
the basis for this conclusion is not clear, especially for
the persons who had died long before the study was conducted.
Wagner et al. (1960) reported on 33 cases of diffuse pleural
mesothelioma that were observed in South Africa during the years
1956-1960. All but one of the cases involved probable exposure to
crocidolite asbestos as a result of occupational exposure (4 cases)
or residence near the Cape asbestos mine fields (28 cases).
The authors reported that during the same period of time, diffuse
pleural mesothelioma was rarely diagnosed in other (nonmining)
areas of South Africa.
Although this study had no concurrent controls, the occur-
rence of diffuse pleural mesothelioma appears to be a sufficiently
rare event that the results would undoubtedly be statistically
significant if the population rates could be computed. Air
11-36
-------
Products and Chemicals Corp. (1980), in a critical review,
raised the question of whether natural outcroppings and weathering
of ore bodies could have been the source of asbestos exposure
rather than mining activities. However, in either case it
seems likely that airborne asbestos was the causative factor.
According to Bohlig et al. (1970), Dalquen et al. (1969)
reported an increased incidence of mesothelioma in the neighbor-
hoods surrounding an asbestos processing factory in Hamburg,
West Germany. Dalquen et al. (1969) reportedly found that
while the total incidence of mesothelioma among the general
population was 0.056% for the years 1959-1969, the incidence
in the residential area near the factory was 0.96%. However,
no test of statistical significance was reported. There are
also several case reports (Tayot et al. 1966, Bohlig et al. 1970,
Stumphius 1969, Wagner 1971, and Tabershaw et al. 1970) of
what appear to be environmentally related cases of mesotheli-
omas among residents in neighborhoods near shipbuilding areas.
Hammond et al. (1979), in the largest of the neighborhood
studies, examined the mortality of residents in the vicinity
of an asbestos factory in Riverside, a district in Paterson,
New Jersey. From city directories for 1942-1954, all male
residents of Riverside and Totowa, a second neighborhood that
served as the control, were identified. These individuals
were traced until 1976. During the period 1962-1976, no signi-
ficant differences were noted in total deaths: 780 (43.8%)
of Riverside subjects and 1735 (46%) of Totowa subjects had
11-37
-------
died. Specific causes were cancer at all sites [163 (9.2%)
vs. 353 (9.4%)] and lung cancers [41 (2.3%) vs. 98 (2.6%)].
One pleural mesothelioma in a Riverside male was reported in
1966. Although this single case is not sufficient to support
the hypothesis generated by the case reports, the duration
of follow-up may not have been sufficient to have detected
environmentally related mesotheliomas. Newhouse and Thompson
(1966) found that the mean length of time between first exposure
and death for mesothelioma cases living in the neighborhood
of an asbestos factory was 48.6 years (vs. 29.4 for factory
workers).
Although the most extensive study was thus inconclusive,
the rarity of mesotheliomas in individuals not exposed to asbes-
tos gives considerable weight to the less well controlled studies
and case reports of mesotheliomas among residents in neighbor-
hoods surrounding asbestos mines and factories. However, these
studies yielded no specific evidence for exposure other than
location of residence. Environmental exposure to asbestos
also results from other activities (e.g., wearing out of brake
linings in automobiles). In one study of urban dwellers, nearly
all (96%) had asbestos fibers in their lungs (Churg and Warnock
1977). This suggests that asbestos from diverse sources, parti-
cularly airborne asbestos, may be an important problem requiring
additional study.
11-38
-------
3. Vinyl Chloride
Cases of the rare cancer, angiosarcoma of the liver (ASL),
have been reported among individuals living near vinyl chloride
fabrication, or polymerization, plants. Brady et al. (1977)
studied the cases of ASL reported to the Tumor Registry of
the Cancer Control Board of the New York State Department of
Health during the years 1958-1975. For each of these cases,
a matched control with an internal malignant tumor other than
primary liver cancer was selected from the registry. Cases
and controls were matched on age (same 5-year age group), race,
sex, county of residence, and vital status. Relatives of both
the subjects and matched controls were interviewed in order
to obtain information on potential exposure to vinyl chloride
(VC), arsenic (As), or thorium oxide (ThCO, as well as medical,
familial, residential, and occupational histories. Of the
26 cases of ASL diagnosed during 1958-1975, 7 had direct exposure
to VC, As, or ThO2 (p<0.02). Of the remaining 19, 5 lived
within one mile of a VC fabrication or polymerization plant.
Although this is suggestive of an association, no statistical
test of the possibility of this finding arising by chance was
reported. Owing to the small number of cases and the lack
of monitoring data directly demonstrating exposure, no firm
conclusions are possible.
Infante (1976) studied the mortality patterns of residents
of four Ohio communities with polyvinyl chloride (PVC) production
facilities. Using data for the Ohio white population as the
11-39
-------
standard, SMRs were calculated for central nervous system (CNS)
cancer, leukemia and aleukemia, and lymphomas. He found that
in these four communities the number of observed CNS cancers
for both sexes combined during 1958-1973 was significantly
greater than that expected (38 observed vs. 24.07 expected;
p<0.00l). SMRs were also calculated for each of the counties
excluding the areas surrounding the PVC facilities, but no
significant excesses were found.
This study was reviewed by Air Products and Chemicals
Corp. (1980), which commented that interpretation of this study
is complicated by the fact that (1) the increase in CNS tumors
was observed primarily in males and (2) most of the excess
occurred in one part of the study area (Painesville). They
argued that these factors seriously challenge any conclusions
that vinyl chloride is associated with community cancer risks.
To these criticisms should be added the study's failure "to
control for occupational exposure, race, and socioeconomic
status.
Infante (1976) has also been criticized by the Society
of the Plastics Industry (1980) for including North Ridgeville
in the study group, while not including other cities located
as close or closer than North Ridgeville to the PVC facilities
(e.g., Mentor, Ohio). If North Ridgeville is excluded from
the study group, the excess in CNS tumors remains significant
(p<0.05, one-sided test), however.
11-40
-------
4. Petrochemical and Other Chemical Emissions
A number of studies have indicated that workers exposed
to a wide range of industrial chemicals are at increased risk
of developing cancer (Althouse et al. 1980). An increased
risk of bladder cancer has been reported among workers exposed
to benzidine (Case et al. 1954) and paints (Cole et al. 1972).
Exposure to polycyclic aromatic hydrocarbons (found in crude
petroleum, catalytically cracked oils, soot, and other pyrolysis
products) has been associated with increased incidence of cutaneous
and pulmonary cancers in workers (Doll et al. 1972, Lloyd 1971,
Hammond et al. 1976, Fraumeni 1975).
Blot et al. (1977) studied cancer mortality patterns for
1950-1969 in the U.S. counties where the petroleum and petro-
chemical industries are most heavily concentrated. Using methods
similar to those of Blot and Fraumeni (1975) described above,
it was found that male residents of these counties experienced
significantly higher rates for cancers of the lung, nasal cavity
and sinuses, and skin compared to male residents of counties
with similar demographic characteristics but with no petroleum
industry. Lung cancer rates for white females in petroleum
industry counties were also significantly elevated. Owing
to the lack of information on occupation and smoking, however,
the specific reasons for these associations are ambiguous and
somewhat debatable. Similarly, the causes of increased mortality
rates for cancer of the bladder and liver among males and females
(increased lung cancer mortality for males only) in U.S. counties-
11-41
-------
with chemical industries are not identifiable without additional
data. However/ the finding of increased rates for both males
and females suggests that factors other than occupational expo-
sures are likely to be involved. Blot et al. (1977) noted
that if occupational exposures in males and females were solely
responsible for these increases, the worker risks would be
substantially above those of the general population and should
be easily detectable.
In a similar large-scale study conducted in Japan, Minowa
et al. (1981) reported associations between lung cancer rates
in males and the proximity of oil refineries, metal refineries,
steam power plants, coal mines, lignite mines, and fishing
ponds. These associations were controlled for urbanization.
However, like the studies of Blot and Fraumeni (1975) and Blot
et al. (1977), these associations were correlational only,
were not controlled for occupational exposures, and were incom-
pletely controlled for smoking.
Capurro (1979) studied the mortality experience of a popu-
lation of 117 people exposed to solvent vapors from a chemical
plant for more than 5 years. These individuals were followed
for a 6-year period (1968-1974). During this time there were
14 deaths (vs. 6 expected), 7 of which were due to cancer.
In particular, there were 4 cases of lymphoma (3 reported on
death certificates). The ratio of observed to expected deaths
(based on Maryland death rates) was 3.0/0.0187 = 160. The
incidence of new cases of cancer of the larynx was also elevated
11-42
-------
61-fold (2 observed vs. 0.033 expected on the basis of incidence
rates from the Connecticut Tumor Registry data). These high
relative risks are based on few cases, and the authors noted
that all 4 individuals with lymphoma were previously employed
at a paper mill that closed in 1948. Questions also remain
on the nature of the study population and the suitability of
using state rates for comparison, particularly because two
different sets of rates—Connecticut for incidence and Maryland
for mortality—were used.
Hearey et al. (1980) compared estimated age-adjusted cancer
incidence rates (1971-1977) among Kaiser Foundation Health Plan
(KPHP) members living near petroleum and chemical plants in the
Contra Costa area of the San Francisco Bay region to incidence
rates among KFHP members living in the remainder of the bay
area. Comparisons of rates for the two areas showed no evidence
of increased cancer risk in KFHP members in the area near the
plants. However, questions remain on the composition of the
study population and whether the individuals enrolled in the
KFHP were representative of the entire Contra Costa population.
It is unclear whether the controls were suitable for studying
the relationship between industrial emissions and cancer.
No adjustments were made to account for possible differences
in occupation, duration of residence, socioeconoraic status,
and smoking; and it is not clear from the written report that
the study was controlled for race. There is also some question
whether there were sufficient differences in potential exposure
11-43
-------
levels between study and comparison populations to produce
an effect large enough to detect.
5. Steel Manufacturing
Elevated rates of cancer have been reported in counties
where steel is manufactured. Perry et al. (1978) reported
that among the female residents of Johnstown, Pennsylvania,
the age-adjusted mortality rates of several types of cancer
(oral, respiratory, breast, urinary, central nervous system,
and peritoneal and other digestive system cancers) were signi-
ficantly elevated over those of residents of the county living
outside Johnstown. Rates in men, with the exception of digestive
system cancers (and breast cancer), were also elevated in the
community. Carnow (1978), in examining data from Allegheny
County, Pennsylvania, and Lake County, Indiana, which are large
steel production areas, also found increased lung cancer mortality
rates among both males and females. Cecilioni (1972, 1974)
analyzed the cancer mortality rates in Hamilton, Ontario, a
steel manufacturing city, from 1966 to 1970. He found the
highest rates in districts close to the steel mills. Similarly,
Lloyd (1978) found significantly elevated lung cancer rates
among male residents living near and downwind of a Scottish
steel foundry in Scotland. This clustering could not be wholly
accounted for by cigarette smoking or occupation.
11-44
-------
D. Migrant Studies
This section summarizes several studies that have reported
differences in site-specific cancer rates between native and
foreign-born populations in South Africa, New Zealand, and the
United States.
Haenszel (1961) found that mortality from lung and bronchial
cancer was higher for English and German immigrants to the
United States than for native Americans, but lower than the
rates in their countries of origin. The results suggest that
immigrants bring some of their greater liability to cancer
with them, possibly because of living conditions experienced
earlier. Yet, by leaving their native countries, they lose
some of the still greater risk existing among people remaining
at home. This might imply that migration involves reduction
in exposure to some "native" carcinogens. Dean (1964) observed
that the lung cancer rates for British subjects migrating to
South Africa were intermediate between those of native-born
South Africans and those of British subjects who remained in
Great Britain. Eastcott (1956) found that immigrants from
the United Kingdom had a 35% higher risk of contracting lung
cancer than native New Zealanders if they came from the United
Kingdom before the age of 30, and a 75% higher risk if they
migrated after the age of 30. The per capita consumption of
cigarettes was higher in New Zealand and South Africa than
in the United Kingdom. Differences in smoking habits are,
therefore, not likely to account for these findings.
11-45
-------
Among Norwegians living in Norway, where air pollution
levels are generally low, the lung cancer rate is also low.
Among the U.S. urban populations, where air pollution levels
are higher, the rate is twice as high. For Norwegians who
have migrated to the United States, the rate is midway between
these (Reid et al. 1966).
In a study of male residents of Cuyahoga County, Ohio,
the risk of lung cancer for Italian immigrants was found to
be lower than that for U.S.-born residents and similar to the
rate in their native country. Immigrants from England and
Wales showed a lung cancer mortality that was similar to the
rate for natives of the United States but lower than the rate
for their peers in England and Wales (Mancuso and Coulter 1958;
see also Mancuso and Sterling 1974). Adjustments for smoking
were not made.
These studies of migrants suggest that early environmental
exposure (in addition to smoking) is important in determining
the risk of lung cancer later in life. In most of the studies
discussed, the frequency of lung cancer among migrants is inter-
mediate between the rates in the original country and the adopted
country. The epidemiologic evidence that risk is higher for
migrants from countries with high pollution levels (and lower
for migrants from countries with low pollution levels) is con-
sistent with the hypothesis that polluted air is a contributing
factor in the etiology of lung cancer.
11-46
-------
If it can be assumed that the exposure of emigrants from
a particular country is representative of the general popula-
tion exposure, these findings would indicate that long-term
exposure to ambient air pollutants increases an individual's
risk of lung cancer. However, there are several problems with
the interpretation of these studies. First, it is not clear
that the statistics on cancer rates in the different countries
and on persons of different national origins in the same country
were collected in the same way and were rigorously comparable.
For example, in most studies cancer rates for immigrant commu-
nities were compared with national rates in their native and
adopted countries. Second, none of the studies was controlled
or even stratified for smoking habits, occupation, socioeconomic
status, or urbanization in the country of origin. Migrants
constitute self-selected populations that have experienced
unsatisfactory conditions in their country of origin; it "is
a matter of conjecture to what extent these conditions may
have involved occupational exposures, residence in polluted
areas, or other factors that may have increased their cancer
risks. Third, none of the studies reported actual measures
of the air pollution levels to which the population groups
were exposed, either in their country of origin or their country
of adoption. Although it is a reasonable hypothesis that air
pollution levels were generally low (in the relevant period
prior to 1940) in New Zealand, South Africa, and Norway, inter-
mediate in the United States, and high in Great Britain, there
11-47
-------
were presumably overlooked variations in exposure within each
country. Thus, although these studies are consistent in sug-
gesting that migrants from one country to another carry part
of their risk (or lack of risk) with them, the studies do not
permit rigorous tests of the hypothesis that early exposure
to air pollution was a critical factor contributing to this
risk.
E. Urban-Rural and Other Geographic Studies
1. Introduction
Geographic patterns of cancer have been studied more exten-
sively than specific industrial emissions. Of particular rele-
vance to the problem of air pollution and cancer is the compar-
ison between cancer rates in polluted areas and those in nonpol-
luted areas.
Many such comparisons have been made, both directly and
indirectly. For nearly all monitored pollutants, urban areas
have higher levels of pollution than rural areas. If the common
constituents of air pollution increase the risk of developing
cancer, it would be expected that cancer rates in polluted
areas would be higher than those in areas with relatively little
pollution (all other factors being equal). When rates in urban
areas are compared to rates in rural areas, this relationship
is observed. A number of investigators (Table II-l, Appendix A)
have reported that for lung and other forms of cancer, incidence
and mortality rates are higher in urban areas than in rural
areas. For example, Table II-2 summarizes data on age-adjusted
cancer mortality rates in the United States between 1950 and 1969.
11-48
-------
TABLE II-2
URBAN-RURAL COUNTY RATIOS OF U.S. AGE-ADJUSTED
CANCER MORTALITY RATES, WHITE POPULATION, 1950-1969
Male
Site
Esophagus
Larynx
Mouth and Throat
Rectum
Nasopharynx
Bladder
Colon
Lung
All Malignant
Neoplasms
Urban-Rural
Ratio
3.08
2.96
2.88
2.71
2.17
2.10
1.97
1.89
1.56
Female
Site
Esophagus
Rectum
Larynx
Nasopharynx
Lung
Breast
Bladder
Other Endocrine
All Malignant
Neoplasms
Urban -Rural
Ratio
2.12
2.11
1.92
1.66
1.64
1.61
1.58
1.52
1.36
SOURCE: Goldsmith (1980), Table 1, p. 206
11-49
-------
The ratios between overall rates in counties classified as
urban and in those classified as rural were 1.56 for all malig-
nant neoplasms in males and 1.36 for all malignant neoplasms
in females; these ratios exceeded 1.5 at 10 individual sites
(Goldsmith 1980). Table II-3 summarizes data from six studies
of lung cancer mortality in the U.S. in the period 1947-1951.
The urban-rural ratios observed in these studies varied between
1.2 and 2.8 (Shy and Struba 1982). Table II-l (in Appendix A)
summarizes the results of 44 other studies, of which at least
39 reported higher rates of cancer in urban and/or industrialized
areas than in rural and/or nonindustrialized areas.
So consistent are the findings of an urban-rural difference
in cancer risk that no one seriously questions their validity,
and most researchers speak of an "urban factor." However,
when different researchers have tried to explain this urban
factor or other geographic differences, disagreements have
arisen. Potential risk factors in addition to air pollution
used to explain the differences include smoking patterns, occu-
pational exposures, population density, life-style, socioeconomic
differences, and/or several other factors. In the following
sections, we review the evidence for air pollution as a factor
associated with geographic variations in cancer rates.
2. Air Pollution as a Factor in Geographic Variation in
Cancer Rates
It is a plausible hypothesis that air pollution is respons-
ible for some fraction of the urban factor or other geographic
variations in cancer. As will be discussed in Chapter III,
11-50
-------
n
I
M
M
H
0
EH
(O
U
s
EH
(0
Q
U
EH
M
D
U
E
EH
M
i
«
o
E
a
|
u
i
3
o
53
O
tj
0
k_l
DISTR]
u
EH
55
M
05
g
fa
.
*j
m
§
D
r i
E
(0
0)
fi
^
C rH O
A) AJ -H
,Q 0 >H 4J
rf 4-> 3 AJ
D K K
m
£•£
4-1 O>
•H 0)
38
O rH
S A)
h
W 3
0) «
O
C -O
AJ C
U A)
01 C
C AJ
3X5
D
,_4
AJ - fi
U CO 0
•H M -rt
£ AJ 4-»
ft 0) AJ
cn AJ ft
0 4J 0
0) AJ CU
u o
*w
. . o
TJ Q) (D
3 M 5>
4J 3 <
0
0
in
CM
C
AJ
.a
°
GO
CM
CM
in
o
rH
n
0)
L
fl) ,Q
4-1 H
•H D
*
in
rH
g
AJ
•H
•H
rH
•H
O
•o
A3
C
A)
e
>H •
M-4 CO
0 •
KD
C
O
>r*
H^
AJ
rH
CM
CM VO ^
rH rH
m
o
"•v.
n
0)
1
A) **H id AJ
3 S H 3
M C D «
25
V
4J
0)
3
•0
AJ
ft
in
M rH n CO
• • * •
H n oo r-
n cs eg rH
m
CO 05
m
m
o
n
0)
(0
c
AJ 4J
4->
. -H fi
+ IH id
O 0 4J
O Ol-rl
O 0 rH Q
» M O AJ
O 4J D4 OJ
m oj o M
S n AJ
09 4->
tt) r» OJ rH
•H fl) g AJ
•P A3 C Vl
•H JJ 0 3
u 5 a «
rH CM CO
in
o
n
o>
0)
rH
At
m
c
C 0
AJ -P
+ rH AJ
O O 4-»
CO CL *r4
O 0 rH (0
» IH O AJ
O 4J ft Q)
in a) o
3
A)
0)
0) IH 0) rH
(1)
CM n
*—• >o
m o)
in 4-1
{J^ CO
•H 3
—^ T1
•o
AJ
rH
A3 0)
01
4J 4J AJ
tt) 3 *-*
O rH
•o -H m Q
rH *> CA Q)
0 O ^H O<
* 0) I AJ
a c r-
•H C ^" rH
U O 0> rH
O U-t <
4J
o
a)
r(
•H
•O
O
C
(U
a) ,a
IH AJ
AJ IH
A)
tn CM
CM CO
to
n
4J -H
C -P
3 C
W
Q)
0 3 -H
U 0 -P
U
C
A)
C
3
C O
A) U
0 3 A3
ft XI
O M >H
M OJ 3
4-> ,C C
rH CM CO
n
s~*
in
in
r-l
•
^H
AJ
4J
0)
0
n
3
0 0
C -H
A>
0)
vO
m
CM
V
a
0)
rH rH
m AJ
O\ G
rH
1 0)
r^ 4J
^* iH
11-51
-------
•o
0)
3
"••4
4J
C
8
*-*
4J
O
53
C rH O
(0 (0 -rt
Xl O U 4J
rl 4J 3 (0
D « «
n
>i 0)
Xl -H
i .
M
>i O
4-1 cn
•H 0)
rH 4J
It) (0
4J U
^
0 rH
S 10
U
H 3
ft\ f£
U/ ^M
U
C -O
10 c
U at
Cn C
C id
3. XI
t-3 M
O
rH
10 - C
U CO 0
•H M -H
fl (0 4J
ft 0> *l rH
M 3
Cn nj ft
O 4J O
0) (0 O*
O Q
H-t
- - o
>i id
T3 0) 0)
3 H Cn
+
0
0
m
- c
CN O
•iH
II 4J
id
C rH
10 3
X) ft
3 &
•— ' M4
r>
«
CN
00 rH
CN CN
CO rH
in
0
rH
CO
0)
t i
10 C rH
O. irH
4J 3
>1-H ft
4J O O
C ft
3 id
O
O
in
CO
>
+
O
C O
- C
tN O
•H
II 4J
(0
C rH
(0 3
CO
>
CN
(N
•a
0)
4J
0)
3
•p-i
•O
(0
(U
Cn
(0
m co
cu CM 0)
rH rH cn
13 i 10
C • cr>
•H C0<* rH
H • CM rH
ft DrH rtj
11-52
oo
•
tN
CO
•
0)
C -H W
3 4-> 0)
O C -H
O 3 4J
^ o c
•< 0 3
CO O
£ < 0
CQ CO
rH § CO
4J O K
o o a
0)
rH
• • • (0
rH CN CO S
tN CM CO r<
14 3 * +
3 M CO CO
0) 0) C C
•H -H 3 3
4-) 4-1 O 0
C C U U
3 3
O O C C
O O fO (d
4-> 4-1
C C "H -H
«3 10 rH rH
4J 4J O O
•H -H ft ft
rH rH 0 O
O O I* V4
ft ft-P 4->
O 0 0) <1)
M H g g
4-* 4J C C
CD 0) O O
s s a a
* • * •
rH CN n •*
in
o
rH
^v.
(Q
0)
4J
&
9)
rH
(0
B
0)
£
CN If!
• C i
n n fi F
C r
JS!
C rH 14 !
10 io r? s
XI ~<
»H 5
3 r< f) i
d> '
* <^, 1V"* t
CO 05 i> 4
0) 0) C t
•IH ••-) 3 r
4J 4J O (
C C 0 (.
3 3
o o c c
O O (0 ft
4J 4.
C G°*'<
10 (0 F"! '-
4J 4J 0 C
•H -H ft C
«H rH 0 C
O O M ^
ft ft4J 4.
o o o> a
M M g E
4-» 4J c r,
n I <0
•H CM
CU
••3
(U CM
-------
the urban atmosphere contains many chemical compounds, several
of which are known to increase the risks of cancer among persons
exposed to them in the work place or via personal exposure.
Many other chemicals found in ambient air are known to cause
cancer in experimental animals, and mixtures of pollutants
extracted from ambient air have been found to be carcinogenic
and mutagenic in experimental tests. The issue to be addressed
is whether exposure of the general population is sufficient
to lead to significant increases in cancer risk. This section
of the report reviews the epidemiologic evidence on this question-
i.e., whether the effects that may exist are large enough to
be detected against the variations in cancer rates imposed by
other factors. Quantitative estimates of the possible magnitude
of the contribution of air pollution are discussed in Chapter IV.
Table II-l (in Appendix A) summarizes the results of 48
studies in which geographic patterns in rates of lung cancer
and other cancers have been compared to geographic differences
in air pollution and other risk factors. The most significant
of these studies are also summarized and discussed in the text.
In a number of studies, various measures of air pollution have
been reported to be correlated with the geographic distribution
of lung cancer, and these results are consistent with the hy-
pothesis that air pollution is a factor contributing to an
increased risk of cancer. However, each individual study has
had limitations that preclude a definitive test of this hypoth-
esis. These limitations are also noted in Table II-l and
are discussed in the text.
11-53
-------
The most common problem with most of these studies is
the inability to control fully for factors that may confound
or interact with ambient air pollution, such as industrial
air pollution, cigarette smoking, or other personal exposures.
Attempts have been made in many studies to control for one
or another of these confounding factors, but the completeness
of this control is often a matter of dispute. Only two studies
(Hammond and Garfinkel 1980, Vena 1982) have provided enough
data to attempt to control for the two confounding variables
simultaneously. In the absence of a systematic, multivariate
study, scientific judgment on the possible role of air pollution
has to be made on the basis of the evidence provided by a number
of different studies in which single confounding factors are
controlled. This section reviews studies of this kind.
a. Smoking
Many of the studies of geographic variations in cancer
summarized in Tables II-l and II-3 did not take into account
possible differences in smoking habits between the study and
comparison populations. As a result, urban-rural differences
in smoking patterns cannot be ruled out in these studies as
a possible explanation of the urban factor. As mentioned in
Chapter I and Section II.B, however, there are a number of
ways in which smoking may interact with air pollution or other
factors. When data on smoking habits have been taken into
account, smoking has usually been treated as a confounding
factor. If there are synergistic interactions between smoking
and another factor, controlling for the effect of smoking as
11-54
-------
a confounding factor would tend to overestimate the role of
smoking and underestimate the role of any factor with which
it interacts. Controlling for smoking tends to submerge the
portion of cancers due to the interaction into the portion
due to smoking acting alone (Walker 1981). Smoking was taken
into account in several studies, however, and the corrected
residual urban lung cancer rates were higher than those in
rural areas (Dean 1966; Stocks and Campbell 1955; Dean et al. 1977,
1978; Hammond and Garfinkel 1980; Haenszel et al. 1962; Haenszel
and Taeuber 1964; Buell and Dunn 1967). The main scientific
issue to be discussed in reviewing these studies is whether
the ways in which smoking was taken into account were sufficiently
complete and precise to rule out smoking as a complete and
sufficient explanation of the urban-rural difference (see Doll
and Peto 1981).
The simplest, and possibly best, way to control for the
effects of smoking is to limit the analysis to data on cancer
in nonsmokers. One of the earliest available urban-rural com-
parisons of cancer rates has recently been presented by Logan
(1982), who summarized and republished the results of a mortality
survey conducted in England in 1881. A breakdown of comparative
mortality by occupational status and by large districts yielded
the following data on cancer rates (standardized per 1,000
cancer deaths in the total population):
All males 47
Occupied males 44
in London 59
in industrial districts 48
in agricultural districts 40
11-55
-------
A similar survey conducted in 1901 led to similar results, with
a ratio of 1.69 between cancer rates in London and in agricul-
tural districts. These data are important because they refer
to a period long before cigarette smoking became widespread;
hence, the urban-rural differential cannot have been signifi-
cantly affected even by passive smoking. (However, there was
no control for occupation or other urban factors, and the reli-
ability and completeness of diagnosis and data collection are
not clear.)
Haenszel et al. (1962) and Haenszel and Taeuber (1964)
obtained smoking and residence histories for a 10% sample of
all lung cancer deaths in white females in the United States
in 1958 and 1959, and for a 10% sample of all such deaths in
white males in 1958. These data were compared to similar informa-
tion from a very large sample of the general population. Because
of the large sample sizes, these studies provide the best avail-
able information on lung cancer by location of residence in
nonsmokers (individuals who had never smoked). Furthermore, it
is possible to control for the effects of migration by restrict-
ing attention to lifetime residents of either rural or urban
areas. The results of this comparison are presented in Table II-4.
Pike and Henderson (1981) suggested that the urban-rural ratio
in men is spuriously high, because the lung cancer rate for
rural men was actually lower than that in rural women. However,
even the ratio in women is significantly higher than unity.
11-56
-------
TABLE II-4
AGE-ADJUSTED LUNG CANCER RATES OF INDIVIDUALS
WHO HAD NEVER SMOKED BY LOCATION OF LIFETIME RESIDENCE
Location
of Lifetime
Residence
Urban
Rural
Males
Lung Cancer
Mortality
Rate/100,000
12.5
3.9
Females
Relative
Risk
3.2
1.0
Lung Cancer
Mortality
Rate/100,000
8.4
5.0
Relative
Risk
1.7
1.0
SOURCE: Haenszel and Taeuber (1964), retabulated by Pike and
Henderson (1981)
11-57
-------
Shy and Struba (1982) summarized the results of six other
studies in which lung cancer rates in nonsmokers were stratified
according to location of residence. Another set of data is avail-
able from the study of Dean et al. (1977, 1978). These data are
summarized in Table II-5. Five of these studies (Stocks and
Campbell 1955; Dean 1966; Buell et al. 1967; Hammond and Horn 1958;
Dean et al. 1977, 1978) showed a marked urban excess of lung
cancers in nonsmokers, whereas two (Hitosugi 1968, Cederlof
et al. 1975) did not. A general problem in interpreting these
data is the low frequency of lung cancer in nonsmokers, which
resulted in small numbers of cancer cases (see discussion above),
and the wide variability in reported nonsmoker rates from study
to study. Doll and Peto (1981, Appendix E) have drawn attention
to variations in estimates of lung cancer rates in nonsmokers,
which they attributed to confusion in some studies between ex-
smokers and lifelong nonsmokers. However, Haenszel and Taeuber's
(1964) study was not subject to these limitations because it
was based on a large sample of lifelong nonsmokers. Hence,
this study (Table II-4) provides the most compelling evidence
for an urban-rural difference independent of smoking.
In evaluating the studies of geographic patterns of cancer
rates in smokers, it is important to consider first whether
urban-rural differences in smoking patterns do indeed exist
and, if so, whether such differences have been of sufficient
magnitude to explain the observed excesses in urban cancer
mortality. It is generally agreed that cigarette smoking first
11-58
-------
TABLE II-5
URBAN-RURAL DIFFERENCES IN LUNG CANCER
MORTALITY RATES IN NONSMOKERS
Study, Data Years,
Age of Population
Areas of Residence
Lung Cancer
Mortality Rates
per 100,000
Nonsmokers
Stocks and Campbell (1955)
1952-1954
Ages 45-74
Dean (1966)
1960-1962
Ages 35+
Hitosugi (1968)
Ages 35-74
Buell et al. (1967)
Age-standardized
Hammond and Horn (1958)
1952-1956
Age-standardized
Cederlof et al. (1975)
1963-1973
Age-standardized
1. Urban Liverpool 131
2. Mixed 0
3. Rural 14
Ratio 1:3 9T7
1. Inner Belfast 36
2. Outer Belfast 40
3. Other Urban 21
4. Rural Districts 10
Ratio 1:43.6
1. High pollution 4.9
2. Intermediate pollution 3.8
3. Low pollution 11.5
Ratio 1:3 ~OT?
1. Los Angeles 28
2. San Francisco Bay area 44
3. All other counties 11
Ratio 1+2:3 3T3~
1. US cities 50,000+ 14.7
2. US towns 10,000-50,000 9.3
3. US towns <10,000 4.7
4. Rural areas 0.0
Ratio 1:4 oo
Males
1. Large cities 0
2. Other towns 10
3. Rural areas 16
Ratio 1:3 ~~U
Females
1. Large cities 3
2. Other towns 10
3. Rural areas 16
Ratio 1:3 OTZ
11-59
-------
TABLE II-5 (continued)
Lung Cancer
Mortality Rates
Study, Data Years, per 100,000
Age of Population Areas of Residence Nonsmokers
Dean et al. 1978 Males
1. Eston 60
2. Stockton 56
3. Rural areas 35
Ratio 1+2:3 T77
Females
1. Eston 15
2. Stockton 19
3. Rural areas 20
Ratio 1+2:3 0.85
11-60
-------
became prevalent in cities (Doll 1978, Doll and Peto 1981,
Wilson et al. 1980). There are very few quantitative data,
however, on differences in the proportions of individuals who
smoke or the number of cigarettes smoked. Doll (1978) referred
to a survey done by the Tobacco Research Council, which indi-
cated that in 1970 men and women residing in "conurbations"
smoked twice as many cigarettes as men in "truly" rural parts
of Great Britain. A 1955 national survey in the United States
(Haenszel et al. 1956) also indicated that differences existed
between urban and rural-farm residents (see Figure II-l).
Doll and Peto (1981, footnote 37) cited without reference a
survey conducted by Fortune magazine in 1935, which
found the respective percentages of men and women
who smoked any form of tobacco to be 61 and 31%
in large cities, as against 44 and 9% in rural areas.
Since many rural men smoked only pipes and/or cigars
(which have relatively much less effect on lung
cancer than cigarettes), the urban-rural differences
between the percentages who smoked cigarettes between
World Wars I and II were probably very marked among
the young of both sexes.
More recent data (Table I1-6) indicate that the percentage
of farm workers who are current, regular cigarette smokers is
similar to that of white-collar workers (USDHEW 1979). However,
a higher percentage of blue-collar workers (craftsmen, opera-
tives, and nonfarm laborers) is classified as current, regular
cigarette smokers. Also, men smoke more than women, although
this difference is not as great as it was 20 years ago (USDHEW
1979), and many of the cigarettes advertised specifically for
women contain less tobacco than the average cigarettes and are
11-61
-------
FIGURE II-l
PERCENTAGE OF PERSONS 18 YEARS OF AGE AND CURRENTLY SMOKING
CIGARETTES REGULARLY, BY SEX, WITH ADDITIONAL DETAIL ON CURRENT
DAILY RATE, FOR URBAN, RURAL NONFARM, AND RURAL FARM POPULATION
PERCENT REGULAR SMOKERS
0 15 30 45 60
URBAN
RURAL NONFARM—
RURAL FARM
HMORE-THAN 1 PACK
Ql/2 PACK AND OVER
^3 ANY AMOUNT
SOURCE: Redrawn from Haenszel et al. (1956), Figure 13, p. 30
11-62
-------
TABLE II-6
ESTIMATES OF THE PERCENTAGE OF CURRENT, REGULAR CIGARETTE SMOKERS,
ADULTS AGED 20 YEARS AND OVER, ACCORDING TO FAMILY INCOME,
SELECTED OCCUPATION GROUPS, AND MARITAL STATUS,
UNITED STATES, 1976
Category
Male
Female
Family income
Under $5,000
$5,000 to 9,999
$10,000 to 14,999
$15,000 to 24,999
$25,000 or. more
Occupation groups
White collar
Professional, technical,
and kindred workers
Managers and administrative,
nonfarm
Sales workers
Clerical and kindred workers
Blue collar
Farm
. Currently unemployed
Not in labor force
Marital status
Never married
Currently married
Widowed
Separated
Divorced
42.5
42.5
42.5
40.4
34.7
36.6
30.0
41.0
39.9
40.4
50.4
36.9
56.8
32.9
40.1
41.1
32.6
63.3
59.9
33.5
32.5
32.5
33.0
35.1
34.3
29.1
41.6
38.1
34.8
39.0
31.3
40.0
28.2
28.3
32.4
20.4
45.1
54.8
Craftsmen and kindred workers, operatives including
transport, nonfarm laborers.
SOURCE: USDHEW (1979), p. A-16
11-63
-------
often also relatively low in tar. Current cigarettes contain
substantially less tobacco per cigarette than did earlier cigar-
ettes .
To consider whether these differences in the prevalence of
smoking are likely to account for observed urban-rural differ-
ences in lung cancer mortality, we can follow the approach of
Schlesselman (1978). To do this calculation, we assume that
the relative risks of lung cancer mortality among males were
12 for current or occasional smokers and 6 for ex-smokers,
(derived from data in USDHEW 1979, Chapter 5, Table 1). These
assumptions are likely to overestimate the relative risks because
they are similar to the values reported for male veterans (Kahn
1966), whereas Haenszel et al. (1956) found that veterans smoked
more than males in the general population in all age categories.
For women, we assumed that the relative risks for current or
occasional smokers and for ex-smokers were 4.4 and 2.2, respec-
tively. These too are probably overestimates. For the pro-
portions of smokers, we used Haenszel et al.'s (1956) data
on whites, which are broken down into urban, rural nonfarm,
and rural farm categories (Figure II-l). We weighted the rural
categories according to their relative proportions in the U.S.
population in 1960 (U.S. Bureau of the Census 1980 ).
Using Schlesselman's (1978) Table 1, we obtained estimates
of the urban-rural ratios in lung cancer rates that would be
And Deare, D., U.S. Bureau of the Census; personal communication,
1981.
11-64
-------
expected to result from the differences in the prevalence of
smoking in 1955, in the absence of any other urban-rural differ-
ences in risk factors. These estimates are presented in Table II-7
and are much smaller than the observed ratios tabulated in
Table 11-2. (The comparison is not precise, because the observed
ratios are for the period 1950-1969, whereas the smoking data
are for 1955.)
There is a problem with the use of the Schlesselman approach,
however. This formula for estimating spurious (confounding)
effects is based on the assumption that the several effects
act independently. As discussed earlier—and in view of the
multistage theory of cancer causation—this is not likely to
be true. In the presence of interactions, the Schlesselman
formula will tend to overestimate the contribution of the con-
founder (in this case, smoking), but the precise contribution
of the confounders to an apparent association cannot be calcu-
lated.
In addition to differences in the proportion of smokers
and in the number of cigarettes smoked, Doll and Peto (1981,
pp. 1246-1247) have drawn attention to the potential importance
of other characteristics of smoking behavior:
The reasons for uncertainty deserve some detailed
discussion, for if they are overlooked a misleading
impression of the hazards of air pollution may be
engendered. The key observation is that lung cancer
risks among cigarette smokers in middle and old
age depend very strongly on the exact age at which
cigarette smoking began. For example, delay of
the onset of cigarette smoking in the late teens
or early twenties by just a couple of years may
11-65
-------
TABLE II-7
ESTIMATED RELATIVE RISKS OF LUNG CANCER MORTALITY
EXPECTED FROM DIFFERENCES IN THE PREVALENCE OF SMOKING
IN 1955 BETWEEN URBAN AND RURAL POPULATIONS
Expected Urban-Rural Ratio
Observed Urban-Rural (based on differences in
Ratio (adjusted for age smoking between urban and
Sex but not for smoking) rural residents)
Men 1.89 (See Table I1-2) 1.06
Women 1.64 (See Table I1-2) 1.15
11-66
-------
reduce the risk of lung cancer at age 60 or 70 by
as much as 20% (see text-fig. El on page 1292).
Therefore, lung cancer risks in cities and in rural
areas depend strongly not only on what old people
now smoke, but also on what they smoked in early
adult life half a century or so ago. If cigarette
smoking by young adults was somewhat more prevalent
(in terms of percentages of serious cigarette smokers
or numbers of cigarettes per smoker) in cities than
in rural areas during the first half of this century,
this alone would engender a substantial excess of
lung cancer today when cigarette-smoking city dwellers
are compared with cigarette-smoking country dwellers.
The smoking of substantial numbers of cigarettes
was an extremely uncommon habit in all countries
in about 1900, while by 1950 it had become common
throughout the developed world.
While any new habit is in the process of becoming
adopted by society (e.g., the use of various drugs
today), it is likely that its prevalence among young
adults will be greater in cities than in rural areas.
In appendix E we discuss in detail the effects of
differences in cigarette usage in early adult life
on the lung cancer risks many decades later among men
who would all, in later life, describe themselves
as "long-term regular cigarette smokers of one pack
of cigarettes per day." Because of such effects,
one must anticipate, even if air pollution were
completely irrelevant to the carcinogenicity of
cigarettes, to find that urban smokers now have
greater lung cancer risks than do apparently similar
rural smokers, at least in studies of populations
who still live in the type of area (urban or rural)
where they grew up. This should, of course, also
hold in countries other than the United States,
and it is noteworthy that urban-rural differences
in countries such as Finland and Norway where the
cities have not been heavily polluted are of a similar
size to the urban-rural differences in Britain and
the United States.
Doll and Peto also drew attention to effects of the amount
of each cigarette that is smoked and the depth of inhalation
(Appendix G). However, few data are available to test their
hypothesis that urban-rural differences in age at starting
smoking may have contributed substantially to urban-rural differ-
ences in lung cancer mortality.
11-67
-------
Haenszel et al. (1956) concluded that no important differ-
ences existed between urban and rural populations in age at
starting smoking. Their data collected in 1955, are presented
in Table II-8 and show no important differences between urban,
rural nonfarm, and rural farm residents in the age distribution
of starting smoking in any cohort of either sex.
In contrast to this, Weinberg et al. (1982) surveyed smoking
habits in two areas of Allegheny County, Pennsylvania, and
found substantial differences in this and other characteristics
of smoking (Table II-9). These data support Doll and Peto's
hypothesis that variations in these characteristics of smoking
parallel differences in prevalence of smoking. However, the
two areas Weinberg et al. examined were not urban and rural,
but urban and inner suburban; and they did not provide an unbiased
measure of geographic differences in patterns of smoking because
they were selected on the basis of having the highest and lowest
rates of lung cancer in the county. Thus, Weinberg et al.'s data
appear to reflect socioeconomic differences in patterns of
smoking and do not necessarily conflict with those of Haenszel
et al. Dean et al. (1977, 1978) investigated patterns of smoking
in urban and rural areas of northeastern England, obtaining
data for lung cancer cases and controls on age of starting
smoking, number of cigarettes smoked, types of cigarette, and
inhaling habits. The results/ reproduced in Appendix H, show
no important differences between urban and rural areas in any
of these aspects of smoking behavior except the number of cigar-
ettes smoked. Correspondingly, Dean et al. found that the
11-68
-------
OS
W
&
O
2
CO
ri CO
" §
cj O
W £
C9 * t*
M w a
u Is! °
0*
< Q ,g
§X fa
06 W
CQ y
o g
00 £ *° «
M 8 a 9
a !&-
1 §^i
c^ ££ QJ PN
W0U
?!
ft W 2
fa §
W S g
EH O
23 rH ^
W 05 §
CJ C^ rn
S §
w §
M ^
|
5
u
(0
Q)
^^
CN
rH CN in CO CN
rH CN CO ^
VO rHO
CN 00 CO
rH CO
C
.a
D rHin^o^t
rH rH rH
VO VO CN CO CN
CN
vo in vo co co
co in vo vo
CN f* CN
O O CN CN ^f
in vo f» m vo
O CN in rH ^
rH rH
^ r~ co co in
o vo ^ co cr>
rH CN CN
O VO co CO ^J"
CN CN CO <4* CT>
rH CN CO CO
g co r^ CN
l* ...
(0 rHCN ^
<4-l rH CN
C
&
rH VO VO O ON rH
3 rH rH CN CN
tf
r» cr> rH t^ a\
r^ in o o CN
CN *t in m
VO CN in VO CN
r^ CN co CN m
CO ^ VO VO
rH <5C O VO CN
r- CN o co vo
co in vo vo
vo co cj^ r*^ ^*
vo vo ^* m vo
co in vo vo
in rH O\
...
vo co cr>
CO ^
in co o in o
rH rH CN CN CO
CN CN VO VO
o o o d
in in r» in in
O O O rH rH
CN co o in o
o o CN co in
in r» o r- co
o CN m co IH
rH
O ^* CN CO ^
CN 00 m O CN
rH CN CN
CN CO CO
€ ...
M rH CO CO
<0 rH
fc
rH
M Tt O VO CN O
fti CO rH ^ CTi CN
^ H rH CN
rH ON CO CN rH
^C in vo vo cr>
rH CN CO CO
CN ^ CO CN VO
VO VO CO CN ^
CN co in in
CN o H in m
CO in CN rH (N
co in vo vo
in ON r^- in ^*
f*- O rH CO ^*
co in vo vo
CN in vo
...
^* ON ON
CN CO
m co o m o
rH rH CN CN CO
11-69
-------
to
EH
i
D
in C
O» Wt -H
^
4J 4> <0
in
o>
.. d>
CD H
H «J
n5 cr>
a> -1-1
(0
>1
m O
0)
>
0)
O
c
0)
V-l
3
r-
co
tn
•
oo
00
CM
0) O
tic
a) a)
•" S M 5? o
co cr> " Jl vo
id
n en
U^
0)
rHX
id id
4-> -P
0 C
EHM
CT>
0) 4-> C
O> W^
rf5|
SM£
0) 4->
o
r-
CM
CM
in
C O
0) U
4->
C -P
0 C
4-1
C
(d
O
•H
4-1
•H
C
01
•H
(0
id
o
to
•H
4J
(d
4->
0)
id
in
0) C
04 3
T3 O
•* m £
o o
B CP-*
01 C
a)
0)
4->
0)
0)
o
c
0)
VJ
0)
0)
CO
•
o
CO
en
o
00
•
vO
01 -P
fl) VI
id ,
43
01
id
•H
•d
o>
4J
4-»
oo
r~
CO
CPO O
•H C
O -H f*
01 4J
(0
3
id
o
o>
P
0)
(0
0)
id
4J
0)
01 ON
W <-*
0) 00
0)
CJ C
m
CO
m
\
in
<*
vo
i
m
in
in
11-70
m
0)
cr
id
O
id
id
Ed
§
to
-------
urban-rural risk ratios did not change greatly when these factors
were controlled for (independently or together). The data
of Haenszel et al. (1956) and Dean et al. (1977, 1978) thus
provide strong evidence against Doll and Peto's suggestion
that these factors significantly distort urban-rural ratios
in cancer rates.
Nevertheless, it would be desirable to calculate the likely
contribution of urban-rural differences in age at starting to
smoke on the urban-rural differential in lung cancer mortality.
However, to do so would necessitate combining data that are
not strictly comparable. For a rough theoretical calculation,
we use Peto's (1977) generalization that the incidence of lung
cancer is proportional to the 4th power of the duration of
exposure to cigarette smoke. Then, for two groups of men who
started smoking at ages 17 and 21 and whose smoking habits
were otherwise similar, the incidences of lung cancer at age
65 would be in the ratio (65-17)4/(65-2l)4, or 1.416. This
figure is consistent with data on U.S. veterans, summarized
by Doll and Peto (1981, Figure El). Incorporating this ratio
into the calculation summarized in Table II-7, we obtain an
estimate of 1.48 for the urban-rural ratio that would be expected
on the basis of the observed differences in the prevalence
of smoking in 1955, combined with an assumption that the mean
age of starting to smoke was 21 in rural areas and 17 in urban
areas, in the absence of urban-rural differences in other risk
factors. Although this calculation involves a number of more
11-71
-------
or less doubtful assumptions, it suggests that the hypothesized
difference in mean age at starting would have to have been
much greater than 4 years to account for the observed urban-rural
differences in cancer frequency. Although Table I1-9 indicates
a difference of about 4 years between residents of two districts
in one county. Table I1-8 does not indicate a systematic differ-
ence of even 1 year between urban and rural areas.
The most detailed and comprehensive attempt to control
for urban-rural differences in cigarette smoking habits is
that of Dean et al. (1977, 1978), already referred to above.
The primary objective of the study was to "determine the changes
that had occurred in mortality from lung cancer and bronchitis
since 1963 and to see how far these were related to changes...
in the smoking habits of the population and in air pollution
levels." Dean et al. compared data on a sample of 616 males
and 150 females who had died from lung cancer in Cleveland
County, England, between 1963 and 1972 with data on 2,666 living
males and 3,039 living females over 35 years of age and inter-
viewed in 1973. Data on the smoking habits and other character-
istics of the lung cancer victims were obtained from relatives
and from hospital records; data on the living samples were
obtained directly in five interviews. In addition to obtaining
information on characteristics of smoking habits, data were
acquired on social class, occupation, exposure to dust or fumes,
location of residence, and a number of other variables. Data
on air pollution were used to classify locations of residence
11-72
-------
as areas of high, medium/ or low pollution, even within the
areas classified as urban. For analysis, data were stratified
by age and various combinations of other variables, and age-
adjusted relative risks were calculated using maximum likelihood
methods.
The major conclusions of Dean et al. were:
After standardizing for age and smoking habits,
and after adjusting for differential population
movements in the three pollution zones, male residents
living at addresses within Stockton classified as
having high smoke and sulphur dioxide pollution
had over twice the relative risk of dying of lung
cancer as had residents at other addresses. An
excess mortality, based on far smaller numbers of
deaths, was also found for females.
Secondly, ... only a small part of the marked
excess lung cancer mortality rates (among residents
of urban areas) would be explained by (smoking pat-
terns) or because they tended to be of lower social
class.
Dean et al. attempted to standardize for amount smoked, age
at starting to smoke, type of cigarettes smoked (plain or filter),
and inhalation patterns. They noted some anomalies in relation
to age at starting to smoke, which they believed may be due
to errors in estimating this age by the relatives of deceased
lung cancer patients who supplied the information. However,
they added:
it seems unlikely that, had age of starting to smoke
been perfectly accurately assessed in the decedents,
it could have explained the urban-rural mortality
difference.
The third observation was that
between 1952/62 and 1963/72, the lung cancer rates
of men aged over 55 who were reported never to have
smoked increased significantly. This difference,
11-73
-------
about three-fold, could not plausibly be attributed
to changes in standards of diagnosis. Equally,
it could not be explained in terms of current exposure
to pollutants as there has been a downward trend
in levels of all the pollutants studied between
these two periods. However this difference might
be explicable, at least in part, in terms of air
pollution if lifetime exposure to pollutants is
of importance, as due to the fact that some of the
sources of pollution in the area have existed only
for 50 years or less, older people in 1963/72 may
have had a greater life-time exposure than people
of similar age in 1952/62.
...we feel that, taking the facts together in combina-
tion it seems reasonable to conclude that air pollution
makes a significant contribution towards lung cancer
mortality. This conclusion is consistent with the
results from Dean's study which showed that, after
standardising for age and smoking habits, male inhabi-
tants of Inner Belfast had 3.3 times the lung cancer
mortality, and 4.4 times the chronic bronchitis
mortality of inhabitants of truly rural areas of
Northern Ireland (Wicken 1966).
...smokers of filter cigarettes have a markedly
lower relative risk of lung cancer and chronic bron-
chitis mortality than smokers of plain cigarettes.
In view of the national switch towards smoking filter
cigarettes, and in view of the reductions in air
pollution that have followed the Clean Air Act of
1956, it was to be expected that, in due course,
overall mortality from both these causes would de-
crease, and indeed marked reductions have been noted
for chronic bronchitis mortality. But for the fact
that, at present, successive generations of the
older age groups, who have the highest mortality
rates, tend to contain an increasing proportion
of people who have smoked for a very long time,
and who are thus at higher risk, overall mortality
rate trends from lung cancer would already show
more clear improvement than they do at present.
However, if trends in lung cancer mortality rates
are studied separately by age-group, the improvements
expected from the switch to filters and reduced
air pollution can be seen. In 35-39 year old males,
for example, national lung cancer rates have dropped
38% between 1956-60 and 1971-75, and increases can
now only be seen in men over 70. Male bronchitis
rates show an even more marked improvement, with
a 30% reduction in overall death rates between 1968
and 1975 and rates declining at all ages except
11-74
-------
in men 70 or over where they have levelled off (Todd
et al. 1976). Lee (1977) has calculated, using
Peto's formula...that even in the age groups at
which mortality rates are still rising/ the rises
are markedly less than would have been expected
based only on knowledge of distribution of duration
of smoking habits, and ignoring the switch to filters
and the reduction in air pollution levels. Of course,
if standards of diagnosis of lung cancer are still
improving...then the benefits of the switch to filters
and the reduction in air pollution are even greater
than the data suggest.
The major conclusions of the study by Dean et al. (1978)
are summarized in Table 11-10. After standardizing for age,
smoking classification, and age at starting to smoke, urban-rural
ratios in lung cancer mortality were 1.50-2.02 for males and
1.46-1.77 for females. Other analyses in the paper by Dean
et al. (1978) show that these urban-rural ratios were not strongly
affected by differences in the type of cigarette smoked (filter
or nonfilter), the depth of inhalation, or by differences in
social class. Moreover, there were significant correlations
between lung cancer frequency and measured air pollution levels
within the urban area.
This study is of particular importance because it controlled
simultaneously for so many aspects of cigarette smoking behavior.
It has two major limitations. First, although data were collected
on occupation and on occupational exposure to dusts and fumes,
these factors were not controlled for in the analysis. Standardi-
zation for social class probably controlled indirectly for
some of the effects of occupational exposure, at least within
the urban areas, but a rigorous analysis would be needed to
11-75
-------
TABLE 11-10
RELATIVE RISK OF MORTALITY FROM LUNG CANCER,
STANDARDIZED FOR AGE, SMOKING CLASSIFICATION,
AND AGE AT STARTING TO SMOKE, 1963-1972
Area Males Females
Eston 2.02 1.77
Stockton 1.50 1.46
Rural districts 1.00 1.00
SOURCE: Dean et al. (1978)
11-76
-------
establish this. Second, the data on smoking habits and other
characteristics of decedents were collected primarily from
surviving relatives and hence are subject to bias in relation
to those collected directly from the living controls. The
authors discussed this source of bias and presented evidence
that it was not great. In addition, the bias is likely to
have existed in both urban and rural areas, so that the urban-
rural ratios may not have been seriously affected.
Weinberg et al. (1982) reported a comparative study of
smoking habits and lung cancer rates in two geographic areas
within Allegheny County, Pennsylvania. The areas were selected
on the basis of large reported differences in lung cancer rates
in white males, combined with evidence for a high proportion
of whites in the local populations and for stability of demo-
graphic and socieconomic characteristics. Information on socio-
economic characteristics, occupation, and smoking habits was
collected in interviews with a total of 988 men from 1,847
randomly selected households; the response rate was 82%. The
differences in smoking habits are tabulated in Table II-9.
The population of the Lawrenceville area scored markedly higher
on the socio-economic indicators than that of the South Hills
area, and this variation accounted for most or all of the differ-
ences in smoking habits. Weinberg et al. attempted to calculate
the effect of the differences in smoking habits by using a
simple multiplicative risk model, in which current smokers
were assigned a relative risk (vs. nonsmokers) of 14, ex-smokers
11-77
-------
were assigned a relative risk of 4, and the differences in
other characteristics of smoking were assumed to yield an addi-
tional relative risk of 1.5 for Lawrenceville (vs. South Hills).
Applying this model to estimates of the proportions of smokers
and assuming no effects from other variables, the authors calcu-
lated that the overall lung cancer rate in Lawrenceville should
be 1.98 times that in South Hills. By varying the assumptions
about relative risks, a range of estimates between 1.64 and
2.09 was generated for this ratio. Observed ratios varied
between 2.07 and 2.27. The authors concluded that "almost all
of the observed difference in risk between areas was attributable
to cigarettes."
This study was soundly conducted and its results appear
valid within their stated assumptions. Nevertheless, it has
several limitations as a contribution to assessing the relation-
ship between air pollution and urban-rural differences in lung
cancer rates. First, it was a descriptive study; no information
about smoking habits or other characteristics of lung cancer
decedents was collected. Second, smoking was the only causative
variable included in the model. Although one measure of air
pollution (dust fall) was considerably higher in Lawrenceville,
the authors stated that relevant measures of air pollution
were lacking and that estimates of the possible excess risk
due to air pollution would be speculative. Third, the design
of the study—a comparison of two areas in the same county
selected for maximum contrast in lung cancer rates—was such
11-78
-------
as to minimize the effects of geographic variables and maximize
the effects of within-population variables. Fourth, the model
used for calculating expected rates was simplified, and the
parameter values selected were open to question. Fifth, even
with the range of parameter values tested, differences in smoking
habits explained only between 76% and 96% of the observed differ-
ences in lung cancer rates. Thus, although the results of
this study are useful in showing the magnitude of differences
in lung cancer rates between areas that can result from differ-
ences in smoking habits, they do not necessarily conflict with
those of the more statistically rigorous study of Dean et al. (1978)
A related study was conducted in Denmark by Borch-Johnsen
(1982) in which the author came to the conclusion that "the
risk of lung cancer (in Copenhagen) is by 10-40% and 50-140%
higher than would be anticipated on account of smoking habits
in the youngest (1914-23) and oldest (1894-1903) generations,
respectively." While finding that smoking did not account
for the urban-rural differences, the author concluded that
"occupational factors are believed to have a greater contribution
to the urban factor than diffuse environmental factors...after
elimination of smoking". This study is available in English
only in abstract form, and a critical review is not possible
at this time.
Doll and Peto (1981, footnote 39) briefly reported unpub-
lished data from their earlier study of mortality in male British
doctors (Doll and Peto 1976). Their results are summarized
11-79
-------
in Table 11-11 and show a much smaller urban-rural ratio than
other studies that have controlled for smoking habits. However,
as Doll and Peto pointed out, all the doctors had been educated
in big cities and may have lived as children in areas different
from those they inhabited in 1951. The method of standardization
for smoking was not stated.
In each study in which the confounding effects of smoking
were controlled, except for that of Doll and Peto (1981), urban
residents were found to be at increased risk of cancer even when
differences in smoking habits were taken into account. Summa-
rizing these findings and pointing out the interaction effects,
Wilson et al. (1980) stated that most of the data
agree that there may be a small increase in lung
cancer among (urban) nonsmokers due to air pollution;
this is at most half the total incidence among non-
smokers which is already small. The increase of
lung cancer among (urban) smokers due to air pollution
is 4 times greater than the increase among nonsmokers
and is statistically significant.
However, Wilson et al. (1980) did not present a statistical
analysis to support the last statement, which is of particular
importance. The results of studies by Haenszel et al. (1962),
Dean (1966), Dean et al. (1978), and Cederlof et al. (1975)
indicate that cigarette smoking and air pollution probably
interact synergistically. A possible mechanism for this apparent
synergism was demonstrated by Cohen et al. (1979), who found
that smoking inhibits the action of cilia in long-term dust
clearance from the lungs.
11-80
-------
TABLE 11-11
LUNG CANCER MORTALITY IN MALE BRITISH DOCTORS,
STANDARDIZED FOR SMOKING AND AGE,
STRATIFIED BY LOCATION OF RESIDENCE
Location of
Residence in 1951
Conurbations
Large towns
(50,000-100,000)
Small towns (<50,000)
Rural areas
Expected
Deaths*
153.65
88.04
109.46
78.85
Observed
Deaths*
152
94
108
76
Ratio
0/E
0.99
1.07
0.99
0.96
*Period of observation unspecified.
SOURCE: Doll and Peto (1981), footnote 39
11-81
-------
Interactions between smoking and air pollution would account
for some of the differences between men and women in patterns
of lung cancer. If interactions of this nature do occur/ then
we should expect that larger urban-rural differences would be
seen for males, who smoke more than women and who generally
started smoking earlier. Such differences have been observed
in several studies (see Tables II-2, II-3, II-4, II-5, and
11-10). Similarly, if such interactions do occur, urban-rural
differences for female smokers should be larger than those
for female nonsmokers.
The increase in the urban-rural difference among women
smokers (relative to nonsmokers) expected on the basis of an
assumption of interaction has not been consistently observed,
however. Haenszel and Taeuber (1964) reasoned that this may
be due to the relatively small proportion of female smokers
before the 1950s, which leads to large sampling variation in
the estimated risks and the slopes of the smoking class gradient.
They also noted the effects of having small numbers of women
smokers are compounded by the smaller "effective" exposures
among women smokers relative to their male counterparts (i.e.,
women do not inhale as deeply as men and tend to smoke low-
tar cigarettes and cigarettes with less tobacco). The other
studies in which women's smoking habits were recorded (Dean
1966, Dean et al. 1978, Hitosugi 1968, Cederlof et al. 1975)
suffer from similar problems. Of these studies, only Cederlof
et al.'s (1975) results are consistent with the hypothesis
that there is an interaction effect among women.
11-82
-------
b. Occupational exposure
Several investigators have also postulated that much of
the urban excess of lung cancer can be accounted for by exposure
to carcinogens in the work place. In some situations, studies
have provided support for this hypothesis. For example, an
excess of lung cancer deaths was observed among white males
in south central Los Angeles County during the years 1968-1972
(Menck et al. 1974). Lack of a clear indication that smoking
or occupational factors accounted for the excess led the authors
to conclude that ambient air pollution was the causative factor.
A later case-control study was undertaken (Pike et al. 1979),
and its authors concluded that increased risks associated with
occupation could account completely for the observed excess.
However, Pike et al. (1979) in fact found associations between
lung cancer and both smoking and occupational categories; on
the basis of these associations, they calculated that the differ-
ences in smoking habits and occupations between the areas of
Los Angeles County originally studied by Menck et al. (1974)
would account for a relative risk of 1.26. This is smaller
than the relative risk of 1.40 that Menck et al. (1974) origin-
ally observed. Hence, there is still a portion of this differ-
ence that is unexplained by smoking and occupation. The sensi-
tivity of both studies was limited by the observation of Pike
et al. (1979) that most of the cases had migrated into the
area during the preceding 20-40 years.
11-83
-------
The data of Hammond and Garfinkel (1980) also suggested
that occupational exposure may account for part of the urban
excess. The excess of lung cancer deaths in urban and rural
areas in their study was reduced when occupational exposure
(defined in the study questionnaire as exposure to dust, fumes,
gases, or X-rays) was taken into account. This reduction was
evident in almost every residence category. Their definition
of occupational exposure is not precise, of course. The study
population had a larger proportion of whites, white-collar
-workers, and better educated individuals than the U.S. population
as a whole, which could lead to an underestimate of the effects
of both air pollution and occupational exposure. When lung
cancer mortality vs. location of residence is plotted separately
for occupationally exposed and nonoccupationally exposed men,
separate effects of both occupation and residence are apparent
(see Figure II-2). Hammond and Garfinkel reported that these
data were corrected for cigarette smoking.
Doll and Peto (1981) provided a quantitative interpretation
of these data, noting that after standardizing for smoking,
the mortality from lung cancer was only 14% greater in men
who reported a history of exposure to dust, fumes or mists
(including asbestos) than in men who did not. Since only 38%
of lung cancer deaths occurred in men who gave a positive history
of occupational exposures, Doll and Peto calculated that the
total contribution of these factors to the production of lung
cancer in the study population was 4.6%. However, Doll and
11-84
-------
FIGURE II-2
RATIO OF OBSERVED/EXPECTED LUNG CANCER DEATHS IN .
MEN BY RESIDENCE AND OCCUPATIONAL EXPOSURE, 1959-1965°
Ratio of
Observed/Expected
Deaths
1.4 —
1.3 —
1.2 —
1.1 —
1.0
0.9
0.8 —
Not Occupationally
Exposed
Smaller Non-
Rural Places
Large City
Areas
(1,000,000 •»•)
Adjusted for age and smoking
SOURCE: Plotted from data reported by Hammond and Garfinkel
(1980), Goldsmith (1980).
11-85
-------
Peto pointed out three factors that could contribute to making
an estimate of this kind too low: the diluting effect of random
errors, the possibility that the study population was biased
by the inclusion of proportionately few blue-collar workers/
and the possibility that undiscovered carcinogenic risks may
occur in industries in which there is no recognized dust, mists,
or fumes. Doll and Peto proposed (on the basis of admittedly
subjective and "stop-gap" methods of estimation) that the fraction
of lung cancer deaths ascribable to occupational hazards in
the U.S. in 1978 was about 15% in males and 5% in females.
At least in males, this fraction included some cases also attrib-
uted to cigarette smoking. However, Doll and Peto did not
discuss possible interactions with air pollution, and they
did not discuss or estimate the contribution of occupational
factors to the urban-rural ratio, except to quote Hammond and
Garfinkel (p. 1247).
The difficulty in separating occupational and air pollution
factors was also recognized by Greenberg (1979). He attempted
to determine the relative importance of different risk factors
for male lung cancer. He found that by adjusting air pollu-
tion indexes to take into account wind direction and distance
from the air monitoring site, the relative contributions of air
pollution compared to occupation increased. He later concluded,
however, that the high degree of intercorrelation among high-
risk lung cancer indicators (smoking, air pollution, occupation,
11-86
-------
etc.) makes it infeasible to pull apart the separate contribu-
tions made by personal, occupational, and local environmental
risk factors. Greenberg considered it likely that there are
interactions among air pollution, occupation, and smoking.
In a case-control study of white male lung cancer patients
from Erie County, New York, from 1957 to 1965, Vena (1982) was
able to study the effects of age, smoking, occupation, and air
pollution and their combinations. Air pollution was stratified
into pollution zones by means of air sampling data for particu-
lates collected from 1961 to 1963 and,by a historical review
of point sources. Exposure to air pollution was indexed by the
number of years of residence in a zone of high or medium air
pollution. Occupational exposure was defined as the number of
years in a job category with potential exposure to respiratory
carcinogens or with documented elevations in risk for lung
cancer. Smoking was defined in terms of years smoked, weighted
by four categories for amount smoked (less than 0.5 packs per day;
0.5-1 pack per day; 1-2 packs per day; and 2 or more packs
per day). Data on age at starting, type of cigarettes, and
degree of inhalation were not available. Although misclassifica-
tion may have occurred and smoking may still be a confounding
factor, this study by Vena (1982) is among the most detailed
available, especially in that the simultaneous influences of
air pollution, occupation, and smoking were assessed.
When exposure to air pollution was defined as exposure
to high or medium pollution for 50 or more years, occupation
as exposure in high risk jobs for 20 or more years, and smoking
11-87
-------
as exposure for 40 or more pack years/ it was evident that
occupation and probably air pollution interact with cigarette
smoking to modify its effect. Significant (p<0.05) age-adjusted
relative risks were observed for smoking (3.30), air pollution
and smoking (4.73), occupation and smoking (6.37), and all
three combined (5.71). When the data were stratified by age
to separate those born after the turn of the century from those
born before, the under 60 years of age category showed signifi-
cant associations between cancer risk and each of the three
individual variables (smoking, occupation, and air pollution)
and each of the combinations between variables. The results
for the over 60 years of age category paralleled the associations
observed for the overall, age-adjusted relative risks.
When Vena (1982) adjusted the relative risks for age,
occupation, and smoking, he observed a small (and nonsignificant)
unexplained lung cancer risk for the medium or high air pollution
areas (compared to the low pollution areas) of 1.03 for residence
of 30 to 49 years and 1.26 for residence of more than 50 years.
Vena (1982) cautiously interpreted this study as indicating
that air pollution should not be dismissed as a risk factor
in lung cancer because of the apparent synergism of air pollution
with smoking and with the combination of smoking and occupation.
He concluded, however, that his findings do not support the
hypothesis that air pollution alone significantly increases
the risk for lung cancer.
11-88
-------
Other investigators have reported their belief that occupa-
tion is not a major factor contributing to the urban excess.
Doll (1978) stated that occupational hazards were "unlikely
to be a major factor as the known and suspected hazards...affect
only a small proportion of the total urban population." As
mentioned earlier, Blot et al. (1977) made much the same point,
noting that if the higher cancer rates in petroleum counties
were the result of occupational exposure, the relative risk
to these workers would have to be substantially higher than
the general population; but this has not generally been observed.
c. Migration
Concerns have been raised that migration can have the
effect of increasing the apparent geographic variability because
(l) it may produce areas in which the age distribution of the
population differs considerably from the U.S. average and (2) per-
sons who migrate are likely to have a different health status
from that of those who remain behind.
Mancuso and Sterling (1974) reported that much of the
differences in lung cancer mortality rates that he found in
Ohio were a result of the very high rates observed in migrants
to Ohio from the rural areas of the southeastern United States.
Blot and Fraumeni (1982) have recently reported that the lung
cancer mortality rates in the southeast now exceed those in
the northeast and Great Lakes states. Mancuso interpreted
his findings as implying that a prior initiating exposure was
more likely to have occurred to the migrants (in contrast to
11-89
-------
lifelong residents) and that later, promoting exposure then
had a greater effect on migrants than on lifelong residents.
The first problem can be avoided when enough data are
available to calculate age-specific and age-adjusted mortality
rates. In the studies based on the mortality data for U.S.
counties compiled by the National Cancer Institute (e.g., Blot
and Fraumeni 1976), appropriate standardization has already
been performed.
The second problem, the possibility of selective migration
into or out of an area, might be corrected for if detailed
statistics were available on duration of residence. By studying
only those individuals who have remained in an area for 20 to 30
years, a more accurate assessment of environmental effects
could be obtained. In most studies of urban-rural differences,
such data are not available. It is possible that a small per-
centage of the urban-rural difference might be due to the migra-
tion of chronically ill persons to areas (generally urban)
with better medical facilities or of healthy individuals out
of these urban areas. Migration between geographic areas,
however, is generally expected to reduce the sensitivity of
geographic studies, as the distinction between exposed and
unexposed is gradually lost. As such, the statistical power
of such studies might be grossly overestimated if migration
were not taken into account. The longer the latency period
of disease, the larger this dampening effect of migration is
likely to be.
11-90
-------
As noted earlier, Polissar (1980) has estimated that 40%
to 50% of the relative excess risk is not reflected in the
estimated risk for most cancers when rates between exposed
and unexposed counties are compared and migration has taken
place during a 30-year latency period. This finding is consist-
ent with the results reported by Haenszel et al. (1962), who
found that the urban-rural gradient for the standardized lung
cancer mortality ratios (adjusted for age and smoking) increased
with the duration of residence. The role of urban air pollution
in explaining this trend, however, is unclear because the SMR
for urban residents declined with the duration of residence.
d. Population density and other factors
Demopoulos and Gutman (1980) labeled a series of cities
as "clean" and "dirty," based on a qualitative characterization
of the nature of the local industries but not on direct measures
of the nature or intensity of ambient air pollution. They
concluded that when areas with comparable population'densities
were compared, general air pollution (i.e., in dirty cities)
and work place exposure (in regions of heavy industry) were
not associated with cancer risks. This conclusion led them
to the speculation that much of the urban excess might be due
to higher population density. However, their designations
of clean and dirty cities were not related to any measured
distinctions between areas of low and high air pollution.
Their presumption that heavy industries should be more likely
to be associated with cancer risks than light industries may
11-91
-------
not be true. Major carcinogenic hazards have been recognized
in a number of light and service industries. Thus, their char-
acterization of clean and dirty cities is unsatisfactory even
as a surrogate measure of either air pollution or of occupational
exposure. Among other problems with this study, no attempt
was made to standardize for smoking or other risk factors,
and the basis for selecting the sample of cities was unclear.
Population density is strongly correlated with a number
of other factors and may represent a proxy measure of air pollu-
tion and a variety of other variables. In studying the relation-
ships between population density, vehicle density (as an indi-
cator of motor vehicle emissions), and total cancer mortality,
Robertson (1980) concluded that vehicle density rather than
some other correlate of population density is associated most
strongly with cancer mortality. Vehicle density, of course,
implies air pollution from burning fossil fuels in mobile sources
Robertson found that the number of motor vehicles per square
mile does not increase linearly with population density but
levels off in the more densely populated cities where public
transportation is often more readily available. He reported
that cancer rates do not increase linearly with city size but
do appear to be linearly correlated with motor vehicle density.
Robertson (1980) concluded that "motor vehicles appear to be
a substantial part of the 'urban factor' in cancer." However,
he failed to control for potential differences in several other
important factors (such as smoking, occupation, and migration).
11-92
-------
Currently available data are insufficient to estimate the rel-
ative contributions of mobile sources and stationary sources
of air pollution. It is likely that in some areas, the largest
source of conventional air pollutants is the automobile (e.g.,
Los Angeles), while in others, industrial sources are more
important (e.g., Charleston, West Virginia). In their recent
review, Wilson et al. (1980) came to much the same conclusion.
However, the relative contribution of mobile and stationary
sources to atmospheric concentrations of carcinogenic air pol-
lutants is not known.
e. Socioeconomic Status
Another confounding factor in geographic studies of cancer
rates is socioeconomic status (SES). Many studies have found
that lung cancer rates are inversely correlated with SES, at
least in males. Weinberg et al. (1982) found marked differ-
ences in smoking habits among occupational groups with different
SES and concluded that differences in SES were sufficient to
explain large variations in smoking habits between two areas
that had been selected for study because of wide differences
in lung cancer rates. Brown et al. (1975) reported that smoking
was inversely correlated with SES among men but positively
correlated with SES among women in Buffalo, New York. Lung
cancer rates were inversely correlated with SES among males,
but the same trend in women was not significant. The authors
concluded that differences in smoking habits could explain
the lung cancer-SES correlation in men, but not in women.
11-93
-------
In studies where attempts have been made to control for
population density, SES, and other confounding factors, the
correlations between variables such as air pollution and popu-
lation density may seriously distort the estimated effects
of air pollution. There is some evidence that the onset of
population-wide cigarette smoking paralleled industrialization.
If that is the case, regression analyses that attempt to estimate
the effects of air pollution may be distorted by controlling
for factors that are correlated with air pollution. Air pollu-
tion has also been found to be inversely related to SES (Bozzo
et al. 1979, Lave and Seskin 1977). Since low SES groups (who
are usually heavier smokers) are exposed to higher pollution
levels than high SES groups, the true effects of air pollution
are likely to be underestimated when the effects of SES and/or
smoking are controlled for.
F. Summary
This chapter summarizes epidemiologic studies of cancers
in the human population and the relation of cancer to air pollu-
tion and other factors. Section II.B introduces the four prin-
cipal types of epidemiologic study and discusses issues that
arise in applying these methods to the cancer/air pollution
problem. Although there is evidence that air pollutants may
be associated with cancers at a number of anatomic sites, only
lung cancers have been studied in sufficient detail to permit
critical analysis. Air pollution is a complex mixture of agents,
and most available measurements are of conventional pollutants
11-94
-------
that are unlikely to be carcinogenic in themselves. Furthermore,
the use of a single component, such as benzo(a)pyrene, as a
surrogate measure of the carcinogenic potential of polluted
air may not be entirely satisfactory. Significant exposure
to some air pollutants occurs in indoor environments, where
monitoring data are scanty. The long latent periods for human
cancers mean that current cancers should* be associated with
exposures in past decades, when some pollutants were present
at higher levels and others at lower levels. The most pervasive
difficulty encountered in the conduct and interpretation of
epidemiologic studies is the control of confounding factors,
especially cigarette smoking. Other problems that arise include
the interpretation of sexual and racial differences in patterns
of cancer mortality, the insensitivity of many studies, and
the selection of appropriate comparison populations.
Section II.C summarizes source-specific or "neighborhood"
studies. A number of studies have reported apparent elevations
in cancer rates in the vicinity of industrial facilities of
various types. Some of these studies were of the large-scale
"ecological" type, whose results are usually regarded as no
more than suggestive. Most other studies in this category
had substantial limitations, including problems in identifying
appropriate control populations; in controlling for smoking,
occupation, and demographic factors; and in verifying exposure.
The more persuasive evidence of this kind is the finding of
rare types of cancer characteristic of exposure to vinyl chloride-
11-95
-------
and asbestos near putative sources of these materials. However,
there are conflicting negative studies in each of these cases.
Section II.D summarizes several studies that suggest that
migrants from one country to another with higher (or lower) air
pollution levels continue to experience cancer rates character-
istic of their native countries. However/ the rigor of the
statistical comparisons of cancer rates is questionable, and
the differences were not related to specific data on exposure
to air pollution.
Section II.E summarizes urban-rural and other geographic
studies. Table Il-l (Appendix A) tabulates 48 epidemiologic
studies (reported in 43 papers) of cancers of the lung and
other sites in human populations. In 28 of these studies,
a statistical association was reported between cancer rates
and one or more (direct or indirect) measures of air pollution;
and most of the rest reported excess frequencies of cancer
in urban areas relative to rural areas. Only seven or eight
studies reported finding no association between cancer rates
and either urban location or measures of air pollution. However,
all of the studies were subject to various limitations, which
complicate their interpretation.
The most pervasive and difficult problem in these studies
is the control of confounding effects, of which cigarette smoking
is the most important. Ten studies of lung cancer rates in
nonsmokers have shown rather consistent urban-rural differentials
in males but not in females. However, all but one of these
11-96
-------
studies were limited by small sample size, and none controlled
for occupational exposures. In a number of studies/ urban-rural
differentials and statistical associations between cancer rates
and air pollution remained significant after attempts were made
to control for the effects of smoking, by using data on smoking
habits in cancer victims or population groups. However, the
completeness of the control for smoking in these studies is
disputed. Some scientists have argued that differences in
aspects of smoking such as age at starting to smoke and depth
of inhalation cannot be controlled for. However, actual data
on these aspects of smoking do not confirm that they would
contribute significantly to urban-rural differentials.
Only a few studies have controlled for the effects of
occupational exposures. One such study revealed significant
urban-rural differentials in both occupationally exposed and
unexposed groups, after controlling for smoking. Other studies
have suggested that there are interactions between the effects
of occupation and air pollution.
11-97
-------
III. EXPERIMENTAL EVIDENCE AND MONITORING DATA
A. Introduction
This chapter reviews and summarizes the evidence that air
contains substances capable of causing or contributing to the
incidence of cancer in humans. Monitoring studies have shown
that air contains substances known on the basis of human and
animal studies to cause cancer. In addition, extracts of air
pollution particulates have been shown to be both mutagenic
and carcinogenic in laboratory studies.
Air pollutants arise from both anthropogenic and natural
sources, such as vegetation, weathering, agriculture, volcanoes,
and fires. Air pollutants of anthropogenic origin can be placed
in three broad categories: vapor-phase organic chemicals, such
as volatile emissions from industrial processes; particulate
organic matter, which includes products of fossil fuel combus-
tion and vehicle emissions; and inorganic substances, such
as compounds of the metals lead, nickel, and arsenic, and the
mineral asbestos. The amount of vapor-phase organics emitted
in the United States has been estimated to be 1.9 x 10 g/yr,
with particulate organics being one-fiftieth to one-tenth of
this amount (Hughes et al. 1980, citingDuce 1978). Estimates
of the amount of anthropogenic inorganic pollutants are difficult
to make because of the wide variety of possible sources and
the large contribution of natural sources to the levels found
in ambient air. Of the three categories of pollutants, however,
III-l
-------
the particulate fraction of air pollution has been subjected
to the most investigation and is of most concern in terms of
long-term human health effects. This concern stems from the
known biological activity of many of the constituents of par-
ticulate matter, such as the polycylic aromatic hydrocarbons
(PAHs), and because particulate matter occurs at high local
concentrations around sources in populated areas.
A sample of polluted air is a complex and dynamic mixture
that can contain over 300 compounds. It can consist of chemi-
cals in the vapor or gaseous phase, relatively pure aerosols
or particulates of specific substances, or heterogenous parti-
cular aggregates of many substances. The relative distribution
of chemicals between the vapor and particulate phases is highly
dependent upon their source, their vapor pressure and polarity,
and the ambient air temperature. Although particulate matter
may be thought of as a collection of solid or liquid particles,
vapor-phase organics may be adsorbed under a range of conditions
into the particulate content of polluted air, thereby changing
their chemical composition (Hughes et al. 1980). In addition,
air pollutants, especially reactive species such as NOX and
ozone (itself derived from precursor pollutants), can undergo
photochemical or spontaneous reactions to produce new compounds
that may have more or less biological activity than their pre-
cursors. All of these factors complicate the identification
of the components of polluted air and their relation to the
III-2
-------
biological activity that is measured by in vivo or in vitro
studi es .
An additional consideration in reviewing the experimental
evidence associating air pollution and cancer is the difficulty
in determining the substances and the levels to which people are
actually exposed. This difficulty stems, first, from problems
in sampling air for pollutants and, second, from the complicated
and largely uninvestigated processes through which inhaled
materials affect humans. One of the problems in sampling is
that although some monitoring stations can sample air contin-
uously over long periods of time, most samples are obtained
over a limited period of time and therefore may not represent
all the pollutants in an area that result from changing weather
conditions and pollution sources. Also, sampling is usually
performed at roof level or close to a known source of emissions;
neither location accurately reflects the air quality at street
level that most people experience. Although advances have
been made in the design of personal sampling devices to provide
more accurate samples of the air that people breathe, most
of the studies of the biological activity of air pollution
and its chemical characterization have used samples that were
whose collection was limited in both time and location. There-
fore, they may not be representative of the actual toxicity
and content of ambient air. In addition, determination of
the effect of airborne substances on human health must take
into account the physiological processes that take place between
III-3
-------
the inhalation of a substance and the ultimate site of its
toxic effect. The effect of an inhaled carcinogen depends
on its distribution in the lungs, its retention and absorption,
its possible metabolism by lung tissue, its distribution via
the circulatory system, and the concurrent presence of irritating
substances. Some studies have investigated these factors,
and they are discussed below.
Because of these and other limitations of the monitoring
data, it is not possible to characterize human exposure to
carcinogenic substances in polluted air quantitatively with
any degree of confidence. Even if such estimates were available,
extrapolating experimental data on the potency of such carcingens
to predict likely magnitudes of human risks would be a very
uncertain process, especially for complex mixtures. Accordingly,
this chapter is primarily concerned with the qualitative evidence
for the presence of carcinogenic substances in ambient air.
Quantitative estimates of the possible magnitudes of the risks
are derived in Chapter IV and are based exclusively on epidemi-
ologic data.
B. Experimental Evidence
Experimental evidence for the presence of carcinogens in
ambient air has been provided by both in vivo and in vitro
testing of extracts of airborne material. This testing, how-
ever, has been limited to particulate material. Because of the
volatility, relatively low concentrations, and rapid degrada-
tion of vapor-phase organic substances, no methods are currently
III-4
-------
available for collecting these chemicals from ambient air and
testing them in vivo or in vitro. The car ci nogeni city of these
substances can be assessed by testing them in pure form at
high concentrations, and this type of evidence is discussed
in the section on monitoring data. The basic approach to deter-
mining the biological activity of airborne particulate matter
is to collect the particulates that are suspended in the air
or released from an emission source on filters, extract this
material with organic solvents, and apply the extract to the
test system.
The composition of these extracts depends on the chemical
and physical nature of the original particulates—specifically,
whether they were homogeneous, were aggregates, or contained
adsorbed organic chemicals—and on the ability of the fraction-
ation and extraction system to solubilize the chemicals that
are present. Because of the constraints posed by this approach
and the dilute nature of air pollution, the quantity of material
available for testing is usually limited. Researchers have
worked around this problem either by making extracts of more
readily available material, such as soot and tar that condense
from combustion emissions, or by using a small number of animals
in assay systems that are sensitive to carcinogens. These
systems include the painting of test material on the skin of
mice, injection into neonatal mice, and instillation into the
lungs of hamsters and rats. Alternatively, researchers have
tested extracts in cell cultures that are capable of detecting
III-5
-------
chemicals that cause mutations or cell transformation, although
they do not directly measure carcinogenic activity. Both of
these biological effects are considered predictors of carcinogenic
potential.
1. In Vivo Tests of Extracts of Air Pollution for Carcinogenicity
As mentioned above, the dilute nature of air pollution
limits the amount of material available for in vivo testing.
In the earliest studies, investigators prepared extracts of
soot, coal tar (a condensate resulting from the combustion
of coal under low oxygen conditions), and particulate matter
and applied them repeatedly to the skin of mice. In a review
of these studies, Shabad (1960) cited several investigations
in which skin tumors and adenomas of the lung were induced
by extracts of coal tar. Also, when dichloroethane extracts
of soot were painted on mice three times weekly, pa pi llamas
(benign skin tumors) appeared after 10 weeks, and after 25 weeks
of treatment, multiple foci of skin cancer were observed.
It was found that 37.5% of the treated animals ultimately devel-
oped malignant tumors and in about 50% of these animals the cancer
metastasized to sites in the lungs and lymph nodes (Shabad 1960).
Shabad also reported that extracts made from airborne particulates
induced malignant tumors in 8% of the test animals when the
same protocol was used.
In another dermal application study, Hoffman (1964) applied
an acetone solution containing 12.5% organic matter from an
extract of polluted air that was measured as having 20 @g of
III-6
-------
organic material per m to the skin of female mice. After
9 months, 23 of the 30 mice had developed multiple papillomas,
and 10 had carcinomas; after 3 more months of treatment, a
total of 19 of the mice had malignant tumors. Animals in a
group that were being concurrently treated with a solution
of a mixture of PAHs at a concentration equal to that of the
air extract had 4 tumors, half of which were malignant. Another
group painted with an equivalent amount of BaP did not develop
any tumors.
Gasoline engine condensate (GEC) and diesel exhaust conden-
sate (DEC) were examined for carcinogenicity in a skin-painting
study with female CFLP mice (Misfeld 1980). In addition to these
materials, BaP and a mixture of 15 PAHs at the same proportions
as found in GEC were tested. Each material was tested at three
concentrations in 80 mice per concentration. GEC, DEC, BaP,
and the PAH mixtures all yielded positive responses and positive
dose-response relationships. The largest response to GEC was
83% in the high concentration group. The high response to
DEC was 13%. It was calculated that GEC was 42 times as potent
as DEC, and the PAH mixture only accounted for 41% of the GEC
activity. Calculations indicated that BaP contributed 9.6%
and 16.7% of the activity found in response to GEC and DEC,
respectively.
Most recently, Nesnow et al. (1982) investigated the tumor-
initiating and tumor-promoting abilities of extracts of emissions
from automobiles with gasoline and diesel engines, from a coke
III-7
-------
oven, from roofing tar, and from a residential furnace that
burned diesel fuel. The animals used were Sencar mice, which
have been bred for their sensitivity to dermally applied carcino-
gens and are widely used in studies of the mechanism of carcino-
genesis. The collected emissions were extracted with dichloro-
methane, which was removed by evaporation; and the resulting
material was applied as a solution in acetone according to
one or more of four protocols in doses ranging from 100 to
10,000 ng/mouse. Under the tumor initiation protocol, each
dose was applied once topically, followed after 1 week by twice
weekly applications of the tumor promoter, tetradecanoylphorbol-
13-acetate (TPA). To determine the ability of the extracts
to act as complete carcinogens, samples were administered weekly
for 50 weeks. Under the tumor promotion protocol, the mice
were treated with one dose of BaP and then weekly for 34 weeks
with the sample. To test for cocarcinogenic activity, both
the test material and BaP were applied initially, followed
by TPA twice weekly.
These studies indicated that BaP and extracts from emissions
of coke ovens, roofing tar, and one type of diesel-powered
automobile were potent initiating agents. The emissions from
the other diesel automobiles and the gasoline engine automobile
showed some initiating activity. BaP, coke oven emissions,
and roofing tar emissions were also shown to be complete car-
cinogens. None of the diesel emissions from the automobiles
or furnace gave positive results in the complete carcinogenesis
III-8
-------
assay; the authors hypothesized that this result may have been
due to the cytotoxic effect of these extracts when applied
chronically. BaP and coke oven and roofing tar emissions also
demonstrated turn or-promoting ability; none of the diesel extracts
was tested in this protocol. Because of the positive results
for BaP in all of the protocols, the authors considered the
possibility that the activity of the emissions extracts may
have been due to their BaP content. However, analysis of the
samples for BaP and a comparison of the tumor-initiating ability
of the amounts of BaP found with that of the extract as a whole
indicated that the BaP content did not account for all the
activity of the extracts.
Depass et al . (1982) have also recently reported the results
of their skin-painting study. In this study, the initiating,
promoting, and complete carcinogenic activity of diesel exhaust
particulate (DP) and dichloromethane extracts of diesel exhaust
particulates (DCM) were examined using C3H strain mice. The
study was to end with the death of all mice, but the reported
interim results covered 714 days of treatment, with some mice
still alive in most groups. Groups of 40 mice were treated
with two concentrations of DP or four concentrations of DCM
for the complete car ci no genes is study, one concentration of
DP or two concentrations of DCM for the promotion study, and
one concentration of each for the initiation study.
In the study on complete carcinogenesis of DP and DCM,
only one tumor was found in a treated mouse. This mouse was
III-9
-------
in the high-dose DCM group. A slight response was also the
result in the promotion study. One animal in each DCM dose
group had a squamous cell carcinoma, and a second low-dose
DCM animal had a papilloma. Three mice in the DP group and
three mice in the DCM groups had tumors in the initiation study.
Tumors were found, however, in one acetone-initiated control
group mouse and two phorbol 12-myristate 13-acetate (PMA) initiated
control group mice. PMA was used as a promoting agent in the
promotion study. The difference in responses between the studies
of Depass et al. (1982) and Nesnow et al. (1982) may have resulted
from differences in the source of test substances, in the mouse
strain or sex, or in the treatment regimen.
Extracts of polluted air have also been administered to
test animals by subcutaneous injection. Hueper et al. (1962)
prepared benzene extracts of city air, concentrated them by
evaporation, and injected 1% (weight per volume) solutions
into C3H or C57 mice monthly for periods of up to 2 years.
This treatment induced local tumors in 2-18% of the animals,
with a latency period of 9 to 24 months. These results were
distorted, however, by substantial mortality in the test group
because of the toxicity of the extracts. Epstein et al. (1966)
developed a more sensitive assay, giving neonatal mice one
to three injections of the test material during the first week
of life and sacrificing the animals up to 1 year later. Extracts
of air partic ulates still caused mortality in the test group,
but the survivors developed hepatomas, lymphomas, and solitary
111-10
-------
and multiple pulmonary adenomas at rates significantly greater
than those for the control group.
In a later study, Rigdon and Neal (1971) collected air
pollutants in the vicinity of petrochemical plants, made benzene
extracts, and injected these once into 30- to 50-day-old CFW
mice. They observed the animals for up to 1 year, noting when
tumors appeared. The treatment induced as much as a 60% inci-
dence of local, nonmetastatic fibrosarcomas. This rate was
greater than that resulting from the injection of mice with
an amount of BaP equal to that in the extracts. This suggested
to the authors that multiple carcinogens or cocarcinogens were
present in the extracts.
Asahina et al. (1972) used Epstein's neonatal mouse assay
to test 10 fractions of an extract of New York City air. Signi-
ficant increases in the number of tumors, including pulmonary
adenomas and lymphomas, were found for 4 of the fractions.
More recently, Epstein et al. (1979) reported a dose-response
relationship between total tumor incidence and the cumulative
total dosage of the extracts injected into mice. The extracts,
which were found to contain PAHs, quinolines, and acridines,
induced solitary and multiple pulmonary adenomas and lymphomas
in both sexes and he pa toe ell ul a r carcinomas in males.
More recently, Pott et al. (1980) collected airborne parti-
culate matter from urban and rural locations, prepared organic
solvent extracts, and analyzed fractionated extracts for BaP
and other PAHs. The extracts were then injected subcutaneously
III-ll
-------
and chronically into mice in a range of doses based on BaP
content. Extracts with BaP contents of 0.37-1.1 pg induced
tumors at rates up to 30%, and a dose-response relationship
was seen with the fractions that predominantly contained PAHs.
Other fractions/ containing primarily polar substances, had
some carcinogenic activity.
A few investigations have been performed to test the capa-
city of fractions of polluted air to induce cancer in lung
tissue. In these experiments, the test material was instilled
into the trachea of anesthetized animals, from which it is
easily distributed into the lung. Bogovski et al. (1970) re-
ported that a benzene extract of oil shale soot containing
0.01% BaP induced lung cancer in rats after this type of intra-
tracheal instillation. Mohr (1976) instilled a condensate
of automobile exhaust into the trachea of hamsters at 2-week
intervals for 30 or 60 weeks. The condensate, which contained
a small amount of BaP (1.7 pg/anima-l), induced pulmonary adenomas
in all of the hamsters, a response the author could not attribute
to the BaP content alone. In a similar study, Kommineni and
Coffin (1976) applied a gelatin suspension of air particulates,
BaP, or particulates and BaP to the trachea of hamsters once
a week for 8 weeks. All three groups showed progressive and
severe inflammatory changes in the lungs; the third group,
which was treated with the particulates and BaP, showed evidence
of the formation of bronchial polyps. In addition to these
studies, researchers at the Health Effects Research Laboratory
111-12
-------
of the U.S. Environmental Protection Agency are completing
studies of the effects of the long-term inhalation of diesel
exhaust in mice and hamsters (Pepelko 1980).
Studies of the biological activity of extracts of air
pollution in animals do not provide data that are directly
applicable to predicting health effects in humans. Differences
in the routes of exposure and the use of high concentrations
limit the extent to which the results may be extrapolated to
human exposures, while the toxic/ noncarcinogenic effects of
the extracts limit the sensitivity of the tests in detecting
carcinogenesis. In summary, however, they do indicate that
ambient air, or materials released into air, contain compounds
that by themselves or acting together have the ability to induce
cancer in mammals.
2. In Vivo Studies of the Irritant Effects of Particulates
The ultimate effect of an inhaled carcinogen, which may
be either in the form of particles or adsorbed on particulate
material, depends on several interrelated factors: the distribu-
tion of the carcinogen in the lungs, its retention and absorption,
and the concurrent presence of respiratory irritants.
The size of a particle determines the extent to which
it penetrates the respiratory tract. In nasal breathers, parti-
cles from 12.5 nm to 2.5 ym in diameter are capable of penetra-
ting the alveolar region of the lungs. Particles greater than
2.5 ym in diameter are mostly removed in the nasal chambers,
and those less than 12.5 nm remain suspended in tidal air and
111-13
-------
are exhaled (Kotin 1968, Shannon et al. 1974). Studies have
also shown that retention of particulate matter in the lungs
is greatest at 1.0 vim in diameter and falls off sharply for
sizes greater than 2 ]im or less than 0.25 vim (Kotin and Falk
1963). With mouth breathing, the size of particles deposited
in the alveolar region of the respiratory tract can be up to
10 vim. In addition, particles up to 15 ym may be deposited
in the tracheobronchial portion of the respiratory tract.
Clearance of very large particles in the alveolar region is
slower than for smaller particles (USEPA 1982). Polluted urban
air contains particles in the range of 12.5 nm to 2.5 urn; par-
ticles of this size are also produced by the burning of solid
fuels and are present in the exhaust of gasoline and diesel
engines. Particle size may also influence the rate and extent
of elution of carcinogenic chemicals from the particles on
which they are adsorbed. Falk and Kotin (1962) found that
the lower size limit for PAH release from particles in physio-
logical conditions in vitro was 100 nm in diameter. Therefore,
particles from 100 nm to 10 vim in diameter are probably of
the greatest biological significance because they can readily
penetrate and be retained in the respiratory tract and because
adsorbed carcinogenic substances can be released from them.
The role of the penetration and retention of particles
in the lungs in inducing cancer has been investigated in a
number of studies. Inhaled ferric oxide (Fe203) dust is an
example of particulate material that although not carcinogenic
111-14
-------
to laboratory animals by itself (Oilman 1962), enhances the
effects of known carcinogens. This observation was initially
made by Saffiotti et al. (1968, 1972a,b) in studies in which
ferric oxide particles and various carcinogens were concurrently
instilled into the tracheas of hamsters.
Peron et al. (1972) showed that the tumori genie effect of
diethylnitrosamine in the hamster respiratory tract was increased
by a factor of 3 when the chemical was instilled in hamsters
with ferric oxide particles in solution. The enhancing action
of the ferric oxide particles has been attributed to their
ability to increase the penetration and retention of carcinogenic
substances that are bound to them. This possibility was investi-
gated by Sellakumar et al. (1973), who reported that the adhesion
of fine particles of BaP to the same sized particles of ferric
oxide was critical for tumor induction by intratracheal instilla-
tion. Without the physical adhesion to the ferric oxide dust,
much higher doses of BaP were needed to induce tumors in hamsters.
Henry et al. (1975) confirmed these results and, by micro-
scopic comparison of the lungs from the hamsters treated with
ferric oxide particles coated with BaP to the lungs of those
administered a mixture of the dust and the carcinogen, deter-
mined that the particles of the mixture were removed from the
lungs more rapidly. However, other studies have shown that
the ability of injections of the carcinogen di ethyl nitros ami ne
to induce lung tumors in hamsters was increased by the tracheal
instillation of ferric oxide particles (Montesano et al. 1970,
IIT-15
-------
Nettesheim et al . 1975). These results suggest that the parti-
cles may have a tumor-promoting effect in addition to enhancing
carcinogen penetration and retention.
In addition to ferric oxide, other particulates have been
shown to enhance the action of carcinogens. Studies have demon-
strated this effect with BaP and particles of asbestos (Miller
et al. 1965, Pylev and Shabad 1972), titanium oxide, aluminum
oxide, carbon (Stenback et al. 1976), and india ink (Pylev
1963). Although the mechanism of these actions is unknown,
Lakowicz and Hylden (1978) demonstrated that asbestos fibers
increase the lipid solubility of BaP and hence could increase
its cellular uptake.
Respiratory irritants present in polluted air also may
increase the carcinogenic effect of airborne substances by
changing the function and structure of the respiratory epithe-
lium and increasing the retention of these substances. These
irritants interfere with ciliary activity and with the flow
of the mucous stream. Air pollutants that act as irritants
to the lining of the respiratory tract include sulfur oxides,
nitrogen oxides, ozone, chlorine, ammonia, pollen, and allergens
(Kotin 1968). Laskin et al. (1970) demonstrated in rats that
simultaneous inhalation of the respiratory irritant sulfur
dioxide and the carcinogen BaP resulted in the production of
squamous cell carcinomas of the lung. Experiments performed
by Richters et al . (1979) suggested that exposure to respiratory
irritants also increased metastasis to the lung. The authors
111-16
-------
injected melanoma cells, which readily metastasi ze to the lung,
into mice that had been exposed for 10 weeks to an atmosphere
containing nitrogen dioxide at 0.4 ppm. Ten and 21 days after
infusion, the exposed animals had significantly more melanoma
nodules in the lungs than did the controls, which had breathed
filtered air.
3. In Vivo Mutagenicity and Genotoxiclty Testing
The mutagenicity and genotoxicity of air pollutants, most
notably diesel exhaust, have been studied in several animal
models. These in vivo tests are usually short term, and their
use of the whole animal is an obvious advantage over in vitro
assay systems. In addition, the test compound may be adminis-
tered by appropriate routes. However, these in vivo assays
usually are less sensitive and less easily quantified than
in vitro assays where the cells come into direct contact with
known amounts of test compound.
In a series of genotoxicity studies on diesel and gasoline
exhaust, as well as coke oven and roofing tar emissions, several
investigators used a variety of in vivo tests, which included
the sex-linked recessive lethal test on Drosophila melanogaster,
metaphase analysis, micronuclei assay, sperm morphology assay,
sister chromatid exchange assay, chromosomal abnormalities
assay, and a liver foci assay.
Schuler and Niemeier (1980) examined the effect of exposure
to Nissan diesel engine exhaust gases in producing sex-linked
recessive mutations in Drosophila melanogaster. The flies were
111-17
-------
exposed to a fivefold dilution of exhaust gases for 8 hours.
The exposed male Oregon-R strain flies were mated with Muller-5
strain females. Two broods of the F- generation and one F~
generation brood were examined for sex-linked recessive muta-
tion. No mutagenic activity was observed. The authors pointed
out that a more thorough assessment would necessitate testing
at higher exposure doses.
Pereira et al. (1980a) examined the genotoxicity of dies el
engine exhaust in female Swiss mice using metaphase analysis
and a micronuclei assay. The mice were exposed for 8 hours per
day, 5 days per week, for 1, 3, and 7 weeks. The exhaust was
diluted 18-fold and contained a final particulate concentration
of 6 to 7 mg/m . Bone marrow cells were used for the metaphase
analysis, which involved examination of cells in metaphase.
This assay can identify compounds capable of breaking chromo-
somes and chromatids. Only the animals exposed for 7 weeks
were examined, and no effects were observed. The micronuclei
assay was done on animals at all three exposure periods. Poly-
chromatic erythrocytes in bone marrow were examined. This
assay can also detect chromosome breakage and disruption of
the spindle apparatus. No significant increases in micronuclei
were found for any of the exposure periods. BaP was used as
a positive control in these studies and was given at a dose
approximating that expected in the dies el exhaust. In both
assays, BaP also had no effect, suggesting that the sensitivity
of these assays was too low for the exposure conditions.
111-18
-------
Pereira et al. (1980b) also conducted a micronuclei assay
using Chinese hamsters exposed to dies el exhaust for 8 hours
per day for 6 months. In this study, they found a significant
increase in the percentage of polychromatic erythrocytes with
micronuclei. The difference between the mouse and hamster
study was not explained. In the same study with hamsters,
chromosomal abnormalities in bone marrow cells were also exam-
ined. As in the mouse metaphase analysis, no increase in chro-
mosomal abnormalities was observed.
In addition to the other two assays, Pereira et al. (1980b)
conducted a sister chromatid exchange (SCE) bioassay with the
bone marrow from the exposed hamsters. SCEs are produced because
of DNA lesions induced by mutagens and may be related to recombi-
national or postreplicati ve repair of DNA damage. In this study,
there was no significant change in the frequency of SCE. There
was, however, a decrease in the mitotic index. Guerrero et
al. (1980) examined SCE in lung cells of Syrian hamsters treated
either by intratracheal instillation of one dose of diesel
exhaust particles at 0 to 20 mg/animal or by inhalation exposure
to diesel exhaust with a 6 mg/m particle concentration, 8 hours
a day for 3 months. Twenty-four hours after the intratracheal
instillation or following the 3-month inhalation exposure,
the animals were killed, and primary lung cell cultures were
established. When the cultures had colonies of 50 cells or
more, SCE analysis was performed. A positive dose-response
relationship was found for intratracheal doses between 0 and
111-19
-------
20 mg/aniraal. Animals exposed by inhalation to diesel exhaust
had no increase in SCE. When the total amount of particles
that were expected to be inhaled by the latter group of animals
was calculated and compared to the amount administered by intra-
tracheal instillation, it was found to be below the levels
that gave positive responses when administered by intratracheal
instillation.
It has been shown that exposure of mice to known mutagens
and carcinogens leads to an increase in the frequency of abnormal
sperm. Pereira et al. (1980c) exposed male strain A mice to
18-fold diluted Nissan diesel exhaust with 6 mg/m particle
concentration for 31 or 39 weeks; these time periods represent
approximately six and eight complete spermatogenic cycles/
respectively. No detectable changes in sperm morphology were
found after either time period. Pereira et al. (1980b) also
examined sperm shape abnormality in Chinese hamsters exposed
to diesel exhaust for 6 months. In this study, there was a
significant increase in abnormal sperm. The authors caution
that this result was obtained from a small group and should
be viewed as preliminary.
Pereira et al. (1980d) also used a rat liver foci assay
to examine the genotoxicity of diesel exhaust. This assay
is similar to the two-stage mouse skin model for carcinogenesis.
Rats were given a partial hepatectomy to enhance the rate of
cell proliferation and then were exposed to diesel exhaust emis-
sions for 3 or 6 months. During exposure the rats were fed a
111-20
-------
choline-devoid diet, inducing a dietary deficiency that acts
as a promoter. At 3 or 6 months the rats were sacrificed, and
their livers were histologically examined for foci of hepatocytes
containing gamma glutamyl transpeptidase, which is used as
a marker for cancerous hepatocytes. No increase in foci was
detected after 3 or 6 months of exposure.
4. In Vitro Tests of Extracts of Air Pollution
Extracts of polluted air and of air emissions have been
tested for mutagenic and genotoxic activity in a wide range of
in vitro systems. These tests are performed more quickly and
inexpensively than whole animal studies and can effectively
utilize the small amounts of test material usually available in
air pollution extracts. In addition, a large number of fractions
of the extracts that have been separated on the basis of chemical
structure or particle size can be tested concurrently, allowing
for the identification and isolation of the substances respon-
sible for the mutagenic or genotoxic activity. Direct extrapola-
tion of the results of in vitro tests to potential human health
effects is not yet possible, although several studies have been
performed that have established a high degree of correlation
between mutagenic and carcinogenic activity for some classes
of chemicals.
Of the wide range of in vitro tests, those that have been
used in testing air pollutants can be placed in four groups.
Gene mutational assays utilize bacterial or mammalian cell
cultures to detect single or multiple base changes (mutations)
111-21
-------
in genes. Larger scale damage to the DNA, in the form of DNA
strand breaks and exchanges between chromosomes, is detected
in assays using cultured hamster embryo cells, liver cells,
hamster ovary cells, and mammalian (human) lymphocytes . The
ability of chemicals or extracts to cause aberrations in chromo-
somes, such as breaks, deletions, and trans locations, is tested
in both hamster cells and human leukocytes. Transformation
assays measure the degree to which substances can alter normal
cultured cells to states in which they more closely resemble
cancer cells.
Transformation of cells in culture is considered analogous
to (although not identical to) transformation of cells in vivo.
These transformed cells in culture may have morphological and
biochemical traits similar to cancer cells. Most important,
when a cell that has been transformed in culture is implanted
in a syngeneic host, it will form a tumor. A variety of cells
has been used in transformation assays, including cells from
established cell lines and cells freshly isolated. There are
actually two types of cell transformation assays. In one assay,
the test compound produces the transformation, while in the other
assay, the test compound enhances a virally induced transforma-
tion of the cell. This latter assay is considered more sensitive
than the first one. The results of both of these assays correlate
well with the results of other tests for car ci no genes is and
mutagenesis .
111-22
-------
Several studies of these types "have been conducted with
extracts of air pollution and emission particulates. Freeman
et al. (1971) tested benzene extracts of the city air particu-
lates for their capacity to transform rat and hamster embryo
cells in culture. Transformation was considered complete if
the cells treated with extracts formed tumors when transplanted
into neonatal mice. The authors found that the extracts did
not transform rat embryo cells but did transform cells that
had previously been infected with Rauscher leukemia virus.
In these cultures of virus-infected cells, the extracts were
600 times more effective in inducing transformation than an
equal amount of pure BaP. In addition to the results seen
in rat embryo cells, the extracts transformed both infected
and uninfected hamster cells. The infected hamster cells were
as sensitive as the virus-infected rat cells; the uninfected
cells were one-tenth as sensitive as the virus-infected rat
cell cultures.
In another study, Gordon et al. (1973) first removed the
PAHs by benzene extraction from particulates collected from
Los Angeles air. The residue was further extracted with meth-
anol, and the transforming ability of this fraction was tested
in cell cultures of Fischer rat embryos and Swiss albino mouse
embryos. (The mouse cells, but not the rat cells, had been
infected with leukemia virus.) Results were positive in both
systems, indicating to the authors that non-PAH carcinogens
were present in the extract.
111-23
-------
Curren et al . (1981) investigated the transforming activity
of dichloromethane extracts of particulates from several types
of diesel engines, a gasoline engine, and coke oven and roofing
tar emissions. They used the BALB/c 3T3 cells in their assay
systems, some of which included and some of which excluded
the metabolic-activating system from rat liver. Several of
the extracts showed significant transforming activity, but
no clear dose-response relationships were found. The metabolic-
activating system reduced the transforming activity of some
extracts and did not greatly increase the activity of any extract;
this suggested that there were direct-acting agents in the
extracts. The most potent extracts came from coke oven emissions
and the gasoline engine. These were followed by extracts from
a Nissan light diesel engine exhaust and then roofing tar emis-
sions. Essentially, no activity was found in extracts of exhaust
from an Oldsmobile light diesel engine and a heavy diesel engine.
Using the same extract material, Castro et al. (1981)
were unable to show any transforming activity in their assay
system using Syrian hamster embryo cells. However, when the
cells were first infected with simian adenovirus SA7, several
extracts were capable of enhancing the viral transformation
of the cells. Ranking the extracts according to the lowest
effective concentration yields the following: extract of roofing
tar emission, coke oven emission, cigarette smoke condensate,
Nissan light diesel engine, and a gasoline engine and a VW
111-24
-------
diesel engine. Extracts from the Oldsmobile light diesel engine
and the heavy diesel engine produced little or no activity.
Many other assays have been developed to identify carcino-
genic compounds using mammalian cell cultures. Several of these
assays have been used to examine the genotoxic or mutagenic
activity of diesel engine particulate exhaust extracts and
extracts of participates from other emission sources. Mitchell
et al. (1981) used L51784 mouse lymphoma cells to examine the
mutagenicity of these extracts. The assay was done with and
without a metaboli c-activating system. All extracts tested gave
positive results in the presence and absence of the metabolic-
activating system, indicating the presence of direct-acting
mutagens in the extracts. The extract of the gasoline engine
exhaust emission was the most potent one tested. Castro et al.
(1981) examined the same extracts for mutageni city using Chinese
hamster ovary cells. In this system, extracts of emissions
from the Nissan and Volkswagen diesel engines, the gasoline
engine, and the coke oven yielded positive results. Unlike
the results of Mitchell et al. (1981) with mouse lymphoma cells,
extracts of emissions from a heavy diesel engine, the Oldsmobile
light diesel engine, roofing tar, and cigarette smoke were
not found to be mutagenic. Curren et al. (1981) used mouse
BALB/c 3T3 cells in a mutageni city assay and found extracts
of emissions from roofing tar, the Nissan light diesel engine,
the gasoline engine, and the coke oven to be mutagenic and
the heavy diesel engine and the Oldsmobile light diesel engine
not to be mutagenic.
111-25
-------
Using Syrian hamster embryo cells, Castro et al. (1981)
examined whether the extracts would cause DNA fragmentation.
This type of damage induced by chemical agents correlates fairly
well with their carcinogenic potential. Only coke oven and
gasoline engine emission extracts caused detectable breakage
of the cellular DNA.
Mitchell et al. (1981) examined whether these extracts
would increase sister chromatid exchanges (SCE) in Chinese
hamster ovary cells. Without metabolic activation, all extracts
tested showed some activity. Coke oven emissions and extracts
of the exhaust from Nissan light diesel engines were the most
active.
Lockard et al. (1981) examined whether extracts from air-
borne particulates would increase SCE in human lymphocytes or
V79 fibroblasts from Chinese hamster lungs. They used extracts
from samples of airborne particulates, collected over a 5-month
period on the campus of the University of Kentucky in Lexington.
There was a linear, dose-related increase of SCE in human lympho-
cytes, with 60 to 80 ng of extract necessary to induce a doubling
in the number of SCE. Several extracts that were positive
with human lymphocytes failed to induce an increase of SCE
in V79 cells; however, other extracts did cause an increase.
BaP was used as a positive control and increased SCE in both
cell types. The increase of SCE in human lymphocytes by BaP
did not occur in the presence of a metabolic-activating system,
although BaP generally needs to be metabolically activated
111-26
-------
to be effective. The amount of BaP, 8 yg, needed to induce
a doubling of the SCE in these cells was much larger than was
likely to be present in the extracts. Therefore, the extracts
must have contained active compounds other than BaP.
The gene mutational assay most widely used in testing
extracts of air pollution is the Ames assay (Ames et al. 1973,
1975), which measures the rate at which special strains of
the bacterium Salmonella typhimurium mutate or revert to a
less specialized form. The assay uses either the test material
directly or the test material in combination with a biochemical
preparation of liver or lung tissue that metabolizes the test
material, thereby testing for the possibility of in vivo genera-
tion of mutagens. The correlation of positive results in the
Ames assay with positive results in long-term carcinogenicity
assays has been found to be between 80% and 90%, depending
on the class of chemical being tested (McCann et al. 1975,
Commoner et al. 1976). A recent international study with 42
chemicals found the false positive rate, i.e., the rate at
which a positive result was obtained for a noncarcinogen in
bacterial assays, to be 5-10% (Bridges et al. 1981).
Gene mutational assays have been used to test air pollu-
tion from a number of sites and sources. Using the Ames assay,
investigators have detected mutagenic activity in extracts
of particulates from residential and urban air (Talcott and
Wei 1977; Pitts et al. 1977; Tokiwa et al. 1977, 1980; Commoner
et al. 1978; Teranishi et al. 1978; Salamone et al. 1979; Moller
111-27
-------
and Alfheim 1980; Lockard et al. 1981; Walker et al. 1982),
in fly ash from coal-fired power plants (Fisher et al. 1979),
in particulates collected from air in tunnels (Ohnishi et al. 1980),
and in exhaust from gasoline- and diesel-powered automobiles
(Ohnishi et al. 1980, Wang et al. 1981, Huisingh 1981, Lewtas 1982).
One study (Tokiwa et al. 1977) reported higher mutagenic activity
in samples taken from an industrial area than in samples from
a residential area. In most of the studies, a linear dose-
response relationship was observed between the amount of material
tested and the level of mutagenic activity (Tokiwa et al. 1976,
1977; Pitts et al. 1977; Teranishi et al. 1978; Commoner et
al. 1978; Salamone et al. 1979; Moller and Alfheim 1980; Ohnishi
et al. 1980; Walker et al. 1982).
Because the extracts of air pollution are composed of
a heterogenous mixture of substances, it is unlikely that the
mutagenic activity can be attributed to a single chemical or
class of chemicals. Most of the tests, however, have indicated
that the airborne mutagens cause the same type of mutation.
Tokiwa et al. (1977), Teranishi et al. (1978), Salamone et
al. (1979), Moller and Alfheim (1980), Ohnishi et al. (1980),
Claxton (1980), and Walker et al. (1982) have reported the
highest levels of activity of their samples were in the Salmonella
strains most sensitive to frameshift mutations.
BaP and other PAHs have been identified in extracts of
air pollutants by Talcott and Wei (1977), Tokiwa et al. (1977),
Commoner et al. (1978), Dehnen et al. (1978), Salamone et al.
111-28
-------
(1979), Holler and Alfheim (1980), Ohnishi et al. (1980), and
Tokiwa et al. (1980). Several studies have indicated that PAHs
require metabolic activation by the liver tissue preparation to
have a mutagenic effect (Wislocki et al. 1976, Wood et al. 1976).
Talcott and Wei (1977) found that 75% of the mutagenicity of
their urban air samples was due to an enzyme-activated fraction;
this activity was substantially reduced when an inhibitor of
the PAH-metabolizing enzymes was added to the culture. However,
Moller and Alfheim (1980), Lockard et al. (1981), and Salamone
et al. (1979) found extracts from their air pollutant sample
usually gave similar results with and without a metabolic-
activating system.
Pitts et al. (1980) recently demonstrated that BaP deposited
on a glass fiber filter in the presence of ambient levels of
ozone is transformed into strong mutagens in the Ames test.
This suggests that airborne BaP may not always require metabolic
activation to exert a carcinogenic effect, but that it can
be chemically activated in the atmosphere by ozone. The finding
indicates that some mutagens found in the particulate extracts
may be artifacts of the method of collection, and as indicated
below, direct-acting mutagens are also found in the extracts.
Analyzing air samples from residential areas, Talcott
and Wei (1977), Moller and Alfheim (1980), Salamone et al. (1979),
and Tokiwa et al. (1977) observed mutagenic activity that did
not require enzyme activation. In later research, Wang et al.
(1978) found that the lead content of extracts of nonindustrial
111-29
-------
airborne particulates correlated well with mutagenic activity,
suggesting to the authors that the source of the mutagens was
vehicular emissions. Further, they detected direct-acting
mutagens in extracts of automobile exhaust, although they did
not isolate the compound or compounds responsible. Wang et al.
(1981) found that extracts from diesel exhaust particulates
were mutagenic and that the mutagenicity of the extract was
not dependent on metabolic activation by liver homogenate.
They actually showed that this activity was reduced by addition
of the homogenate. The reduced activity was found to be not
from enzymatic activity but from nonspecific binding of the
mutagens to the protein in the homogenates instead of the DNA
of the bacteria. They showed that glutathione, a natural con-
stituent of the body that can bind to electrophilic compounds,
reduced the mutagenicity of the extract, thus suggesting that
the mutagens are direct alkylating agents. Claxton (1980) also
found that the majority of the mutagenic activity in extracts
of diesel exhaust was direct-acting. Mutagens in gasoline
engine exhaust extracts were partially di rect-act ing, but meta-
bolic activation did increase the mutagenic activity of th»
extracts. Whether any of the direct-acting mutagenic activity
discussed here is artificial, because of the method of collection
used, is not known; this makes assessments of the extracts
more difficult.
In a study designed to determine the size of the particles
associated with airborne mutagens, Talcott and Harger (1979)
111-30
-------
detected the highest activity in particles less than 2 pirn in
diameter and found that this fraction contained alkylating
agents. Fisher et al. (1979) and Tokiwa et al. (1980) also
compared particle size and mutagenic activity. Fisher et al. (1979)
found that fly ash particles of 3.2 \im diameter had the greatest
mutagenic activity, and Tokiwa et al. (1980) found the highest
mutagenic activity and PAH content in particles with diameters
of 0.3 to 1.0 ym. Particles of these sizes readily penetrate
lung airways (Kotin 1968).
C. Monitoring Data
A number of substances known to cause cancer in humans
or laboratory animals have been detected in ambient air. These
substances include PAHs, aza-heterocyclic compounds, vinyl
chloride, asbestos, metals, pesticides, N-nitroso compounds,
carbon tetrachloride, and many other industrial chemicals.
Table III-l (in Appendix B) is a compilation of suspected
and known carcinogens found in air pollution. This list contains
PAHS, pesticides, and inorganic compounds.
The presence in air of some of the suspected or known
carcinogens listed in Table III-l has not been established
by monitoring, but it is highly probable. These compounds
are used in industry or are industrial by-products. Because
of their volatility or association with fume-producing processes,
they are likely to enter the air. Alkylating agents such as
bis(chloromethyl)ether and chloromethyl methyl ether are potent
carcinogens in rodents (Laskin et al. 1971, Leong et al. 1971)
111-31
-------
and humans (Albert et al. 1975, Lemen et al. 1976, Pasternack
et al. 1977, Sakabe 1973). The presence of these substances in
ambient air has not been determined/ but in moist air bis(chloro-
methyl)ether remains stable for at least 18 hours (Collier
1972), a period of time long enough for human exposure to occur.
The presence of carcinogenic substances in the ambient air
strongly suggests that humans are exposed. However, monitoring
data alone are generally inadequate to determine the extent
of exposure of individuals. Given that the average person
inhales from 10 to 20 m /day of air, one can estimate the quan-
tity of the inhaled material to be in the microgram to milligram
range.
Particulate air pollution is an important contributing
source of known and suspected carcinogens in the air. In addi-
tion to the organic compounds, particulate air pollution contains
arsenic, beryllium, cadmium, chromium, lead, nickel, and asbestos.
As discussed in a review of particulate air pollution by USEPA
(1982), there is a multimodal distribution in the size of the
particulates. Particles less than 0.1 ym are in the nuclei
(Aitken) mode and typically originate from combustion sources.
These particles are short-lived because of coagulation of the
.particles into particulates in the 0.1 to 2.5 ym range; the
newly formed particles are considered to be in the accumulation
mode. Particles making up these two modes are termed fine
particles. The final category is composed of particles greater
than 2.5 ym, making up the coarse mode. These particles are
111-32
-------
usually derived from mechanical processes or wind erosion and
are not usually formed to any great extent from fine particles.
Fine particles/ because of their long residence time and atmos-
pheric formation, can build up far from their source, while
coarse particles normally occur only near strong source emis-
sions .
As a general historical perspective, 10% or less of the
total suspended particulate in New York City in the early 1960s
was made up of benzene-soluble organic material. Control pro-
grams put into effect between the early 1960s and the mid-1970s
produced a substantial reduction in total suspended particulates.
With the reduction of particulates there was a marked decrease
in the concentration of benzene-soluble organics and trace
elements (USEPA1982).
A large number of gaseous air pollutants are suspected or
known carcinogens. The concentrations of these compounds are
usually harder to measure than those associated with particulates
because of the difficulty in collecting sufficient amounts to
quantify them. A number of gaseous air pollutants have been
measured by investigators, and these estimates are summarized
by Sawicki (1977) and Brodzinsky and Singh (1982). Singh et al.
(1982) have recently reported the results of a 3-year study
on gaseous air pollutants. They measured 44 different organic
chemicals in 10 cities throughout the United States. In general,
they found a number of known bacterial mutagens and suspected
carcinogens. Most of the compounds measured were in the subparts
111-33
-------
per billion concentration, although concentrations of aromatic
hydrocarbons and formaldehyde averaged 5-20 parts per billion
(ppb). The concentrations of anthropogenic compounds were
generally one or two orders of magnitude higher in urban air
than in rural or clean remote air. Diurnal variations were
observed and depended on source strength and prevailing meteo-
rology. Afternoon mixing led to sufficient dilution to produce
minimum concentrations of several primary pollutants.
D. Multimedia Exposure
In addition to exposure to airborne carcinogens by inhala-
tion, studies of the environmental distribution of air pollutants
indicate that human exposure can also occur through routes other
than inhalation. There is evidence that some substances released
into the air, if unaltered chemically, ultimately end up in soil
and water or on plants, including edible plants.
Arsenic and lead have been studied for their environmental
distribution. Lindau (1977) found arsenic in drinking water
(0.08-3.0 Mg/liter), soil (5-15 mg/kg), and vegetables and
grains (0.1 yg/g). Levels of arsenic measured in the vicinity
of a copper smelter were 500 ug/liter (water), 30 mg/kg (soil),
and 0.06 ng/g (barley). Levels were considerably lower in
samples taken 40 km from the plant. Studies in 32 areas of
the United States showed a correlation between the amount of
lead in rainfall in a given locality and the amount of gasoline
used in that locality (Lazrus et al. 1970). Numerous other
studies have demonstrated an inverse relationship between the
111-34
-------
lead content of grasses and soil and their distance from highways
(NAS I972a). Studies of crop plants indicated that although
the lead content of exposed parts was proportional to air lead
concentrations, the levels of lead in the seeds and roots (the
edible portions) were unaffected (Motto et al. 1970). After
review of this and other studies, the Committee on Lead in
the Human Environment of the National Academy of Sciences con-
cluded that most of the lead content of plants, possibly as
much as 90 to 99%, originates from atmospheric pollution.
They added, however, that this estimate cannot be applied yet
to the edible portions of crop plants (NAS 1980).
Kotin and Falk (1963) demonstrated that BaP is stable in
the atmosphere, both in its crystalline form and when it is
adsorbed on soot. Lunde and Bjorseth (1977) showed that BaP
can be transported long distances in the air. In the United
States, BaP was found in higher concentrations in soil around
petroleum and chemical plants (Menck et al. 1974); in the Soviet
Union, it was found in higher concentrations in soil around air-
fields, coke ovens, and oil refineries (Shabad 1980). According
to Shabad (1980), levels of BaP in water in the Soviet Union
are also higher in industrial areas. Santodonato et al.'s
(1981) summary of multimedia human exposure to polycyclic aro-
matic hydrocarbons (PAH) is found in Table III-2.
Atmospheric deposition of PAH onto food and into water
cannot be considered the only source of PAH exposure via these
routes, however, since food preparation and local effluent
sources may add to PAH levels.
111-35
-------
E. Summary
This chapter compiles and summarizes experimental evidence
and monitoring data. A substantial number of studies have
shown that extracts of airborne materials from polluted air
and materials emitted from motor vehicle engines and stationary
sources are frequently carcinogenic and mutagenic when tested
in experimental bioassay systems. Results of in vivo tests
have included the induction of skin cancers/ lymphomas/ fibro-
sarcomas, liver tumors/ and lung tumors in mice; lung tumors
in rats and hamsters; and chromosome damage and sister chromatid
exchange in hamsters. Respiratory irritants present in polluted
air may also enhance the effects of other carcinogenic agents.
Results of in vitro tests have included the induction of point
mutations in bacteria and Drosophila melanogaster, malignant
transformation of mammalian cells in culture/ and sister chro-
matid exchange and DNA fractionation in cultured mammalian
cells/ including human cells. Positive results in these in vitro
tests are generally correlated with the potential for carcino-
genicity.
Table III-l (in Appendix B) lists more than 50 chemicals
that have been detected in ambient air and that are known or
suspected to be carcinogenic in humans or experimental animals.
Where comparative data are available/ concentrations of these
chemicals tend to be higher in urban areas than in rural areas
and higher still in industrial emissions. There is evidence
111-36
-------
TABLE III-2
ESTIMATED HUMAN EXPOSURE TO PAH FROM VARIOUS AMBIENT SOURCES
(pg/day)
Source
Air
Water
Food
BaP
0.0095-0.0435
0.0011
0.16-1.6
Carcinogenic
PAH3
0.038
0.0042
b
Total
PAH
0.207
0.0270
1.6-16
aTotal of BaP, BjF, and indenod, 2, 3-cd)pyrene
No data available
SOURCE: Santodonato et al. (1981)
111-37
-------
of significant multimedia exposure to several pollutants after
their release into ambient air. Despite the qualitative findings
summarized in this chapter/ incomplete information on exposure
levels precludes quantitative estimates of possible risks.
111-38
-------
IV. QUANTITATIVE ESTIMATES
A. Introduction
Chapters II and III have reviewed the qualitative evidence
for and against an association between air pollution and cancer
rates. This chapter reviews and summarizes estimates of the
possible magnitude of this association—i.e., the number of
cancers that might be "attributable" to exposure to air pollution.
It should be emphasized that quantitative estimates of this
kind can be made (with caution, of course) even if the qualitative
evidence for the association is not regarded as fully conclusive.
The question addressed in this chapter is the following: If
air pollution is a causative factor in human cancer, what esti-
mates can be made of the fraction of human cancers to which
it contributes?
It should be emphasized that the word "contributes" in
this question does not imply that air pollution would operate
independently as a single causative factor. As emphasized
earlier, most cancers have multiple causes, and there is evidence
that air pollution may act in conjunction with other factors
to increase the risk of cancer. Some reviewers have recognized
this by assigning a certain fraction of cancers to more than
one causative factor. One way that has been used to develop
estimates of the fraction of cancers "attributable" to air
pollution is to "subtract out" the effects of other factors.
This is almost certain to lead to underestimation of the contri-
IV-1
-------
button of air pollution, by subtracting out the cancers attributa-
ble to the joint action of these other factors with air pollution
and attributing them solely to the other factors.
B. General Estimates
Because of the limitations inherent in the epidemiologic
studies, estimates of what percentage of human cancers may
be attributable to air pollution have been the subject of dis-
agreement. Several participants who commented on EPA's proposed
airborne carcinogen policy cited estimates by Higginson and
Muir (1979) and Wynder and Gori (1977) that no more than 1%
of total cancer deaths are attributable to air pollution.
The data on which these estimates are based were not fully
described, but these estimates appear to be "subtracted out"
estimates, since in both reviews the fractions of human cancer
rates attributed to various factors add up to 100%. Hence,
these authors implicitly excluded multiple causation.
The most extensive recent review of data providing evidence
for associations between cancer and environmental factors is
that of Doll and Peto (1981). In the conclusion of their review
(Table 20), they proposed 2% as the "best estimate" of the
percentage of all cancer deaths attributable to pollution of
all kinds, with a "range of acceptable estimates" extending
from less than 1% to 5%. Although the basis for these figures
is not completely clear, they appear to have been based on
the estimates of Pike et al. (1975) and Cederlof et al. (1978),
both of which were cited to substantiate the conclusion that
IV-2
-------
urban air pollution (as characterized by BaP) might have con-
tributed to about 10% of lung cancer in big cities, i.e., about
1% of all cancers in the country as a whole. The effects of
industrial emissions were regarded as negligible, and cigarette
smoking was considered sufficient to account for most, if not
all, of the patterns of variation in lung cancer rates, including
urban-rural differences. A critique of this secondary review
paper is presented in Appendix G.
Shy and Struba (1982) presented another review of the
scientific evidence on air pollution and cancer. While citing
some of the epidemiologic and experimental evidence reviewed
in this report, they concluded that "firm conclusions about
air pollution and lung cancer are simply not warranted by the
current state of knowledge." Although they did not make quanti-
tative estimates of the possible magnitude of the association,
they discussed attempts to estimate the risks from exposure
to BaP by linear extrapolation from data on occupationally
exposed workers and concluded that such extrapolation "would
support an extremely low risk (0.1 to 0.01 of a two to threefold
excess) for ambient air." A risk of this magnitude would account
for between 1% and almost 20% (between 0.01(2 - 1) and 0.1(3 - 1))
of all lung cancers, and thus would fall within the range of
other estimates discussed in this chapter. Further comments
on Shy and Struba's review are presented in Appendix G.
IV-3
-------
C. Estimates Based on the Analysis of Epidemiologic Data
A number of investigators have attempted to derive estimates
of the quantitative relationship between lung cancer rates
and air pollution, using BaP and other substances as indices.
Although we believe that BaP has become less and less useful
as an indicator of generalized air pollution over time, we
report 12 published estimates:
• NAS (I972b) based on the data of Carnow and Meier (1973)
• Pike et al. (1975) based on the data of Doll et al.
(1965, 1972)
• Pike et al. (1975) based on the data of Stocks (1957)
• Wilson et al. (1980) reviewing estimates of Pike et
al. (1975)
• Pike and Henderson (1981)
• Lave and Seskin (1977)
• Doll (1978)
• Cederlof et al. (1978)
• Wilson et al. (1980) based on the data of Hammond et al.
(1976)
• Wilson et al. (1980) based on a review of several of
the above estimates
• GAG (1978)
• GAG (1982)
We also present an independent estimate, based on an analysis
of data of Hammond and Garfinkel (1980), as reassembled by
Goldsmith (1980). This is an estimate of the urban-rural differ-
ence in lung cancer rates, after allowance for smoking and
occupational exposures. The residual difference is divided by
IV-4
-------
an estimate of the urban-rural difference in ambient concentrations
of BaP to obtain an estimate of the quantitative relationship
between air pollution and lung cancer rates. Although Hammond
and Garfinkel (1980) reported data on air pollution levels
for some subpopulations in their study, these data were obtained
after the mortality from lung cancer had occurred. These after-
the-fact data are not used in our independent computation.
An earlier review by the National Academy of Sciences'
Committee on Biologic Effects of Atmospheric Pollutants (NAS
1972b, p. 246) laid out the argument for using BaP as an air
pollution indicator:
It appears/ then, that there is an "urban factor"
in the pathogenesis of lung cancer in man. The poly-
cyclic organic molecule mentioned most prominently
in this report has been benzo(a)pyrene. It was felt
that benzo(a)pyrene could be used as an indicator
molecule of urban pollution, implying the presence
of a number of other polycyclic organic materials of
similar structure that may also have some carcinogenic
activity. The standard measure of benzo(a)pyrene con-
centration in the air is the number of micrograms per
1,000 m of air. On the basis of epidemiologic data
set against information regarding the benzo(a)pyrene
content of the urban atmosphere, one can develop a
working hypothesis that there is a causal relation
between air pollution and the lung cancer death rate
in which there is a 5% increase in death rate for
each increment of urban air pollution. In this study,
an increment of pollution corresponded to 1 pg of
benzo(a)pyrene per 1,000 m of air. On the basis
of this assumed relation, a reduction in urban air
pollution equivalent to 4 benzo (a )pyrene units (i.e.,
from 6 ug/1,000 m to 2 yg/1,000 m ) might be ex-
pected to reduce the lung cancer death rate by 20%.
These data, however, are not to be interpreted as in-
dicating that benzo(a)pyrene is the causative agent
for lung tumors. There is much to support the idea
of synergism or cocarcinogenesis, especially with
respect to cigarette smoking. In addition, the car-
cinogenic significance of other polycyclic organic
molecules in urban air pollution should be determined.
IV-5
-------
However, BaP seems to have become less useful, with time,
as a general indicator of air pollution. A recent review of
problems associated with air pollution (Karolinska Institute
Symposium on Biological Tests in the Evaluation of Mutagenicity
and Carcinogenicity of Air Pollutants, 1982), subsequently
published in Environmental Health Perspectives (Holmberg and
Ahlborg 1983) came to the conclusion:
At the present time there is no way to quantitate
how changes in air pollution levels may have reduced
mortality from lung cancer because there has been
a lack of a completely reliable indicator of air
pollution carcinogenicity.
The Karolinska 1982 review repeated the conclusions of an earlier
review (Cederlof et al. 1978) that "combustion products of
fossil fuels in ambient air, probably acting together with
cigarette smoke, have been responsible for cases of lung cancer
in large urban areas, the numbers produced being of the order
of five-ten cases per 100,000 per year" and indicated no basis
for any revision of the conclusions drawn by NAS (1972b) in
view of current data. Five to 10 cases per 100,000 per year
was about 12% to 23% of all lung cancer cases in the mid-1960s
(Table IV-1) .
There is evidence that BaP levels have decreased in the
United States (CEQ 1980) in the past 20 years, without a propor-
tional decrease in all other air pollutants—thus making BaP
a poor index of current trends in air pollution levels. In
1958-1959, the median level of BaP measured in urban air was
about 6 ng/m (range, 1-60 ng/m ) and that in rural air was
IV-6
-------
about 0.4 ng/m3 (Sawicki et al. 1960). By the mid-1960s, the
median level at urban sites was reduced to 3.2 ng/m , and by
the mid-1970s it was reduced to below 1.0 ng/m (Wilson et
al. 1980, CEQ 1980, Shy and Struba 1982). Although earlier
measurements are not available for the United States, Shy and
Struba (1982) suggested that levels in the 1930s and 1940s
would have been several-fold higher. Wilson et al. (1980)
cited data compiled by Ludwig et al. (1971), which showed that
dustfall rates declined by a factor of about 2 in Pittsburgh,
Cincinnati, and Chicago between 1935 and 1958, and by the same
factor in New York City between 1945 and 1958. Since much
of the dustfall in these urban areas was associated with incom-
plete combustion of fossil fuels in these periods, these rates
may provide a surrogate measure of likely changes in BaP levels
However, there was no marked change in dustfall rates between
1958 and 1966, a period in which the data cited above suggest
a substantial decrease in BaP levels. Levels of BaP in the
United Kingdom in the mid-1970s were several times higher than
in the United States, probably in the range of 3 to 5 ng/m
(Lawther and Waller 1978, Wilson et al. 1980).
One consequence of the changes in BaP levels since the
1950s (or earlier) is that differences between regions (e.g.,
between urban and rural areas) have been reduced (CEQ 1980),
so that associations between air pollution (as measured by
BaP) and cancer rates are more difficult to demonstrate. Com-
paring cancer rates to contemporaneous BaP levels is likely
IV-7
-------
to overestimate the strength of the relationship between them
because cancer rates are actually influenced by exposures 20
or more years earlier, when BaP levels (and differentials)
were higher. Also, as discussed in Section II.B.S.d., levels
of other carcinogenic components of ambient air have probably
increased, while those of polynuclear aromatic hydrocarbons
(of which BaP serves as an index) have decreased. Hence, apply-
ing the relationship of present-day cancer rates to BaP levels
at some time in the past will underestimate both the hazards
posed by present-day ambient air and the contribution of present-
day exposures to future cancer risks.
Recognizing these difficulties, we have summarized in
Table IV-1 the estimates made by others of the dependence of
lung cancer rates on BaP levels. Most of these estimates were
obtained by linear regression techniques (i.e., calculation
of the linear relationship between differentials in lung cancer
rates and differentials in BaP levels). Hence, the dependence
of lung cancer rates on air pollution levels is expressed in
units of incremental lung cancer rate per ng/m of BaP.
The 13 estimates reviewed in this chapter are listed in
the second column of Table IV-1. In comparing these estimates,
it should be recognized that they fall into two categories
that are not strictly comparable. The estimates by GAG (1978,
1982), Pike et al. (1975) based on data of Doll et al. (1965,
1972), Pike and Henderson (1981), and Wilson et al. (1980)
based on data of Hammond et al. (1976) were based on studies
IV-8
-------
CQ
d
>
3
S
D4
CO
o
l-H
Jj
^
B
EH
l-l
3
w
•^»
tH W
> 0
^ rn
M 52
j
3 CO
S
El*
ri
o
^
U
O
rn
w
tz
fa
O
CO
e*
5
|
C-t
S
w
^J1 rH
o m in
O 4-1 VO
O H-4 O ON
O EH rH
0
O d*> • Q)
•HO) W 4J
14 41 *2 ^ K
41 M
04 2
w c
id id
41 .Q rH
>* M on
3 > S
41 c >-5 cn
O4 -rl C
c
09 Id ^*
r3 41 •
id 04
4) 4J Irl IB
Q < 0 0
^^
4-rH O4
C 00 M
m rf CQ
•H 4JO
O IQ S
1 -H 4)^-
41 i
•n to 3
< K^ 4)
a
^
o^
id
44-H £0
C CO
41 ,ffO
•n 4J B
O ,
ON
rH
M
4)
iH
41
£
•O
ft
id
0
X co
0 ON
C rH
3 «
id
CM 4)
r-
ON C
rH O
a
CO rH
2 S
ON
VO
CM
m
o
x
,
o
m
i
0
x
d
4-1
41
rH
VM
O
id
id
•o
m
ON CN;
ON
• rH
rH
« u?
4i VO
41 ON
rH
41
X*^
04 m
00
rH
CM
,
in
-
09
M
*— g)
09 X
M Q
4> 8
X 09
o cJ
8 O
09 C
in in
o'o
3-t
• •
rH 0
^,
a
rf-K Q)
to ,&4
VM O
0) 0
££ 00
O C
S O
n d
in in
l i
00
x x
rH O
^i*
p-
in
rH rH
in
r*- 09
S|
rH CO
id
4-> O
4)
id
4) 4J
fiftf iTf
•rt TJ
cu *-*•
00
•H
n
f
in
in
o
&
•
o
in
i
o
X
oo
d
^,
in
f>«
ON
rH
,
rH
id
0 4J
CO 4)
ON
X
• -rl
rH 04
id
4-> O
41
S
C 41
O -n
CO >
rH 4)
•H M
00
rH
CN
•
in
in
O
X
oo
o
in
l
o
X
co
o
§
•H
0
09
4)
T3
I
•a
c
id
4)
X
O4
CO
CO
1
co
CO
i
CM
in
'o
cS
•
i-H
1
•
0
in
i
o
eg
in
l
CO
•
rH
f"»
r-
rH
C
IH
X
09
a
ON ON
rH
in CM
• •
^* CM
^^
09
14
^. 41
cox
rl O
4) B
X 09
§C
0
•US
in in
00
x x
r~ in
• CO
0 •
O
jf^
09
M
*-* 4)
09 X
V4 O
4) B
X 09
0 fl
8 O
09 C
in in
l l
0 0
x x
ON in
CM rH
oo
ON
rH
-H
rH
8
rH C
4) 0
10 4J
B 41 41 Id
•rl B C CM
X O In 41 4) 3
•*•* rn ~~. ocno-csa
*t " ' « 4J M .Q. 4J O
rHCM'j1 coco x a4inca
i CM 41 a o -H a -H
t- 4J Id C Id --H iH
•H x id
4) • to W ,C to to U
01 ~ Ifl 4J V4 -rl 41
to 4.1 3t -rt 01 l-i C
— X « 3 X 41
4) *-%x-» 0 CO ON
CN UN 4J C U T3 B cn
. VO O 41 CO 41 41
CMOrH Oin in 3410ON4IC
ON in a a to c ,C O
rH Tt o a to o 4J a
ON a5f « B .a
TJ 1 41
41 C CO rH >O
41 4) r? O 09 01 4)
ff U 4J -rl 01 B 4-1 U
4-> C 4J O 41 Id 41 '
•HOidXOBU —
Ct04JrH4IX-rlC 4J
•H 3 41 X id X
gJiQjvq) OO "41
a4J ,e a x 4^
,C-r>lCnOt71klCO *— ' -r4
«rH. rH. ldrHld41C4)4l *>
O • O • O r?4l— 'r^OiHXU tO 01
1 O * O 4J X 4J 4J B O |H3
co* * -HinsBvMVXl
4)rHVOcnkl0909O aS
O ooNcOn idO
* O BO 41 01
4-1 VM Cidid>iB-n rH 4)
•H IM XI 0 M Id *J
09 3 O X C 4)
(0 O tO O 10 - CO -HrH
* rH 01 o ccntn cn a
rH 4J -"O O C 41 -rl 8
41 id ifl o oi -n o u t-i o
u ^>oa4JBx oo
3OCOrHOidid4> C
09 4-1 rH rH C -rl
aT3 14 41 3 X 4) -H
CN 4) o> > a to ,c VH
x 4-1 a 01 o -H +j T3 o
41 N -rl TJ Q4 M 41
O C 01 4J • 4J 09
in in in inm O'O 4J>idn,cO4)o
1 1 1 1 1 •n4l41ldldM4)4-» 013
ooo oo 4J -ri ,e 01 ,c o> o xoi'O
xxx xx i-ia oi 01 o> c 13 a v-i
«*rHCM ^-r* 3-nC^SON-rlC a
• inJ • 1-4 • Q. It ^^ ^»J rt) ') | (M (M
CN-O -r-i 9-t -oi^ajjo
i o * o • a 3 4i TS NX: M 41 a id 4J
CM* * C B4JCrH4JX-MOCifl-oiam
4J-H01 09 -rl ,£ 0] 01 0 tO
«4JO>i4l01 4UB01MQ
rH OVO 4JrH4)X O^-VH O,
9 4) 0^ iH (fl rH OB *O *^4 M
O,VM rH rH B 10 -r< C 4J 01 01 41
^ OVM 10 e vj fl a c o)
vo T) a 4) >, 4J Cn s O n to
p"% g ,_^ 1^ C ^( W (Q 3 ^H (tj iM
ON o • 4J Vj 0 O O to 01 0) iO O 01
rH B COtOtdSgMHVrHCO X
S >r?4l Id OldldOOrHlH
. id D 4J u Minid4)
00 VM C«CrHCOld4l4lE>IM
r» 44 o— ON-ricDid-B.C'O.c 410
ON O CU O rH B ^ O O -*H 4J C 43 C 0}
rH 00 Ifl 00 3 01 O 4J Ifl -r! CO
ON >O 4J ON 0 CO I-H o O C C 4)
rH o -a (0 c o ij 01 > ro -a
Id '1 — -H T3 ld4J4)4ICOrH-HX:'O3
rH S 4) 41 >i A id aT) 01 ifl (3 4-1 41 4J
4J 10 Ifl 4-1 X NAMrlldrHB 4-1 tO
41 93 >nC -H 3-nnBid vuid
4-1 O-H T3 T3 ax! SM'OrHC
VM 00 4) VM CM a*** M-4J4I094JCA Oca 3O
0 r>» 0 <0 4)M id4-iON4)id4IC|MUc«>O
rHON C C^ Mid a4tXCOld4l
4) 094-1 09 ld-(-l4)4l3OiBO OCO
"O O "Hid O -HIO 4J«>gfO4->«jidStaoifll
CJ U S^^ OtHid rHCNcot in*
IV-9
-------
of workers occupationally exposed to products of incomplete
combustion. in these studies/ BaP was used as an index of
exposure to these products. The remaining estimates were based
primarily (or entirely) on studies of the general population
exposed to ambient air, and BaP was used as an index of exposure
to a wider mixture of materials. The fact that the worker
studies yield lower estimates of dose-response coefficients
(0.1-0.8 x 10~5) than the population studies (0.8-5.0 x 10~5)
(p<0.01, Mann-Whitney test) suggests that products of incomplete
combustion may be associated with only a part of the excess
of lung cancers in urban areas/ thus making BaP a poor indicator
of total air pollution.
The principal limitation in quantifying the population
studies is that they all relate cancer deaths observed in the
period 1959-1975 to BaP levels measured or estimated for the
period 1958-1969. If levels of BaP and related products of
incomplete combustion were higher in the 1930s and 1940s (when
there were more coal-burning emissions but fewer automobiles),
these studies would overestimate the dose-response coefficient
between lung cancer rates and BaP levels. As an illustration
of the likely magnitude of this effect, we present in the third
column of Table IV-1 adjusted estimates of the dose-response
coefficient, derived by assuming that the effective population
exposure to polluted air for cancers developing in the 1960s
and later was in the period 1935-1945 and that levels of BaP
at that period (using dustfall rates as a surrogate index of
IV-10
-------
likely BaP levels, as discussed above) were about twice those
measured in the early 1960s. For the estimates based on European
studies (Lave and Seskin 1977, Doll 1978, Cederlof et al. 1978),
the figures in the second column were based on a value of 3.5 ng/m
for the urban-rural differential in BaP levels, which was appro-
priate for the mid-1970s. We have adjusted these European
figures by a factor of 4 to account for the assumed reduction
in BaP levels since the period 1935-1945. Estimates based
on studies of occupationally exposed workers have not been
adjusted.
To place these estimates of dose-response relationships
in quantitative perspective, the last two columns in Table IV-1
present calculations of the number of lung cancer deaths that
could be attributed to air pollution characterized by 6.4 ng/m
of BaP. This is approximately twice the average level of BaP
to which the U.S. population was exposed in the mid-1960s,
and hence is the average level of BaP to which we have assumed
the urban U.S. population was exposed in the period 1935-1945.
The figures in the last two columns of Table IV-1 are derived
by multiplying the "adjusted" dose-response coefficients in
the third column by 6.4 ng/m . These figures give estimates
of the number of lung cancer deaths in the 1960s associated
with air pollution levels in prior decades. The estimates
based on studies of the general population fall into the range
between 2 and 8 deaths per year per 100,000 people, or between
8% and 33% of the lung cancer rates in the mid-1960s. The
IV-11
-------
estimated median from these population studies is 4.5 deaths
per year per 100,000, or about 19% of the lung cancer deaths
in urban areas of the United States in 1965.
These estimates of the association of BaP-indexed air
pollution with lung cancer rates in the 1960s are not sensitive
to changes in our assumption about BaP levels in the period
1935-1945. If we had assumed a higher figure for average BaP
levels in that period, our estimates of adjusted dose-response
coefficients in the third column of Table IV-1 would have been
lower, but the multiplier used to derive the estimates in the
last two columns would have been correspondingly higher.
Despite the relative stability of these estimates, they
unfortunately cannot be used to generate reliable estimates
of the future effects of present air pollution, or even to
make firm estimates of the contribution of past air pollution
to current cancer rates. This is because BaP is not a stable
index of the carcinogenicity of polluted air. Although the
general population exposure to BaP and to other products of
incomplete combustion has decreased considerably since the
1950s, it appears unlikely that the carcinogenicity of polluted
air has decreased in direct proportion. The fact that BaP
levels relative to other air pollutants have changed with time
implies that all of the estimates in Table IV-1 are time-depen-
dent. Hence, they cannot be used to predict the future conse-
quences of present-day air pollution, using BaP levels as a
surrogate for all air pollution.
IV-12
-------
Despite these limitations, each of the studies listed
in Table IV-1 is considered in more detail below.
Carnow and Meier (1973) estimated the risk of lung cancer
mortality by relating deaths in 1960 to levels of BaP in 1968.
Wilson et al. (1980) reduced this estimate by half to 1.0 death/105
males per ng/m of BaP. Wilson cited monitoring data from
28 sites in 1959 that seemed to indicate that levels of BaP
had declined. There are few data on levels of BaP before 1966,
and it is not possible to establish whether or not Wilson et
al.'s correction was appropriate. The more complete monitoring
data available from 1966 to 1977 indicate that levels from
1966 to 1969 were steady or slightly increasing (CEQ 1980)
and then declined. Thus, we do not know whether or not Wilson
et al.'s correction of Carnow and Meier's estimate is the same
as the adjustment we have applied in Table IV-1 to allow for
the likely reduction in BaP levels prior to 1959. For this
reason, either of the two adjusted estimates may be appropriate.
Pike et al. (1975), assuming a linear relationship between
exposure and carcinogenic response, extrapolated the results
of a study of gas workers by Doll et al. (1965, 1972) to the
general population:
The carbonization workers were exposed to an estimated
2,000 ng/m BP [BaP] for about 22 percent of the
year (assuming a 40-hour working week, 2 weeks paid
leave, 1 week sick leave); very roughly, the men were
exposed to the equivalent of 440 (2000 x 0.22) ng/m
BP general air pollution. This exposure caused
an extra 160/10 lung cancer cases, so that we may
estimate, assuming a proportional effect, that each
ng/mj BP causes 0.4/10^ (160/10^ divided by 440)
extra lung cancer cases per year. A city with 50
IV-13
-------
ng/m BP air. pollution might/ therefore, have an
extra 18/10 lung cancer cases per year. These
numbers are not negligible, although they are small
when compared, say, to smoking a pack of cigarettes
every day.
(Pike et al . 1975, p. 231)
Thus, based on the experience of carbonization workers, Pike
et al. (1975) estimated the risk of lung cancer mortality as
0.4 deaths/10 persons per ng/m of BaP.
Wilson et al. (1980) reexamined this estimate by Pike
et al. (1975) and included a doubling factor to correct for
the fact that the gas workers were not exposed for all of their
lives, leading to an estimate of 0.8 deaths/10 persons per
ng/m BaP. However, neither Pike et al. (1975) nor Wilson
et al. (1980) made any further adjustment for the fact that
the gas workers were not all followed up to their deaths.
Because the incidence of human lung cancers increases in propor-
tion to the 4th or 5th power of age (or duration of exposure)
the possibility exists that Pike et al. (1975) and Wilson et al.
(1980) have underestimated the full lifetime cancer risks,
probably by a factor of 3 or more. For example, comparing
exposures beginning at birth and continuing for a lifetime
with industrial exposures beginning at age 20, and assuming
4
a 73-year life span, implies a risk ratio of (73/53) =3.6.
Thus, Pike et al.'s original estimate may be as low as one-sixth
or one-seventh of the appropriate estimates. However, we have
not amended either estimate to take account of this factor.
Pike et al. (1975) also used the data of Stocks (1957)
to obtain a second estimate.
IV-14
-------
Second, there should be an increased lung cancer
rate in high PAH-polluted areas (25); the effect
is magnified in most studies when we consider the
joint effect of urbanization and cigarette smoking.
Table 2 presents data (26) comparing rates in Liver-
pool to those in rural North Wales. This study
by Stocks was done in an area of "stable" air pollu-
tion. A fair summary of these data is that the
urban effect produces an excess of 28/10 lung cancer
deaths in nonsmokers and 100/10 such deaths in
smokers, the latter figure being independent of
the actual amount smoked. We might refer to this
increase as a modified additive effect. The differ-
ence in BP levels in the air between the two areas
was estimated to be 70 ng/m (77 ng/m compared
to 7 ng/m ); thus, we may very crudely estimate
the air pollutiongeffeet in.the presence of cigarette
smoking at 1.4/10 per ng/m BP or 0.4/10 per ng/m
BP in nonsmokers (Table 3).
(Pike et al. 1975, pp. 231-232)
Based on the prevalence of smoking in the United States in
the recent past (i.e., approximately one-third of all adults
are smokers), this estimate leads to a risk of lung cancer
5 3
mortality of 0.8 deaths/10 persons per ng/m of BaP. (if
based on earlier smoking habits, this estimate would be higher;
Wilson et al. (1980) listed this estimate as 1 death/10 persons,
possibly because it was based on past smoking habits.) This
estimate may be low if, as appears likely/ the estimated average
difference in BaP levels between urban and rural areas is great.
Pike and Henderson (1981) estimated the quantitative rela-
tionship between lung cancer risks and exposure to cigarette
smoke (data from various sources), coke oven emissions (data
from Lloyd 1971 and Redmond et al. 1972), and hot pitch fumes
(data from Hammond et al. 1976). They calculated that exposure
to about 15 ng/m BaP could be equated to smoking 1 cigarette
per day, and hence estimated the "single cause lifetime jrisk"
IV-15
-------
of lung cancer to age 70 resulting from ambient air exposure to
3 5
1 ng/m BaP as 73x10" . This corresponds to an age-standardized
lung cancer rate of about 0.8x10" deaths per year per ng/m BaP.
Carnow (1978) suggested a number of factors that may have
led Pike et al. (1975) to underestimate the risk. Although Pike
et al.'s estimates of the risk of lung cancer mortality may be
low, it is likely that 3.5 is a reliable estimate of the ratio
of the risk for smokers to that for nonsmokers (1.4/0.4 = 3.5),
although 3.5 is lower than the usual estimates of the relative
risks for smokers. Wilson et al. (1980) reported that this
difference (3.5) is statistically significant. Knowing this
value, plus making some reasonable assumptions, permits the
estimation of the average risk to the general population from
data on males alone (see Appendix D). The general population
excess is about 82% of the male excess.
Based on extensive regression analyses of lung cancer
mortality and air pollution levels, Lave and Seskin (1977)
suggested that
if the quality of air of all boroughs (England)
were improved to that of the borough with the best
air, the rate of death from lung cancer would fall
by between 11 and 44 percent.
This corresponds to 4.4 to 17.6 deaths/10 persons at British
levels of pollution (assumed to be 3.5 ng/m of BaP) or 1.3
to 5.0 deaths/10 persons per ng/m of BaP.
Doll (1978) estimated that the risk of lung cancer mortality
attributable to urban air pollution in Europe was no more than
10 deaths per 10 smokers and no more than 5 deaths per 10 non-
IV-16
-------
smokers. Based on current U.S. smoking habits, this estimate
corresponds to 6.7 deaths/10 persons or 1.9 deaths/10 persons
per ng/m / taking average levels of BaP to be 3.5 ng/m in
Europe. Doll (1978), however, provided data indicating that
levels of BaP ranged much higher than 3.5 ng/m in highly urban
areas in Britain. He also estimated the attributable risk
in smokers to be twice the risk in nonsmokers, an estimate
which is lower than the 3.5-fold ratio derived by Pike et al.
(1975). However, it is not possible to ascertain whether the
former is too high or the latter is too low since the ratio
cited by Doll (1978) was a personal estimate by the author and
was not based on any specific calculation or data. No data
were cited to support Doll's estimates of attributable risk.
Cederlof et al. (1978, p. 9), summarizing the conclusions
of a conference on air pollution and long-term health effects,
stated:
Combustion products of fossil fuels in ambient air,
probably acting together with cigarette smoke, have
been responsible for cases of lung cancer in large
urban areas, the numbers produced being of the order
of 5-10 cases per 100,000 males per year (European
standard population) The actual rate will vary
from place to place and from time to time, depending
on local conditions over the previous few decades.
This estimate was a synthesis of material presented at a con-
ference, and the basis for it was not provided in detail.
Taking the risk to the general population as 82% of the risk
to males (see Appendix D) and average European levels of BaP
as 3.5 ng/m , this estimate corresponds to 1.2 to 2.4 deaths/10'
persons per ng/m BaP.
IV-17
-------
The Carcinogen Assessment Group (GAG 1978) of the Environ-
mental Protection Agency reviewed a number of epidemiological
and animal studies in an attempt to estimate the "excess lung
cancer incidence" resulting from lifetime exposure to polycyclic
organic compounds. For their overall estimate, GAG (1978)
took the geometric mean of the estimates derived from four
epidemiologic studies. (Using the geometric mean produces
lower estimates than using the arithmetic mean. If risk is
linearly related to exposure, the arithmetic mean is more appro-
priate.) This overall estimate was expressed as 0.28% excess
lung cancer "incidence" (a slight misnomer since all the studies
were mortality studies) per ng/m of BaP. As a percentage,
this estimate is a ratio of the estimated excess lung cancer
mortality rate to the background rate. This would correspond
to about 0.11 deaths/10 persons per ng/m BaP.
By expressing the estimate in this way, GAG (1978) assumed
that the effect of exposure to each ng/m of BaP is dependent
on the background rate of lung cancer mortality in the exposed
population. This means that if the background rate were high,
the effect would be large; but if the background rate were
low, there would be little or no effect. It is reasonable
to expect that the magnitude of the effect attributable to
BaP will vary as a function of the presence or absence of sub-
stances (such as cigarette smoke or other carcinogenic air
pollutants) that interact with BaP in the induction of cancer.
However, it is not clear why this variation should otherwise
depend on the background mortality rate of lung cancer.
IV-18
-------
In 1982, GAG (1982) updated one of the 1978 estimates.
Reviewing the results of epidemiologic studies of workers exposed
to coke oven emissions, they estimated that the unit risk (for
males) of dying from lung cancer as a result of a working life-
_ c o
time of exposure to BaP is 9.25x10 per ng/m of BaP. This
corresponds to a rate of about 0.14x10" deaths per year per
ng/mg of BaP. However, this estimate is not comparable with
some others in Table IV-1 because it was calculated exclusively
for exposure to products of incomplete combustion (as indexed
by BaP), whereas others were calculated for air pollution (as
indexed by BaP) with other pollutants assumed to be present
in proportion to the BaP values. The latter type of calculation
includes the effect of compounds of air pollution other than
products of incomplete combustion (such as asbestos and synthetic
organic chemicals), whereas CAG's 1982 estimate does not.
CAG's 1978 estimate appears to have included and averaged esti-
mates of both types.
Wilson et al. (1980) derived an estimate of lung cancer
mortality of 0.2x10 per ng/m BaP using the data of Hammond
et al. (1976) on a group of roofers and waterproofers working
with pitch and asphalt. The estimate appears to be too low,
primarily because the comparison group was made up of other
members of the workers' own trade union. This would tend to
underestimate the risk if other members of the trade union
were already at increased risk of lung cancer, as seems likely
from other occupational studies.
IV-19
-------
The estimates made by Carnow and Meier (1973), Pike et
al . (1975), Hammond et al. (1976), CAG (1978), and the authors
of some animal studies of BaP, which were assembled by Wilson
et al. (1980), indicate that the effect of BaP in the animal
studies is much smaller than the "enhanced" effect attributable
to BaP from occupational or urban epidemiologic studies. The
arithmetic mean of the estimates from the epidemiologic studies
led Wilson et al. (1980) to what they called a "best estimate"
of 0.5 deaths/10 persons per ng/m of BaP. There are several
problems with this "best" estimate, not least of which was
that several of the separate estimates (Carnow and Meier 1973,
Pike et al. 1975, Hammond et al. 1976) appear to have entered
Wilson's calculations more than once.
The estimates derived by CAG (1978) differ from those made
by Wilson et al. (1980, Table 5-4) from the same studies.
For example, Wilson et al. (1980) estimated the Carnow and
Meier (1973) response coefficient as 1.0 death/10 persons
per ng/m of BaP. CAG (1978) reduced this estimate to less
than one-tenth of this figure. Also, as indicated earlier,
the estimate of Pike et al. (1975) based on the data of Doll
et al. (1965, 1972) was modified by Wilson et al. (1980) to
0.8 deaths/105 persons per ng/m of BaP. In the CAG (1978)
analysis, this figure was given as 0.57 deaths/10 persons
per ng/m3 of BaP (160/105 divided by 283 ng/m3 of BaP) and
was then converted to a percentage by dividing it by an anoma-
lously high background rate of lung cancer mortality (0.57/10
IV-20
-------
divided by 200/105=0.285%). This final CAG estimate is close
to CAG's overall figure of 0.28% and was converted by Wilson
et al. (1980) to 0.12 deaths per 105 persons (0.28% x 40 deaths/105
persons = 0.11/10 ). (Note that Wilson et al. used a background
rate of 40/10 persons/ while CAG used a background rate of
200/10 persons—a five-fold difference. The age-adjusted
mortality rate in the United States for all respiratory cancers—
i.e., lung cancer plus others—was 45.9/10 persons in 1979.)
The last estimate listed in Table IV-1 was developed for
this report and takes account of the criticisms and suggestions
made concerning earlier estimates (Clement 1981, Karch and
Schneiderman 1981). The detailed derivation of this estimate
is given in Appendix E. The estimate follows from the lung
cancer mortality data of Hammond and Garfinkel (1980) as reas-
sembled by Goldsmith (1980), standardized for age and smoking,
and stratified by occupational exposure and location of residence.
These data show that urban residence and occupational exposure
have significant independent effects, and we calculate an
attributable risk of 13% for occupationally exposed and 12%
for nonexposed categories. It is likely that these figures
Hammond and Garfinkel (1980) and Goldsmith (1980) expressed
the opinion that these data did not show a convincing effect
attributable to air pollution. However, neither set of authors
analyzed the data in the way presented here (in Appendix E)
to test the effect of urban residence. Hammond and Garfinkel
(1980) reported no statistical association between lung cancer
rates in the 1960s and measures of air pollution that were made
in 1968. They apparently assumed that no change in pollution
(relative or absolute) had taken place between the 1940s—when
the cancer cases that appeared in the 1960s were initiated—
and 1968 when their two air pollution measures were made.
IV-21
-------
are biased downwards (possibly by factors between 1.4 and 3.3,
as discussed on p. E-8) because of selection bias in the study
population. The population studied by Hammond and Garfinkel
was more suburban, had a higher percentage of whites/ had a
lower percentage of blue-collar workers, and was more educated
than the U.S. population as a whole. However, no attempt is
made to correct for this bias here.
D. Summary
This chapter summarizes attempts to estimate the possible
magnitude of the association between lung cancer mortality
rates and air pollution levels. The index of air pollution
most commonly used has been the average atmospheric concentration
of benzo(a)pyrene (BaP) . Using this index, however, creates
problems because average levels of BaP in the United States
have declined considerably since 1966 and probably were higher
still prior to 1966. However, it is not clear that the overall
hazards posed by air pollution should have declined proportionately
because there is evidence that levels of other potential carcin-
ogens have increased since 1940. Thus, BaP is no longer a
stable index of the carcinogenicity of polluted air, and estimates
made for one time period cannot be applied directly to others.
Therefore, the estimates based on studies of lung cancers in
the past cannot be used directly to predict future effects
of current pollution.
Recognizing this problem/ Table IV-1 tabulates 13 estimates -
(but they are not based on 13 independent studies) of the quanti-
IV-2 2
-------
tative relationship between lung cancer rates and air pollution
levels/ as indexed by BaP concentrations. Estimated slopes
(regression coefficients) of this relationship range from O.lxlO"
to 5.0x10" lung cancer deaths per year per ng/m BaP. Some
of these figures should probably be adjusted downwards by factors
of 2 to 4 to take account of the likely reduction in BaP levels
since the 1930s and 1940s, when most effective exposures took
place. The estimates derived from studies in the general popu-
lation (0.8-5.0x10" ) are significantly higher than those derived
from studies of workers exposed to products of incomplete combus-
tion (0.11-0.8x10" ). This difference suggests that incomplete
combustion products are associated with only part of the excess
lung cancer rates observed in urban areas. Most of the studies
were based on lung cancer mortality data from the 1960s, and
the results are consistent with the hypothesis that at that
time factors responsible for the urban excess in lung cancer
were associated with about 19% of lung cancers in urban areas
of the United States. In the one study in which both cigarette
smoking and potential industrial exposure were taken into account,
this estimate was about 23%. These quantitative estimates
can be derived without resolving the issue of whether the un-
explained urban excess of lung cancer can or cannot be attributed
confidently to air pollution, which depends on an interpretation
of the data summarized in Chapter II.
IV-2 3
-------
REFERENCES
AIR PRODUCTS AND CHEMICALS CORP. 1980. Comments on Environ-
mental Protection Agency Proposed Rulemaking: National
Emission Standards for Identifying, Assessing and Regulating
Airborne Substances Posing a Risk of Cancer (44 FR 58642).
OAQPS Document No. 79-14
ALBERT, R.E., PASTERNACK, B.S., SHORE, R.E., LIPPMANN, N.N.,
and FERRIS, B. 1975. Mortality patterns among workers
exposed to chloromethyl ethers—a preliminary report.
Environ. Health. Perspect. 11:209-214
ALBERT, R., SELLAKUMAR, A., LASKIN, S., KUSCHNER, M., NELSON, N.,
and SNYDER, C. 1982. Gaseous formaldehyde and hydrogen
chloride induction of nasal cancer in the rat. JNCI 68:597-
603
ALTHOUSE, R., HUFF, J., TOMATIS, L. , and WILBOURN, J. 1980.
An evaluation of chemicals and industrial processes asso-
ciated with cancer in humans based on human and animal
data. IARC Monographs Vols. 1-20. Cancer Res. 40:1-12
AMERICAN CANCER SOCIETY (ACS). 1980. Cancer Facts and Figures,
1980. New York
AMERICAN INDUSTRIAL HEALTH COUNCIL (AIHC). 1981. Posthearing
Comments on Environmental Protection Agency Proposed Rule-
making: National Emission Standards for Identifying,
Assessing and Regulating Airborne Substances Posing a
Risk of Cancer (44 FR 58642). OAQPS Document No. 79-14
AMES, B.N., DURSTON, W.E., YAMASAKI, E., and LEE, F.D. 1973.
Carcinogens are mutagens: A simple test system combining
liver homogenates for activation and bacteria for detec-
tion. Proc. Natl. Acad. Sci. USA 70:2281-2285
AMES, B.N., MCCANN, J., and YAMASAKI, E. 1975. Methods for
detecting carcinogens and mutagens with the Salmonella/
mammalian-microsome mutagenicity test. MutaFT Res. 3T:
347-364
ANDERVONT, B.H., and SHIMKIN, M.B. 1940. Biological testing
of carcinogens: II. Pulmonary tumor induction technique.
J. Natl. Cancer Inst. 1:225
ASAHINA, S., ANDREA, J., CARMEL, A., ARNOLD, E., BISHOP, Y.,
JOSHI, S., COFFIN, D., and EPSTEIN, S.S. 1972. Carcino-
genicity of organic fractions of particulate pollutants
-------
collected in New York City and administered subcutaneously
to infant mice. J. Cancer Res. 32:2263-2268
ASARCO. 1980. Comments on Environmental Protection Agency
Proposed Rulemaking: National Emission Standards for
Identifying, Assessing and Regulating Airborne Substances
Posing a Risk of Cancer (44 PR 58642). OAQPS Document
No. 79-14
AXELSON, O., EDLING, C., and KLING, H. 1981. A case-referent
study on the possible impact of exposure to radon and
its daughters in dwellings. Scand. J. Work Environ. Health
5:10-19
BADGER, G.M., COOK, J.W., HEWETT, C.L., KENNEWAY, E.L., KENNEWAY,
N.M., MARTIN, R.H., and ROBINSON, A.M. 1940. The produc-
tion of cancer by pure hydrocarbons. Proc. R. Soc. Lond.
129:439
BEAN, J.A., ISACSSON, P., HAHNE, R.M.A., and KOHLER, J. 1982.
Drinking water and cancer incidence of Iowa. II. Radio-
activity in drinking water. Am. J. Epidemiol. 116: 924-
931
BIDLEMAN, T.F., and OLNEY, C.E. 1974. Chlorinated hydrocarbons
in the Sargasso Sea atmosphere and surface water. Science
183:516-518
BINGHAM, E., and FALK, H.L. 1969. Environmental carcinogens:
The modifying effect of cocarcinogens on the threshold
response. Arch. Environ. Health 19:779
BLOT, W.J., and FRAUMENI, J.F. 1975. Arsenical air pollution
and lung cancer. Lancet, July 26, 1975, pp. 142-144
BLOT, W.J., and FRAUMENI, J.F. 1976. Geographic patterns
of lung cancer: Industrial correlations. Am. J. Epidemiol.
103:539-550
BLOT, W., and FRAUMENI, J., JR. 1982. Changing patterns of
lung cancer in the United States. Am. J. Epidemiol. 115:664-673
BLOT, W.J., BRINTON, L.A., FRAUMENI, J.F., and STONE, B.J.
1977. Cancer mortality in U.S. counties with petroleum
industries. Science 198:51-53
BOGOVSKI, P., VOSAMAE, A., and MIRME, H. 1970. Co-carcinoge-
nicity Studies in Oil-Shale Processing Products. In Tenth
International Cancer Congress. The International Union
Against Cancer
-------
BOHLIG, H., DABBERT, A.F., DALQUEN, P., HAIN, E., and HINZ, I.
1970. Epidemiology of malignant mesotheliomas in Hamburg:
A preliminary report. Environ. Res. 3:365-372
BORCH-JOHNSEN, K. 1982. Urbanization and lung cancer: A
quantitative and qualitative assessment of the urban factor.
Wgeskr. Laeger. 144:1713-1718
BOYLAND, E., and SIMS, P. 1967. The carcinogenic activities
in mice of compounds related to benz(a)anthracene. Int.
J. Cancer 2:500
BOZZO, S.R., NOVAK, K.M., GALDOS, F., HOKOOPIAN, K., and HAMILTON,
L.D. 1979. Mortality, migration, income and air pollution:
A comparative study. Soc. Sci. Med. Medical Geography
130:95-109
BRADY, J., LIBERATORS, F., HARPER, P., GREENWALD, P., BURNETT, W.,
DAVIS, J., BISHOP, M., POLAN, A., and VIANNA, N. 1977.
Angiosarcoma of the liver: An epidemiologic survey.
J. Natl. Cancer Inst. 59:1383-1385
BRIDGES, B.A., MACGREGOR, D., ZEIGER, E., BONIN, A., DEAN,
B.J., LORENZO, F., GARNER, R.C., GATEHOUSE, D., HUBBARD,
S., ICHINOTSUBO, D., MACDONALD, D., MARTIRE, S., MATSUSHIMA,
T., MOHN, G., NAGAO, M., RICHOLD, M., ROWLAND, I., SIMMON,
V., SKOPEK, T., TRUMAN, R., and VENITT, S. 1981. Summary
report on the performance of bacterial mutation assays.
In de Serres, F.J., and Ashby, J., eds. Progress in Muta-
tion Research. Vol. I: Evaluation of Short-Term Tests
for Carcinogens: Report of the International Collaborative
Program. Elsevier/North-Holland, New York. Pp. 49-67
BRODZINSKY, R., and SINGH, H.B. 1982. Hazardous Organic Chemi-
cals in the Atmosphere: An Assessment of Available Data.
Prepared for U.S. EPA, Contracts Management Division,
Contract 68-02-3452. Prepared by SRI International, Menlo
Park, CA
BROWN, S.M., SELVIN, S., and WINKELSTEIN, W. 1975. The associa-
tion of economic status with the occurrence of lung cancer.
Cancer 36:1903-1911
BRYAN, W.R., and SHIMKIN, M.B. 1943. Quantitative analysis
of dose-response data obtained with three carcinogenic
hydrocarbons in strain C3H male mice. J. Natl. Cancer
Inst. 3:503
BUELL, P., and DUNN, J.E. 1967. Relative impact of smoking
and air pollution on lung cancer. Arch. Environ. Health
15:291-297
-------
BUELL, P., DUNN, J.E., and BRESLOW, L. 1967. Cancer of the
lung and Los Angeles-type air pollution. Cancer 19:
2139-2147
BUSER, H.R., BOSSHARDT, H.P., and RAPPE, C. 1978. Identifica-
tion of polychlorinated dibenzo-p-dioxin isomers found
in fly ash. Chemosphere 2:165-172
CAPURRO, P. 1979. Cancer in a community subject to air pollu-
tion by solvent vapors. Clin. Toxicol. 14:285-294
CARCINOGEN ASSESSMENT GROUP (GAG). 1978. Preliminary Report
on Polycyclic Organic Compounds Exposures. Environmental
Protection Agency, Office of Research and Development/
Washington, D.C.
CARCINOGEN ASSESSMENT GROUP (CAG). 1982. The Carcinogen Assess-
ment Group's Review of Clement Associates' Document Entitled
"Review and Evaluation of the Evidence for Cancer Associated
with Air Pollution." Office of Health and Environmental
Assessment, U.S. Environmental Protection Agency, Washington,
D.C. April 1982
CARNOW, B. 1978. The "urban factor" and lung cancer: Cigarette
smoking or air pollution. Environ. Health Perspect. 22:17-21
CARNOW, B., and MEIER, P. 1973. Air pollution and pulmonary
cancer. Arch. Environ. Health 27:207-218
CASE, R.A.M., HOSKER, M.E., MCDONALD, D.B., and PEARSON, J.T.
1954. Tumours of the urinary bladder in workmen engaged
in the manufacture and use of certain dyestuff intermedi-
ates in the British chemical industry: Part I. Br. J.
Ind. Med. 11:75-104
CASTRO, B.C., HATCH, G.G., HUANG, S.L., LEWTAS, J., NESNOW,
S., and WATERS, M.D. 1981. Mutagenic and carcinogenic
potency of extracts of diesel and related environmental
emissions: In vitro mutagenesis and oncogenic transfor-
mation. Environ. Int. 5:403-409
CECILIONI, V.A. 1972. Lung cancer in a steel city: Its poss-
ible relation to fluoride emissions. Fluoride 5:172-187
CECILIONI, V.A. 1974. Further observations on cancer in a
steel city. Fluoride 7:153-156
CEDERLOF, R. et al. 1975. The Relationship of Smoking and
Some Social Covariables to Mortality and Cancer Morbidity:
A Ten-Year Follow-up in a Probability Sample of 55,000
Swedish Subjects Age 18-69. Department of Environmental
Hygiene, The Karolinska Institute, Stockholm
-------
CEDERLOF, R. , DOLL, R., FOWLER, B., FRIBERG, L., NELSON, N., and
VOUK, V. 1978. Air pollution and cancer: Risk assessment
methodology and epidemiological evidence. Environ. Health
Perspect. 22:1-12
CHEMICAL INDUSTRY INSTITUTE OF TOXICOLOGY (CUT). 1980. Pro-
gress Report on CUT Formaldehyde Studies. Research Tri-
angle Park, North Carolina
CHURG, A., and WARNOCK, M.L. 1977. Correlation of quantitative
asbestos body counts and occupation in urban patients.
Arch. Pathol. Lab. Med. 101:629
CLAXTON, L.D. 1980. Mutagenic and carcinogenic potency of
diesel and related environmental emissions: Salmonella
bioassay. In Pepelko, W.E., Banner, R.M., and Clarke,
N.A., eds. Health Effects of Diesel Engine Emissions:
Proceedings of an International Symposium. Vol. 2. EPA
Publication No. 600/9-80-057b. Pp. 801-809
CLEARY, G.J. 1963. Measurement of polycyclic aromatic hydro-
carbons in the air of Sydney using very long alumina columns
for separation. Int. J. Air Water Pollut. 7:753
CLEMENT ASSOCIATES, INC. 1981. Review and Evaluation of the
Evidence for Cancer Associated with Air Pollution. Prepared
for U.S. Environmental Protection Agency. Washington,
D.C.
COHEN, D., ARAI, S.F., and BRAIN, J.D. 1979. Smoking impairs
long-term dust clearance from the lungs. Science 204:514
COLE, P., HOOVER, R., and FRIEDELL, G.H. 1972. Occupation
and cancer of the lower urinary tract. Cancer 29:1250-1260
COLLIER, L. 1972. Determination of bis(chloromethyl)ether
at the ppb level in the air samples by high resolution
mass spectroscopy. Environ. Sci. Technol. 6:930-932
COMMONER, B., HENRY, J.I., GOLD, J.C., REDING, M.J., and
VITHAYATHIL, A.J. 1976. Reliability of Bacterial Muta-
genesis Techniques to Distinguish Carcinogenic and Noncar-
cinogenic Chemicals. Final Report to the U.S. Environmental
Protection Agency. EPA-600/1-76-022
COMMONER, B., MADYASTHA, P., BRONDSON, A., and VITHAYATHIL, A.J.
1978. Environmental mutagens in urban air particulates.
J. Toxicol. Environ. Health 4:59-77
COOPER, W.C., MURCHIO, J., POPENDORF, W., and WENK, H.R. 1979.
Chrysotile asbestos in a California recreational area.
Science 206:685-688
-------
COUNCIL ON ENVIRONMENTAL QUALITY (CEQ). 1980. Eleventh Annual
Report of the Council on Environmental Quality. U.S.
Government Printing Office, Washington, D.C.
CROCKER, T.T., CHASE, J.E., WELLS, S.A., and NUNES, L.L. 1970.
Preliminary report on experimental squamous carcinoma
of the lungs in hamsters and in a primate. In Nettesheim, P.,
Hanna, M.G., and Deatherage, J.W., eds. Morphology of
Experimental Respiration Carcinogenesis. AEC Symposium
Series, No. 21
CURREN, R.D., KOURI, R.E., KIM, C.M., and SCHECHTMAN, L.M.
1981. Mutagenic and carcinogenic potency of extract from
diesel related environmental emissions: Simultaneous
morphological transformation and mutagenesis in Balb/c
3T3 cells. Environ. Int. 5:411-415
CUTLER, S.J., and YOUNG, J.L., eds. 1975. Third National
Cancer Survey: Incidence Data. Natl. Cancer Inst. Monogr.
41:1-454
DALQUEN, P., DABBERT, A.P., and HINZ, I. 1969. Zur Epidemio-
logie der Pleuramesotheliome. Prax. Pneumonol. 23:547-558
DAVIS, D.L., and MAGEE, B.H. 1979. Cancer and industrial
chemical production. Science 206:1356-1358
DAVIS, D.L., BRIDBORD, H., and SCHNEIDERMAN, M. 1982. Cancer
prevention: Assessing causes, exposures, and recent trends
in mortality for U.S. males, 1968-1978. Teratogenesis,
Carcinogenesis, and Mutagenesis 2:105-135
DEAN, G. 1964. Lung cancer in South Africans and British
immigrants. Proc. Rep. Soc. Med. 57:984-987
DEAN, G. 1966. Lung cancer and bronchitis in Northern Ireland,
1960-1962. Br. Med. J. 1:1506-1514
DEAN, G., LEE, P.N., TODD, G.F., and WICKEN, A.J. 1977. Report
on a second retrospective mortality study in North East
England. Tobacco Research Council Research Paper 14,
Part I. Tobacco Research Council, London
DEAN, G-, LEE, P.N., TODD, G.F., and WICKEN, A.J. 1978. Report
on a second retrospective mortality study in North East
England. Tobacco Research Council Research Paper 14,
Part II. Tobacco Research Council, London
DEHNEN, W., PITZ, N., and TOMINGAS, R. 1978. The mutagenicity
of airborne particulate pollutants. Cancer Lett. 4:5-12
-------
BELLA PORTA, G., TERRACINI, B., and SHUBIK, P. 1961. Induction
with carbon tetrachloride of liver cell carcinoma in ham-
sters. J. Natl. Cancer Inst. 26:855
DEMOPOULOS, H.B., and GUTMAN, E.G. 1980. Cancer in New Jersey
and other complex urban/industrial areas. J. Environ.
Pathol. Toxicol. 3:219-235
DEPASS, L.R., CHEN, K.C., and PETERSON, L.G. 1982. Dermal
carcinogenesis bioassays of diesel particulates and dichloro-
methane extract of diesel particulates in C3H mice. Dev.
Toxicol. Environ. Sci. 10:321-327
DOLL, R. 1978. Atmospheric pollution and lung cancer. Environ.
Health Perspect. 22:23-31
DOLL, R. 1982. Trends in mortality from lung cancer in women.
In Magnus, K., ed. Trends in Cancer Incidence. Hemisphere
Publishing Corp.,, Washington, D.C. Pp. 223-230
DOLL, R., and PETO, R. 1976. Mortality in relation to smoking:
20 years' observations on male British doctors. Br. Med.
J. 2:1525-1536
DOLL, R., and PETO, R. 1981. The causes of cancer: Quantita-
tive estimates of avoidable risks of cancer in the United
States today. JNCI 66:1191-1308
DOLL, R., FISHER, R.E.W., GAMMEN, E.J., GUNN, W., HUGHES, G.O.,
TYREE, F.H., and WILSON, W. 1965. Mortality of gas workers
with special reference to cancers of the lung and bladder,
chronic bronchitis, and pneumoconiosis. Br. J. Ind.
Med. 22:1
DOLL, R., VESSEY, M.P., BEASLEY, R.W.R., BUCKLEY, A.R., FEAR,
E.G., FISHER, R.E.W., GAMMON, E.J., GUNN, W., HUGHES,
G.O., LEE, K., and NORMAN-SMITH, B. 1972. Mortality
of gasworkers—final report of a prospective study. Br. J.
Ind. Med. 29:394-406 (As reported by Pike et al. 1975)
DORN, H.F., and CUTLER, S.J. 1959. Morbidity from Cancer
in the United States: Parts I and II. Public Health
Monogr. 56:1-207
DOW CHEMICAL CO. 1978. Trace Chemistries of Fire. Unpublished
DUCE, R.A. 1978. Speculations on the budget of particulate
and vapor phase nonmethane organic carbon in the global
troposphere. Pure Appl. Geophys. 116:244-273
EASTCOTT, D.F. 1956. The epidemiology of lung cancer in New
Zealand. Lancet 1:37-39
-------
EDLING, C., COMBA, P., AXELSON, 0., and FLODIN, U. 1982.
Effects of low-dose radiation--a correlation study. Scand.
J. Work Environ. Health Suppl. 1:59-82
ENSTROM, J.E. 1979. Rising cancer mortality among non-smokers.
JNCI 62:755-760
ENTERLINE, P.E. 1979. Attributability in the Face of Uncer-
tainty. Revised paper presented at International Conference
on Occupational Lung Disease, Session on Asbestos and
Cancer. San Francisco, February 27, 1979
ENVIRONMENTAL DEFENSE FUND/NATURAL RESOURCES DEFENSE COUNCIL
(EDF/NRDC). 1980. Comments on Environmental Protection
Agency Proposed Rulemaking: National Emission Standards
for Identifying, Assessing and Regulating Airborne Sub-
stances Posing a Risk of Cancer (44 FR 58642). OAQPS
Document No. 79-14
EPSTEIN, S.S. 1967. Carcinogenicity of organic extracts of
atmospheric pollutants. J. Air Pollut. Control Assoc.
17:728-729
EPSTEIN, S.S. 1978. Statement at Public Hearing: Proposed
Regulations of the U.S. Occupational Safety and Health
Administration for the Identification, Classification,
and Regulation of Toxic Substances Posing a Potential
Occupational Carcinogenic Risk
EPSTEIN, S.S., JOSHI, S., ANDREA, J., MANTEL, N., SAWICKI, E.,
STANLEY, T., and TABOR, E.G. 1966. Carcinogenicity of
organic particulate pollutants in urban air after adminis-
tration of trace quantities to neonatal mice. Nature 212:
1305-1307
EPSTEIN, S.S., FUJII, K., and ASAHINA, S. 1979. Carcinogenicity
of a composite organic extract of urban particulate atmos-
pheric pollutants following subcutaneous injection in
infant mice. Environ. Res. 19:163-176
ESCHENBRENNER, A.B. 1945. Induction of hepatomas in mice by
repeated oral administration of chloroform, with observa-
tions on sex differences. J. Natl. Cancer Inst. 6:325-
341
FALK, H.L., and KOTIN, P. 1962. An assessment of factors
concerned with the carcinogenic properties of air pollut-
ants. In Analysis of Carcinogenic Air Pollutants. Natl.
Cancer Inst. Monogr. No. 9, pp. 81-89
FERON, V.J., EMMELOT, P., and VOSSENARR, T. 1972. Lower respira-
tory tract tumors in Syrian golden hamsters after intra-
8
-------
tracheal instillations of diethylnitrosamine alone and
with ferric oxide. Eur. J. Cancer 8:445-449
FINE, D.H., ROUNBEHLER, D.P., FAN, T., and ROSS, R. 1977.
Human exposure to N-nitroso compounds in the environment.
In Hiatt, H.H., Watson, J.D., Winsten, J.H./ eds. Origins
of Human Cancer. Book A: Incidence of Cancer in Humans.
Cold Spring Harbor Laboratory. Cold Spring Harbor, New
York. Pp. 293-307
FISHER, G.L., CHRISP, C.E., and RAABE, O.G. 1979. Physical
factors affecting the mutagenicity of fly ash from a coal-
fire power plant. Science 204:879-881
FORD, A.B., and BIALIK, 0. 1980. Air pollution and urban
factors in relation to cancer mortality. Arch. Environ.
Health 35:350-359
FRAUMENI, J.F., ed. 1975. Persons at High Risk of Cancer:
An Approach to Cancer Etiology and Control. Academic
Press, New York
FREEMAN, A.E., PRICE, P.J., BRYAN, R.J., GORDON, R.J., GILDEN, R.V.,
KELLOFF, G.J., and HUEBNER, R.J. 1971. Transformation
of rat and hamster embryo cells by extracts of city smog.
Proc. Natl. Acad. Sci. USA 68:445-449
FRIBERG, L., and CEDERLOF, R. 1978. Late effects of air pollu-
tion with special reference to lung cancer. Environ. Health
Perspect. 22:45-66
GARFINKEL, L. 1981. Time trends on lung cancer mortality
among nonsmokers and a note on passive smoking. JNCI
66: 1061-1066
GERSTEIN, D.R., and LEVISON, P.K., eds. 1982. Reduced Tar
and Nicotine Cigarettes: Smoking Behavior and Health.
National Research Council/National Academy of Sciences.
National Academy Press, Washington, D.C
OILMAN, J.P.W. 1962. Metal carcinogenesis: II. A study of
the carcinogenic activity of cobalt, copper, iron and
nickel compounds. Cancer Res. 22:158-162
GOLDSMITH, J. 1980. The "urban factor" in cancer: Smoking,
industrial exposures, and air pollution as possible explana-
tions. J. Environ. Pathol. Toxicol. 3:205-217
GORDON, R.J. 1978. Survey for airborne nitrosamines. NTIS
PB-293 506/2GA. ARB-R-A6096-30-78-93
-------
GORDON, R.J., BRYAN, R.J., RHIM, J.S., DEMOISE, C., WOLFORD,
R.G., FREEMAN, A.E., and HUEBNER, R.J. 1973. Transforma-
tion of rat and mouse embryo cells by a new class of car-
cinogenic compounds isolated from particles in city air.
Int. J. Cancer 12:223-232
GREAVES, W.W., ROM, W.N., LYON, J.L., VARLEY, G., and WRIGHT, P.O.
1981. Relationship of lung cancer and distance of residence
from nonferrous smelter stack effluent. Am. J. Ind. Med.
2:15-23
GREENBERG, M. 1979. Spatial distribution of cancer mortality
and of high and low risk factors in the New Jersey-New
York-Philadelphia metropolitan region, 1950-1969: Part 1.
Department of Environmental Protection, Trenton, N.J.
GREENBERG, M., McKAY, F., and WHITE, P. 1980. A time series
comparison of cancer mortality rates in the New Jersey-New
York-Philadelphia metropolitan region and the remainder of
the United States: 1950-1969. Am. J. Epidemiol. 111:166-174
GRIMMER, G. 1968. Cancerogene Kohler wasserstaffe in der
Umgebung des Menshen. Dtsch. Apath. Ztg. 108:529
GRISWOLD, M.H., WILDER, C.S., CULTER, S.J., et al. 1955.
Cancer in Connecticut, 1935-1951. State Department of
Health. Hartford, Connecticut
GUERRERO, R.R., ROUNDS, D.E., ORTHOEFER, J. 1980. Sister
chromatid exchange analysis of Syrian hamster lung cells
treated in vivo with diesel exhaust particles. In Pepelko,
W.E., Danner, R.M., and Clarke, N.A., eds. Health Effects
of Diesel Engine Emissions: Proceedings of an International
Symposium. Vol. 2. EPA Publication No. 600/9-80-057b.
Pp. 951-969
HAENSZEL, W. 1961. Cancer mortality among the foreign-born
in the United States. J. Natl. Cancer Inst. 26:37-132
HAENSZEL, W., and TAEUBER, K.E. 1964. Lung cancer mortality
as related to residence and smoking histories: II. White
females. J. Natl. Cancer Inst. 32:803-838
HAENSZEL, W., SHIMKIN, M.B., and MILLER, H.P. 1956. Tobacco
Smoking Patterns in the United States. U.S. Department
of Health, Education and Welfare. Public Health Monograph
No. 45, Public Health Service Publication No. 463
HAENSZEL, W., LOVELAND, D.B., and SIRKEN, M.G. 1962. Lung
cancer mortality as related to residence and smoking his-
tory: I. White males. J. Natl. Cancer Inst. 28:947-
1001
10
-------
HAKIM, S.A.E. 1968. Sanquinarine--a carcinogenic contaminant
in Indian edible oils. Indian J. Cancer 5:183
HAMMOND, E.G. 1972. Smoking habits and air pollution in rela-
tion to lung cancer. In Lee, H.K., ed. Environmental
Factors in Respiratory Disease. Academic Press, New York.
Chapter 12
HAMMOND, E.G., and GARFINKEL, L. 1980. General air pollution
and cancer in the United States. Prev. Med. 9:206-211
HAMMOND, E.G., and HORN, D. 1958. Smoking and death rates--
report on forty-four months of follow-up of 187,783 men.*
J. Am. Med. Assoc. 166:1294-1308
HAMMOND, E.G., and SEIDMAN, H. 1980. Smoking and cancer in
the U.S.A. Prepared for Conference on the Primary Preven-
tion of Cancer: Assessment of Risk Factors and Future
Directions, New York, June 7-8, 1979. Prev. Med. 9:169-173
HAMMOND, E.G., SELIKOFF, I.J., LAWTHER, P.L., and SEIDMAN, H.
1976. Inhalation of benzopyrene and cancer in man. Ann.
NY Acad. Sci. 271:116-124
HAMMOND, E.G., GARFINKEL, L., SELIKOFF, I.J., and NICHOLSON, W.J.
1979. Mortality experience of residents in the neighborhood
of an asbestos factory. Ann. NY Acad. Sci. 330:417-422
HEAREY, C.D., URY, H., SIEGELAUB, A., HO, M.K.P., SALOMON, H.,
and CELLA, R.L. 1980. Lack of association between cancer
incidence and residence near petrochemical industry in
San Francisco Bay Area. JNCI 64:1295-1299
HENDERSON, B.E., GORDON, R.J., MENCK, H., SOOHOO, J., MARTIN,
S.P., and PIKE, M.C. 1975. Lung cancer and air pollution
in southcentral Los Angeles county. Am. J. Epidemiol.
101:477-488
HENRY, M.C., PORT, C.D., and KAUFMAN, D.G. 1975. Importance
of physical properties of benzo(a_)pyrene-ferric oxide
mixtures in lung cancer induction. Cancer Res. 35:207-217
HICKEY, R.J., SCHOFF, E.P., and CLELLAND, R.C. 1967. Relation-
ship between air pollution and certain chronic disease
death rates: Multivariate statistical studies. Arch. En-
viron. Health 15:728-738
HICKEY, R.J., BOYCE, D.E., HARNER, E.B., and CLELLAND, R.C.
1974. Environmental pollution and human health. In Berry,
B.J.L., and Horton, F.E., eds. Urban Environmental Manage-
ment: Planning for Pollution Control. Prentice-Hall,
Englewood Cliffs, New Jersey. Pp. 317-340
11
-------
HIGGINS, I.T.T. 1974. Trends in respiratory cancer mortality
in the United States and in England and Wales. Arch. Environ
Health 28:121-129
HIGGINSON, J., and MUIR, C.S. 1979. Environmental carcinoge-
nesis: Misconceptions and limitations to cancer control.
JNCI 63:1291-1298
HIRAYAMA, T. 1981. Non-smoking wives of heavy smokers have
a higher risk of lung cancer: A study from Japan. Br.
Med. J. 282:183-185
HITOSUGI, M. 1968. Epidemiological study of lung cancer with
special reference to the effect of air pollution, inst.
Public Health Bull. (Tokyo) 17:237-256
HOFFMAN, D. 1964. Studies in air pollution carcinogenesis.
Proc. Am. Assoc. Cancer Res. 5:27 (Abstract)
HOFFMAN, D., and WYNDER, E.L. 1962. Analytical and biological
studies on gasoline engine exhaust. Natl. Cancer Inst.
Monogr. No. 9
HOFFMAN, D., and WYNDER, E.L. 1966. Beitrag zur carcinogenen
Wirkung von Kibenzopyrenen. Z. Krebsforsch. 68:137
HOFFMAN, E.F., and GILLIAM, A.G. 1954. Lung cancer mortality:
Geographic distribution in the United States for 1948-1949.
Public Health Rep. 69:1033-1042
HOLMBERG, B., and AHLBORG, U. 1983. Editorial comments.
In Consensus report: Mutagenicity and carcinogenicity
of car exhausts and coal combustion emissions.
Environ. Health Perspect. 47:1-30
HOOVER, R., and FRAUMENI, J.F. 1975. Cancer mortality in
U.S. counties with chemical industries. Environ. Res. 9:
196-207
HOOVER, R., MASON, T.J., McKAY, F.W., and FRAUMENI, J.F. 1975.
Geographic patterns of cancer mortality in the United
States. In Fraumeni, J.F., ed. Persons at High Risk
of Cancer: An Approach to Cancer Etiology and Control.
Academic Press, New York. Pp. 343-360
HUEPER, W.C., KOTIN, P., TABOR, E.G., PAYNE, W.W., FALK, H.,
and SAWICKI, E. 1962. Carcinogenic bioassays on air
pollutants. Arch. Pathol. 74:89-116
HUGHES, T.J., PELLIZZARI, E., LITTLE, L., SPARACINO, C., and
KOLBER, A. 1980. Ambient air pollutants: Collection,
chemical characterization and mutagenicity testing. Mut.
Res. 76:51-83
12
-------
HUISINGH, J.L. 1981. Short-term carcinogenesis and mutagenesis
bioassays of unregulated automotive emissions. Bull. NY
Acad. Med. 57:251-262
INFANTE, P.P. 1976. Oncogenic and mutagenic risks in communi-
ties with polyvinyl chloride production facilities. Ann.
NY Acad. Sci. 271:49-57
INFANTE, P.P., WAGONER, J.K., RINSKY, R.A., and YOUNG, R.J.
1977. Leukaemia in benzene workers. Lancet 2:76-78
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (lARC). 1973.
IARC Monographs on the Evaluation of the Carcinogenic
Risk of Chemicals to Man. Vol. 3: Certain Polycyclic
Aromatic Hydrocarbons and Heterocyclic Compounds. World
Health Organization, Lyon, France. Pp. 22-44
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC). 1974a.
IARC Monographs on the Evaluation of the Carcinogenic
Risk of Chemicals to Man. Vol. 5: Some Organochlorine
Pesticides. World Health Organization, Lyon, France.
Pp. 83-156
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC). 1974b.
IARC Monographs on the Evaluation of the Carcinogenic
Risk of Chemicals to Man. Vol. 7: Some Antithyroid and
Related Substances, Nitrofurans and Industrial Chemicals.
World Health Organization, Lyon, France. Pp. 203-221
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC). 1976.
IARC Monographs on the Evaluation of the Carcinogenic
Risk of Chemicals to Man. Vol. 11: Cadmium, Nickel,
Some Epoxides, Miscellaneous Industrial Chemicals and
General Considerations on Volatile Anaesthetics. World
Health Organization, Lyon, France. Pp. 75-112
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC). 1977.
IARC Monographs on the Evaluation of the Carcinogenic
Risk of Chemicals to Humans. Vol. 14: Asbestos. World
Health Organization, Lyon, France. Pp. 1-106
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC). 1978.
IARC Monographs on the Evaluation of the Carcinogenic
Risk of Chemicals to Humans. Vol. 17: Some N-Nitroso
Compounds. World Health Organization, Lyon, France
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC). 1979.
IARC Monographs on the Evaluation of the Carcinogenic
Risk of Chemicals to Humans. Vol. 19: Some Monomers,
Plastics and Synthetic Elastomers, and Acrolein. World
Health Organization, Lyon, France. Pp. 377-437
13
-------
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC). 1980. IARC
Monographs on the Evaluation of the Carcinogenic Risk of
Chemicals to Humans. Vol. 23: Some Metals and Metallic
Compounds. World Health Organization, Lyon, France
ITO, N., NAGASAKI, H., ARAI, M., MAKIURA, S., SUGIHARA, S., and
HIRAO, K. 1973. Histopathologic studies on liver tumori-
genesis induced in mice by technical polychlorinated bi-
phenyls and its promoting effect on liver tumours induced
by benzene hexachloride. J. Natl. Cancer Inst. 51:1637-1646
JANIS, J. 1982. Descriptive and statistical methodology for
age-period-cohort analyses with application to lung cancer.
University of North Carolina at Chapel Hill, Chapel Hill,
N.C. Cited in ICRDB Cancergram Series CK15, No. 82/11,
November 1982. P. 1
JEGIER, Z. 1969. Pesticide residues in atmosphere. Ann.
NY Acad. Sci. 160:143-154
KAHN, H.A. 1966. The Dorn study of smoking and mortality
among U.S. veterans: Report on eight and one-half years
of observation. Natl. Cancer Inst. Monogr. 19:1-25
KARCH, N.J., and SCHNEIDERMAN, M.A. 1981. Explaining the
Urban Factor in Lung Cancer Mortality. A Report to the
Natural Resources Defense Council. Clement Associates,
Inc., Washington, D.C.
KATZ, M. 1964. Air pollution and lung cancer. Med. Serv. J.
Can. 20:665-675
KAUFMAN, D.G., and MADISON, R.M. 1974. Synergistic effects
of benzo(a)pyrene and N-methyl-N-nitrosourea on respira-
tory carcThogenesis in Syrian golden hamsters. In Karbe,
E., and Park, J.F., eds. Experimental Lung Cancer: Car-
cinogenesis and Bioassays. Springer-Verlag, Heidelberg.
Pp. 207-218
KIMBROUGH, R.D., and LINDER, R.E. 1974. Induction of adenofi-
brosis and hepatomas of the liver in BALB/cJ mice by poly-
chlorinated biphenyls (Aroclor 1254). J. Natl. Cancer
Inst. 53:547-549
KIMBROUGH, R.D., SQUIRE, R.A., LINDER, R.E., STRANDBERG, J.D.,
MONTALI, R.J., and BURSE, V.W. 1975. Induction of liver
tumours in Sherman strain female rats by polychlorinated
biphenyl Aroclor 1260. J. Natl. Cancer Inst. 55:1453-
1459
KING, R.B., ANTOINE, A.C., FORDYCE, J.S., NEUSTADTER, H.E.,
and LEIBECKI, H.F. 1977. Compounds in airborne particu-
14
-------
lates: Salts and hydrocarbons. J. Air Pollut. Control
Assoc. 27:867-871
KLEIN, M. 1963. Susceptibility of strain B6AF/J hybrid infant
mice to tumorigenesis with 1,2-benzanthracene, deoxychloric
acid, and 3-methylcholanthrene. Cancer Res. 23:1701
KOCIBA, R.J., KEYES, D-G-, BEYER, J.E., CARREON, R.M., WADE,
C.E., DITTENBER, D.A., KALNINS, R.P., FRAUSON, L.E., PARK,
C.N., BARNARD, S.D., HUMMEL, R.A-, and HUMISTON, C.G- 1978.
Results of a two-year chronic toxicity and oncogenicity
study of 2,3,7,8-tetrachlorodibenzo-p-dioxin in rats.
Toxicol. Appl. Pharnfacol. 46:279-303
KOMMINENI, C., and COFFIN, D.L. 1976. The effects of crude
air particulate material (CAP) on hamster respiratory
system. Toxicol. Appl. Pharmacol. 37:115 (Abstract)
KOTIN, P. 1968. Carcinogenesis of the lung: Environmental
and host factors. In Liebaw, A.A., and Smith, D.E., eds.
The Lung. Williams and Wilkins, Baltimore
KOTIN, P., and FALK, H.L. 1963. Atmospheric factors in patho-
genesis of lung cancer. In Haddow, A., and Weinhouse, S.,
eds. Advances in Cancer Research. Vol. 7. Academic
Press, New York. Pp. 475-514
KOTIN, P., and FALK, H.L. 1964. Polluted urban air and related
environmental factors in the pathogenesis of pulmonary
cancer. Dis. Chest 45:236-246
KVB, INCORPORATED. 1980. An Inventory of Carcinogenic Sub-
stances Released into the Ambient Air of California.
Prepared for Science Applications, Inc., Los Angeles.
Report No. ARB/R-80/120
LAKOWICZ, J.R., and HYLDEN, J.L. 1978. Asbestos mediated
membrane uptake of benzo(a)pyrene observed by fluorescence
spectroscopy. Nature 275:446-448
LASKIN, S., KUSCHNER, M., and DREW, R.T. 1970. Studies in
pulmonary Carcinogenesis. In Hanna, M.G., Jr., Nettesheim,
P., and Gilbert, J.R., eds. Inhalation Carcinogenesis.
Oak Ridge National Laboratory, U.S. Atomic Energy Commission
Symposium Series No. 18, Conference 691001. Pp. 321-351
LASKIN, S., KUSCHNER, M., DREW, R.T., CAPPIOLLO, V.P., and
NELSON, N. 1971. Tumors of the respiratory tract induced
by inhalation of bis(chloromethyl)ether. Arch. Environ.
Health 23:135-136
15
-------
LAVE, L.B., and SESKIN, E.P. 1977. Air Pollution and Human
Health. The Johns Hopkins University Press, Baltimore
LAWTHER, P.J., and WALLER, R.E. 1978. Trends in urban air
pollution in the United Kingdom in relation to lung cancer
mortality. Environ. Health Perspect. 22:71-73
LAZRUS, A.L., LORANGE, E., and LODGE, J.P., Jr. 1970. Lead
and other metal ions in United States precipitation.
Environ. Sci. Technol. 4:55-58
LEE, A.M., and FRAUMENI, J.F. 1969. Arsenic and respiratory
cancer in man: An occupational study. J. Natl. Cancer
Inst. 42:1045-1052
LEMEN, R.A., JOHNSON, W.M., WAGONER, J.K., ARCHER, V.E., and
SACCOMANNO, G. 1976. Cytologic observations and cancer
incidence following exposure to BCME. Ann. NY Acad. Sci.
271:71-80
LEONG, B.K.J., MACFARLAND, H.N., and REESE, W. 1971. Induction
of adenomas by chronic inhalation of bis(chloromethyl)ether.
Arch. Environ. Health 22:663-666
LEVIN, M.L., HAENSZEL, W., CARROLL, B.E., GERHARDT, R., HANDY, V.H.,
and INGRAHAM, S.C. 1960. Cancer incidence in urban and
rural areas of New York State. J. Natl. Cancer Inst. 24:1243-
1257
LEWTAS, J. 1982. Mutagenic activity of diesel emissions.
Dev. Toxicol. Environ. Sci. 10:243-264
LILIENFELD, A.M., PEDERSON, E., and DOWD, J.E. 1967. Cancer
Epidemiology: Methods of Study. The Johns Hopkins Press,
Baltimore
LILLIAN, D., SINGH, H.B., APPLEBY, A., LOBBAN, L., ARNTS, R.,
GUMPERT, R., HAGUE, R., TOOMEY, J, KAZAZIS, J., ANTELL, M.,
HANSEN, D., and SCOTT, B. 1975. Atmospheric fates of
halogenated compounds. Environ. Sci. Technol. 9:1024-1048
LINDAU, L. 1977. Emissions of arsenic in Sweden and their
reduction. Environ. Health Perspect. 19:25-29
LLOYD, J.W. 1971. Long-term mortality study of steelworkers:
V. Respiratory cancer in coke-plant workers. J. Occup.
Med. 13:53
LLOYD, O.L. 1978. Respiratory cancer clustering associated
with localised industrial air pollution. Lancet 1:318-320
16
-------
LLOYD, W. 1979. Cancer Epidemiology. 18th Annual Medical-
Legal Industrial Symposium, Milwaukee, November 9, 1979
LOCKARD, J.M., VIAU, C.J., LEE-STEPHENS, C., CALDWELL, J.C.,
WOJCIECHOWSKI, J.P., ENOCH, H.G., and SABHARWAL, P.S.
1981. Induction of sister chromatid exchanges and bac-
terial revertants by organic extracts of airborne particles.
Environ. Mutagenesis 3:671-681
LOGAN, W.P.D. 1982. Cancer mortality by occupation and social
class. International Agency for Research on Cancer, Lyon,
France. IARC Scientific Publication No. 36. Also published
in*London, England, by Her Majesty's Stationery Office:
Studies on Medical and Population Subjects, No. 44
LUDWIG, J.H., MORGAN, G.B., and McMULLEN, T.B. 1971. Trends
in urban air quality. In Matthew, W.H., Kellogg, W.W.,
and Robinson, G.O., eds. Man's Impact on Climate. MIT
Press, Cambridge, Massachusetts. Pp. 321-338
LUNDE, B., and BJORSETH, A. 1977. Polycyclic aromatic hydro-
carbons in long-range transported aerosols. Nature 268:
518-519
LYON, J.L., FILLMORE, J.C., and KLAUBER, M.R. 1977. Arsenical
air pollution and cancer: Letter in response to Blot
and Fraumeni's article. Lancet, October 22, 1977, p. 869
LYON, J., FILLMORE, J., and SMITH, T.J. 1978. Arsenical air
pollution and lung cancer: Reply to Stellman and Rabat's
letter. Lancet, January 28, 1978
LYONS, M.J. 1959. Vehicular exhausts: Identification of
further carcinogens of the polycyclic aromatic hydrocarbon
class. Br. J. Cancer 13:126
McCANN, J., CHOI, E., YAMASAKI, E., and AMES, B.N. 1975.
Detection of carcinogens as mutagens in the Salmonella/
microsome test: Assay of 300 chemicals. Proc. Natl.
Acad. Sci. USA 72:5135-5139
McCONNELL, G., FERGUSON, D.M., and PEARSON, C.R. 1975. Chlori-
nated hydrocarbons and the environment. Endeavor 34:13-18
MACDONALD, E.J. 1976. Demographic variation in cancer in
relation to industrial and environmental influence. En-
viron. Health Perspect. 17:153-166
McGANDY, R.B., KENNEDY, A.R., TERZAGHI, M., and LITTLE, J.B.
1974. Experimental respiratory carcinogenesis: Interaction
between alpha radiation and benzo(a)pyrene in the hamster.
In Karbe, E./ and Park, J.F., eds. Experimental Lung
17
-------
Cancer: Carcinogenesis and Bioassays. Springer-Verlag,
Heidelberg. Pp. 485-491
MALTONI, C., COTTI, G., MORISI, L., and CHIECO, P. 1977. Car-
cinogenicity bioassays of vinylidene chloride: Research
plans and early results. Med. Lav. 68(4):24l-262
MANCUSO, T.F., and COULTER, E.J. 1958. Cancer mortality among
native white, foreign-born white, and non-white male resi-
dents of Ohio: Cancer of the lung, larynx, bladder, and
central nervous system. J. Natl. Cancer Inst. 20:79-105
MANCUSO, T.F., and STERLING, T.D. 1974. Relation of place of
birth and migration in cancer mortality in the U.S.A.—
study of Ohio residents (1959-1967). J. Chronic Dis. 27:
459-474
MANCUSO, T.F., MACFARLANE, E.M., and PORTERFIELD, J.D. 1955.
Distribution of cancer mortality in Ohio. Am. J. Public
Health 45:58-70
MANTON, K.G., STALLARD, E. , and REGGAN, W. 1982. Strategies
for analysing ecological health data: Models of the bio-
logical risk of individuals. Stat. Med. 1:163-181
MASON, T.J., MCKAY, F.W, HOOVER, R., BLOT, w.j., and FRAUMENI, J.F.
1975. Atlas of Cancer Mortality for U.S. Counties: 1950-
1969. National Institutes of Health, Washington, D.C.
DHEW Publication No. (NIH) 75-780
MATANOSKI, G.M., LANDAU, E., TONASCIA, J., LAZAR, C., ELLIOTT,
E.A., McENROE, W., and KING, K., 1981. Cancer mortality
in an industrial area of Baltimore. Environ. Res. 25:8-29
MENCK, H.R., CASAGRANDE, J.T., and HENDERSON, B.E. 1974.
Industrial air pollution: Possible effect on lung cancer.
Science 183:210-212
MILLER, L., SMITH, W.E., and BERLINER, S.W. 1965. Tests for
effects of asbestos on benzo(ajpyrene carcinogenesis in
the respiratory tract. Ann. NY Acad. Sci. 132:489-500
MINOWA, M., SHIGEMATSU, I., NAGAI, M., and FUKUTOMI, K. 1981.
Geographical distribution of lung cancer mortality and
environmental factors in Japan. Soc. Sci. Med. 15:225-
231
MISFELD, J. 1980. The tumor-producing effects of automobile
exhaust condensate and of diesel exhaust condensate.
In Pepelko, W.E., Danner, R.M., and Clarke, N.A., eds.
Health Effects of Diesel Engine Emissions: Proceedings
of an International Symposium. Vol. 2. EPA Publication
No. 600/9-80-057b. Pp. 1012-1025
18
-------
MITCHELL, A.D., EVANS, E.L., JOTZ, M.M., RICCIO, E.S., MORTELMAN,
K.E., and SIMMON, V.P. 1981. Mutagenic and carcinogenic
potency of extracts of diesel and related environmental
emissions: In vitro mutagenesis and DNA damage. Environ.
Int. 5:393-401
MOHR, U. 1976. Investigations on the carcinogenic burden
by air pollution in man: XIV. Effects of automobile
exhaust condensate on the Syrian golden hamster lung.
Zentralbl. Bakteriol. (Orig. B) 163:425-432
MOLLER, M., and ALFHEIM, I. 1980. Mutagenicity and polycyclic
aromatic hydrocarbon-analysis of airborne particulate
matter. Atmos. Environ. 14:83-88
MONTESANO, R., SAFFIOTTI, U., and SHUBIK, P. 1970. The role
of topical and systemic factors in experimental respiratory
carcinogenesis. In Hanna, M.G., Nettesheim, P., and Gilbert,
J.R., eds. Inhalation Carcinogenesis. Oak Ridge National
Laboratory, U.S. Atomic Energy Commission Symposium Series
No. 18, Conference 691001. P. 358
MONTESANO, R., SAFFIOTTI, U., FERRERO, A., and KAUFMAN, D.G.
1974. Synergistic effects of benzo(a)pyrene and diethyl-
nitrosamine on respiratory carcinogenesis in hamsters.
J. Natl. Cancer Inst. 53:1395-1397
MOTTO, H.L., DAINES, R.H., CHILKO, D.M., and MOTTO, C.K. 1970.
Lead in soils and plants: Its relationship to traffic
volume and proximity to highways. Environ. Sci. Technol.
4:231-238
MURCHIO, W.C., COOPER, W.C, and DELEON, A. 1973. Asbestos
fibers in ambient air in California. March 1, 1973.
California Air Resources Board Report 4-054-1
NATIONAL ACADEMY OF SCIENCES (NAS). I972a. Lead: Airborne
Lead in Perspective. Committee on Biological Effects
of Atmospheric Pollutants, Division of Medical Sciences,
National Research Council, NAS, Washington, D.C.
NATIONAL ACADEMY OF SCIENCES (NAS). I972b. Particulate Poly-
cyclic Organic Matter. National Research Council, NAS,
Washington, D.C.
NATIONAL ACADEMY OF SCIENCES (NAS). I972c. Radiation: Biolo-
gical Effects of Ionizing Radiation. National Research
Council, NAS, Washington, D.C.
NATIONAL ACADEMY OF SCIENCES (NAS). 1976. Vapor Phase Organic
Pollutants. Committee on Biological Effects of Atmospheric
Pollutants, Division of Medical Sciences, National Research
Council, NAS, Washington, D.C.
19
-------
NATIONAL ACADEMY OF SCIENCES (NAS). 1979. Polychlorinated
Biphenyls. National Research Council/ NAS, Washington,
D.C.
NATIONAL ACADEMY OF SCIENCES (NAS). 1980. Lead in the Human
Environment. Committee on Lead in the Human Environment,
Environmental Studies Board, Commission on Natural Re-
sources, National Research Council, NAS, Washington, D.C.
NATIONAL CANCER INSTITUTE (NCI). 1976. Bioassay of Trichloro-
ethylene for Possible Carcinogenicity. Washington, D.C.
DHEW Publication No. (NIH) 76-802
NATIONAL CANCER INSTITUTE (NCI). I977a. Bioassay of Chlordane
for Possible Carcinogenicity. NCI Carcinogenesis Technical
Report Series No. 8. DHEW Publication No. (NIH) 77-808
NATIONAL CANCER INSTITUTE (NCI). I977b. Bioassay of Tetrachloro-
ethylene for Possible Carcinogenicity. NCI Carcinogenesis
Technical Report Series No. 13. DHEW Publication No.
(NIH) 77-813
NATIONAL CANCER INSTITUTE (NCI). 1978a. Bioassay of Aroclor
1254 for Possible Carcinogenicity. NCI Carcinogenesis
Technical Report Series No. 38. DHEW Publication No.
(NIH) 78-838
NATIONAL CANCER INSTITUTE (NCI). I978b. Bioassay of 1,2 Dibromo-
ethane for Possible Carcinogenicity. NCI Carcinogenesis
Technical Report Series No. 86. DHEW Publication No. (NIH)
78-1336
NATIONAL RESEARCH COUNCIL (NRC). I98la. Indoor Pollutants.
National Academy Press, Washington, D.C.
NATIONAL RESEARCH COUNCIL (NRC). 1981b. Formaldehyde and Other
Aldehydes. National Academy Press. Washington, D.C.
NATUSCH, D.F.S. 1978. Potentially carcinogenic species emitted
to the atmosphere by fossil-fueled power plants. Environ.
Health Perspect. 22:79-90
NELSON, K.W. 1977. Industrial contribution of arsenic to
the environment. Environ. Health Perspect. 19:31-34
NESNOW, S., EVANS, C., STEAD, A., CREASON, J., SLAGA, T.J., and
TRIPLETT, L.L. 1982. Skin Carcinogenicity studies of
emission extracts. Dev. Toxicol. Environ. Sci. 10:295-320
NETTESHEIM, P., CREASIA, D.A., and MITCHELL, T.T. 1975. Car-
cinogenic and cocarcinogenic effects of inhaled synthetic
smog and ferric oxide particles. J. Natl. Cancer Inst.
55:159-168
20
-------
NEWHOUSE, M.L., and THOMPSON, H. 1966. Mesothelioma of pleura
and peritoneum following exposure to asbestos in the London
area. Br. J. Ind. Med. 22:261-269
NEWMAN, J.A., ARCHER, V.A., SACCAMANO, G., KUSCHNER, M., AUER-
BACH, O., GRONDAUHL, R.D., and WILSON, J.C. 1976. His-
tologic types of bronchogenic carcinoma among members
of copper mining and smelting communities. Ann. NY Acad.
Sci. 271:260-268
NICHOLSON, W.S., and PUNDSACK, P.L. 1973. Asbestos in the
environment. In Bogovski, P., Gilson, J.C./ Timbrell, V.,
and Wagner, J.C., eds. Biological Effects of Asbestos.
IARC Scientific Publications No. 8. International Agency
for Research on Cancer, World Health Organization, Lyon,
France
NOVEY, H.S., and MARTEL, S.H. 1969. Asthma, arsenic, and
cancer. J. Allergy 44:315-319
NOYES, W.F. 1969. Carcinogen-induced neoplasia with metastasis
in a South American primate, Squainus oedipas. Proc. Soc.
Exp. Biol. Med. 131:223
OCCUPATIONAL SAFETY AND HEALTH ADMINISTRATION (OSHA). 1980.
Identification, classification, and regulation of potential
occupational carcinogens. Fed. Reg. 45:60001. (January 22,
1980)
OFFICE OF TECHNOLOGY ASSESSMENT (OTA). 1981. Assessment of
Technologies for Determining Cancer Risks from the Environ-
ment. U.S. Congress, Washington, D.C.
O'GARA, R.W., KELLY, M.G., BROWN, J., and MARTEL, N. 1965.
Induction of tumors in mice given a minute single dose
of dibenz(a,h)anthracene or 3-methylcholanthrene as new-
borns: A dose-response study. J. Natl. Cancer Inst.
35:1027
OHNISHI, Y., KACHI, K., SATO, K., TAHARA, I., TAKEYOSHI, H.,
and TOKIWA, H. 1980. Detection of mutagenic activity
in automobile exhaust. Mutat. Res. 77:229-240
OHTA, T., MORITA, M., and MIZOGUCHI, I. 1976. Local distribu-
tion of chlorinated hydrocarbons in the ambient air in
Tokyo. Atmos. Environ. 10:557-560
OLSON, W.A., HABERMAN, R.T., WEISBURGER, E.R., WARD, J.M.,
and WEISBURGER, J.H. 1973. Induction of stomach cancer
in rats and mice by halogenated aliphatic fumigants.
J. Natl. Cancer Inst. 51:1993-1994
21
-------
OSMOND, C., and GARDNER, M.J. 1982. Age, Period, and Cohort
Models Applied to Cancer Mortalty Rates. Statistics in
Medicine, Vol. 3, pp. 245-259
OTT, M.G., HOLDER, B.J., and GORDON, L.H. 1974. Respiratory
cancer and occupational exposure to arsenicals. Arch.
Environ. Health 29:250-255
OTT, M.G., TOWNSEND, J.C., FISHBECK, W.A. , and LANGNER, R.A.
1978. Mortality among individuals occupationally exposed
to benzene. Arch. Environ. Health 30:22-25
PASTERNACK, B.S., SHORE, R.E., and ALBERT, R.E*. 1977. Occupa-
tional exposure to chloromethyl ethers. J. Occup. Med.
19:741-746
PEPELKO, W.E. 1980. Overview of the Health Effects Research
Laboratories (Cincinnati) Diesel Exhaust Health Effect
Study. In Pepelko, W.E., Danner, R.M., and Clarke, N.A. ,
eds. Health Effects of Diesel Engine Emissions: Proceed-
ings of an International Symposium. Vol. 2. EPA Publi-
cation No. 600/9-80-057b. Pp. 673-680
PEREIRA, M.A., CONNOR, T.H., MAYNE, J., and LEGATOR, M.S.
I980a. Metaphase analysis micronuclei assay and urinary
mutagenicity assay of mice exposed to diesel emissions.
In Pepelko, W.E., Danner, R.M., and Clarke, N.A., eds.
Health Effects of Diesel Engine Emissions: Proceedings
of an International Symposium. Vol. 2. EPA Publication
No. 600/9-80-057b. Pp. 924-933
PEREIRA, M.A., SABHARWAL, P.S., KAUR, P., ROSS, C.R., CHOI,
A., and DIXON, T. I980b. In-vivo detection of mutagenic
effects of diesel exhaust by short-term mammalian bioassays
In Pepelko, W.E., Danner, R.M., and Clarke, N.A., eds.
Health Effects of Diesel Engine Emissions: Proceedings
of an International Symposium. Vol. 2. EPA Publication
No. 600/9-80-057b. Pp. 934-950
PEREIRA, M.A., SABHARWAL, P.S., GORDON, L., and WYROBEK, A.J.
I980c. The effect of diesel exhaust on sperm-shape abnor-
malities in mice. In Pepelko, W.E., Danner, R.M., and
Clarke, N.A., eds. Health Effects of Diesel Engine Emis-
sions: Proceedings of an International Symposium. Vol. 2.
EPA Publication No. 600/9-80-057b. Pp. 977-992
PEREIRA, M.A., SHINOZUKA, H., and LOMBARDI, B. I980d. Test
of diesel exhaust emissions in the rat liver foci assay.
In Pepelko, W.E., Danner, R.M., and Clarke, N.A., eds.
Health Effects of Diesel Engine Emissions: Proceedings
of an International Symposium. Vol. 2. EPA Publication
No. 600/9-80-057b. Pp. 970-976
22
-------
PERSHAGEN, G. , ELINDEN, C.G., and BOLINDER, A.M. 1977. Mor-
tality in a region surrounding an arsenic emitting plant.
Environ. Health Perspect. 19:127-130
PERRY, W.W., GOLDSMITH, R., MATTSON, M.E., URBANEK, M.A. , BAILEY,
C.B., and BERNEY, B.W. 1978. Investigation of Selected
Correlations between Industrial Activity and Community
Disease. U.S. Environmental Protection Agency, Washington,
D.C. EPA-560/6-78-004
PETO, R. 1977. Epidemiology, multistage models and short-
term mutagenesis tests. In Hiatt, H.H., Wilson, J.D.,
and Winston, J.A., eds. Origins of Human Cancer. Cold
Spring Harbor Laboratory, Cold Spring Harbor, N.Y. Pp.
1403-1428
PIKE, M.C., and HENDERSON, B.E. 1981. Epidemiology of poly-
cyclic hydrocarbons: Quantifying the cancer risk from
cigarette smoking and air pollution. Polycyclic Hydrocar-
bons & Cancer 3:317-334
PIKE, M.C., GORDON, R.J., HENDERSON, B.E., MENCK, H.R., and
SOOHOO, J. 1975. Air pollution. In Fraumeni, J.F.,
ed. Persons at High Risk of Cancer: An Approach to Cancer
Etiology and Control. Academic Press, New York. Pp. 225-
239
PIKE, M.C., JING, J.S., ROSARIO, I.P., HENDERSON, B.C., and
MENCK, H.R. 1979. Occupation: "Explanation" of an ap-
parent air pollution related localized excess of lung
cancer in Los Angeles County. In Breslow, N.E., and Whit-
temore, A.S., eds. Energy and Health. SIAM, Philadelphia.
Pp. 3-16
PINTO, S.S., ENTERLINE, P.E., HENDERSON, V., and VARNER, M.O.
1977. Mortality experience in relation to a measured
arsenic trioxide exposure. Environ. Health Perspect.
19:127-130
PITTS, J.N., GROSJEAN, D., MISCHKE, T.M., SIMMON, V.F., and
POOLE, D. 1977. Mutagenic activity of airborne particulate
organic pollutants. Toxicol. Lett. 1:65-70
PITTS, J.N., LOKENSGARD, D.M., RIPLEY, P.S., VAN CAUWENBERGHE, K.A.,
and VAN VAECK, L. 1980. "Atmospheric" epoxidation of
benzo(a)pyrene by ozone: Formation of the metabolite
benzo(a)pyrene-4,5-oxide. Science 210:1347-1349
POLISSAR, L. 1980. The effect of migration on comparison
of disease rates in geographic studies in the United States.
Am. J. Epidemiol. 111:175-182
23
-------
POLISSAR, L., and WARNER, H. 1981. Automobile traffic and
lung cancer: An update on Blumer's Report. Environ.
Sci. Technol. 15:713-714
POLLACK, E.S. 1980. Cancer Incidence Trends in the United
States: Some Methodological Problems. Presented at UICC
Symposium on Trends in Cancer Incidence, Oslo, Norway,
August 6, 1980
POLLACK, E.S., and HORM, J.M. 1980. Trends in cancer incidence
and mortality, 1969-76. JNCI 64:1091-1103
POTT, P., TOMINGAS, R., BROCKHAUS, A., and HUTH, F. 1980.
Studies on the tumorigenic effect of extracts and their
fractions of atmospheric suspended particles in the sub-
cutaneous test of the mouse. Zentralbl. Bakteriol. (Orig. B)
170:17-34
PRINDLE, R.A. 1959. Some considerations in the interpretation
of air pollution health effects data. J. Air Pollut.
Control Assoc. 9:12-19
PYLEV, L.N. 1963. Induction of lung cancer in rats by intra-
tracheal insufflation of cancerogenic hydrocarbons. Acta
Univ. Inter. Contra Cancrum 19:688-691
PYLEV, L.N., and SHABAD, L.M. 1972. Some results of experi-
mental studies in asbestos carcinogenesis. In Biological
Effects of Asbestos. International Agency for Research
on Cancer, World Health Organization, Lyon, France. IARC
Scientific Publications No. 8
RALL, D.P. 1978. Statement at Public Hearing: Proposed Regula-
tions of the U.S. Occupational Safety and Health Administra-
tion for the Identification, Classification, and Regulation
of Toxic Substances Posing a Potential Occupational Carcino-
genic Risk
REDMOND, C.K., CIOCCO, A., LLOYD, J.W., and RUSH, H.W. 1972.
Long-term mortality study of steelworkers: VI. Mortality
from malignant neoplasms among coke oven workers. J. Occup.
Med. 14:621
REID, D.D., CORNFIELD, J. , MARKUSH, R.E., SEIGEL, D., PEDERSON, E.,
and HAENSZEL, W. 1966. Studies of Disease among Migrants
and Native Populations in Great Britain, Norway, and the
United States: III. Prevalence of Cardio-Respiratory
Symptoms among Migrants and Native-Born in the United
States. Natl. Cancer Inst. Monogr. 19:321-346
24
-------
RESEARCH TRIANGLE INSTITUTE (RTl). 1977. Identification and
Analysis of Ambient Air Pollutants Using the Combined
Techniques of Gas Chromatography and Mass Spectrometry.
Research Triangle Park, North Carolina. April 7, 1977.
Quarterly Report 1
RICHTERS, A., KURIATIS, V., and SHERWIN, R.P. 1979. Air pollu-
tant NO~ inhalation and cancer metastasis. Lab. Invest.
70:280 ^(Abstract)
RIGDON, R.H., and NEAL, J. 1971. Tumors in mice induced by
air particulate matter from a petrochemical industrial
area. Tex. Rep. Biol. Med. 29:109-123
ROBERTSON, L.S. 1980. Environmental correlates of intercity
variation in age-adjusted cancer mortality rates. Environ.
Health Perspect. 36:197-203
ROBINSON, W.S. 1950. Ecological correlation and the behavior
of individuals. Am. Sociol. Rev. 15:351-357
ROGOT, E., and MURRAY, J.L. 1980. Smoking and causes of death
among U.S. veterans: 16 years of observation. Public
Health Reports, Vol. 95, pp. 213-220
ROM, W.N., VARLEY, G., LYON, J.L., and SHOPKOW, S. 1982.
Lung cancer mortality among residents living near the
El Paso smelter. Br. J. Ind. Med. 39:269-272
ROSSI, L., RAVERA, M., REPETTI, G., and SANTI, L. 1977. Long-
term administration of DDT on phenobarbital-Na in Wistar
rats. Int. J. Cancer 19:179-185
ROTHMAN, K.J. 1975. Causes. Am. J. Epidemiol. 104:587-592
ROTHMAN, K.J., and BOICE, J.D., Jr. 1982. Epidemiologic Anal-
ysis with a Programmable Calculator. 2nd ed. Epidemiology
Resources, Chestnut Hill, Massachusetts
SAFFIOTTI, U., CEFIS, P., and KOLB, L.H. 1968. A method for
the experimental induction of bronchogenic carcinoma.
Cancer Res. 29:104-124
SAFFIOTTI, U., MONTESANO, R., SELLAKUMAR, A.R., CEFIS, F.,
and KAUFMAN, D.G. 1972a. Respiratory tract carcinogenesis
in hamsters induced by different numbers of administrations
of benzo(a)pyrene and ferric oxide. Cancer Res. 32:1073-1081
SAFFIOTTI, U., MONTESANO, R., SELLAKUMAR, A.R., and KAUFMAN, D.G.
1972b. Respiratory tract carcinogenesis induced in hamsters
by different dose levels of benzo(a)pyrene and ferric
oxide. J. Natl. Cancer Inst. 49:1199-1204
25
-------
SAKABE, H. 1973. Lung cancer due to exposure to bis(chloromethyl)
ether. Ind. Health 11:145-148
SALAMONE, M.F., HEDDLE, J.A., and KATZ, M. 1979. The mutagenic
activity of thirty polycyclic aromatic hydrocarbons (PAH)
and oxides in urban airborne particulates. Environ.
Int. 2:37-43
SANTODONATO, J. , HOWARD, P., and BASU, D. 1981. Health and
Ecological Assessment of Polynuclear Hydrocarbons: 5.5.
Human exposure from various media. J. Environ. Pathol.
Toxicol. 5:162-177
SAWICKI, E. 1977. Chemical composition and potential "geno-
toxic" aspects of polluted atmospheres. In Air Pollution
and Cancer in Man. International Agency for Research on
Cancer Scientific Publication No. 16, Lyon, France. Pp. 127-
157
SAWICKI, E., ELBERT, W.C., HAUSER, T.R., FOX, F.T., and STANLEY, T.W.
1960. Benzo(a)pyrene content of the air of American com-
munities. Am. Ind. Hyg. Assoc. J. 21:443-451
SAWICKI, E., MEEKER, J.E., and MORGAN, M. 1965a. The quantita-
tive composition of air pollution source effluents in
terms of aza heterocyclic compounds and polynuclear aromatic
hydrocarbons. Int. J. Air Water Pollut. 9:298
SAWICKI, E., MEEKER, J.E., and MORGAN, M. I965b. Polynuclear
aza compounds in automotive exhaust. Arch. Environ. Health
11:773
SAWICKI, E., McPHERSON, S.D., STARLEY, T.W., MEEKER, J., and
ELBERT, W.C. 1965c. Quantitative composition of the
urban atmosphere in terms of polynuclear aza heterocyclic
compounds and aliphatic and polynuclear aromatic hydro-
carbons. Int. J. Air Pollut. 9:515-524
SAWYER, K., PITTS, J.N., CROCKER, T.T., GORDON, R.J., SHIMKIN, M.B.,
and WINKELSTEIN, W. 1979. Final Report of the Ad Hoc
Panel on Atmospheric Carcinogens to the Air Resources
Board. California Air Resources Board Report No. 77-8-4
SCHLESSELMAN, J.J. 1974. Sample size requirements in cohort and
case-control studies of a disease. Am. J. Epidemiol. 99:381-
384
SCHLESSELMAN, J.J. 1978. Assessing the effects of confounding
variables. Am. J. Epidemiol. 108:3-8
SCHLESSELMAN, J.J. 1982. Case Control Studies: Design, Con-
duct, Analysis. Monographs in Epidemiology and Statistics.
Oxford University Press, New York
26
-------
SCHNEIDERMAN, M.A. 1978. Post-hearing Comments: Proposed
Regulations of the U.S. Occupational Safety and Health
Administration for the Identification, Classification,
and Regulation of Toxic Substances Posing a Potential
Occupational Carcinogenic Risk. U.S. Department of Labor,
Occupational Safety and Health Administration
SCHNEIDERMAN, M.A. 1979. NCI statement before the Subcommittee
on Health and Scientific Research of the Senate Committee
on Human Resources, March 5, 1979
SCHNEIDERMAN, M.A., and BROWN, C.C. 1978. Estimating cancer
risks to a population. Env. Health Perspect. 22:115-124
SCHULER, R.L., and NIEMEIER, R.W. 1980. A study of diesel
emission on Drosophila. In Pepelko, W.E., Danner, R.M.,
and Clarke, N.A., eds. Health Effects of Diesel Engine
Emissions, Proceedings of an International Symposium.
Vol. 2. EPA Publication No. 600/9-80-057b. Pp. 914-923
SELIKOFF, I.J., and HAMMOND, E.G. 1975. Multiple risk factors
in environmental cancer. In Fraumani, J.F., ed. Persons At
High Risk of Cancer. Academic Press, New York. Pp. 467-484
SELLAKUMAR, A.R., MONTESANO, R., SAFFIOTTI, U., and KAUFMAN, D.G.
1973. Hamster respiratory carcinogenesis induced by benzo-
(a)pyrene and different dose levels of ferric oxide.
J. Natl. Cancer Inst. 50:507-510
SHABAD, L.M. 1960. Experimental research on the link between
atmospheric pollution and lung cancer. Int. J. Air Water
Pollut. 3:221-230
SHABAD, L.M. 1980. Circulation of carcinogenic polycyclic
aromatic hydrocarbons in the human environment and cancer
prevention. JNCI 64:405-410
SHANNON, L.J., GORMAN, P.G., and PARK, W. 1974. Feasibility
of Emission Standards Based on Particle Size. March 1974.
Report No. EPA-60015-74-007
SHUBIK, P., CLAYSON, D.B., and TERRACINI, B. 1970. The Quantifi-
cation of Environmental Carcinogens. Union Internationale
Contre le Cancer, Technical Report Series, Vol. 4, Geneva
SHY, C.M., and STRUBA, R.J. 1982. Air and water pollution.
In Schottenfeld, D., and Fraumeni, J.F., eds. Cancer
Epidemiology and Prevention. Saunders, New York. Pp. 337-363
SIEMIATYCKI, J., DAY, N.E., FABRY, J., and COOPER, J.A. 1981.
Discovering carcinogens in the occupational environment:
A novel epidemiologic approach. JNCI 66:217-225
27
-------
SIMMONDS, P.G., KERRIN, S.L., LOVELOCK, J.E., and SHAIR, F.H.
1974. Distribution of atmospheric halocarbons in the
air over the Los Angeles basin. Atmos. Environ. 8:209-216
SINGH, H.B., FOWLER, D.P., and PEYTON, T.O. 1976. Atmospheric
carbon tetrachloride: Another man-made pollutant. Science
192:1231-1234
SINGH, H.B., SALAS, L.J., STILES, R., and SHIGEISHI, H. 1982.
Measurements of hazardous organic chemicals in the ambient
atmosphere. Environmental Science Research Laboratory,
Office of Research and Development, U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina
SIVAK, A. 1979. Cocarcinogenesis. Biochim. Biophys. Acta
560:67-89
SOCIETY OF THE PLASTICS INDUSTRY. 1980. Comments on Environ-
mental Protection Agency Proposed Rulemaking: National
Emission Standards for Identifying, Assessing and Regulating
Airborne Substances Posing a Risk of Cancer (44 FR 58642).
OAQPS Document No. 79-14
STANLEY, C.W., BARNEY J.E., II, HELTON, M.R., and YOBS, A.R.
1971. Measurement of atmospheric levels of pesticides.
Environ. Sci. Technol. 5:430-435
STANLEY, T.W., MORGAN, M.J., and GRISBY, E.M. 1968. Application
of rapid thin-layer chromatographic procedure to the deter-
mination of benzo(a)pyrene, benz(c)acridines, and 7H-benz-
(d,e)anthracene-7-one in airborne particulates from many
American cities. Environ. Sci. Technol. 2:699
STANTON, M.F. 1974. Fiber carcinogenesis: Is asbestos the
only hazard? J. Natl. Cancer Inst. 52:633-634
STELLMAN, J.M., and RABAT, G.C. 1978. Arsenical air pollution
and lung cancer: Reply to Lyon et al.'s 1977 letter.
Lancet, January 28, 1978
STENBACK, F., ROWLAND, J., and SELLAKUMAR, A. 1976. Carcino-
genicity of benzo(a)pyrene and dusts in hamster lung (in-
stilled intra-tracheally with titanium oxide, carbon,
and ferric oxide). Oncology 33:29-34
STERLING, T.D. 1975. A critical reassessment of the evidence
bearing on smoking as the cause of lung cancer. Am. J.
Pub. Health 65:939-953
STEVENS, R.G., and MOOLGAVKAR, S.H. 1979. Estimation of rela-
tive risk from vital data: Smoking and cancers of the
lung and bladder. JNCI 63:1351-1357
28
-------
STOCKS, P. 1957. Cancer in North Wales and Liverpool Regions.
Br. Empire Cancer Campaign Ann. Report. (Supplement)
STOCKS, P. 1958. Report on cancer in North Wales and Liverpool
Region. In British Empire Cancer Campaign 35th Annual
Report, 1957. Supplement to Part II. British Empire
Cancer Campaign, London
STOCKS, P. 1960. On the relations between atmospheric pollution
in urban and rural localities and mortality from cancer,
bronchitis, and pneumonia, with particular reference to
3,4-benzopyrene, beryllium, molybdenum/ vanadium and ar-
senic. Br. J. Cancer 14:397-418
STOCKS, P. 1966. Recent epidemiological studies of lung cancer
mortality, cigarette smoking and air pollution, with discus-
sion of a new hypothesis of causation. Br. J. Cancer 20:
595-623
STOCKS, P., and CAMPBELL, J.M. 1955. Lung cancer death rates
among non-smokers and pipe and cigarette smokers. Br. Med. C
2:923-928
STUMPHIUS, J. 1969. Asbest. in een Bedrijfsbevolking. Van
Gorcum, Amsterdam
SULLIVAN, J.L., and CLEARY, G.J. 1964. A comparison of polycy-
clic aromatic hydrocarbon emissions from diesel- and petrol-
powered vehicles in partially segregated traffic lanes.
Br. J. Ind. Med. 21:117-123
SUTA, B.E. 1978. Human Exposures to Atmospheric Arsenic.
Prepared for U.S. Environmental Protection Agency, Contracts
68-01-4314 and 68-02-2835. SRI International, Menlo Park,
California
SZALAI, A. 1972. The Use of Time: Daily Activities of Urban
and Suburban Populations in Twelve Countries. Mouton,
The Hague and Paris
TABERSHAW, I.R., COOPER, W.C., and BALZER, J.L. 1970. A labor-
management occupational health service in a construction
industry. Arch. Environ. Health 21:784-788
TABOR/ E.G. 1966. Contamination of urban air through the
use of insecticides. Trans. NY Acad. Sci. 28:569-578
TALCOTT, R., and WEI, E. 1977. Airborne mutagens bioassayed
in Salmonella typhimurium. J. Natl. Cancer Inst. 58:449-451
TALCOTT, R., and HARGER, W. 1979. Mutagenic activity of aerosol
size fractions. NTIS, Springfield, Virginia. PB-294
732/3GA: EPA/600/3-79/032
29
-------
TAYOT, J., DESBORDES, J. , ERNOULT, J.C., and POTOINE, B. 1966.
Mesotheliome pleural et asbestose. J. Fr. Med. Chir .
Thorac. 7:757-774
TERANISHI, K., HAMADA, K., and WATANABE, H. 1978. Mutagenicity
in Salmonella typhimurium mutants of the benzene-soluble
organic matter derived from airborne particulate matter
and its five fractions. Mut. Res. 56:273-280
THORPE, E., and WALKER, A.I.T. 1973. The toxicology of dieldrin
(HEOD): II. Comparative long-term oral toxicity studies
in mice with dieldrin, phenobarbitone, a-BHC and p-BHC.
Food Cosmet. Toxicol. 11:433-442
TODD, G.F., LEE, P.N., and WILSON, M.J. 1976. Cohort analysis
of cigarette smoking and of mortality from four associated
diseases. Tobacco Research Council, London. Occasional
Paper No. 3
TOKIWA, H., TAKEYOSHI, H., MORITA, K., TAKAHASHI, K., SARUTA, N.,
and OHNISHI, Y. 1976. Detection of mutagenic activity
in urban air pollutants. Mutat. Res. 38:351 (Abstract)
TOKIWA, H., MORITA, K., TAKEYOSHI, H., TAKAHASHI, K., and OHNISH,
Y. 1977. Detection of mutagenic activity in particulate
air pollutants. Mut. Res. 48:237-248
TOKIWA, H., KITAMORI, S., TAKAHASHI, K., and OHUISHI, Y. 1980.
Mutagenic and chemical assay of extracts of airborne particu-
lates. Mut. Res. 77:99-108
TOWNSEND, J.L. 1978. Smoking and lung cancer: A cohort data
study of men and women in England and Wales, 1935-1970.
J. R. Stat. Soc. A 141:95-107
TRICHOPOULOS, D., KALANDIDI, A., and SPARROS, L. 1981. Lung
cancer and passive smoking. Int. J. Cancer 27:1-4
U.S. BUREAU OF THE CENSUS. 1980. Statistical Abstract of
the United States. 101st ed. U.S. Government Printing
Office, Washington, D.C.
U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES (THE SURGEON GENERAL)
(USDHHS). I982a. The Health Consequences of Smoking:
Cancer. Public Health Service, Office on Smoking and
Health, Washington, D.C.
U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES (USDHHS). 1982b.
Health United States 1982. DHHS Publication No. (PHS)
83-1232
30
-------
U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES (USDHHS). 1983.
Health United States 1983. USDHHS Publication
U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE (USDHEW).
1979. Smoking and Health: A Report of the Surgeon General.
Office of Smoking and Health. DHEW Publication No. (PHS)
79-50066
U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE (USDHEW).
1980. Health United States 1979. DHEW Publication No.
(PHS) 80-1232
U.S. ENVIRONMENTAL PROTECTION AGENCY (USEPA). 1974. Preliminary
Assessment of the Environmental Problems Associated with
Vinyl Chloride—Report on the Activities and Findings
of the Vinyl Chloride Task Force. September 1974
U.S. ENVIRONMENTAL PROTECTION AGENCY (USEPA). 1975. Sampling
and Analysis of Selected Toxic Substances: Task Il-Ethy-
lene Dibromide. Final Report. Office of Toxic Substances.
September 1975. EPA 560/6-75-001
U.S. ENVIRONMENTAL PROTECTION AGENCY (USEPA). 1978. An Assess-
ment of the Health Effects of Arsenic Germane to Low Level
Exposure. Revised External Review Draft. OAQPS-"Z9-8-II-A-6.
Prepared by Criteria Development and Special Studies Div-
ision, Office of Health and Ecological Effects, ORD
U.S. ENVIRONMENTAL PROTECTION AGENCY (USEPA). 1979. SIR-Human
Exposure to Atmospheric Concentrations of Selected Chem-
icals. Office of Air Quality Planning and Standards,
Research Triangle Park, North Carolina
U.S. ENVIRONMENTAL PROTECTION AGENCY (USEPA). 1980. SAI-Human
Exposure to Atmospheric Concentrations of Selected Chem-
icals, Vols. I and II. Office of Air Quality Planning
and Standards, Research Triangle Park, North Carolina
U.S. ENVIRONMENTAL PROTECTION AGENCY (USEPA). 1981. Technical
Document: Formaldehyde. Office of Toxic Substances,
Office of Pesticides and Toxic Substances, U.S. Environ-
mental Protection Agency, Washington, D.C. November 16,
1981
U.S. ENVIRONMENTAL PROTECTION AGENCY (USEPA). 1982. Review of
the National Ambient Air Quality Standards for Particulate
Matter: Assessment of Scientific and Technical Information.
Office of Air Quality Planning and Standards. Staff Paper,
Research Triangle Park, North Carolina. EPA Publication
No. 450/5-82-001
31
-------
VENA, J.E. 1982. Air pollution as a risk factor in lung cancer
Am. J. Epidemiol. 116:42-56
WAGNER, G. 1971. Cancer diseases and cancer mortality.
Therapewoche 21:644-651
WAGNER, J.C. 1972. The significance of asbestos in tissue.
In Grandmann, E./ and Tulinias, H., eds. Current Problems
in the Epidemiology of Cancers and Lymphomas: Symposium
on the Epidemiology of Cancer, Lymphomas, Duesseldorf,
November 4-6, 1971. Recent Results in Cancer Research.
Vol. 39. Springer-Verlag, Berlin
WAGNER, J.C., STEGGS, C.A., and MARCHAND, P. 1960. Diffuse
pleural mesothelioma and asbestos exposure in the North-
western Cape Province. Br. J. Ind. Med. 17:260-271
WALKER, A.M. 1981. Proportion of disease attributable to
the combined effect of two factors. Int. J. Epidemiol.
10:81-85
WALKER, R.D., CONNOR, T.H., MACDONALD, E., TRIPF, N.M., LEGABOR,
M.S., MACKENZIE, K.W., JR., and DOBBINS, J.G. 1982.
Correlation of mutagenic assessment of Houston air particu-
late extracts in relation to lung cancer mortality rates.
Environ. Res. 28:303-312
WANG, Y.Y., RAPPAPORT, S.M., SAWYER, R.F., TALCOTT, R.E., and
WEI, E.T. 1978. Direct-acting mutagens in automobile
exhaust. Cancer Lett. 5:39-47
WANG, Y.Y., TALCOTT, R.E., SEID, D.A., and WEI, E.T. 1981.
Antimutagenic properties of liver homogenates, proteins,
and glutathione on diesel exhaust particulates. Cancer
Lett. 11:265-275
WEINBERG, G.B., KULLER, L.H., and REDMOND, C.K. 1982. The
relationship between the geographic distribution of lung
cancer incidence and cigarette smoking in Allegheny County
Pennsylvania. Am. J. Epidemiol. 15:40-55
WEISS, W. 1978. Lung cancer mortality and urban air pollution.
Am. J. Public Health 68:773-775
WESOLOWSKI, J.J. 1975. Asbestos in the California environment.
AIHL Report No. 164A
WICKEN, A.J. 1966. Environmental and personal factors in
lung cancer and bronchitis mortality in Northern Ireland,
1960-62. Tobacco Research Council, London. Research
Paper No. 9
32
-------
WILSON, R., COLOME, S.D., SPENGLER, J.D., and WILSON, D.G.
1980. Health Effects of Fossil Fuel Burning: Assessment
and Mitigation. Ballinger Publishing Co., Cambridge,
Massachussets
WINKLESTEIN, W., KANTOR, S., DAVIS, E.W., MANERI, C.S., and
MOSHER, W.E. 1967. The relationship of air pollution
and economic status to total mortality and selected respira-
tory system mortality in Men: I. Suspended particulates.
Arch. Environ. Health 14:162-172
WISLOCKI, P.O., WOOD, A.W., CHANG, R.L., LEVIN, W., YAGI, H.,
HERNANDEZ, O., DANSETTE, P.M., JERINA, D.M., and CONNEY, A.H.
1976. Mutagenicity and cytotoxicity of benzo(a)pyrene
arene oxides, phenols, quinones/ and dihydrodiols in bac-
terial and mammalian cells. Cancer Res. 36:3350-3357
WODINSKY, I., HELINSKY, A., and KENSLEV, C.J. 1964. Suscep-
tibility of Syrian hamsters to induction of fibrosarcomas
with a single injection of 3,4,9,10-dibenzpyrene. Nature
203:308
WOOD, A.W., WISLOCKI, P.G., CHANG, R.L., LEVIN, W., LU, A.Y.H.,
YAGI, H., HERNANDEZ, O., JERINA, D.M., and CONNEY, A.H.
1976. Mutagenicity and cytotoxieity of benzo(a)pyrene
benzo-ring epoxides. Cancer Res. 36:3358-3366
WORLD HEALTH ORGANIZATION (WHO). 1969. Health effects of
air pollution. WHO Chron. 23:264-274
WYNDER, E.L./ and HOFFMAN, D. 1959a. A study of tobacco carcino-
genesis: VII. The role of higher polycyclic hydrocarbons.
Cancer 12:1079
WYNDER, E.L., and HOFFMAN, D. 1959b. The carcinogenicity
of benzofluoranthenes. Cancer 12:1194
WYNDER, E.L., and HOFFMAN, D. 1963. Ein experimenteller Beitrag
zur Tabakrauchkanzerogenese. Dtsch. Med. Wochenschr. 88:623
WYNDER, E.L., and GORI, G.B. 1977. Guest Editorial: Contribu-
tion of the environment to cancer: An epidemiological
exercise. J. Natl. Cancer Inst. 58:825-832
WYNDER, E.L., MABUCHI, K., and BEATTIE, E.J., Jr. 1970. The
epidemiology of lung cancer: Recent trends. JAMA 213:
2221-2228
YOBS, A.R., HAWAN, J.A., STEVENSON, B.L., BOLAND, J.J., and
ENOS, E.F. 1972. Levels of selected pesticides in ambient
air of the United States. Presented at the American Chem-
ical Society Symposium on Pesticides in Air. Boston,
Massachussetts. April 11, 1972
33
-------
YOUNG, J.L., ASIRE, A.J., and POLLACK, E.S., eds. 1978. SEER
Program: Cancer Incidence and Mortality in the United
States: 1973-1976. National Cancer Institute, Bethesda,
Maryland. DHEW Publication No. (NIH) 78-1837
ZEIDBERG, L.D., HORTON, R.J.M., and LANDAU, E. 1967. V.
Mortality from diseases of the respiratory system in re-
lation to air pollution. Arch. Environ. Health 15:214-24
ZIMMER, E.G., and HAENSZEL, W. 1956. Cancer in Iowa. Public
Health Service Publication No. 466
34
-------
APPENDIX A
TABLE II-l
Urban-Rural and Other Geographic Studies
of Cancer: Code to Comments
a. Limited information on types/ duration, and intensity
of exposure
b. Smoking habits not taken into account in design or analysis
c. Occupational exposures not taken into account in design
or analysis
d. No information on socioeconomic variables
e. Dilution effect occurs owing to migration
f. Dilution effect occurs owing to labeling all residents
of certain geographic areas as "exposed" or "not exposed"
g. Cause of death as recorded on death certificate may be
inaccurate
h. The Standard Mortality Ratio (SMR) may be biased when
numerators (counts of death) are based on death certificates
and denominators (population counts) on census data
A-l
-------
OS
a
u
§
o
CO
w
o
M
aw
a, a
33
§p
H
CO
°J
« <
U U
a M
E-I u
O Q
Q O
OS
D
3
3
on
4J
C
0
Q.
a
i Ol
•H C
o -H
C -0
•rl C
a
0
h 1-4
0 XI
f. «
S-p,
0
*>
X l«
w o
«
4J
a
•p« 09
01 0
>
g
I
a
U >H
1-1 .C
a
0
-^ « a w
C a •O •
o e a. M oo
Ix53.cc
5 5-0*3 -S**
cflCtnac
O
oo-rtaiioai04J
i8cc*j«av4«f
S5-i5«8*j x: -w 3
M 4 0>S « M
OrHiHM0S - « 0
cfl4,a>ioa.-ic-*4
a 1 B M-H 0 10 ia M
U 3 (9 kl -r« O 14 4.)
0 0
i
ki i
•• t 0 *i
XW ft C
4J 0 0
•H ft •- ki
a B -U - 0
•v.0?
ax: -H
w o vi
0 10 3
x 0 -u
14 O 00
Se 10 0
ft « cn
C B 0
• 3 _ **
i >o
vi a>
O r^
a
O X
4J
4J
>r* •«
e 0
o a w a
c 0 • 0
4J0
I
1 O> 0
0 g—H C £ B
U 0 O
X £ O _^,
18 O "O O
>i o ^3 en o wi 3 D o.
«w M C «TC 3-0 U 0
«-H^'H-H»-ioCid-r4
won ««ooo
4j-oa)>i
« 3 « o o a c *> a a a
x c a>
0 «>,C
a ---i -u
at)
c
ai
0>
s c
m « o
U
c<« o
-H«W
w >-
ft 3
Srtupu
C
10 -i C •** Ol
-
fiCiOO)C
o)3Sn3
«o«)i-o
41
C H
0 3
10 C
rH 0
3 O
ft U
£ft
•V K-l M PJ
rH PH O 0
o x-p« e >w 0
o xt a 01
A 10 -U *J iO
o "0 <« c e *>
(0 H U VJ
0 O ft ft ft
ft.?
4J
4) B
M C
0 C-H
X -rt
14 B
0X3
S -u 41
C --I
. 3 0
fl a
4J 0
V4 4J
M C*l P-I
0 CM I
o o
C wm
0 B
"-» ft 3 B
0 T3 0
rHON d -*4
iO n -p4 4J
B C
-S3
•• 0 3 O
41 pH 4) O
.Uf-
0 Ol
0 •
sjM
CQiO
C-p4~CCBTfpHO
0 C 10 -rt -H • ff> •
UB-HO CPH0PH
0 *j •*
" ' ~ ft C 4->
a> ^
Ol
» C 4J
0 -H C
ax 3
•H O O
ft a a
a 10
0 14 C
0 -p4 >Q *O
u o 0 0
10 OX
o »
«-i
*)O
cc
°1
-------
K
W
CJ
3
i^
"
0
w
Cft
w
S "n*?
CO Q)
3
UJ ^
M "*"*
S ^4
*** n
w. y
2£
W OJ
cs Ed
W D
£ S
0
Q !er
30
< M
J S
Irf S
D CJ
1
§ .
m tf
2
*
1
M
U
Ofl
*3
t«
*
co
4J
C
M
C
o
•H
4-1
id
rH
3
a
£
4)
4-1
S
^^
O
3.
HJ
*,
0 rC
rQ Ol
C
0 1 1
IH i o a>
4J C '01 1 CO -H
id 0 C en c « rH
rH -in o id O id id
41 4J -H c en
r4 3 4J S O B
rJrHiaO-O-H^SO)
O rH .4 rC C •" 4-1 Ifl
0 £ D> 0) rt-JJ-J ^
4) 8 0) O 4) O
> e 4J cn-H a c oi
4-1 4) O -rl -r) 1 -H
iH 3 "O O 4) 4J 4J
C04->T3C4>4J|OC
O 4) C -H Q|iH .H Id
OtrQighnaoc
,%
4)
o it1
C rH
•r* -H
,0
M B
iH
a c
id o
<*J -H
4J
C 10
Id rH
•rl 3
•d Q.
g&
1
•H
*J
CO 4)
4) rH
•iH
*H a
0
o1
B ™
0 T| tH 4J
4) Q.4-1 O
e m n id
41 01 M
4J M tT* 4-1 CQ
c < -H a
cm M ifl
Ol 8 - Id rH
•H 01 >i c a.
rH Id r4 -rl 0
:§ &4J 3 c
4) CO
rH •
id S
a a
.. «)
•» 0) •*•*
41 rH 4-1
4J Ifl C
•H a 3
£ 4) 0
S C r4
3. 3 -H 4)
rH O
rH rH* C
O m id
•w a o
0 0
CO 4-> >1
4) 18 M
n "O IB
id 4) 0) B
4) C 4J 0
Vi 4) 18 B
O r4 rH rH
C >,4) 3
M Q. -1 ft
1 >H
u 0
a. n
^~ n
n n]
4) 4)
10
com
4)
4J 01
41 >
M 0
10
•H 0 DI
cj w «
a 4)
4) C
D1 01
10 M V4
V4 rr1* 01
rH V4 ^0
§• BrH
•••
4)
v a
r: o) o
It rH 3
C 10 tJ>
O 8 -r*
C 41 4-1
01 ~>- 0 4)
4J 01 O -P
•H rH 18
SS5U
•Q t*>
Sf-
id cri
rH
O u
C 4)
14 -r(
V4
id oi
C 4J O
O *r4 C
4-1 £4 1 Ifl
id c o
O O -H
•H tn
>H u o* a
4J 4) O 3
0) ftCN rH
BTJ 0 4)
t— 1 ,
3 4)
co c u m
o en
O «-H M
18 O 0)
41 4J >
S o."?
w « CO)
U CJ CO O Ol
18 TJ » id
5J1 V4 C r4
•H 4) 18 4) <4J
CJ Di a Oi 0
«
M
I id
l id o 4)
C 0 -H >,
O M^~,
O V4 4J (Q
rH^ag
4) to CO
IH O -"H 0) fJ»
•O 4J -^>. 1
S3 -r< 0 ^ ^|
C rH rH
0 « 1
a 4J co
rH U m
fi a rH
i".c
4J 4J
*2 O C
3 2 10
e a
001 e c 41
C rH 18 18 -H
^- O 0) M
4) •» -H CX 4J
4-> 4) .4 O C
•n -H 0) M 3
O >H
•. «*
rQ 4) fl
fc •. S,
10 -d en
r4 01
01 1 -H
rC C r4
(2 J«| | t Q ^j ^( fl)
01 o) -H u 4J o o
4) O 01 -H 4J 3 C
3 C CO O 4) id
4-*(8lO *CiOr2O
4) O C 0) 0) iw
rQ Id rH "O H *O
4) rQ 4J O 4)
•H« 2.2 S S*1 «
rCB^r^rQOtT)^
C 4) (H > D MH -r4 rH
O 4J ^H 4J 4)
**^ *^ »»^i C C C
4J,C4JO"O13CO
IdS-rt-rtC-HO-
rH rH 4J O O O >1
01 rH 10 3 -H O0)01
V4ld4^rHO)4J<0 QB4)
4J ,4 rH 3 CO >H 4J
OOOOrHOCCUld
^
4J
•H
0)
C
•U
c
id
,Q
r4
D
4) TJ *
rH 0) 1 0)1
§>i co c id co
O 3 4) 0 4) 3
rH "O Cn-fl rJ "O
>M a c id JJ id c
0> C rH C
4) s 4) 3 -H >!
Cn 4) >i O O< >
4J H 18 4) Oi C 4)
C 0 41 D, -H rC
41 i*H ,C 41 13
0 X S •• 4J ^ ,C
4)OC,44-14)-r(M
ftS-H4JIH.4*4J
1
14
O
8
«.
M CO
0) 01
O 4J
Id rJ<0
o ^
XrH
rH 4J 1
Ifl -rt O
4-> rH "1
O Id C«
EH 4-1 rH
• «
0)
0)
rH
aS
4)
-4-1 CO
4) 0) 4)
4J 18 4J
•H 4) Id
0 C
0 g
rH B
3 4->
I3o
a TJ ,
4) 4-)
O kl V4
C 41 -H
18 rS JJ
^H
4) rS C
en -H
18 a
V4 01 C
4) 4J id
> 18 rC
< IH 4J
•o
0)
>,
0
rH >,
Q.-<
8 4J
0) a
3
DO t3
4) C
rH fH
§>,
>
IK id
0 4)
(rC
o c
1
Cn n
•H Q)
CO -r-t
0) 4J W
*O •*-* 0
o
4) :
4J 4) :
O C rH >1
41 O O 4J
•m>H : V4
Sid W *O
C id =
O
1 0)
C (fl rC
id c o oi i
o id u CM
* S D w
i« 01 o o 01
o e 4J c -i
3 o> a
O - r4
£• V-l X "™H i
VI 0) 14 rH
O U 18 T3 H f*
MO Ol 10 M-l rH
4)"
4J
rC W
X •
C D
O
CI~
0) .. J)
4J 4) -
-------
B)
O
b
O
to
u
to 0
u c
i'i
28
H to
o u
a: S
H 3
§CJ
M
I"
U
J
I
*
to
C
01
I
So
en
i^ C
U fi
n
0
to i-i
0 A
£ id
to
O
4-1
X 10
to u
•H -H
« |Q
c
0
id
to en
o> ••••
4- 4J
•H 09
to 0
o
id
I-l
3
&
to
O
U MJ
A 0
tj *o en
I
3
a
4-1
•o
0
0 0
4J to
en
•o
o id
4J
N to
^H 0
n c
id 0
Qj4J .
0
CO.
id
I-l
0
a-
0 C
0 .H
a o,
09 Q
id P.
U
a en
id , -H 4J x 41 en nj
M 13 4J 3 0-Hi-lh
0 0 3 A >I-H .C 0 0
4J 4J iH 4J O to U
idiOi-l • -H 0 •• h C
CU to '
09 I-l W 4J to 4J
3 U U
« 09 S
c?
I
l
HJ -
en tn
•^ id
B " C B
0 CO 3 O •- 0
C to U C U
0 O - C O id
O 4J >I-H •* fc X
4) ifl 4-1 4J 0)
•^-rt 0 4J O C
o -a > c 3 o -a
0 C 0 0 "O-H C
W -H (X S 0 4J 10
c>-iC4J-ic4J
0 0 -H
O 4J r-l tO
0 U uci-i
'•O O 4J
09 4J 3
S-SI8
id
4J 00
U 4) M
O 4J 0
E -H o
iH O
O *J
ia
aii
-Ho
o D
B
id o
MC
09
0
0 U
a 0
o c
o
S N
o
Id
en c
c o
•H -H
09 4J
« 3
id *•• r-i
• 0 0 o*
to ema *o
•a
a c
id
in u
"«lO
to
O
to —> o N
i-t 0 0 id C Q •-<
3 tnt) N3 -H^
-o
•a 1-10
IH Id 00
- -nm
m -i
£
jD Cn
id MJ
•o
0) MJ
4J 0 0) 03
4) >ti-i to c to mo
•-I 4J O 0 0 C -H -H
«J-i-IOfct))ld4Jr-l4J
O 0) 4J 4-* 4J
4J Cn 0 ,C O 4-1 3
•,J C 4J 4J IJ (4'Ot3
I
0 <*!
I
3
r-t^
•H >J ft - 0) 09
> - id « 4J >,
fi O > ki M Id r-l
U) t4 0) 0 0) 10
r-l 4-1 10 10 10 O 10
id o c x o
d id 0 f» in »j
O O
CO 4-> C r-l 4J
14 rH IB --H A O
4) id Oi eo fl 10
O<4J 3 41 C i-l MJ
rf 0 O A -H id >i
•-I H O ~ •' ~ ~
-d
oi
c
O 0 C
O 0
O -H
£ O i-l 0 U
gg id M-ir- I a O 0
3 r-i in en u 4J
0)
U 0
C U
id to
3
"O ^1 09
0 0 W V
C X ••* U
M o id -i-i "O
3 Dj *O fl^
£ C C O,
O
o >
en rn <^ M-I -r-i -a
-H B 1 rH o
T3
oidO)a>4Jia
> -H 4J 3
&3- ^
— rH
« 4^ 4J 3 m
C 3 (0
6 cr x^
e« 0 A cu
ai
i
(0
o
to
W00 - -Cr-lrH
to to to 4J >I-H •->
0
•- C
O 4J
(0 to h
09 0 X 10
id C n o
jj 0 iQ ..«
04 to -H - IB
0
••O
B to
0 0
0 X I C
4J 0 I O O
•H - z a to -H
A 0) 09 r-4 4-1 O>
X •-! 0 Cn-H 0 0
C 0 C 09 I 10 C
. 0 to 0 A 4J
4J 0 O Di 09 Oi-H
en
I
d) ON
Si-
lj OA
i
§
o
A-4
-------
OS
a
o
§
0
b
O
CO
H
Q
S *~7
C^ "O
Wfll
V
3
w C
fG jj
££
28
U CO
u u
OS Q
W G
£ &
o
Is
j X
2 j
S o
Ski
§ .
1<
D
1
BJ
3
£-•
*
01
4-1
C
41
E
B
O
U
rH
S, en
•H C
U i-l
C "O
•H C
as
0)
4)
(U Xt
ff (0
O'M
0
4J
w o
•H -H
os-o
HI
•o
4)
±)
10
a en
4) i-l
4J 4J
i-l 0)
CO 4)
M
C
0
•H
4J
10
rH
3
Q.
£
4)
&
"^.
hi
O
4->
g
.. „
0 tn
XI **H
«, h
(0 41 .£*
•
01 01
4) 01
rH O 13
Cn 41 M 01 1 41 1
C C Cn O Oi *H 41 01 M rH ,C rH VH HI 01 —
O rH Q 3 O rH Q) Wl rH -rl CO ON
4J rH O fl O Cn 41 01 01 rH
(0 tfl (0 -H-HOC01C3K
i4Cinj<:cj4-i4iOoioi'O<>J •
O4->HO(OU>-rlCrVlrH
•O B C -r( 01 Q, 4J « 1-1 (0
4) 10 4) VJ 3 3 O 01 Cn
n U rH CJ • XI JS MJ C 4J
4)4l^01M04llH4) 4J
U O 4J 41 4) -n 4J O ^ C 4J 41
C (0 0 X 4) C O -H 10 -rt -rl B -rt
M o -H
•rt to 4) B
K rH Ifl
01 4) 4-1
01 > rH
« 10 (0 C
rH C 10
0 C 0-H
O i-l TJ
rH 4J 4) 4)
10 -O (0 B B
•H 4) O 0
o 01 3 -a o
O 10 T5 C C
CO A « (0 -H
*
O 1
0 1
01 1H M
r-t 10 10
4) U
0) (0 H-—
^311
C 3 «—
41 C O
•H >,-H a
XI rH 4J C
§ Q ifl 0
< 0. BX1
•a
1 C
__ C (0
Cn S< » 1
B X 00
rH S SON
4J K rH
O > rH -
10 I-l
M O - 4)
41 01 T3
O rH 10 "O CN
C (0 4) (0 1-
A V4 k4 rH ON
0 O 0 XI rH
a
a 4)
VO 4J rH
4) CO U 4)
rH IT) It Cn
10 rH M C
8 4J S
- 4) B ffl >1
41 rH 3 Q 4J
4J 10 o i-3 c
•H B C 3
g^S.S8
§
a
M •
4) rH
>O ifl "1
c f-
41 4J ON
B 4) rH
* «
'O & *O rC
1. % •» *
o en u en
•. «. •* *
X5 H-f P M-i
•. * •«« * »
IT) -l 01 C 4J
•H H) - >O O T3 Itl rH 4J 10 U 4)
4J4)SC-HrH IOC* MO «>>
CiOr-;3(flT3'O4) COV< H O Q< r-i-H
O rH .H C kl (0 -H - 4) C IS 14 4) (04J
•HOIM-IT3-lfleOlH 4J 14 41 O 41 OW 4J (0
4-> M O IT) 4) 4) O Cl04)4JC41'O<4H C 41 Cn
10 >H C CO 01 rH O Or-)U(Ol02C 4J O S 41
rH O 01 10 4) 41 XI •H4)C'4-IO4Jl0010-H C
4IO01>C4JlO«l 01 H 10 rH 41 4J C 0] C
rl O 10 (fl -H -H 01V4CJ34->Xll4rH 01 0 >1
VjcnioHHcni-iH^:^ 41 O w 01 41 3 01 01 rH
8C4IOC fl 4J rH M CJ Cn lO-OOJ-lV401rH
On fl M > 4J O< C-O4lw 4)
C04-> O - T3 ^
c co c id ot e 1-1 NOI c it
41 4) 4J 4) 4J O 4J Ifl ON
•H UH XI - fl l-llOCOlfl rH
XI 0 01 rH XI M rH -
B -0)3 B 4-i - 3 Ol -
10 01 0) rH O 10 C 4J rH IH VO 10 C VH rH 4J
41 41 4) 0 id ON 41 0 >*-i fl 3 41
z rH a o ftrH E o o £i 01 a
a u a
>1 0> 10 3 rH
0)OO)M3rH -n CnC V4I04J" 4JrH —
4-14-IHOCJfl'OrH C-rl fl 4J 01 01 (0(00)
JOlOO)CnO)CCrH 34JL4 C*r4lflO)MO)4-lC %
kiwuo>iao)ioo) rHao) >i*ri CV^OITJ-TIO en
*r4 C 4^ ^ M S 4) M< V4 M Q) k4 iH r| (^ .f«4 ^}
SOl'OlOO * u-i-i-t P)O rl l-nf^ fi^ [im J|l il_i_i^»
WOICJO-H-OIOI 0 C 3 4-> oocnifl(OVO
*r*0) lH010]4-! UO-MBBSrHt-OlOM rH
fl 01 10(0(04) 01VJ4J-n4IOI4lfl OIUOOII
4JB4JOO414IX1 O 4J CJ Q44J 4J 4J O 01 1-4 C
0) Cft •- rH iH •
rH rH 0) (fl <
S4I B 01 •
0 01 < W
co a « *J
4J 4J
0) 1** 0) ^
vo r>
S9v SON
0) rH D rH
X X
CJ • O •
33 10 B3 (0
A-5
-------
OS
u
h
o
CO
U
CO 41
a. c
28
8-
M ca
o w
« a
w p
E £
sg
J O
I
a <
I
*
03
4-1
S
0)
a
o
u
a I
I
09
- .01
•H C
O -H
c -o
•H C
kl -H
a
4)
kl i-l
41 rQ
rC 10
kl
0
j^ JO
00 O
•rl -H
OS 13
c
•a
4)
10
09 O1
4) -H
4J 4J
•H 09
M 41
o
I
£
0
O O!
,a >w
fl 4) ff
kl
41 I
e id ki -H ' 09 4i
O O kl >t,C P 4) 03 -H
"W 41 0 id 41 ^TJ r-l fl
> O -H kl 41 ft-H »
rC I 4-> 03
>lOOC4)4lrCrH
kl -H -H -rH -
«e
id -H 3 3 ki
•4J C O -O 4J
> -H o u a o
4J 03 -H C 09
C4I-H -H4) C N 4J 4) O ki
N 3 O 03 C O4J-H-Hid,G «3
-co
id o c c id
O O ^ 3,0
I C C U
C 3 41 id
01 -H
id ai
si
o »
id a
ft v
10 03
0 . 3 *
B 00 V M^
ki ki 4) id U'-'
O 4) Hfl C
09 ki C
41 01 fl - O>rH
•u o J5 n a
fl c u 10 y "0
M fl 3 rH rH C
O 3 fl
>i cow
0 +-i-P C
N >i-H 003
b
£
-H cio
4J ^= 1-1 u ^
0 -H
O-^
-H B -H C
3 O id 3 C 3
09 "W 01 4->
4J 4J 8 U
c c c c >i id
4) 10 fl 01 0 tJ
U -H-H O iH 3
ki 13 T3 M ft C
4) 41 41 S id
S B ft 41 B
0>
CU
,§c
• 4J ,C O
O C O-H
C 3 4-1
0 C U
•a o-H 3
O rH 3 C kl r* -H
CflrHC-H4)3rHT3
O
C
O
3
kl
a
I
kl
00
4)
to
4) VO
Sci
S rH
C 10
o a
e 4) i
4-1
09
41 I O
rC C 6
4J w<
09
00 ,C iH -H
4) U 10
•o
41
fl
ki
•HrHff>3 41B1C7IC
£2"pCrC3-Hp
41 I
u c
C ki
09 Id kl
4) 0
ft m o
>< a
4J fl 01
rH kl
09 10 O
ntJ id
DON
CM W -H
f-l OI'H
•a c 4->
»rC O
03 O ft
to a
rH 4J
0>O 3
-Hkl
nj l-i
0) O
-H
O
0
W
id
kl
3
h
03
>
r3
kl
01 . 03
4J rH fl C
i-IIA l-t 3 « C 3 rH -
§w ki-Hieog fl CM cn
09 4J a c e c 09
- 4) (*> C l-i 01 .--^ 01
»4l-r
fl t-l
I
C
O
v£>
-
io>iio
kl 41 kl -H
4) > kl 4J
C 4J O rH
10 00 rH
o 01 4J g
en c ft
iH -rt 10
4-1 -H iH *O
O -W "4-1 fl 0\
4J O-H rt
kl kl_O>4J •O
fe o 03 a 10
ff fl4ICrQC4IO
idkiCidkiO-Hft
ki o a c w o
•H —« C kl
- ifl C -H kl 3 -U
O ki 4J f-ir-
C >— O 00 C C
fl 0 ki • o ifl id
4J 4)^- 4-1 -H 4J
0^ C (0 rH Id fH 00
3,OkiB4) O|4)>H
•HklOl rHiH04J4J
ki 414J e ki 4-> o) 3
boidrQ^ktacnu
(0
3 in
QW
O 0 4J 0 03
fl 4-> ft kl
0 4J rH O 0) 4J
3 O 4J 10
4) ki ft O ifl <0
OIOIQ S rH £,H C7>
O C 1-1 03 - 4J 4)
kiOiOCXiOC
0) -H o 01 -rt 0) O
0- 4J >,T3 B J= Z
C <4-l I
«0. (
- 01 01 3
C rH U
id 4) >o -H
01 > C 4J
S 4) id ki
IB
01 ft
a s 4>
3 3 4J -O
S a Id 4) 09
•H-H
09 >i>O 4) O»
C 01 O id C
•a « « o s
rH kl 4J Id
id
kl
I
3
S rH 00
§2<
C 41 W
>il5
•H iH f*
SS3
41 S X -H J<
Ifl M-l O 3 .0
OJ 4)
4J O
41 V0
C rH
4) rH
J ro
ft
I
o
o
4)
s
o
41
s
A-6
-------
os
H
o
3
B&
Pu
o
CO
w
D
C «T
CO 0
U B
ZJ **
0, B
sss
W CQ
O M
fltf Q
w o
gg
o S
2 CJ
fl
2 o
M4 W
2
8*
D
1
w
!rf*
&•
*
(0
4-*
B
0
s
a
0
o
id oo
cu en
•H B
tl -H
V f*
G "O
ft B
w
a
to i-l
0 fl
fl «
•
M
0
4-1
£6 id
00 O
•rl -H
OS -0
c
•0
0
4->
id
01 en
0 -H
4J 4J
•H 09
W 0
^
c
HI
§
•H
id
3
a
&
0
•^
M
0
4->
3
•c
O O
3 Cn o< io ft c
to O 0 O M fl O
o rn s - a « >.
0 4J oo 0 B B 0
B O*3 4J C fl fl
0MB fl a 0 -H M
•H d) 1 Cn-H 0 M *H
OB4Ja900
v-ft'W04*OiQlu w
0
1
M-l
O
0 ><
O to
a 4J
0 00
01 3
SI
B o
id i"
c o>
0 ft B
44 .H Id -
id id to
to a N 0
M 0
Nrt O C
4J id 4J id
•H 4J Id O
iH O M
Id 4-> ft iH
4j o. id eft
M M n 4J VO
60000*
^
0 B
4J -H •
•H 1 -•»» fl X
fl M 0 4J 0
2 3 rtin -H E«
B a id i-i S
OB
B fl " a 0 B • -H 00
B C 3 U 0 4J <
•H >,-H -O Cn to B >i»<
4J > c q u 3 fl
00 i-l i-l 00 O 13 0
10 id NO oi
O O i-l > MO
•H a oi o > c 4J 4J
*> 0 Cn* ^ ° 0 *
OOBBidid4JM4J
H«wld**ldMOQUO
1 1
00 M Id
3 0 O
4J > 3
10 10 -O
4J 0
oo B
o-o
O E —
S 0 41
0 00 0 >
c fl a 01
Ofl 0 -i
0 0
0 X B i-l
o 0 -H id
ft . C rt
3 ^~0
c o— oo
>1-H S
iH 4J < "O
0 fl 04 C
o. s^- id
i S
0 iH
S 1
00
MU>
0 C^
O rt
B
id -
0 N
4J
Ctl-H
3 *
N
00 4J
v •as
rH >O 00 O
«M ft B O
S "0 -H
•- 00
- 0 fl a 0
4J id i-i o 0
•H S (fl *«H CT>
fl 0 0 M B
S'Hfl 4J <
4J *
Cft
X rH
O
C •
0 i-l
S id
^
0
t3 fl
•. «
O en
•t *
id i
B rH C « 4J 0
0 Id B -H - 4-> 1 09
ft to OS«-^-H4JM-l
idi0 •* id Cn
B -O O E
0 •-! O ft
a o fl
x fl M a
0 0 id ft
a 0 w M
-i-i ME 0 id
4J 0 id ft — ft
SO 0 oo *) a >o
BE - 0 « id 0
•H 0 a c B 4J
C'Ooi4J-rfcnoi3
fl 01 0) fl 09 0 iH
DMflQ,OT3flQ<
I r-
0 -I
S 1
w
0 en
O •-!
C
id -
0 N
4J
Cn-H
B rH
3 «
J 4J
E
g,
0* O fD
•H-H a
id a 3 E
a « o -H
0fl 8
>w s O s
•s^ f* M
^^* B 00
0r- O
M (0 C Oi
«irt M
iu ^
E O 0 0
•H E
-------
§
CJ
0
w
CO
w
Q
e^
Cfl 4)
ri 2
w C
W »i^
X 4r>
^j C
*^ o
K CJ
8, ^.
U CO
U H
(H
OS 0
M r~\
P4 W
£ w
f UJ
J
Q 3
1-4 «;
§r^
^3
E2 o
SB
§
9**
K
D
*
rt
1
M
_q
9
P
*
01
•P
C
Oi
E
E
0
CJ
1-1
Qt Oi
•H C
CJ-H
C t>
•H C
££
a
4)
hi rH
41 A
fl id
-P iH
0 hi
£
M
o
4J
J* id
o> O
S?
M
•O
4)
V
id
a D>
4) iH
4J 4J
•H 09
CO 4)
>
M
C
0
Tt
4J
10
rH
3
Qt
8,
4)
4J
^
rt
o
4J
— —
O
I
c ai i -
4) r< 10 >1
1) -H rH hi 41
C X rH C 1> O rH rH H
•H4J«-Cldhl4Jld4>4> ••
o>B3Anidoi33:c
4) A id hi 0 hi Oit-i O
U 0) B 3 O -H Id 01 -H
C hi hi 4) Q,fl 09 41 a
4>0>ldOhiO>aQi3O3
hi O CO-POIOO-i-iiH
OlCrHOI'rlldhlBiH'OO
•w id id M hi OIMCC
M-l O M 1) -P I) C Id-rlO
•H 3 14 C TJ 4) *0 & CJ
>O>ihi
Qi-iCiHO09D>i04)3hi
*a.id.*.p4icBOB4>
4J 03 A O O O O C C 0)
J0)M8:r,X-rl4JldOA
4J09OOO
>,
4J
•rl
09
C
4)
•O
C
o o>
•H N
4-> -H
« B
i-H
3 >i
Q,4J
ff-rl
Ot U
rH
4)
3 rH
rH
iH O --rl rl 10
4J -H a 4J oi a
IT) 4J id id CO
N a 01 a> c
•HE fl - O
C 3 . 09-rl
10 B rH 0) 4J 44
fl C 4) 8-n id
hi 0 3 0 C 4->
D U n io
i-H id 10 3 1
O 09 - fl — 4J
3 IdBtfl-H
ht fl 41 O O rH
4) U fl 09 Id Id
o c o oi e -u
C O id O hi
IT) U rl •*•! -P O
U A -P 0 01 E
•v- co)
4) «*> id fl
rH VO 4J 4J
Id rH -H
8 rH C
•- o -n
.- 4) o,
4) rH 0 09
4J Id hi Id •
iH 3 4J 4) CO
fl 4) 0) ht •
S«H 8 id D
01
rH
TJ
CO
•H in
hi 0*
Ot rH
Qt
3 C
01 O 01
•H 0)
0 -P 01
•u id id
•rl 01
id o to c
*» O iH 0
Id 09 *O *rt
TJ B 4J
id 0) 3
-M CO rH
C rH 0) rH
4) id fl 0
iH DO 4J %
0 3
•H at c hi
>W O 4) -H
i
Qt tn 0) 3 4J fl 1 *»
C O "O C «J 4J -H -H
-•rlhlO-rlB hlrH fl
>iT3 4) he •HO)idid>i4J
4J-HQ4Q,grHhl>4J4J-rl
-tg9 3OO ht-HS
01 4) - o-H S >i O 01 >,
G hi in 10 M T3 4J H CT3 -P
4) *O (dCC-H 4)0)-ri
ft* ^ fl\ (*•( A jM ._j f j t. n-* 1 1 *n
w w v* "o ,u at *rt u w u +* w
o rH ai id -i oi idc
0)TH >,--rHllODrHai
rH4^COn>iQ|4-1CrHO)lO
OMiHrHhlrHMCldOrl
•rtrH QiOlOtOI-HOiHrlC
fl3>l8XQl fl 0 0
4) Q.-P 4>hi3Ol>C4io-H
> 0 C 0 O 4J fl -n > 4J
D.3 0 X 4J >i id
hi O -P hi T) ChlrHrH
O4->OrHCO)C3rH-H4JCnQ|
00>C'O-riidOid4JO-nQ
rO-riid4J3«WflidZfl Qt
C -
O >n hi n
^t« C 4) 01
4J ffl 0 -P rH
Id rH 0) -H Id fl
rH 0) 4J S 10
3 C It id -n
d-rt OlrHO ht
SO) 3 rH Id
>i-o a >
4J O -
•u e u» ax x
CSC 4J 4J
«) o -n •- IH iH
O O 4J 09 09 rH
hi id >i c -H -O O
•H lO-o _e
hi B 4)
0 4J >, C
•P • C 0 0
O O" 4) rH -H
S 09 o a,4J
o
r-
0>
rH
id
- 4J
hi id
0) t3
O
C >i
id **
o-«
rH
rH 10
10 4J
•P hi
^i
09
4)
•-iH
0) 4J
•-i-H iH
09 Id O
01 B
O 01 •
id n co
hl~v. •
01 D
rH rH
210 CO
eo>
c
0
09
4J
hi
4)0
,Q CD
0 Cft
^
a» ,c
o en
— —
A IH
— i i -a
Efl O A C
— -P N >, id
•H C i-H
>1 S 41 0 -
4J A E 0
•H-O C
rH 4) ^* • iH
Id 4J •• S N
4J id <"i 3
M ^*H *^4 *
SOI -rH O
hi >1rH-rl
hi 4J >l C
hi 0 -H rl 01
ot cj a 4> n
O CAM
C >t 4) id 8
Id rH TJ • 3
O O> 4) --rl
C 4) C B TJ
Oi 0 M 4) 3 id
C hi 0 ht C C
3 4J 8 >i 4) at
3 09 B a, -a >
-in >
>irH rH
•P 10 hi
-rl D -rl O
B O> O^-.
C fl O TJ
0) a C
tj 0) 10
rt c ^
C id -H >
OS M
•H hi
4J 4J 0) O
IT) C > 41
rH 0) 0 B
SO B
Q, rl 13 fl
0 4) C rH
Oi Qt id O
g
O hi
•rl Id 1 -
JJ 11 O U
10 rH M X
hi O TJ 0 0)
JJ 3 >i 8 O
C C fl B id
4) >i M
O rH O - -P B
C 0 -rt B *>
0 Q|4J C« C
o id g >H o)
* 843 8
hi 0 hi 13 01
•H IH hi Id C rH
< O id o id 4)
o
-4_> in
en B >iO>
C Id 4J rH
a oi -ri i
rH hi i-H O
A id m
w
•-rH O
a id
4) 8 B C
o o) id hi
0)
ca
*
A-8
-------
Si
w
U
§
U
c3
05
W
Q
CQ (U
0
ffi _i_i
S*
§5
5
u to
CJ U
o£ Q
S D
EH CO
Q
Q (d?
§ Ci
M
j O
2 o
i
1<
3
^
j 1
M
pq
§
*
03
4->
C
41
B
S
O
CJ
iH
fl a
ft V
•H C
U -H
c -a
Q« fa
09
41
IH rH
41 O
fl 10
r>
14
0
X 10
09 O
•H iH
C
rH
I
It
a o>
01 -H
4J 4->
••H 09
CO 41
M
C
o
4-1
10
rH
3
04
I
^4
0
4J
3
* * •.
4) fl t) cn
•> * ^ ^
>O tn u VH
» * * fc
0 vw fl 41 fl
Ol fl 4J >1
C>1 1 4-1 rH C B 4-> 03 4IMr-l
3rH fl Cn-r< (fl 3 O T» -rt <0 (014-1 4141
1 1 -3 Ol 4J CO -rl X .H O -H 4) rH 41 C IH 4J 1 1 O 3
IHC C -Hrl09OO4J4-l|flrl 0 O 41 ft Ifl C MJ
><04) fl O >1 S 4) 1 T3 O O 3 fl 4J fl 1-lglHErHlO
4-> O TD • O M 09 4J 00)04)0310 rHrHkl 4-> fl 3 0) O T3
•H rll04->4JiH>Cje4in4J rHOIOC IflMtPOlrJlH
rH>lOI-ngW -H rH4l«4JrHlOCO O-O4JOO4JiO-H4>flC 0 VJ ft MCUU3-H
rlCS4J0101fla, iHrH — Ij 3 -H M O 0) 0 IJ fl fl rHrH
§O09C r-l 4J 01 O41'iabrl -H U M OU4JrHfl
H OJ T3 fl •^•O S U >i C O >, C fl >i C 3 C) Q01UCT3
4J fl -H C 3 V4 4J fl T3 O 4J 01 r-l fl W CHA 3 3 41 C
U 03 4-1 P fl 01 O rtO-H C -HX T3 4-1 U 4J C VM094llflC
41 -Hg M "O -rt 4>Oooflifl>i09iO C C CO C 3 T3 % O
U~Xfl—4>0>l4 O CO rHC4J IfllfltJI «rHC'OO4J>»--»
Cb fa O 4J 01 C >i 01 fl Z 4J 01 OCC O fl -rt 55 0> 4J 4J
10 'OC Clflfl IflrH'OS— ' B Q • (TliH3lH -HE rH fl -H ft
OTSOI-rt-afl-^ft O OI O fl rl C tJ rH V" 41 >1 O rHg
C 4J BOO09 C 0> 4-1 03 O O! O iH-H 09 -r( 41 4J 03 • 03 fl 3
tn fl fl >i fl oo a>oxcooiiHX4J xcfloi d s -H ccuoo
C rH 4J ^ 03 H BtHO C MH O O O Cn4J 4-1 P14-lrHT3OOrlC
M— 1*09— —"Ifllfl M0900O4JBOOVM CQ 03 2 rl COfl4-lfl4-l4-l5O
C 4-1
0 E ifl 1
•rl O rH
4j -H c m
Q.4-1 41 OH
B Ifl* rH
3 rH Ifl
09 3 4-1 •
41 41 C ft -1 tO
> > 0 0 01 41
0 0 O fttJ g B
£ £ B-n o
ifl 10 4) C 3 4-1 01 -H
4J O 4J 4J
01 09 4-1 03 03 4J ft
10 10 0> "O >! 3 41 B
M O> 01 O^- rl 3
4) 4) fl 03 > -H m fl 03
g g en fl M ri vo DIB
ifl *0 iH^3iOCn 'rio
M CO CjC 09 > rH U O
09
co i I ig
rH rH rl 0
41 »4 — 3 iH
O > 10 1 <0 W rH
101-n 141 -410IO094I 41
fl X M (OX >irHrHt4l^4->OtQl 3
Vi 0 4) Vi 0 4-1 O T5 C C VM
4-lgfl 4-1 B 1H,)3 >,IO 4) -rl CO B C-»
C09U| COO 09*rtCflCSrH3'n> rHO'OO^I
4)09 41 Cfl>iflO>ftOfT> 41-HiH-rlin
O
COS ^ C O 13 -M O **^ CQ Q) (0 M ^^ *W Of CP O< **^
0 4J -H O C ft4J C 03 fl rH e-Hgl
CJ 00 fl 01 O 03 01 ffl fl O 4) > 1 W 3 t-H 3 r-l
CO C r^iH^gflO^ro >Ho90iin
IH 0 13 ft riO OPOMflflCniri r-i C T3 C O\
•H1HE4I -H-H IgMHrJlOlHOICO* OOCOrH
i-H rH CO O 10 O —
09 09
41 41
41 41 4-> 4-1
> > 10 10
0 0 IH rl Vl rl
40 O 01 01
10 « O >iiCf>
C 4J vo c vm
oo 09 ioiHCO ETI4JOO
g g Crim c 1-1 iri
<0 (Q 3OCft 3OCA
W tii J B rH iJgrH
O -O i oiiflifl COIOIH
0) B -i -H TJ U 01 B 4J CO 4) B 0) 0) B 4->
O4)4-tfliH-fH CJ 0) C fl 'O CJ 01 ft O 01 C
iflVMcnaKfl lO^-lscnC ifl: 4->
80 c
vo O
W rH —
09
4J
B
O
U
41
•a
o
o
S
A-9
-------
B*
Da
3
**j
^
&<
o
U
K ^r^
CO fl)
0
U B
H* ,1
M* 4^
*3 5
S o
M CO
O M
^ O
U 3
P w
J
02
sa
o
2 o
gg
i
§ .
m <
«
D
•
*"*
I
M
M
f»1
3
**
*
01
n
41
e
e
0
CJ
iH
&00
_ ?
•H B
O -H
B |Q
•H B
U -rl
a
t)
M rH
4) 43
43 <0
O 1-4
5
0
4-1
X (0
a u
•H «rl
0tf *O
B
•s
4J
a 01
4) -H
•H a
(0 4)
M
B
O
4-1
10
r-l
3
ft
£
4)
4J
a
IH
O
43
4J
•<
>H
O
•,
10 01
^
r-l 41 C
4> a > 4)
Ifl X (0 01 4.) B *rl *rl 4)
0] OJ (1) l-i (0 l-i O 4J O 3
T3 -ri U 01 (8 B 4) -H (8 m 4J O>
a jjj "O r4 (8 Ifl 4)0 4J4JOrH4)0*OB
4JojifliO2.*4>4J-H.-iav4i8-~ x
iO (0 4) 4)O05Cfi^'li4'*-i3iH 8 0P X CD 0
&O V4rO*n^O a i— 1 S
«H rH 4) 4J O> -rl 43 a
(0 0^ S *rt SO) fl)3)OOiOfl)00BlMf0
X >it-i O 4J 0i ifl O Q. oo 45 ra r- E 3 -H
03 rH O B M -rl M (fl 4) O i-l 43
•H 4-1 a 4) O 43 B 01 T3 41 • 4-1 M 4-1 - -r* a
MBB > 4-> iOBBXa3ft 4) 1) M 41
(0 0 >i 4) 4) -H IB O 0) 43 •O 01 U 4-1 O
4)OlOV4O84ia V^2Ol m m rfa uj m uj ft
I'S&o'SBIdadS^oo-JMSSSS^
4J 4)
B 43
00 4) 4J m 43 43
4J >i b X 4) 01 T3 2 4) 01
-r4 43 l-i M-l S a 43-HBOMt-i
43 30O*4*nA 43iQi— 1(043
IB T3 0 rHOrlrlO
434) a 10 1) 4J 4) 10 4) (0 >i
C 01 4J -O 4) 41 >: 4341434)1-1
C8-r«4)iOiO-na4) 10 (0 (fl
•rlMrHia 4J CO B C B 3
X4)ft-rl£XlO 3-HX-HXa
J4-isoo0ioir-inoi a as
4)i04)i-i'i-i4)OB)-H-n^'-riB
1 IM
B 4) O
•H O
B (0 43
41 4) 0) 01
O M H M
B 2 iO 3
4) (fl 43
•O iJ 41 a
i-l H 4J
a 4) r-l 4-1
S5 ^ft
B »
U -H 00
41 4)
U >fiH rH
B 4J ior«.
18 i-l 8*
U I-l r-l
10 4) 1
B ki IH vo
3O 43 ffi
8le M
« *
4) 4) 1 a (0 4J
Sr-l0i4)OUOXOOC
OiHV4B 4J 00 tS 4).
43 (0 4) ^-« -ri*— •'^43 >i
4) M ki 4) "O M a 01 4J •
4-1 O IB V S rH 4) (0 -1 4) C C
!$.5'n-3s g,5 «S38S
V4
41 •
43 "4
B iO W
•H CO
4) 4-1 (Ti
* 4)r4
•a en
•> At
U *M
» •«
43 4) 43
O
jj . 41
B O CO rH
O W 2 4) (8
•n 10 4-i rH 8 >«
3 -H 4) A "M -n 0
43 B 43 -H rH
i-l -rl V r4 O (0 01
M 03 (0 4J 4J I-l
4J <« > hi O
0) M 0 4) O 4J
•H 4> B 1-1 a u
•O O B 0 (0 4)
e o-n a n a
O (0 -H 4J 4)
•H O 4J 3 43 rH 10 H
10 C -H O » O -H
M S U ft 0 rH
fi-H 4J -H O< (8 >1
01 M 4J C -rl 4J
O -n nfl
rH i-l i-l a -^x 10 CO
4J 3 4-1 6 4) M-l "^
4-> CT V4 ftrH r-l 01 X
S^ftiS'e B^Srg
M
Q
S *
n
U 4)
4) 4>
O (0 M
B H r^
(0 ^^
O >irH
4J 1
B rH VO
3(0 9>
4J rH
O
MO
^x. rH 10
4) 4) ^ 1 -H
Ifl (8^-H B ft
8^ b -O -n rH
4) 4)
•- « > 43 0 13
4) rH 0 4-1 4J (0
4J (fl -H O -M
•H B "O 10 -rl -H
43 4) E O h Jp
S >w id 43 *-K
CO
i-l
n
a
•rl
4)
« 4:
O 01
43 >w
1
r-l 4J 43 -rl
(fl B (Oh
43 U 4) 43 (0
O -n 4J Oi
B I-l (fl i-l 2
O -H M 43 O
h B rH
43 01 B rl 4) rH rH
£ O -0 4J 8) ft
|
B
4)
U 1
M-I a
0 4)
4)
O O
U
B 4-1
•rl 0
«
B V4
iO -U 4)
•H O
•o a B
41 3 4)
35 a (Q
1 3
•H 1 0 >1
4J Q) "*^, fH ^
V< 5 O^ 3 vO
&0 3- ^0 O*
0) iH
Oi »-^
•O •*-" C • fl)
fl) fl) H C
ISI^r?
4) 4J *r \
ft Ifl U 04 rH
0) rH -rl-^VO
SSiigS
i .
a
rl 4)
4) 4->
O 10 rH
B b VO
(0 Oi
O >»rH
Oi '. i 9i
B rH *O
M 4J rH
O 4J
rH fi
-•a (0 3
4) C M-l O
rH 10 i 4)
•rl irj IQ
43 O rH B
gin o 10
B ^^
•«H ^O
41 ON
4-1 rH
a
4) •
X (0
B
•H 4J
St 4>
rH
<
•
ft
a
4J
B
4)
a
a
8
o
4J
4)
1
4)
S
*
A-10
-------
U
<
"
§
w
co
U
M
Q
g£
CO 4)
•o
0 3
*-* rH
£§
S8
U CO
O M
ec Q
U — t
S E
K r*
E^ 03
o
,J
Q a
i-4 ^t
20
< M
05 O
= y
oi U
(
Km <
I—I
1
w
CD
9
p
*
to
4-1
c
Q)
B
e
0
U
1-1
ft en
-H C
O -H
C T3
•H C
as
09
4)
VJ rH
0) A
A 10
8U
$
U
o
x a
01 U
•rl -rl
« 13
c
M
•a
01
4-1
a)
o) en
0 -n
4J U
•n 09
CO 4)
c
HI
c
0
•rt
4J
a
rH
ft
0
A
41
4-1
—
M
o
4J
g
<
•t
4) ft
o en
,Q "H
1 >H
•-H 4J 4) -rt
tt C .Q 10 rl
•H At -H
,C O 09 ,C It)
U -H 4) £n
C M-l 4J -H S
O-H 10 ,C 0
rl C M rH
iC Cn H-*
•H c 0 'O
<0 OJ-rt c
C 09 10
10 0 4) 10
C O D C C
en c u o o
C » 4) 10 -H -H
3 V4 l-l 4J 4J
H 4) 4) C 3 3
O rH i-H
14 C "*4 41 rH rH
£<0-H 3 OO
O TJ 4-1 ft ft
fc
09 4)
3 - S 1
4J rH 0 •»•> ,C
10 4) O C 4J r-l
4J C > C 41 -HO
09 10 4) -rt 1 4) 01 -^
•H4)r^^09ft4)Ca
a H 18 -H M 09IQO
O CSiO>3AO
ceoio4>eno4Ji4
o o -H ti x c x: ~^
O 4J -H kl 09
4>ioi0cc>-iiiH
4-1 1
CD>H Ot
C rH •*
3iO
4J i-4
.^
41
4J
*r4 *•
A 4)
5 rt -
C (0 4)
0 S -1
C 4) >H
"V,«W -H
4)^ > •
4J 4) .e c
•i-l iH 01 C
£ 10 10 41
2 a a E-
t^
VO
o\
cniH
14
4) •
X) •-<
•O 10
*H
«) 4J
N 01
rt
•i
ft
09
4J
c
4)
o
U
0
4J
41
•8
U
4)
41
CO
*
A-ll
-------
06
W
O
g
b
O
CO
W
M
Q
g
co
CJ
i
rtl CO
rn M
O Q
CO
jfj S
E^ O
O SC
55
08
MI-RURAL AK
B.
§
D
'
*"*
t
M
CO
3
£*
*
(0
E
41
B
I
^
S. 01
•H E
O -H
E -0
•H E
hi -H
Q. Ctt
09
0
VI >-l
4) A
,a id
4) -H
O M
>
Risk
Indicator
•o
4)
id
a 01
41 -H
CO 4)
E
M
0
•H
4J
id
3
Cb
4)
id
s
P
o
H
-l 4) CO 4) 41 0 4-> -
ViViOldOi4>4>4>0'-ICO
O E 4) -H Vi 3 Oi,E i-l 4)
'W'HflVi.EflOEP.QiE
O O id B < 0, 01
09 00 P* 4) -H
r-l C Ol C 41 Id 00 VI Vl ,E
o o E id a M • q 4) -H
B i« >H B vi r* a
O 3 0 Qi • -H 0 1-1 £t> 09
SlfllIs4>4)iHVi4>
41 < N-H V| -H -H iH O-H
•H 09 E B E -HE
0 Q 3 id >i C 3 -D -H 3
p$ ^^ O fa CQ W u lOU U
VI
4)
O
id
o
Ol
E
3
*E B E
09 Vl O Id
•- V4 O -H O S
09 4) M-I «-H 0
4) ,Q -H -H Vl -H
r-t B "H > 4) CO
Z B CJ Q < M
41 vO
•H iH
r^
1-3
E
10
(Q O
*J 0
Vl 4>
4) 4)
iia
o vi
it.
4) id
3 O 10
O -H 0!
S -P VI
a -H 3
0 UX>
0 01 G
Vl 4) 4)
41 4) Ot|C
Vi X Vl *»
0 O 10 0
09
Vi B »
CJ rH »H
C ffl Oi 0
SS5£
OHO O
C O 0 4->
3 B 4) W
a
*j
•H
,0
10
,B
Oi
B 4)
i* n
08
J B
tfl-H
Residence
(cities,
smaller
towns ,
M
41
a
B
fl
O
Oi
E
3
•>i
4J r-l Vl
Vl 4) O
-S-0!?;
c vi -P
rt 0 *« 10
O-H U
4J ifl 4) T3
E -H 4J E 4?
0 O * 10 O
CJ 0 Vi B
09 Vl 4)
•P 10 E 09
M-j 0 4) Id 4)
< C T3 O M
09
4J
,Q
10
,B
E
2
0
B
CO
09
E - -^»
O 09 09 a
Residence
(conurbati
large town
small town
rural area
Vl
41
U
E
id
o
Ol
B
3
0 E
t -H
f fl
^* -H -H i-l
«*) Vl I*"
09 (0 ff>
•» Vl iH
en 0 -P I
« 0 2m
S "O O •-(
VO
•Of«.
C ^^
id -i
i-l O iH
0 4) O*
Q Q. -l
4)
•O
-
O
id
B
4) O
4) - -P
3 X E 4)
4J E -H Vl
4) T3 >i 41
X3 E Vifl E 3 rH
id id -p 41 o< fl
B rl-HX E W B Vl
OOl S id 3 4) -H 3
•H B • -P i-4 *J Vl
flXB •» «Vi4)E
•H O 3 03 4-> C ,E-H
08^3-HO*O»
0 00 OlfQ -H 4J -H E
09 00 0 Oi 4) fl
EVi4)4>-H-H4)3t)o9
0 fl O A£ 00 U 4) -H fl
ViOiB'O O C C S 4J 41
cooofltouotaufl
B 01
T; **
Jf^
fl
1s
1-1
fl
4) O Vi Vl
IH o MJ « O o 3
o c o + t-io vi
4) O O - A
4) T3 >iO O vi -
O -H 4JO >,m g E
fl o9-H..ni,a 3
v^
41
O
B
10
O
Ol
B
3
•«
09
4) B B
1 % • a
n
>O VI
B O
SSs
X * -H
fH
1
<
•
Oi
*
to
E
41
3
0
O
O
**
4)
•8
o
41
?
A-12
-------
«
a
o
o
Cn
o
CO
u
C6 W
B
c
4)
I
ft
a
01
•H B
O -H
E -a
•rt E
Risk
ndicator
•o
id
a
41 •
•H n
CO 4)
g
ft
£
4)
4J
5
1-1
0
4J
'O
ID
a) 10
4J B
C OI-H
4) 3 id C
4-1
4) T3
fl B C id
10 -H H
B «S^ 4)
0 01 0 "O
4J -H n n
ax B
O B 4J O
O ffl -H
a so
n 4) 4J
id -u B
I B
O O - I
• O M 4) -H 01 tj
01 01 4J B 4-1 4) 4)
id a o) ft 3 E
•O
id
E
n o
o
«i
w a
§
4)
1 o n
01
4)
M
10
B 4J
id 3
01
n
10
u
4)
O
O
10 (0
UM
S
Kn
v
CO
c o
10
O id O 4)
•HQ4JS4J O ^ _
4) O O « U ft 3
N 0) E 3 n -H 4) «< W
•^.H 4> T3 O O U -rt
4J O B n 10 iO •
P. U 4J -rt 01 O 00 4J
4J--O 4J Id 13 VO Id
54) 41 4) O C 10 r-1 4J
l-i <0 tt (71 Z 3
)t3 (d •** O Oi r-liDC'O
4) ft ft 3 >O -H 4)
id * H 41 E &> •-* 9> 3
4) 4-1
l
nSWiw u 4J a>
0 10 O
^ T3 X
O •!-> «
eH
OMB
M'aSid
O 4->a
ij o M-
a> a> « 3
>-iu
4JB>iH
4J il) 4) D
,
S,rH >;
flJ3'Oai>4vjS3<'0 4J O O "^
*J C 4) 4) n tn « SAaasno
u xto so it 4) o 41 n v
0 JS 0 4) C Mr4iO*>4i0oiM -U
a -U H -P -H -U CU^. «n 4) DI-H j: (0 4)
•* 0) 10 3 O> C M 0) 3 « 4J ^! 3
> HJ
tjl 4)-rt
us
01
4) H
c4>-u'dio>cioo a
O * 41 ft O (B M
cnwi04)ifl4)-^ax>
ataBa
V-H >i 0)4)-H
fl O 4J
0) 4) 4)
fl n
o
4J
4)
« O «
e 4J 01
.S*,!1
4-1 M 3
id 3 . .
X -H 3
Ofl O .
•l^^3
O 4J
Di B)
vo
o<
•
«
4)
01 O O
4)-OMfflO
4)
u
B
id
o
O>
- i-l B 4J
01 10 t 01 3 10
4> 801 4J O 4-1
O 4) oo B o n
id «w - 4>
M \CO T3 •"< IT)
4) f» iH J 4i
O 4J O 4-1 M
O -H O 4) id
4) M
BM - U 4J 3 4-1 0
4)rHiH4)01>,OCiOO'^
•HOiiou4Ja>iawK
A>4JlOB^>lDrHO'O'<
SO>Ol-l4)lOCOkJ>1<1<
4)
O
id
u
•o x
i So
B-O uoo
B id en
id O "H
S
i
~ 4> f O 4) 4J
oii-inr> o -4J 01
4)104JI'OC^*-<-2SH
•o
Id ~H
0 S>n
w idOi
to O -H
(X
B
4)
0
o
0
4J
4)
?
o
4)
4)
A-13
-------
s
8
z
3
o
CO
w
£?
M
CO
CJ
M CO
KM
>*4
0* rH
•* O
2 D
§e*
co
w
£%
»* y
s z
to
0 CJ
Q W
z co
?N
*J
M
s
g°'
05
1
<
§
O
1
^*
1
^
^"*
s
1-4
CO
*
*
V)
JJ
c
4)
g
o
u
a to
a cn
•rt C
O-rt
c-o
-4 C
k( >rt
CU b
to
41
rH
XJ
U 10
41 -rt
J2 U
JJ 10
0 >
41
H
JO
a
J^ -rt
tO kl
•rt 10
K >
•O
o
JJ
10
en
jj.
to to
41 4)
JJ >
•rt C
CO M
C
0
rH JJ
O 10
JJ 3
c a
0 0
CJ fti
c
0
•rt
JJ
10
•O 3
3 a
JJ O
cn a.
N^
^4
O
JJ JJ
3 10
< Q
UH
^
41
O
^
10
rH cn
• CO 10 C
to cnrH M O
u JJ C 41 4) M >O
«l O-rt>>jJC 41
O OC*rt J£ 0) 4) C CQ 0) J>
C JJ-rtU OrHIfl-rt V U
10 JJ JJ E 3 Cn O>
ocooooitflrHtaoieta . jj c co
10 . CO -rt rH JJ 0 41 CO 01 3 jQ
•D ONWH rHOTJIOU-lia— I.HCj3rHO
C C rH J3 O -rl JJ O U
ITS 3 41 E JJ rH U >O 10 C 0> C *O O
rH 05 0 lOOCOOINOIJJOCCO
4) » C UH 41 3 -rt u C CO <0 JJ 10
rH CO -n gUJJUCUIO-HaiOCn
no) "o-rt c o JJ J3 cn -rt c u
ErHt04>JJ>i4>CnCkiC4IO~ltl
ugjJcins-Hjj o 41 tn o c
b£22injj44>5<J
1 CO
•rt 0)
MH 10 *O Q* 4)
C U 0 C U -D 01 4) 41 CO C
O41 -rtlO •*0>0kl>rtO)rH
CO JJ ID 1> CUT3 10 10 "O VJ rH
^4 tO id X ^t 3 *rt CO C *rt
4l-rtJJ4IIOtQ UHOO
O4Cn(OtOt^lt44)4>O>CC^i
01 41 cn u £ 3 u
JJft«O4)4} JJO) (OO
x e e o> o cfo jj
01 C iu 10 n cn£ C >, C u-i m
Z-rtOMtO«».rtJJIOOV4
C 0)
.» 10 en Cn
41 rH a M e -
»rH 4) O 3 10
tOIOMCQ OrHU-rtfg
41 E M JJ CO jc O JJ vo
10 UJ •4)10 O 1-1 JC rH
l-i \ Z "D 11 U 41 O 1
01 -< >, o> TJ O C O
rH rH rH CO tJOICOVO
rHIOi-IOIinrH-rtlOklOt
C LJ | r^ 01 *rt en *rt
rH3 10 in . O Jji C T3 ) M 01 O O 01 1 41
Cn C co TJ ••rti-ucg-rtkirvi JJ
CIO CODCCOCOrHjrrHE^lOUOin • JJ
•rt - j* C O iO 10 EH <0 (0 C en — 4)
JJCntOIO*rtJJO)4)fc tfl^>rtrHC4V4
cocra-Hfljcouuc c . 41 ^r- 10
3-rtiOUgH<0 J3OiOCO)Oiki CO w -rt ro
COrH-rtuJJJJU-rtEVjO U IJ O JJ «3 M
w (0 JJ 4) J£3 OucnaicVcn O> 4)
JJ aT O rH C JJ 0 rH O JJ C 0 JJ 4) rH
<«828Jaw°cM'-5oo8ii8o2£
• en - I
-DC l-i 10
4) -rt I (0 J3 •> «
j<: JJ (0 Cn C to 0>J2 4)
O ki Cn-H •rt'OUH COOtVOtOOX:
E 10 -rt O CO 10 JJ X >i-oJJ JJ
(0 JJ O UJ nj »rH JJJJ
CO uO C^O« >i4IC
en JJ uj o •• 0 Cn kiC'OJJkiO
C C JJ O 0)C-rtCH30Ct03
•rtsia 41 D O JJ -H (o cn us 3 cn co
jco aijj>-i"<'rarnEOJJ u
•OJSCJrf-rtWMio -rt -H 3 r^ 9 ij s a
-H JJ 0 O JJ OltiOf-rt rHOIWJJ
tO JJOOOI'UCki.QjJ tJ'Or-ICO'OC
a-SSwmfSSsS-SoESa^ee
4)
0 >i
C JJ
10-rt
10
CnHJ
C u
3 O
1-3 e
cn
C i-H
•rt 10
•D rH U C CO
to 4)u>,D
-4) JJ E H JJ rH >,
rONrHkiOO CO 0) JJ
in TJ E rH C 0 O> 4) 3
-C4)in4)(Oi-iOirHO
CNiou-imcouu-iijOCJ
•O
C
<0 (0
4> Ui
WrHOO CnCtJ CM
oiflj JC'oc'rtc r**
rH E in 3 4) 3 10 «en
10 11 n -rt rH 1-1 rH >, rH
E UJ U TJ 01 01 JJ 1
•O 41 E O > C ro
VCODT3'DOCOI3U3
I-H in cn>-H 10 LI (0 rH O ON
VOrHlOO.CUHOUCJrH
r-
JJ r-
Ol ON
r-l
C 00
10 .|~
4) rH ON
O 19 rH
c
10
£ 1
JJ 10
Q>
CO O
•rt
kl kl
01
kl JJ
01 rH
3 -H
O UH
rH C
o
S w 2
4)
1 04 -O
O O » -rt
>,VO CO 4> tO X
•OCrt * rH 4) 0
3 rH UJ ».fl rH
JJ O 0) 3 O O
10 0 J< H-rt -rt
<0 (N O 0 JJ kl
uair>g(0ki>oki
JZkiONtOCiO CO)
JJIOrHC -rtfllOUH
A-14
-------
U
5
CJ
&4
0
CO
W
Q_^
^^
P t3
n A)
C' (U
rn *j
o-S
Ht *J
a c
cu o
3
rn ^^
X .-
U W
Q
fti y
U P
33 CO
0
§£
*•* A
3 i
3 H
2 3
4°
m •
p*j pj
^
*
^
1
I
*
n
4J
C
01
B
B
O
U
r-l
10 09
Q, en
•H C
U -H
C -O
•H B
Ul -H
& b
CO
4)
r-l
l-l 10
4) fH
4)
rH
10
CO rl
33
T3
4)
4J
01
•H
CO CO
4) 4)
S£
c
o
r-l 4-1
O ifl
rl r-l
•U 3
C Q.
82
c
0
'w
(0
5-..-H
•0 3
3 ft
"Jj-
0
£ 4)
4-1 4-1
33
*
•o
0 ,C
flj M-l
C M-l
Qi rtj 0
H -H
10 ri 4JioccoiC'dC4Jco)4J
Cn E (0 rl Wi 13 4) ,3 C -H 4) lOS •
C04)3iO>H4)iO£ Uri4)*HOiCQU
Ifl O U rl rl4)3 X 104)
rHCHHSV 1H O'O'O-rlOrt <«
UiaOrlCMH-O'V-l CBUCW >i a OI-H a 3 0 4)
Cn O O 4J M CO 10 C 0 B
•rl C 4-1 -rt O 0) 4J 4-1 M CO CO)4)O
X3iOriai-n4Jg90iOiO 10 0) 4J
OrH«0)C4JCO-nl-l 01 • 4J O >1
B UOI03.C r-i v< 01 4) 10 0) 4-1
CO C CUVi-n ViOUUUri ft-H
o< > 4-1
OC*rlOl2iCCr^O jC'HT-tU^J'rl
4J*rll03r?*rl|C3iQC*r14) CUlO.C'O
tOT34JrH0133COi
C rl
•H 0
X 4J
o a
B fH
Wr=
o
•a 4)
ecu
10 O C
•H 4)
0) 4J T3
O 10 -rl
10 V< CO
rH 3 4)
04 T3 kJ
4)
U
C
10
U
0>
I
4)
4) rH
"» > ft
Q) CO *r4 3
4-1 4J 10
ff 4) prl 4J O tl
$ C" C rHO rH
10 rl 0) 10 O 0
VO 4) CO C -,C
rH CO > 4) O <" 4)
lf> 4) 0 rl 1H CN 09
>rH ft4J 3
rH (0 M 4) 10 "« O
(n B 0 -i C 0*
4) •
.. MJ rH M ~
co o 10 • a
4) B l-i DrH
4) ca x ff etft
4J O>4J i-l rH
•rl ((0 rH C 10 -1 -
H 2 15 rH •§ -O ^
rH
4)
N •
CO rH
C (0 (M
4) VO
lO -fci O^
3: 01 rH
4)
»
T?
«.
O
*>
(0 i H4)-H4)-rt C-n
CO 10 "O Cn>W 4J |H CO 4) rH iH CO
-4J 0 ifl O 01 4J (0 4)
4) O) r-l (0 CDC -H rH 3 4J
r-l-rllflCOXB-HrH fflO)C-H4JCnC-rl4J
SC04)tMC4)rllOCOl03lOrHlO
(QrHOigAiOOriO3^3ki4)3
ft
O! 13 "H -H B
4-1 Ol 4-1 10
.C 10 4) 4J O 'O
» > C rHO rH
co o oi 10 o o
C" 4) CO C * ,C
«*> rH U 4) O "1 4)
<*> 10 O U -n 1*1 CO
- S ft 4-1 3
•3- oi m 01 ifl tJ O
a>
ifl ifl U D *O <*>
4)
•s
U
41
0)
to
A-15
-------
£
w
2
**•
tj
o
M
Q *"*
3 m
CO 3
c
O.H
M "ti
£ 5
s °
§03
H
U M
Q
p
oi
§gj
g
fA
_ fl
^2 *•*
2w
9 S
1 C >
I w
CQ •
OS 0
D
rH
f
"
W
1
*
to
1 1
c
01
s
o
u
rH
& a
01
•rt C
O -rt
C -0
•rt C
(0
0
rH
A
0 -rt
8£
0
rH
a
(8
u ,-IJ
(X "ft
to ki
a*
•a
0
4J
<8
Oi
•rt
a a
0 0
4J >
ss
B
O
rH 4->
0 18
4J 3
B a
82
B
0
•rt
4J
18
>l-l
•O 3
3 a
(0 &
• —
rl
o
4-1 4J
s s
UJ
V
•0
o
iO
0) O"* O^
>i M C C
H 4J 1 Oi 10 0 0
0 -ri a , Q, - 41
•rt 18 18 ,C rH rH OX
•rt 09l8Ui4J04?-rt0B
BD100 -rt o CH 10 X 01
OI C 4J r) rH C *rt 0 B
WrHU^flSS 0-Si B
en ^i
5S
X 4J
o 01
. ^»
a 01
1 0 B
•O 4J - 4J 1 -O 0
00Oi0iaMOi00C A
0 -rt >0 01 rH •*-'• BO ^4^0
00-ri0uioo
flOliOOOirtOlriflBO'O
S2o«Sow'Snai!a!o*
rl
0
0
B
18
o
Oi
B
3
0 a
rH 0 1 O
OirH 0 01 2
S 18 S -rl
B T3 rH 1 O 4J
18 >M B 18 >n in -rt
K 0 iO S" *« O
1
•• B B "O -
0 18 -rt C 18
rH O 18 >l
-18 a -H in
0 B Oi 0 -rt 8 VO
to 41 c ra x o en
0 -) B 18 O < Z rH
3
01
O CD
4J VO
KrH
M-J
»
iO
• oi C
IH 0 O T3
B 0 O -rt 0
0 O G 4J to
0 B 0 n) n)
* 0 18 ki B 0 H
4-> O O 0 10 IH 0
0 B W O.E B
•O B -rt X -rt 18
B -rt 3 -o 0 O
O 0 rH -O
•rt 0 rH 0 0 OI
4J rl <*H
Bnaoo • TS (fl o
0>rHO DO* 0
•rt rH 0 rH O BO
•M 1 O 18 * 01 O B
U|4-> 1 rH O 0 3 -rl 4J TJ
3*rt0 *rt>rt01^Cl8*rt
O,B3'O4JrlBOrl01
8SrHCiOO4)B31i
— fl io Q. OiCJ n T3 ri
^^
IH 01
18 B
rH 05
0 18 01 rH A
O rl Q O O rl
C 4J J ffl 3 i-l a)
0 B 0 >i C 4J O
>O0*rH4J>iiaO
n 3 013 go -O
u
0
o
B
18
o
Oi
B
3
CQ O
*O *O >t Di-P M
«) i c f: c arn ^-t c c «w
4J C **^ 4^ O 3 0^ 29*U **^
M_f Ofl) 30BOOOO 0)«H-H*^OW
0 ojiojocooajaju'-'r^c 4J(n a>
*(oo»CiH 4Jw»-iB)Cn f** "Oidf^u*-)
o)xc 41 't)f~iOOO
Oi S
i c -u
^x CS .H C
tt) »n i r»> «o 3
«—( tfi CTI 0* 1 0^ **"* iH 0
flj C^ (QClO^CDtd QOO
S *H 3 CM"*-*
0) iH >1*O CD 4^ *-^ CD rC »H
•HgSMCOH^ 3CP
<-! 0 0 O 18 *rt oi BOB
S'MCfiOj-lrH-rtOl^
B
rH
18
0
0 0^
"" m
S! rH
rH
^
•
Q<
*
a
4J
B
0
i
0
o
o
0
o
0
0
*
A-16
-------
*•
OS
H
u
*t
*c
^-*
IV.
0
03
H
w
Q--N
p-0
P* (11
C' Q)
w -o
a
CJrH
w o
S C
Q* 0
a"
05 ^~
o co
w w
O M
U E-i
O r—J
Q§
§ £
2 £
S
j R
M \J
2 w
o w
«-*»
^
i rt
!
o •
« 0
D
*
1^
I
I***
w
.3
*— »
i
*
0)
4J
£
4)
S
O
U
rH
&n
o>
••H C
O -H
Cirt
^
•H G
to -ri
0. h
n
4)
rH
XI
1H Ifl
4) -rl
A "H
&$
4)
rH
A
n)
it -H
n M
i-l 0)
»>
•O
4>
.p
ifl
O>
-ri
4J
o a
4) 4)
4J >
•H C
W M
C
0
•H
rH 4J
0 ifl
to rH
4J 3
It
C
0
•H
4->
ifl
>irH
"3 3
3 (X
££
^x.
)H
o
p. 4)
+3 w
ss
M-l
•.
T3
V
U
.,
«
1
u c
C -H -i-l 1
41 >n t7> 0
41 M-l U 09
*. «s a m
•u M B ifl a
4) 4-> 4) -
fl .. 4J rH !•>
•a 09 c io &
C C 4) ifl S •
0 10 rH U 4) rH
•H 18 -H M-l
4J U B M-l O
10 4) t-i C -n
•H O C C -H 4J
O C 1-1 O> 10
O ifl -H C to-
ut V Q V O O O
0 3 0 rH -H O "I
g rH > «J O~- 0.
1 MJ
0 0
•H 4)
U U -U O
0-H . 0 C
n S 09 -H 4i
O 3 U-O
. C 4-> -P -n
4) 0 ifl n o
B> O -P -n 41
<
C*l 4)
0 O 0
4) O CF~
01 iH) 4) 4J 4) f«
IB >|MJ g 4) -O 0>
U rH MJ 3 4) -H rH
4) in ifl rH to ca
> 18 M O -W 4) C
< "0 4-» > 00 H -H
Ll
4)
0
c
10
O
CP
c
3
B 1
» 0 r«
rH X >H 18 to I—
O 4) M-i B 4) on
U 09 4) MJ O O rH
•P >iT) U O C C C
C4)rH4)C (0 (8 -rl 4)
SB B 4J 4) >|rH O rH
1000*0414) W 4->
03T34)-H>aCD*JCO
"* CrHOVjC-riniflt^
VOMHl04)C3O^n4)>7i
VOOrlOO-rlOiCaUUrH
•*
- >i» r*
4) rH 4) i4 O>
rH B C 4) rH
18 0 O
B T3 TJ C C -
11 C 41 « 1-1 41
MH 10 V O rH
-^-, H O 09 4->
4) 4) C71 4) -P *>•
rH •* rH C O lO !*•
ifl't 4) 3 ifl 4) W
JE >O 09 rH O W rH
to
4)
Li C
ifl to
09 10
a 9
•H rH
.H -a oo
p c <*
& Ifl rH
Cn
*.
M-l
•k
lO
a
(A 4) C
o> 4) c i o -a
lH C 4) ^O -H-H4I
Ol 41 -H ^ .prH.tt'O-Pm
OC C M 4J -X 00 ii 4J Ql009C3ca
•r«4) iH O M-l C O C 05 OOl -Hi-lrH-rl
4J 4) MlillOOBO-rl^V^lCrl rH S
as o IH oj -n LI CTI ,e o >ioa
•H4J B C-P -P 3«J •Hil'O Ql-rl LI
CP 4) Li 03 •U3T3i8-PODi4)4JC3 "00
rlA-rl COrHCCC£C^3l8-PLl -P
4) IQ'OOArHlO-Hia^-HVrHOOl-HOJO
Cft C-H O flOrjlO i-l -H IB il 18
>i-H iaiQ4JoriOi«a-rii4icjJOMH4> MH
m ff C Ifl M COMH ~H In O^-H A 4J 4J
09iOCOiiOriOU-r< M-l Q C-PiOOX
•PC O33-Hi-i C* OiriCK rCCM
C O D>-H O-O l04JrHO>09rH Qj-H i-l •• U -rt
4) i-l C 4J O -H lflrH-HLlrH3iaaV4 13 Ll
LI 4J -H 3 0 > - Ql IB 00 O fl 0 O OJ rH
Iff .•* V _J . ^^ M it >. i i _f i 1 jit M if*
ly "O i-* ^ ^^ ^» *• *^ H ** IW *J V »J TO
Do.c C CTI-H
O C 1-1 -H LI
a o 4= o
Ifl -H C O>
4) — 4J 0 M 4)
4J Ifl -rt m 4J
•P DI IH 4J 18
4) C 3 B MH O
Li IH TJ O, O
ifl X 3 Jd
tn o n o 4> oo
•n a C O C -n
u a io o o LI
09
4)
Ifl
w
4)ABC4) la MH 3
OO>30iifl4J| O 0
C-n-n-H4)A4)CarH B 03
•4) £ 73 4J TJ B 4) -H IB S 4)
•O 4)3 -O B ^ O 41 rn 4J
-rtlOBrHlB4)O4) i-l-rtflC
09 rH4)CVOLl'OLl>O-H
4)COOLi-naK3CO4)LiO
K-rl^OllBSrHaifl^LlQlQi
IH
4)
0 >i
C -P
10 -H
O rH
10
anj
G to
3g
0 >l
4) -P
rH c in
IB 3 *t\0
B o LI o>
O O -H
4) >H I
*j 4) r-.
•ri 1-1 3 in
S^S-.
•M
00
o»
rH
10
c
i?
r4
<
a
h
a
4J
C
V
B
S
0
u
o
^
4)
•a
0
u
4)
4)
to
«
A-17
-------
-------
X
H
y
T\
fr
&
<
P3
M
njj
W
E
£_t
^
M
CO
u
£5
CO
CQ
D
CO
u
H
23
o
o
2
•H U
M a
IH U
2j
W 3
.J O
§ 55-
§« *
P^
O
Q
W
E-"
U
W
CM
CO
^
CO
fe
o
CO
S3
o
M
C-i
1
U
a
o
u
tn
C
•H
4.1 £J
flj O
U -H
•H 4->
^3 (0
C M
M 4J
C
0)
u
o
0)
rH
(1) ft
u e
M (d
D CO
o
CQ «w
Q
cn
•H
,C
(D >i
•H 4->
rH •*-(
,0 o
Id 'H
4J C
0) 0)
W Cn
0
m c
0> -H
CJ O
C M
0) «)
M U
f^*
E
CT>
3.
in
(T)
1
rH
•
o
rH
id to
•H C
M 0
4-» -H
co m
3 (Q
TJ -r-l
c e
«~^
n
4-1
id
M
cn
J^
^^
rH
0
rH
u
M
«
f-l
1— 1
555
«
^
a •
M rH
fd nJ
4-» 4-1
t -H
0) J^
rH O 0
C 03 -H in
rd vO 3 vo
4J CT> (0 CT\
CO H CO ,H
CO
s ^
•^ e
C" "^«
C
f*H ViO
i
o o
(0 M
0) -H
rH rt)
U
"5 I
id M
04 £3
.— •»
0)
o
•H
c£
«.
(0
id
M
00
vo
OA
f— 1
6
•H
1
u
J
u
U CQ 1
O Q r— I 0)
Q*J 55 o C
W D ^-H
EH O 0 "O
W CU N -H
ffi S CM
1 O 0) 0
•< U CQ rtJ
N
<
• •
• rH rH •
i — i fO id ' — i
(0 id
4J 4-1
4-1 d) d) 4-1
0) 0)
•H -H
ii^ t-^ _^ ^
•H u ^a o o -H
ft o -H in -H m fQ o
3r~ Svo 3vo Dt^
tOrH COrH COrH W iH
00
ro 6 f)
B -*. e
00 "01?^
•^ c c
Cn 00
C O ^D CN
0
rH CN O O
M MM
•H -H -H
id O
id 'O ^l*
COS
4J 0) X
Cn 0 B
rrt avj »wj
"O an "^
S^g
i
"
^
nj
*— c
N -H
C 'O
0) -H
o M
•H O
•
iH
(0
-P
CD
-H
^S^
0 ft
•H in
3 vo
CO rH
•£
0)
(0
Q)
M
CU
0> 4-1
rH 0]
O 3
•H id
,3! ^
> 0)
B-l
-------
-o
0)
3
G
•H
•P
C
0
c;
r-(
1
M
M
M
u
. T
1—4
S
EH
(X
M
<
W
S3
M
CO
w
CJ
EH
CO
ffl
D
CO
CJ
M
ll_
§
I
M
U
S
CJ
g
§
^
«
o
Q
H
EH
U
W
&
CO
D
CO
fa
0
CO
53
O
H-4
lENTRAT
u
£3
O
CJ
Cn
C
•H
4-1 C
id o
0 -H
•H 4->
•a (d
G H
M 4-1
c
0) 0)
0 0
G c
0) O
M U
0)
»H
/It
£
c
o
•H
4J
id
t-i
4->
C
0)
o
a
o
CJ
0)
rH
a> ft
0 g
M (d
3 CO
c? «w
0
en
c
•rl
to >,
•rH 4J
iH -H
ft 0
id -H
4-1 C
CO 0)
W 01
O
co c
0) -H
O O
C H
a) id
* U
0)
«H
#
0$
0)
o
c
(d
4->
CO
X!
3
CO
• •
rH rH
id id
4-1 4->
0) Q)
•H -H
-* ,*
00 0 XJ
•H m -H m
S vo 5 vo
£3 52
CO
S
\
Ol 4-1
c c
•* co
O CO
CD £
V4
"rt
(d cu 4-*
rH CO
C 03
id -H id
Xl & £
£ £ S
P ?> CD
co
VO
• O
r-l r-t *t
id I
C V4 rH
id 4-i e
S O1 •- -H
•O *-» JS
•a id co to
c ffl o
CO Id -H TJ
g - s a
53 M — »«— id —
D a> a) a>
O «C O O 4J O
O< C -H ^ C -H
S >i Sen o B
O 3 »-- H > >—
U
CJ
hH
J i
CJ— - — '
>" "O -o
U CD »•
Q 3 id o>
« C — G
W -H N -H
EH 4-1 C T3
WC 0) -H
SO XJ H
i3 SS
N
<
9
rH
id
4J
VO CD O
r"~ co
lO>
w srs. m
< Id CTi >
S3 CO r-l t>i
CO
00 e
_H <
? % 3-
C ft
O O •<*
^ |
1 1 0
in '-»
>— (d 4J 0)
« 0 G
XJ 4J id
^ co •- e
r-> '—3
en o> co .c
rH 4J G«—
C id
CJ id 6CD
Oi HH 3 r»
< C ,C
M M— rH
U
53
U
tS3
53
W
CO
- VD
vo r^-
r- cr>
O^ i— |
„_ i
f^
•
• -TuJ
V |— 1
rH ffl
ffl
IU
4J
4-1 0)
0)
irt fn
IU *J '
4-1 C
S3
rH
• CO
iH CO
O >H
CU 0)
4-> -^
id to co
•H B 4-t
IH id id
0) H LJ
oi.>_i
1
S
EH
U
EH cq
Q
53"
O OH
CO Q
Bj t4
< S3
U U
o
CO
rH
CO
g
co
€
Oi
CO
CO
rH
id
•H
J-l
1 I
Indust
B-2
-------
^^
T3
cu
D
C
•r4
4J
C
0
u
rH
1
M
M
M
W
r-3
OQ
<
EH
ft
M
<
M
ffi
EH
2i
M
CO
W
y
EH
CO
ffl
D
CO
u
M
a;
M
CJ
§
HH
u
s
u
2J
5
o
2!
«
e<
0
Q
w
EH
U
w
CU
CO
D
CO
&4
O
CO
2.
O
IH
EH
•a id
C to
M 4->
c
cu a)
O CJ
c c
cu 0
>H U
cu
c
cu
o
c
0
u
cu
iH
cu ft
U 6
j-i id
D CO
CO n-i
O
cn
•H
rC
CO >i
•H 4J
•H -H
XI 0
id -H
•P C
CO CU
W tn
O
co C
J rH
XI
U
S
in
rH
1
rH
O
•
0
to
•H
id
C
id
Xi
to
D
in
<*
en vo
rH r-*
o\
to H
0) *-»
C M 0)
C U -P
cu 23 id
M X-l
XJ -
C— •« -
cu cu cu
rC O U
t3 -H-H
CO 6 H
W^v-'
jg
S
o
§
«
o
nJ
rn
u
o
CO
CT>
iH
<
CM
M
CO
D
0
in
I
rH CO
0 6
0 \
• cn
o a
H
id co
•H C
to 0
•P -H
CO CO
3 CO
•O -H
c s
M 5)
*— -
0)
o
•H
e
>«^
VO
r«
cy»
rH
U
«
«:
M
2;
M
/y*
W
Q
£
£
o
a;
s
1C
u
M
CU
u
*.
in
f-
CTi
rH
<
CU CT>
W f»
CO 0\
D rH
CO
CO
•
0
1
cr> co
vo g
o -^>
• cn
O 3.
rH
id co
•H C
to 0
4J -H
CO CO
3 CO
T3 -H
c e
M (U
•
rH
id ^
CO
0) 4-J 4J
u cu id
•H to
e c
*-* O -
co cu
CO O -H
f- S
O> •+. v.-'
^H ^^^
CO CO
M 4J r»
y id o>
53 M rH
W
Q
n
s
§
|i|
00
M
Q
W
53
U
J
EC
W
O
CO
(Ti
H
<
CU
M
CO
D
o
rH
1
m
O CO
0 g
0 ^,
^ ^
O 3
H
id co
•H c
M O
4-> -H
CO CO
3 CO
"O -H
C B
M a)
X-*
CO
4->
id
r4
*»^
VO
r-
rH
0
<^
<
M
M
M
X
0
H
53
Cd
iJ
g
EH
U
- rH
XI CO
rH O>
CO rH
CTi
rH <
CU
tj W
M CO
53 D
CO
•
r-»
•*
I
co co
rH B
0 ^
• cn
O 3.
to co
•H C
id o
•H
C CO
id co
XI -H
to B
D cu
CN
CO
~.
rH CO
4-J
id
rH to
id «—
4J 0
q; co
.. CT>
4J — • rH
IH CO
0) 4-> EH
XI
-------
^^
•n
CD
3
c
•rH
4-1
a
o
u
i-H
1
M
M
M
a
CQ
EH
PH
M
*^I
W
K
E-i
M
CO
U
U
s
EH
CO
CQ
D
CO
U
M
1
M
U
0£
*^t
o
2;
2Jj
o
2
«
o
Q
H
U
W
CM
CO
*~^
CO
6
CO
2
o
M
^J
«
j2
o
53
0
o
tJl
c
4-> C
(0 O
O -H
•H 4-1
T3 (0
C M
M 4J
c
0) (U
U CJ
C C
0) O
M CJ
MH
P4
c
o
•H
^_J
id
4J
c
U
§
U
(V
rH
u e
M id
3 CO
0
CQ M-l
o
Oi
c
•H
e*
CO >i
•rH 4J
i-H -H
,Q O
(0 -H
•P C
co a)
W 01
o
co C
(U -H
o o
C J-i
cu id
M U
0)
£
ON
rH
0
tt
CO <
W M
Q
M
U
CO
CM
Q
EH
2
C5 EH
O O
K
o>
10
ON
H
^
0)
'tTi
0)
rn
S
tTi
-^
o
o
m
t-H
id
•H
•P
C
0)
rrj
•H
CO 1-1
0) -H
« Id
("N.
r-*
0\
1-1
•
rH
jy
^
co id
&z
• •
rH rH
'O id id
C
(0 *3* 4-^ 4-*
r^ CD 0
C ON
Id rH ^i >•)
e CD 1 r-H rH
rH Q) C rH C rH
*O c id r**" id f^*
•H rH 4J ON 4J ON
CQ O CQ H CO rH
CO
e
•^ m
O) B
C ~^
Oi 0 C
0
r» O
O rH
1 'S' n rH
rH 1 B I
o o *>*. m
• • 01
o co c O
MM M
•H -H -H
id id t)
£5 H C
to it) id
Q) V_i ,£3
O 3 M
O « P
•
H
(0
4J
(U
>t
(U
rH
C rH
id r-
4J cy*
CO H
m
g
'N^X
DI
C
CM
*
r***
«N
1
CM
*
rH
M
•H
id
C
id
,Q
P
*^^
0)
u
•H
B
>-^
id
^J1
r**
ON
rH
O
P3
M
C
•H
14
•d
rH
0)
•H
Q
(N
r**
ON
rH
•
rH
10
4_J
0)
CO
ft
o
ro
B
"^
CP
C
^"
•
CO
1
00
CD
IH
•H
•
rH
10
4_l
0)
J>1
(U
rH
C rH
id r-
4-> ON
CQ rH
m
B
"'^
CT*
c
^
•
rH
1
O
M
•H
id
rH
id
IH
s
CO
p"*
ON
rH
*
IH rH
(U id
^
rH 4-1
Id 0)
Is
o — •>
>O -P CD
C rH CJ
id -H
•• B
(U >-*»—
Q< CU
MO"
0 -H r^
EH *-* rH
CU
C
id
•O
c
•iH
t-H1
B-4
-------
«— .
T3
CU
3
G
•H
4->
C
O
u
rH
M
M
M
W
. T
1— 1
CQ
<
EH
«
M
(d
Vl
4->
c
(1)
u
c
0
CJ
0)
•H
CU Q*
o g
vi id
3 CQ
CQ «W
0
Cn
d
•H
rC
CO >i
•H 4J
rH -H
XI 0
(d -H
4-> C
CQ CU
W Oi
o
CQ C
CU -H
o o
C Vl
cu (d
vi cj
cu
M-l
0)
«
cu
o
c
id
4J
CQ
X)
P*
3
x*x
•o
cu
3
c
•H
4-*
c
0
CJ
CQ
W
Q
M
o
M
EH
CQ
W
CM
Q
M
EH
«<
S3
u
o
35
v£>
vp
(Ti
rH
Vl
o
Xi
£
n
g
-^
Cn
d
t
iH
1
VO
•
o
M
•H
id
rH
(d
Vl
3
,—.
0)
0
•H
^g
id
t^-
r-
a\
rH
M
CJ
s
0)
c
id
•o
Vl
o
rH
6
13
c
id -*
r^-
C CT>
id H
g
cu >i
r-l CU
T3 C
•H rH
CQ O
n
g
01
d
m
CN
•
o
Vl
•H
id
d
Si
Vl
D
•o
c
(d
d
(d
g
0)
rH
13
•H
m
r>
rH
•
0
1
in
o
o
•
o
Vl
•H
id
d
id
0)
o
O
•*
i-»
o\
rH
>1
cu
d
rH
O
m
g
\
en
d
CO
w
CO
<
u
Q
jz
1^
CO
u
EH
*$
3
D
CJ
IH
EH
<
oS
CJ
M
js
<•
0
«
o
\y.
ft
M
••
O CD
co r- o
CT> CT\ 00
rH rH O\
rH
U id
tf 4-1 CQ
«< 3 >
M CQ t<
rn n
g g
•^ "^.
Cn 0i
a- d
m ^f
• iH
CN I
1 rH
O V
rH
id co vi
•H d -H
vi o id
4-> -H
CQ CO d
3 co id
•0 -H 43
d g vi
M CU D
>— »
CQ
d
id
g
3
£
o
oo
ffv
rH
CJ
a
M
O
•H
d
0)
CQ
4
T3 ro
d !-• rH
id en nj
•H
d 4->
0 >J
•H 3 0 CT>
53& CJ H
m
g
01 ^f
d on
rH g
O —
0 X CO
•H Vl
I n cu
-H m ,Q
1 -H
o n >w
rH Vl
Id CO -H
•H d id
vi o
4-> -H 4->
CQ CO Vl
3 CQ CU
T3 -H CQ
d g cu
M CU Q
•«
*-^
CO
^ . -N
id co » CQ
Vl -P CD C
*— (d vi id
M cu g
CN v-' 4J 3
r- co ,d
r^ g
rH r» IJJ ••
CTI ^ CQ
Vl rH 4->
CU » -H
d cj cu P
0) BJ o XI
id
CQ
cu
XI
5
•
rH
id
4-> -H
D) x
CO
0 *
•H o
,£ rH
O M O m
vi r^ co r^
3 CT> 0) O>
S rH S H
0
o in
r~ co on
• g rH g
m \ ^.
1 CO X CQ
O Vl Vi
o cu o cu
"1 XI H XI
* ft 1 -H
n «w r«
^
in
o n
-------
^— ^
•o
CD
C
•rH
4J
G
8
rH
1
M
M
IH
W
J
CQ
<
EH
«
M
<
W
g
55
M
CQ
W
U
EH
CO
CQ
D
CQ
CJ
(H
53
U
1
H
u
t§
3
u
£
12
0
*
OS
o
Q
w
EH
u
H
0<
CQ
g
fo
o
CQ
53
o
M
EH
<£
K
EH
U
55
O
CJ
cn
c
*i*4
•P C
id O
O -H
•H 4J
T3 Id
C H
M 4J
c
CD CD
O O
G C
CD O
M CJ
CD
id
M
4J
C
CD
CJ
C
0
CJ
CD
rH
CD O<
o e
M id
3 W
0
CQ m
0
cn
C
•H
rC
^ >1
•H 4J
rH -H
Xl O
id -H
4-> G
CQ CD
W cn
O
CO C
CD -H
O O
G rl
CD id
rl U
CD
«w
CD
«
CD
CJ
c
id
4J
CQ
XI
3
CQ
CQ
H
CQ
<
C5
Q
gj
CQ
H
EH
<
J
D
CJ
H
EH
< "O
O< 0)
3
U C
M -H
53 4J
*3 C
O O
os cj
2V"
53
IH
O
CO
ON
rH
U
§
M
m
•
a\
rH
1
n n
0 E
o \
• Di
O 3
iH
id CQ
•rl C
M O
4-) -H
CQ CQ
3 CQ
13 -H
C E
»H CD
CQ
4->
•H
XI
XJ
id
M
^
CO
4J
id
>H
v»^
o
CO— v
Cft CQ
rH >i
CD
CJA:
§§
" E
E
3
•H
rH
rH
>•,
M
3
00
r~
TJ CA
C rH
id
M-f
cn O
V-l rH
CD rl
X)
rH
1
M
E
3
•H
E
O
M
6
o
CO
rH vo -^
• Oi id • • CD
O 3. — rH VO 3
rH «.
(d CQ CO CQ
•H C C C
»H O O M O
4-1 -H -H CD -H
CQ CQ CO 4-> 0)
3 CQ O CO iH CQ
13 -H 4J -H CD -H
C E 3 E E E
IH Q) rt! CD CO CD
*-*
CO
4J
id
M
*.
CD
U
•H
E
*•—• •
O
CO
o\
•H
S
M
•o
id
%
0
CO
ON
rH
<
&4
U
CQ
O
CO
•
IT)
rH
1
C*
VO
r-» ro
0 E
o -^
• Oi
O 3.
rH
id co
•H C
n 0
4-> -H
CQ CO
3 CQ
T3 -H
C E
M (D
%
CO
4J
id
>H
«>
CD
CJ
•H
E
•»•-•»*
\O -^
r^ to
cr< M
rH CD
4->
O CQ
^ B
»s! id
"rG
rH
CD
AS
CJ
-rl
53
B-6
-------
X-*
rg
0)
3
C
"M
C
0
0
v«x
rH
1
M
M
M
W
m
rf
EH
PH
M
i
•H 4->
rH -H
ft 0
(0 -H
•P C
to o)
W en
0
to c!
M-l
0)
«
H rH
CO
rH rH t~-
(0 (0 cn
rH
[ l | \
a) o
C G U
•H -H 0
fa fa CD
O
CO
rH
P
>
TJ ^
C vo
(0 CTi
rH
C
•H ,Jd
4J H
0 (0
^ fa
r
m
ro ro
e E
^ ^.0
0^ 0^ *
3- C rH
1 ro
& O ro s
ro ^f o ^**
1 1 • cn
rH rH 0 3.
rH
(0 CO V-i
•H 4J -H
>H C 10
43 (V
10 3 C
3 rH (0
T3 M-) .Q H
C M-l IH -H
M (U O (<
^-^
to
(1)
•H
O
(1)
pt
to
CO
CO
r^»
"*
U
Pi
rtj
M
CQ
Q
O 2!
CQ ID
Q 0
03 CU
EH S
M O
23 U
1
CO
•
o
1
ro ro
0 B
O cn
v a
M
•H
03
C
(0
43
D
CQ
s
§
OH
rfj
1
Q
S3
CJ
EH"
_|
•H
10
C
10
>H
J3
.«
^— s
CO 4->
(I) 0)
•H O
o r^ -H
(U O\ 4J
ftrH 43
10 0
• -H
0^ rH M-l
*- (0 MH-
r*-
^ •S'm
U X 43
<^J (0 M
M (4 — •
0)
C
0)
H
jjj
(0
0
M
c
CQ
*o ^*
C *£>
(0 C^
rH
c
•H ,y
43 rH
0 (0
^ fa
0
B
cn
C
CT>
•
rH
1
rH
0
•
o
)_l
•H
(0
rH
«
3
Pi
.,
•**— s
— * to
to _i
rH 01,5
4) O
i— i O V-i
(0 tZ! U
r-
f""x
O^
rH •
t3 *™H
C ^ • (0
cn nj 43
C H (1)
(0 43
> >1 4
-------
13
CU
3
C
•H
-P
C
0
U
»— '
i-H
1
I-H
M
M
U
. T
M
CQ
ff
f*t
EH
K
M
<
W
13
a
H 1
CO
U
1^
EH
CQ
CQ
D
CO
INOGENIC
O
(*
Si
u
1
^
C*
0
Q
EH
r^
w
O4
CQ
D
CO
b
O
CO
ICENTRATION
^
0
U
d
c
*rH
•P C
(0 O
O -H
•H 4->
'O it
C M
M -M
c
CU 0)
O CJ
C C
28
cu
MH
(11
K
C
O
•H
4->
id
Concentr
cu
rH
cu o<
0 B
3 CQ
CQ MH
O
CTi
C
•H
CO >i
•H 4J
rH -H
XJ 0
id -H
•P C
CO CU
W en
O
References
Carcin
(U
CJ
c
id
4J
CO
JQ
3
CQ
CQ
Z
HYDROCARBO
U
M
EH
<
2
Q
« —
< -O
cu
U 3
M C
|4 -H
CJ 4J
h C
U 0
>H U
j_q »_,
00
MD
o>
rH
VH
CU
B
•rl
i,
>H
O
CO
e
o>
c
(N
CO
1
CN
•
CO
4->
CO
3
T3
1
iH
3
ro
t
CJ» ^-v
H CU
•O cj m
C C -H VO
-H nj B
.* »-* rH
B M
•H cu id •
x; -a en rH
to c in ro
>i
T3 12 rH 4J
C CU
l -H MH O -H
S5S-03
1
r— •
x: u
- c
(0
id en
r- 1
c
id >H
B cu
MH -a
n S.
S^1
Present
Vehicle
exhaust
c
i
MH
MH
o
K~
cu
•O 0
C -r)
-H
rH CU
•p
• CQ
rH CU
id w
jj •»,-*
Wodinsky e
(hamsters)
1967 (mice
i
•H
^
id
o
N CU
C C
CU (U
£ n
•H >*
Q ft
m
r~
CT>
rH
U
ftf
<
M
CO
0
CP
00 «n
0 B
CN
1 O
r-H 0
0 0
• »
0 rH
rl
2
£J
i
MH
MH
o
ffi —»
cu
T3 0
C -H
id^S
rl
cu id
•O CTi
C m
>lOS
12 rH
CU
g
cu
M
>1
S
r— i
CU
o
N
c
cu
CQ
(0
.p
cu
C!
•H
cu
4J vD
CO
Q^CTv
frj _j
Oi
\
fTt
O"
a _i
180-570 i
(of tota!
extract )
AI i
Airborne
particu-
lates
B-8
-------
,— »
•o
CU
3
C
•H
-P
C
0
o
*-^
rH
1
M
M
M
W
, T
1— 1
CQ
<
EH
«
M
<
W
ffi
EH
£3
M
CO
u
§
EH
CO
«
D
CO
CJ
M
S3
W
o
o
23
M
CJ
OJ
<
CJ
£3
S
0
£3
«
#
O
Q
M
EH
CJ
W
&
CO
»— t
3
CO
fa
o
CO
£5
O
M
EH
£
EH
M
CJ
£3
0
CJ
tn
c
,pj
4-> C
rC 0
O -H
•H 4->
-d (d
C H
M 4-)
G
CU CU
CJ CJ
c c
CU O
n U
CU
M-l
n\
&
G
0
•H
4->
(d
M
-p
c
cu
o
c
o
u
cu
rH
cu ft
u e
IH ,
•H 4J
rH -H
XI 0
rd -H
•P C
co cu
W en
O
CO G
CU -H
O CJ
C M
cu id
M CJ
cu
M-l
cu
«
cu
o
id
4->
CO
XI
3
CO
CO
£3
«
<
8
tf
Q
fa
£
CJ
M
EH
52
8~
< -o
cu
U 3
M C
r^-H
CJ 4J
>H C
CJ O
S3
.
•
rH
id
4J
n cu
r^
en c
rH -H
CU
CJ 4J VO
c^ co vo
rf* O /T\
r*« >t tji H
C ^^ id'-
Cn 4J 4->
r^ 3-00
4J id
CO O i-|
rH 00 MH 4J
1 «N 0 X
VO 1 ^- CU
m
O
XI M JH (d
r4 -rj (0 rH
CD <3 ft 3
en
vo
CTi
•» rH
cu x
O rH
•H id
6 fa
•o
n c
vo (d
CTi
rH e
id *~~
C rC CU
•H oi O
CU C -H
rH -H S
* m~
CU
c
1 CU
'-' O
id id
"""* V4
N A5
C P
0) C
ffl id
•
rH
id
4J
cu
•H
,*
O id
•H in
S vo
frt /T\
$2
4-1
C
cu
CO
cu
)H
CM
rH
id co
•rH 4J
M C
4-1 CU
CO 3
3 rH
T3 M-l
C M-l
M CU
in
en
co
C
C
o
CO
0)
cu
"0
c
C cu
(d o
•
3- 4-> 4->
^00
o 4-> (d
M
<* MH 4->
I 0 X
O «-' CU
m
Vehicle
exhaust
•H
03
C
(0
ft
£
Airborne
particu-
lates
0)
c
(!)
CO
B-9
-------
T5
3
C
*r^
4J
C
0
u
SM^
•H
|
M
M
M
W
•J
03
EH
PH
M
^1
W
33
EH
25
M
CO
w
u
EH
CO
m
D
CO
u
M
23
w
o
o
25
M
U
CM
r^|
U
JZj
»2
O
J23
«
o
o
w
EH
U
H
O<
CO
D
CO
fa
o
CO
55
O
f-l
3
EH
W
CJ
25
O
CJ
Cn
C
•H
4-> C
It) O
O -H
•H -P
'd rd
c M
M 4J
C
CO CD
O O
C C
CD 0
rl U
(D
4-1
d)
OH
c
0
•H
4J
-l Ifl
3 CO"
S«.
0
en
c
•r-l
^
CO t>i
•rl -p
rH -rl
,0 0
(0 -H
•P C
CO 0)
w en
0
CO C
CO -H
o o
C M
co (0
to CJ
co
co
c*
CO
u
c
id
CO
,Q
CO
en en
r~» r~
en en
rH rH
CO CO
r^ f^
25 25
n
S
Di O
G
0
VO 1
7 0 %j
in O •*-,
• • £pi
0 0 C
to rl
•rl -H
(Q <0
(3 rH
fO (0
1 I
•«
*~* rC
CO M -H (0 CO
r*» ^ M 4->
en t3 v— ' fl
rH C •» rl
ro^in —
• co r*-
rH ff o en (0
(0 en -n rH co
3 JE . ^
CO ,4 rH rH
JQ •* (0
0 B r- H
M S^ rH CO 25
Q
EH
25
OS CO
O J
33 25
CJ W
^t 52
1*1 ^
O M
cu a
CO
r*^.
en
H
ff
O
CO
3
4-*
fd
25
o
iH
1
O
H <*>
S
•H
CN CJ
^1
•H
n)
C
M
D
CO
0)
•I-l
o
CO
CO
o
CN rH
r- (0
CO 4J 4J
CO CO
CO
•a c c
C ron)
O -H -rl
B r^. rH r- H r-
•H en -H en -H en
CO rH r3 rH r-J rH
f\
ft
ft
»Q *• CN O
ft CO 43 B • ft
ft en ft 3 •* ft
n) ft B i
in n -H co t~-
CN CO CN X rH O
• > • (d
rH Ifl *t B O O
CO
CO
CO -H .H
en (C
c
CO
3
rH
M-l
4-4
0)
CO
r»
en
0)
(0 —
A CO
•rl 4->
U (0
O
Q
U
EH
B-10
-------
<— v
•a
CO
•a
3
rH
o
§
u
^-^
rH
1
M
M
M
W
j
hH
«
<
EH
«
M
<
H
EC
EH
»
M
CO
CJ
EH
CO
0)
D
CO
u
M
2!
U
C5
O
S3
HH
CJ
s
CJ
jg
2
§
*
«
O
Q
W
EH
CJ
H
Pu
CO
D
CO
fe
o
CO
&
0
M
EH
S
E^
M
0
*7
§
CJ
Cn
C
-1
•rt
jj d
rd 0
CJ -H
•H 4-t
T3 <0
C M
M 4J
c
CO CO
CJ CJ
C C
CO O
M U
CO
M-l
CO
on
c
0
•H
4->
(0
IH
4->
c
CO
o
c
o
u
CO
rH
CO £4
CJ g
*4 (0
3 CO
CO M-l
o
en
c
•H
rC
CO >i
•H 4->
rH -H
X) 0
(0 -H
•y c
co co
W DI
0
CO C
CO -H
O CJ
C to
CO (0
M CJ
co
M~l
CO
«
CO
o
c
(0
4->
CO
a
3
CO
•
H
(0
4J
0)
rH
rH
CO
C
c
O m
cjr-
O CTi
SrH
CO
g
•^
Cn
C
rH
i-H
M
•H
<
*-*
CO
o
•H
g
V0
f-
rH
u
S3
u
2;
j
EH
U
O
«
s
nc
u
M
OJ
EH
<*
r^
c\
rH
<
&4
M
CO
D
CM
•
ID CO
1 g
VO -v.
• Cn
CN C
rH
to co
•H C
n o
4-> •!-«
CO CO
3 CO
-0 -rH
C g
rH CO
CO
4-»
(0
J^
- CO
CO G
a «j
•H g
g 3
**"^ -C
H
rH CO
4->
CJ CO
ry' c
W S
<3 (0
W rC
M
Q
H
PC
E
U
J
iH
g
M
>
•
1 — 1
nj
4J
CO
C
it)
•H
rH
rH
•iH
J
XI
S1
04
O
C7i
|>»
1
in
9
00
CO
M
•rH
(0
c
5
VH
D
ID
o
o
rH
I CO
s ^
d £
O
CO
CO
co o
C 4J
£ °
0 <0
4-> CO
CO CJ
W H
S3 O
U rH
Q #
EC
U
B-ll
-------
APPENDIX C
CALCULATION OF THE AGE-ADJUSTED RESPIRATORY CANCER RATES
IN MALES AND IN THE GENERAL POPULATION
According to the U.S. Bureau of the Census (1980), the
proportion of white males in the U.S. male population in 1960
was 88.7%, and that of black males was 10.3%, with 1.0% unclassi-
fied. In 1970, the proportion of white males was 87.7% and
that of black males was 10.9%, with 1.4% unclassified. USDHHS
(1983) provided race-specific rates of respiratory cancer mortality
per 10 males, age-adjusted to the 1940 U.S. population:
1960 1970
White males 34.6 49.9
Black males 36.6 60.8
Hence, for the total U.S. male population, the rate of respiratory
cancer mortality per 10 persons in 1960 was:
(34.6X.887) + (36.6X.103)
= 34.8
(.990)
and in 1970 it was:
(49.9) (.877) + (60.8) (.109)
= 51.2
(.985)
C-l
-------
For the entire population (all races/ both sexes combined),
the rate of respiratory cancer per 10 persons was 19.2 in
1960 and 28.4 in 1970 (USDHHS 1983). Data on respiratory cancer
death rates for 1965 have not been tabulated, but for the purposes
of this report, arithmetic means of the 1960 and 1970 rates
are used. These are 43.0 per 100,000 for males (all races)
and 23.8 per 100,000 for the entire population.
C-2
-------
APPENDIX D
CALCULATION OF THE RISK OF LUNG CANCER TO
THE GENERAL POPULATION AS A PROPORTION OF THE RISK TO MALES
The risk of lung cancer mortality per ng/m of benzo[a]-
pyrene, estimated by Pike et al. (1975) based on the data of Stocks
(1958), is 1.4 deaths/10 persons among smokers and 0.4 deaths/10
persons among nonsmokers. As discussed in Chapter II, the
magnitude of these risks per unit exposure is likely to be
underestimated, but the relative difference in risk of 3.5
(1.4/0.4) is probably reliable and is reported by Wilson et
al. (1980) as statistically significant. Because a number
of estimates were made for the risk of lung cancer mortality
among males only, it was necessary to derive the risk to the
general population as a function of risk in males.
USDHEW (1979) provided data on smoking habits in men and
women in 1977: 40% of men and 30% of women are smokers. (This
represents a decline compared to previous smoking habits.)
The recent data of Hammond and Seidman (1980) indicate that
the relative risk of lung cancer mortality among smokers is
8.53 in men and 3.58 in women. We assume that the excess risk
from air pollution is proportionately the same (8.53/3.58 = 2.4).
So if the relative risk among male smokers is 3.5, the relative
risk in female smokers will be [1 + (3.5 - l)/2.4], i.e., 2.05;
we assume that the risk among nonsmokers is the same in males
and females (i.e., 1.0). From the census data (U.S. Bureau
of the Census 1980),
D-l
-------
the fraction of the population that is male is 0.487 and the
fraction of the population that is female is 0.513. Thus,
(0.487) [(3.5) (0.4) + (1) (0.6)] +
(0.513) [(2.05) (0.3) + (1) (0.7)] = 1.65
The relative risk among all males is 2.0, and the relative
risk in the general population is 82.4% of the risk in males
(1.65/2.0).
D-2
-------
APPENDIX E
DERIVATION OF AN ESTIMATE OF THE PROPORTION
OF LUNG CANCERS ASSOCIATED WITH THE URBAN ENVIRONMENT
In this Appendix, we derive an estimate of the proportion
of lung cancers associated with the urban environment. Earlier
versions of this calculation were included in our previous
reports (Clement 1981, Karch and Schneiderman 1981), but these
have been modified to take into account criticisms of these
earlier versions and suggestions by CAG (1982) and other corn-
mentors .
Our estimate is derived from a study by Hammond and Garfinkel
(1980), together with additional information from the same
study presented by Goldsmith (1980). The data in these papers
were standardized for age and smoking habits, and included
information on (self-reported) occupational exposure. (The
authors, however, did not give details of how the "corrections"
were made, or of the age distribution and smoking habits of
their standard population). Although the data presented by
Hammond and Garfinkel (1980) were not fully described or given
in the 1980 paper, they were clearly derived from data obtained
in a survey sponsored by the American Cancer Society (ACS)
from 1959 to 1965 (Hammond 1972). (Table 1 in Hammond and
Garfinkel 1980 is identical to Table 5 in Hammond 1972.)
We used these data as reassembled in another recent paper
by Goldsmith (1980). One can compare lung cancer mortality
E-l
-------
rates among men between urban and nonurban areas as Goldsmith
did by combining some groups to form three categories: "metropo-
litan areas of greater than one million," "other non-rural
places/" and "non-metropolitan rural areas." The results of
Goldsmith's (1980) reassembly are found in Table E-l. These
data are plotted in Figure II-2 (above, p. 11-85) and show
a trend for increasing cancer mortality with greater urbanization
in both occupationally exposed and nonexposed persons, after
correction for smoking. The use of three residence or exposure
categories in this sort of study has been questioned, but is
apparently common practice. See, for example, Hitosugi (1968)
and Vena (1982) who used similar categories.
These results provide a measure of the risk of death from
lung cancer among males that is attributable to an urban effect,
by contrasting the urban and the rural areas (i.e., by combining
the first two categories in Table E-l to compare with the third).
These risk ratios derived for the ACS population can be weighted
according to the proportion of the U.S. population in each
category in 1970 (U.S. Bureau of the Census 1980). The attrib-
utable risk for an urban effect can then be computed.
This computation is illustrated in Tables E-2 and E-3.
The residual urban effect, after correcting for smoking, is
about 13% for the occupationally exposed group of men and 12%
for the nonoccupationally exposed group of men. Applied to
a cancer mortality rate of 43 per 10 (see Appendix C), this
corresponds to a risk of 5.2 excess respiratory cancers per
10 urban men.
E-2
-------
TABLE E-l
LUNG CANCER DEATHS AMONG MEN BY PLACE OF RESIDENCE
AND OCCUPATIONAL EXPOSURES—SMOKING ADJUSTED—1959-1965*
Occupationally Exposed Not Exposed
Observed Expected Ratio Observed Expected Ratio
TOTAL 576 530.5 1.09 934 979.7 0.96
Chi sq. = 6.03, p<0.02
Metropolitan
area city 92 69.1 1.33 168 158.3 1.06
(1,000,000+)
Other non-rural
places 341 315.3 1.08 584 607 0.96
Nonmetropolitan
rural areas 143 146.1 0.98 182 214.4 0.85
Chi sq. = 9.75, p<0.01 Chi sq. = 6.4, p<0.05
SOURCE: Goldsmith (1980), Table 7; Hammond (1972)
*The observed and expected number of lung cancer deaths listed
by place of residence and by occupational exposure (to dust,
fumes, gases, or X-rays), and adjusted for age and smoking
habits, is confined to men who had lived in the same neighbor-
hoods for more than 10 years. The subjects were among a pop-
ulation of 1,064,004 men and women studied by the American
Cancer Society (Hammond 1972).
E-3
-------
TABLE E-2
RELATIVE RISKS IN MEN OF LUNG CANCER MORTALITY
(ADJUSTED FOR AGE AND SMOKING)
BY RESIDENCE AND OCCUPATIONAL CATEGORY
Derived by Comparing Residents of Metropolitan Counties
and Urban Sections of Nonmetropolitan Counties with Residents
of Rural Sections of Nonmetropolitan Counties (25-State Study,
Confined to Men Residing in Same Neighborhood for Last 10 Years)
Occupat ionally
Exposed*
Weighted Relative Risk, Urban**
(U.S. population 1970)
Overall Weighted Relative Risk***
Not
Occupat ionally
Exposed*
Metropolitan Counties
Greater than 1 million residents
Less than 1 million residents
Nonmetropolitan Counties
Urban
Rural
1.26
1.17
0.99
1.00
1.16
1.14
1.18
1.00
1.19 1.16
1.17
*To dust, fumes, gases, or X-rays
**Relative risk of lung cancer mortality in metropolitan counties
and urban sections of nonmetropolitan counties (weighted according
to population data from U.S. Bureau of the Census 1980)
***Weighted by the proportion of men in occupationally-exposed
and nonexposed categories in study population of Hammond and
Garfinkel (1980)
SOURCE: Adapted from Hammond and Garfinkel (1980), Table 1, p. 208
E-4
-------
TABLE E-3
ATTRIBUTABLE RISKS OF LUNG CANCER MORTALITY
(ADJUSTED FOR AGE AND SMOKING)
U.S. MALES (25-STATE STUDY) DUE TO URBAN
FACTOR AS AN INDICATOR OF AIR POLLUTION
COMPARISON
Residents of Metropolitan Counties and Urban Sections
of Nonmetropolitan Counties with Residents of Rural Sections
of Nonmetropolitan Counties; Proportion (p) "exposed" in
U.S. Population in 1970 is 0.816 (U.S. Bureau of the Census 1980)
Occupa t ional ly
Exposed*
Not Occupat ional ly
Exposed*
Overall**
Relative
1
1
1
Risks (RR)
.19
.16
.17
Attributable Risks
13.4%
11.6%
12.2%
(AR)
*to dust, fumes/ gases, or X-rays
**weighted according to the proportions of occupationally exposed
and nonexposed men in the study population of Hammond and Garfinkel
(1980)
E-5
-------
To convert this rate to a rate for the general urban population,
we multiply by 0.82 (see Appendix D) and obtain 4.3 deaths/105 urban
persons. The proportion of the population that is urban according
to the categories established originally by Hammond and Garfinkel
(1980) is the proportion living in metropolitan areas plus
those living in urban sections of nonmetropolitan areas/ that
is, all persons not living in the rural areas of nonmetropolitan
areas. In 1970, this proportion was 81.6% (U.S. Bureau of
the Census 1980), leading to an estimate of 3.5 deaths/10
persons/year in the U.S. population (4.3 x 0.816). Note that
we use the same proportion for occupationally exposed and unex-
posed, although it is more likely that a higher proportion
of occupationally exposed persons are also urban residents.
This estimate of attributable risk has several limitations.
First, the method of standardization for smoking habits was
not described, so it is possible (as suggested by several com-
menters) that the correction was incomplete and that some frac-
tion of the unexplained urban effect was due to differences
in some aspects of smoking habits (such as age at starting
to smoke), for which standardization was not carried out.
However, the ACS study collected information on a variety of
aspects of smoking habits and revealed no effect of age at
starting to smoke. Moreover, as discussed in the text, the
data of Haenszel et al. (1956) revealed no systematic differences
in age at starting to smoke in any age-group studied in 1955.
E-6
-------
Second, the aggregation of the data into three broad res-
idence categories by Goldsmith (1980) and subsequently into
two categories in Table E-2 may have obscured some differences.
Although Hammond and Garfinkel (1980) presented data for more
residence categories, these were aggregated by Goldsmith, and
it was not possible to use the disaggregated data because Hammond
and Garfinkel's residence categories cannot be matched to data
from the U.S. Census. In the absence of specific reasons to
suspect bias, aggregation of data is generally expected to
result in the reduction or masking of associations by pooling
individuals with greater and lesser exposure within each cat-
egory.
Third, the study population in the ACS survey, although
it contained many residents of large urban areas, is not likely
to have been representative of the entire U.S. population (cf.
Sterling 1975). It had a different age distribution and included
more white-collar workers, higher educational levels, and a
higher socioeconomic class on the average than did the general
U.S. population. Thus, the proportion of occupationally exposed,
which was classified on the basis of self-reported exposure
to "dust, fumes, vapors, gases, or X-rays" (Hammond and Garfinkel
1980, p. 4) may be underestimated, and the proportion living
in urban areas with the highest air pollution levels (i.e.,
residents of inner cities) may also be underestimated.
Several attempts have been made to estimate the possible
magnitude and consequences of this selection bias. Karch and
E-7
-------
Schneiderman (1981) suggested that the attributable risk from
urban residence (unexplained urban effect) might have been
underestimated by a factor of about 2.1; this estimate was
based on a comparison of the data of Hammond and Garfinkel
(1980) with those of Haenszel and Taeuber (1964). Doll and
Peto (1981) suggested that the selection bias in the ACS study
had led to underestimation of the effects of alcohol by a factor
of about 2 (footnote c to Table 11), and to underestimation
of the effects of occupation by a factor of about 3.3 (p. 1244).
GAG (1982) matched data on social stratification of the ACS
population to data on the relationship between exposure to
BaP and socioeconomic stratification, and suggested that the
ACS population would have been exposed to average levels of
BaP only 70% of the U.S. average. Although all these estimates
are somewhat speculative, the consensus view is that selection
bias in the ACS study is likely to have reduced the apparent
magnitude of these risk factors by factors between 1.44 and 3.3.
Strictly, our estimates of attributable risk in Table E-3
are estimates of the "unexplained urban effect"—i.e., the
fraction of the excess urban lung cancer rate that is not ex-
plained by standardization for recorded differences in smoking
and occupation. In principle, this "unexplained urban effect"
might include contributions from other factors (such as unrecorded
aspects of smoking behavior) as well as from air pollution.
However, in the remainder of this Appendix we will use our
estimates as a measure of the effects of air pollution. In
E-8
-------
the absence of reliable data on air pollution levels at the
appropriate period in the 1930s and 1940s/ we will relate the
excess cancer mortality in the 1960s to the level of 3.5 ng/m
BaP characteristic of U.S. population exposure in the early
1960s (see CEQ 1980, and discussion in the text). (This pro-
cedure, although questionable, is similar to that used for
other estimates tabulated in Table IV-1, and its consequences
are discussed in the text.). Using CAG's (1982) estimate that
the ACS population was exposed to an average level of BaP only
0.70 times the U.S. average, we estimate the average exposure
of the ACS population to be about 2.5 ng/m BaP.
Related to an average exposure to air pollution character-
ized by 2.5 ng/m BaP, an estimate of 4.3 deaths/10 persons/year
corresponds to a dose-response coefficient of 1.7 deaths/10
persons per ng/m BaP. This is the figure included in Table IV-1,
E-9
-------
-------
APPENDIX F
TIME TRENDS IN LUNG CANCER RATES
In principle, changes in mortality and incidence rates
with time can provide clues as to the causes of disease. Changes
in exposure to a causative agent should, after appropriate
latent periods, be followed by changes in incidence and mortality
of the disease in the exposed cohorts. Thus trends in age-
and sex-specific incidence and mortality rates can provide
supporting evidence for the existence of an association that
is hypothesized for other reasons. Likewise, observed trends
that are not consistent with an hypothesized association may
provide substantial evidence against the hypothesis—or at
least indicate that another causative factor is involved.
To test the hypothesis that air pollution plays a role
in the etiology of cancer, it would be desirable to compare
age- and sex-specific trends in cancer rates to earlier trends
in exposure to air pollution. However, as explained in Section
II.D.2.d of this report, there is insufficient evidence in
trends in exposure to make specific predictions, since downward
trends in the ambient concentrations of some air pollutants
have been offset by upward trends in others. However, data
on trends in cancer rates are of some importance in considering
one specific issue. Doll and Peto (1981) presented arguments
that available data on trends in lung cancer rates could be
adequately explained by the available information on changes
F-l
-------
in smoking habits, without the necessity for invoking other
causative factors. This conflicts with an earlier conclusion
by Schneiderman (1978). in this appendix we review data bearing
on this dispute, including more recent analytical studies by
Manton et al. (1982) and Janis (1982). This review is necessarily
limited to lung cancer, because there are insufficient data
on the contribution of smoking to cancers at other sites.
Examination of the data on cancer deaths in the United
States for the last 30 years reveals a steady increase in the
overall age-adjusted mortality rate (USDHEW 1980). Incidence
rates have also increased, although not as consistently. Between
the First National Cancer Survey in 1937-39 and the Second
National Cancer Survey in 1947-48 (Dorn and cutler 1959), the
overall age-adjusted incidence rate for cancers at all sites
rose by approximately 11%. Subsequently, between the Second
National Cancer Survey and the Third National Cancer Survey
in 1969-71 (Cutler and Young 1975), the age-adjusted incidence
rate declined by 4%. In analyzing data from the Third National
Cancer Survey and the NCI Surveillance, Epidemiology, and End
Results (SEER) program (Young et al. 1978), Pollack and Horm
(1980) concluded that, between 1970 (average of 1969-1971)
and 1976, the overall age-adjusted cancer incidence rate was
again rising. They found an increase of approximately 10%
during that 5-year period. Because age-specific trends in
cancer are not constant across all ages (i.e., decline in youngest
F-2
-------
age groups and increase in older groups), it is important to
examine age-specific rates separately.
There is evidence that the Third National Cancer Survey
produced inconsistent estimates for the 3 years 1969-1971;
1969 appears to have included some prevalence cases/ i.e.,
cases diagnosed earlier than 1969, and 1971 (the last year
of the survey) may have been under-reported. Pollack (1980)
has since reported incidence data derived completely from the
SEER program for 1973-1977, which should be free of these poss-
ible flaws. The SEER data show increases in total (age-adjusted)
cancer incidence of 6.8% in white males, 3.8% in white females,
3.4% in black males, and 2.4% in black females during the 4-year
period.
A major portion of the increase in cancer mortality and
incidence rates is due to an increase in lung cancer. This
increase in lung cancer is a general phenomenon in many countries.
Increases in cancer of the respiratory tract are appropriately
attributed largely to cigarette smoking, and secondarily to
occupational exposure, environmental pollution, or other sources.
In England and Wales, for example, there was a 10-fold
increase in death rates from lung cancer from 1901 to 1930
and an additional 10-fold increase from 1930 to 1960 (Katz
1964). in Canada, the male death rate from lung cancer increased
from 3.0 per 100,000 in 1930 to 24.6 per 100,000 in the 1960
population (Katz 1964). In Switzerland, a 32-fold increase
F-3
-------
occurred between 1900 and 1952 (deary 1963). From 1933 to
i960, the annual lung cancer death rate in Australia increased
from 3.15 per 100,000 to 28.9 per 100,000 for males and from
2.02 per 100,000 to 4.2 per 100,000 for females (Cleary 1963).
In the United States, the lung cancer mortality rate for
males has increased more than 25-fold in 45 years and is now
increasing even more rapidly for women (USDHHS 1982). During
the period between the Second National Cancer Survey and the
Third National Cancer Survey (1947-1970), the incidence of
lung cancer more than doubled in men and women, and in blacks
and whites (Dorn and Cutler 1959, Cutler and Young 1975).
Rates for black males have increased more rapidly than for
white males. Projections for 1981 are 122,000 new cases of
lung cancer and 105,000 deaths (ACS 1980).
A comprehensive study of lung cancer in Western Europe
was made in 1969 by the World Health Organization (WHO) Working
Party on Cancer Statistics. The study revealed that over the
previous 10 years, lung cancer mortality had increased by 8%
for males and 3.1% for females. The conclusion was that the
observed increase in lung cancer death rates was real and not
an artifact of better diagnosis or reporting or of longer life
span.
In West Germany, lung cancer deaths increased from 6,296
in 1952 to 15,000 in 1965. According to Wagner (1971) during
this period there was no significant change in efficiency of
diagnosis or reporting. To determine whether increases in
F-4
-------
lung cancer (in Denmark) were real or due to more accurate
diagnosis, X-rays taken during the course of examinations for
detection of pulmonary tuberculosis were reexamined. The X-rays
did not reveal many misdiagnosed cancers/ and it was concluded
that a true increase in lung cancer incidence had occurred
(WHO 1969).
There is considerable disagreement over the full set of
reasons for these increasing rates. Both direct industrial
exposure and air pollution levels have been suggested as contri-
buting to the increases—as well as cigarette smoking (Davis
and Magee 1979). Doll and Peto (1981) compared age-specific
lung cancer mortality in England and Wales with lung cancer
mortality in the United States, relating each to cigarette
smoking (their tables E5 and text Figure E4, summarized here
in Table F-l). From about 1900 to 1920, British cigarette
sales were higher (per capita older than 15) than U.S. sales
(per capita older than 18). From 1920 to 1940, U.S. and British
sales were almost equivalent; from roughly 1942 on, U.S. sales
have been substantially higher than in Great Britain. In the
youngest age groups (30-34 and 35-39), mortality per million
men in 1978 was almost identical in the two countries. This
appears to be inconsistent with the substantially greater number
of cigarettes consumed after 1940 by U.S. men if cigarette
smoking were the sole cause. For the age groups 40-44 and
45-49, U.S. mortality in 1978 was 25-40% higher than in Great
Britain. For men older than 55, the mortality rates in Great
F-5
-------
£
t^
Ht
|
« co
o w
ss
«B
W CO
U
§g
UEH
LJ
o a
53 D
^j
J
M ^
CO
Q W
•H H 3
i §g
* *
w 1LQ
kJ ^ fcz
3 •— i 5*t
^ s S
i la
tf <
W J
cu o
O W
53
HH «.
« CO
o w
S J
CO i O
a -P -H
Ci i
•rH r-1
(t H rH
U It -H
-P S
Cn >H
COM
D S 0)
it «W -4 ft It "O rH
O, G DlrH 3
DJ O -H O IO
t 0
O 4-* -rl
C -H iH
(t M rH
U It -H
•P S
01 to
COM
3 S 0)
•J a,
tO
It *M (U
S O -P M
•H 4J <1)
X M CU *
O , 1 1
til -P
CU it «w
< B O
C *"**
•H to CO
to r*-
0) it ON
Cn H —
r**. ^i ^\i ^"^ _ j r*^ i rt r^
I ^ i N IN *^J r~^ I ^ U J l^
•nyjoNCDcM'trMin
(H rj< O W3 VO IT)
*• ^ •. ^
r-l iH fN) CO
inoNOvor-onoo
i-trHOOO>vo^fro
rH rH iH rH
r^ncDint^cNinco
rHVDCOCO'S'rHi— IM
rH co o ON ro O
r-i M ro in
ininmMOOmoo
t~>r~vovor>»vo^tro
VOcHVOrHVOrHVOi-H
vovomin^f^coro
ONONONONONONONON
iHrHrHrHi-lrHMiH
VOi-HlOrHVO'HVO'-l
^•^•rOnCMCJi-HrH
ONONONONONONONON
•«j * _j _^ ^j ^j t ^i
f~1 r"1 r-( r-( f— 1 r^ l™1 i™1
^*O>'!l'ON'^ >
o o en
c
00 0) -H
C C -P
O 0 M
to to it
MM 4->
<1) t
•H 4J ^N
M OT O"
CO IT)
TJ
4-> 0) 01
It 4-> C
9) -H -H
M C B
C5 P 3
to
C C W
MM <|
* *
*
in
H
Q}
M
3
Cn
•H
PH
•o
It
**
-------
Britain in 1978 were substantially higher, despite the fact
that average numbers of cigarettes smoked were roughly equal
at the time these men started smoking. The Doll and Peto tabu-
lation ends at age 69. In the United States, the greatest
increases in lung cancer mortality between 1968 and 1978 were
in men aged 75-84 (Davis et al. 1982).
Two possible explanations for these inconsistent results
suggest themselves: (1) other characteristics of smoking,
such as the age at starting or the length of the unsmoked stub
(Doll and Peto 1981), are (or were) substantially lower in
Great Britain than in the United States, and/or (2) other things
in the environment (e.g./ industrial exposure, air pollution)
led to higher rates in Great Britain despite lower smoking
levels than in the United States.
There is at least one other way of looking at the time-
trend (cohort) data. The U.S. Surgeon General, in his report
entitled Health Consequences of Smoking (USDHHS 1982a, pp. 51,
53, and 56-57), has reported smoking data by year of birth
(in 10-year intervals—e.g., 1901-1910) and cancer mortality
for age-specific groups (e.g., 30-34, 35-39, etc.). From these
data it is possible to find birth cohorts with similar cigarette-
smoking patterns and then to compare their lung cancer mortal-
ities at specific ages. (See Figures F-l, F-2, and F-3, derived
from Figures 12, 14, and 16 of that report.) For example,
for men born between 1901 and 1910, 62% was the maximum that
ever smoked. The next cohort with a similar maximum was the
group born between 1931 and 1940. The median age of starting
F-7
-------
FIGURE F-l
CHANGES IN THE PREVALENCE OF CIGARETTE SMOKING
AMONG SUCCESSIVE BIRTH COHORTS OF MEN, 1900-1978
1921-30
MEN
1941-50
1951-60
1900 1910 1920 1930 1940 1950 i960 1970 1980
YEAR
Note: Calculated from the results of over 13,000 interviews
conducted during the last two quarters of 1978, provided
by the Division of Health Interview Statistics, U.S.
National Center for Health Statistics
SOURCE: USDHHS I982a
F-8
-------
FIGURE F-2
CHANGES IN THE PREVALENCE OF CIGARETTE SMOKING
AMONG SUCCESSIVE BIRTH COHORTS OF WOMEN, 1900-1978
WOMEN
1931-40
^fci
tf
1951-60
1941-50
SOO 1910 £20 S30 1940 1950 i960 1970 1980
YEAR
Note: Calculated from the results of over 13,000 interviews
conducted during the last two quarters of 1978, provided
by the Division of Health Interview Statistics, U.S.
National Center for Health Statistics
SOURCE: USDHHS 1982a
F-9
-------
FIGURE F-3
MORTALITY RATES FOR MALIGNANT NEOPLASMS OF THE
TRACHEA, BRONCHUS, AND LUNG, FOR WHITE MEN AND WHITE WOMEN,
BY BIRTH COHORT AND AGE AT DEATH, UNITED STATES,
5-YEAR INTERVALS DURING 1947-1977
1.0000 r—
9000 —
MOO
7000 —
(000 -
MOO
1000
MO
MO
700
WO
40.0
30.0
200
100
10
1.0
7X1
6.0
SJO
40
to
0»
U
07
04
I 0000
9000
aooo
7000
(000
6000
1000
900
no
700
(00
500
400
300
100
90
10
70
60
so
4.0
I I I I I I I I I I I I I
10 —
01 -
at -
0.7 -
06 -
9.S -
i I i i i i i
I i i i i S i I
9IIITM COHORT
s i s i j i
I I \ \
04
WHITE WOMEN
I I I I I I I I I I I I I
I I • 8 I 5 I S S § S I £
I I i t 8 5 I I § S 2 2 I
IIRTH COHORT
Note: Calculated from the results of over 13,000 interviews
conducted during the last two quarters of 1978, provided
by the Division of Health Interview Statistics, U.S.
National Center for Health Statistics
SOURCE: USDHHS I982a
F-10
-------
to smoke was about 17 for the 1901-1910 men, and about 16 for
the 1931-1940 cohort. The lung cancer mortality rates for
men aged 40-44 years born in 1931-1940 were almost double the
rates for men born 1901-1910, whose smoking patterns were similar.
For women the comparable smoking cohorts are 1921-1930 and
1931-1940, separated by only 10 years. The second (more recent)
cohort of women has a 25-60% higher lung cancer mortality rate
at comparable ages (30-44); a 25% increase in 10 years is equiva-
lent to a doubling in 30 years: (1.25)3 = 1.95.
Table F-2 gives the smoking data for men. Similar data
for women can be derived from the Figures F-l, F-2, and F-3
from the Surgeon General's report.
TABLE F-2
SMOKING HISTORY: U.S. MALES
Decade of
Birth
(mid-year)
(1)
1891-1900 (1895)
1901-1910 (1905)
1911-1920 (1915)
1921-1930 (1925)
1931-1940 (1935)
1941-1950 (1945)
1951-1960 (1955)
Maximum
Percent
Smoking
(2)
47
62
72
70
61
58
Possibly
Year of
Maximum
(3)
1924
1938
1946
1952
1962
1968
not yet
Year of
50% of
Maximum*
(4)
1913
1922
1933
1942
1951
1961
reached
Median- Age
Beginning
to Smoke
(4)-(l)
(5)
18
17
18
17
16
16
Inappropriate
*Year of median starting to smoke
F-ll
-------
Calculations attributing increases in lung cancer to a
single cause, such as smoking, ignore the multicausal nature
of carcinogenesis and possible interactions with air pollution
or other factors. Although there is little doubt that cigarette
smoking has played a major causative role in the increase in
lung cancer, not all lung cancer, even among those who smoke,
can be attributed solely to cigarettes.
The discrepancy noted between the trends in lung cancer
mortality rates for U.S males (rate of increase now decreasing)
and U.S. females (rate of increase now increasing) has been
suggested as being incompatible with the argument that air
pollution has a major influence on lung cancer rates. These
trends are said to be more consistent with changes in cigarette
consumption (with a 20-year lag period) (Doll and Peto 1981).
However, rates in black women, who smoke less and who in general
started smoking at a later age, are almost identical with rates
in white women—and have increased equally rapidly.
Schneiderman (1978) attempted to account for the effects
of smoking on trends in cancer rates by estimating the proportion
of lung cancer, as well as several other types of cancer, that
could be attributed to cigarette smoking at different time
periods. When this proportion was subtracted from the total,
he found that there had been a substantial increase in the
residual lung cancer rate, i.e., the fraction of lung cancers
attributable to factors other than smoking, between 1947 and
1969-1971. More recently, Schneiderman (1979), using the data
F-12
-------
of Pollack and Horm (1980), to calculate the increases between
the Third National Cancer Survey and the 1976 SEER survey in
lung cancers not related to smoking, found that the fraction
of lung cancers not attributable to smoking had risen substan-
tially during that period. Schneiderman's methodology is,
however, deficient in at least two respects: (1) he attributed
all "interaction-with-smoking" cancers to smoking alone, and
(2) he neglected cohort effects.
Several more sophisticated attempts have been made to
take cohort effects into account in looking at the time trends
in lung cancer. In one of these, Manton et al. (1982) commented
on their own findings and those of two other published studies:
These results suggest that, at most, we can attribute
between 79 and 92 percent of the increase (from 1950
to 1977) in U.S. white male lung cancer mortality
to corresponding increases in cigarette consumption.
For U.S. white females the pattern is less obvious
with between 62 and 100 percent of the increase in
lung cancer as the maximum attributable to smoking.
Manton cited two cohort studies of British data (Townsend 1978,
Stevens and Moolgavkar 1979) that showed attributable risks for
males at 94% and 89%, and for females at 71% and 94%, respectively,
It was not clear if these attributions were percentages of
total lung cancers, or percentages of changes.
Two additional cohort studies have been recently published
(Osmond and Gardner 1982, Janis 1982). The study by Osmond
discussed lung cancer in women (and bladder cancer in men)
and noted
...that women started smoking later than men is
reflected in the later position of the peak cohort
for lung cancer, 1925/6 rather than 1900/1. Numbers
F-13
-------
of cigarettes smoked by successive generations of
either sex (in the U.K.) have not declined to any
great extent, raising the question as to what has
caused lung cancer decreases (in younger persons).
Reduction of tar content of cigarettes has been
suggested (Doll and Peto 1981), but not unanimously
accepted (Gerstein and Levison 1982). Alternatively,
reductions of air pollution may have been important.
Janis noted that the peak cohort for British and U.S. (white)
males was the same (1900); this implies temporal similarities
in cigarette-smoking patterns in the two countries, which in
turn raises questions as to why age-standardized rates of lung
cancer have begun to fall in Great Britain, but not in the
United States. These several studies raise doubts about the
cohort effect (reflecting between-cohort differences in cigarette
smoking patterns) as the sole reason for the continuing increase
in lung cancer mortality in the United States.
The Manton data, however, indicated a possible U.S. peak
cohort born later than 1891-1900, although at the time of the
Manton review the peak rate had occurred in white men born
about 1900. In contrast to Janis, Manton found that the highest
"susceptibilities" were in the youngest cohort, but that the
rates for these men, in turn, were likely to be modified (down-
ward) by decreasing proportions of regular smokers and by changed
(lower tar) cigarettes. No studies of cohort effects in black
males, who currently have a 40% higher lung cancer mortality
rate than white males despite lower (tar-weighted) cigarette
consumption, have come to our attention.
Janis (1982) reported an independent "year" effect (i.e.,
a temporal effect not associated with a specific cohort effect)
F-14
-------
with increasing risk year-by-year. Manton's model has an opera-
tional counterpart in a measure of "susceptibility." For each
succeeding cohort Manton found increasing "susceptibility"
over time in both men and women. A possible explanation of the
findings of both Janis and Manton is an interaction among envi-
ronmental or industrial pollutants that may have increased
over time, giving an appearance of a "year" effect (or increased
susceptibilities of cohorts). Janis also noted that British
lung cancer rates rose more rapidly than U.S. rates, and have
now begun to fall more rapidly. This, too, suggests an inter-
action with general air pollution (higher in Great Britain),
which has sharply abated in Britain (since the 1950s-1960s).
As noted earlier, U.S. lung cancer rates have not been as high
as British rates, particularly at older ages. Consistent with
the cigarette smoking explanation is the rapid decline in lung
cancer mortality (relative to continuing smokers) after cessation
of smoking. That conditions in Britain are not strictly compar-
able to those in the United States is suggested by the fact
that, among British physicians who have stopped smoking, lung
cancer mortality rates appear to level off (after 15 or more
years cessation) to about twice those of nonsmokers (Doll and
Peto 1976), whereas in the United States it has been reported
that the rates of stopped smokers, after 15 years of not smoking,
reach those of men who never smoked (Wynder et al. 1970).
A recent report of the National Academy of Sciences/National
Research Council (Gerstein and Levison 1982) raised substantial
F-15
-------
doubts about the positive health effects of reduced tar/nicotine
cigarettes. The report concluded
...while some large scale studies have suggested
small gains in health due to using lower T/N (or
filter rather than non-filter) cigarettes, other
population-wide studies do not support this view.
Thus/ the evidence for switching to lower T/N cig-
arettes is doubtfulT"(Emphasis original)
Calculations based on the National Cancer Institute data
for 1973-1977 (SEER), which did not include cohort effects,
suggested that less than 20% of the increased incidence in
cancer in white males, and less than half the increased inci-
dence in white females, were attributable to cigarette smoking
(Schneiderman 1978). These estimates did not take into account
interactions or the reduced proportion of all adults smoking
cigarettes and the reduced tobacco and tar content of the ciga-
rettes sold since 1965 (USDHEW 1979).
The increase in lung cancer incidence and mortality during
the 1970s is of particular interest. Such a change is consistent
with an increase in exposure to some environmental factor or
factors other than smoking during the 1940s or early 1950s.
As noted by Rail (1978), Epstein (1978), and Davis and Magee
(1979), this is the period of the initial rapid growth in the
synthetic organic chemical production, as well as a period
of increased activity in other industries, including the use
of asbestos.
Evidence that there have been increases in lung cancer
independent of smoking habits was given by Enstrom (1979),
who studied lung cancer mortality rates for nonsmokers in the
F-16
-------
United States. He found that these rates had risen considerably
between 1914 and 1968, especially in the oldest age categories
and appear to have doubled during the period between 1958 and
1968. This finding was questioned by Doll and Peto (1981)
on the grounds that Enstrom may have included ex-smokers in
his nonsmoker category. Enstrom's finding is in contrast that
of Garfinkel (1981) who reported no such increase in the pop-
ulation followed by the American Cancer Society.
Garfinkel also cited a similar result from the nonsmokers
in the Dorn study of veterans (Rogot and Murray 1980). On
closer examination, however, both these sets of data exhibit
peculiarities (or fluctuations), due to small numbers or possibly
to reporting errors. Following specific birth cohorts, three
of Garfinkel's groups of male nonsmokers (persons born about
1916, 1901, and 1886) showed declines in age-specific rates
in the third time period—to levels in the 1916 and 1901 cohorts
below, those shown by any of the other cohorts at the same attained
age. (The 1886 cohort could not be used in this comparison
because other cohorts had not attained ages 85-89.) This is
contrary to the general pattern of increase in cancer mortality
rates with increasing age (except for the very oldest persons).
Excluding these aberrant points, which suggest that recent
follow-up may have been incomplete, each succeeding cohort
of males shows a higher lung cancer rate (at the same attained
age) than the preceding cohorts—with only one exception:
men born about 1896 had lower rates at ages 70-74 than did
F-17
-------
men born about 1891 (26.4 vs. 32.3). The data for the women
in the ACS study show similar patterns (with the 1916 cohort
also showing an unexpected inversion in the last follow-up
period). The rates for women nonsmokers/ which are based on
larger numbers/ are otherwise more consistent than those for
men. The Dorn data are also erratic. The 1901 cohort has
lower lung cancer rates reported for ages 60-64 than for ages
55-59. Except for this and one other data point (men born
about 1896, attained age 65-69), the men reported in the Dorn
data show somewhat higher rates for the same birth cohorts
and for the same attained ages than the ACS study. This is
in keeping with the nature of the ACS sample—somewhat less
urban, somewhat less "blue-collar", somewhat higher education
and social class than the United States as a whole. The Dorn
population, while derived only from men healthy enough to have
been in the military, is likely to be closer to the general
U.S. population.
It is worth noting that Dean et al. (1978) also reported
substantial increases in rates among nonsmokers. In contrast,
Doll (1982) apparently assumed no change over time in lung
cancer mortality among nonsmokers in the United States from
1933 to 1977. This is rather surprising because in his Figure 1
(page 224) in which he plotted rates for nonsmokers (age-adjusted)
for 1960-1972 (from Hammond), the nonsmoker rates for several
of the early years are higher than the rates for the total
F-18
-------
populations, also age-standardized--considered separately by
sex.
Attempts have been made to study the trends in cancer
mortality rates following apparent reductions in pollution.
Higgins (1974) was able to account for increases in lung cancer
in the United States and England up to about 1970 by changes
in smoking habits. He found more recent rates inconsistent
with cigarette smoking. He attributed the decline in lung
cancer rates in England, which began as early as 1960, to the
dramatic reduction in air pollution. This relationship is
supported by the finding that the earliest (and greatest) reduc-
tion in lung cancer rates occurred in London where there was
also the earliest and greatest reduction in measured air pollu-
tion. A similar conclusion appears to have been reached by
Lawther and Waller (1978), who found that the lung cancer trends
from 1951 to 1973 in Greater London and the rural districts
of England and Wales were moving in opposite directions. The
rates declined in London, where the Clean Air Acts had been
first put into effect, while they were increasing in the rural
areas. Todd et al. (1976), in analyzing cancer mortality rates
and cigarette consumption in England, found additional evidence
supporting the hypothesis that atmospheric pollution interacted
with cigarette smoking to increase the incidence of lung cancer.
They argued that the finding that the male cohorts with the
highest "cumulative consumption of constant tar cigarettes"
were 5 or 10 years younger than those that experienced the
F-19
-------
highest age-specific lung cancer mortality rates (at all ages
between 30 and 59 years) implied the existence of etiological
agents (in addition to cigarette smoking) that influence the
development of lung cancer in humans.
F-20
-------
APPENDIX G
CRITIQUE OF TWO RECENT REVIEWS
This Appendix discusses two recent reviews which have
concluded that the contribution of air pollution to cancer
risks is small and/or indeterminable. Doll and Peto (1981)
presented a comprehensive review of data on cancer rates in
the U.S. population and their known or presumed association
with various environmental factors. Their final conclusion
(Table 20) was that about 2% of all cancer deaths in the U.S.
(possible range, less than 1% to 5%) could be attributed to
pollution of all kinds. This estimate appears to include about
1% attributed to the effect of urban air pollution on lung
cancer (p. 1248). Although this estimate is consistent with
others reviewed in this report (see Table IV-1), Doll and Peto
expressed considerable reservation about the reliability of
these estimates and the methods used to derive them.
The precise basis of Doll and Peto's conclusions is diffi-
cult to determine from their paper. In their section on air
pollution (pp. 1246-1248) they cited no specific epidemiological
studies of the association between cancer rates and any specific
pollutants, and only two studies of urban/rural differentials.
One of these was their own unpublished study of British doctors,
presented in a footnote (see Section II.B of this report for
discussion). The other was the paper by Hammond and Garfinkel
(1980): they cited this paper as demonstrating an urban/rural
G-l
-------
differential after standardizing for age and six categories
of current smoking. They then added:
These differences do not allow for differences attri-
butable to occupational hazards but even so are
not large, and much or all of them might be due
to the expected effects of early cigarette usage.
The authors allowed for occupation by examining
separately men exposed and not exposed to dust/
fumes, etc. and concluded that their data offer
"little or no support to the hypothesis that urban
*air pollution has an important effect on lung cancer."
It is evident from these statements that Doll and Peto had not
conducted an independent analysis of these data (cf. Appendix E)
Doll and Peto expressed considerable skepticism about
the possibility of detecting effects of urban air pollution
(or other regional effects):
Some investigators have attempted to estimate the
effect of pollutants by comparing the lung cancer
mortality rates in different areas and "making allow-
ance" for differences in smoking habits by retrospec-
tive inquiry of the amount smoked by representative
residents. We doubt, however, whether it is possible
in this way to disentangle the effects of smoking
and environmental pollution, especially in those
studies that have examined cancer rates only within
categories of men with such broadly similar smoking
habits as nonsmokers (including ex-smokers), current
smokers smoking 20 cigarettes a day or less, and
current smokers smoking more. Such broad classes
are hardly likely to take account of differences
in a habit which may affect the incidence of lung
cancer by up to fortyfold sufficiently accurately
for a twofold urban-rural difference to be estimated
with certainty.
They continued by pointing out the difficulty of controlling
for other aspects of smoking, including age at starting, type
of cigarette, depth of inhalation, etc. (see Chapter II).
However, their discussion of urban/rural differences in these
aspects of smoking was speculative, and they did not cite any
G-2
-------
specific data (such as those of Haenszel et al . included in
this report as Table II-4) on urban/rural differentials in
these aspects of smoking. They did not cite the study of Dean
et al. (1977, 1978) in which these factors were measured,
reported, and controlled for.
Much of Doll and Peto's skepticism about the role of air
pollution appears to stem from their conclusion that cigarette
smoking can account for most, if not all, of the geographic
and temporal patterns in lung cancer rates. (They did not
discuss effects of air pollution at sites other than the lung.)
They estimated that as much as 91% of lung cancer in males
and 78% of lung cancer in females was attributable to cigarette
smoking. These figures are higher than most other estimates,
and the method used for arriving at them is subject to upward
bias. Specifically, Doll and Peto used the data from the ACS
survey (Garfinkel 1981) to estimate lung cancer rates in non-
smokers, used these rates to estimate the number of lung cancers
that would have occurred in the United States without smoking,
and attributed all the rest to smoking. However, as pointed out
earlier, the ACS survey was a biased sample of the U.S. popula-
tion. Doll and Peto recognized this bias in their calculation
of risks due to alcohol (Table 11) and occupation (p. 1244),
for which they estimated that the ACS sample underestimated
national risks by factors of 2.0 and 3.3, respectively. However,
they did not take any account of this bias in their estimate
G-3
-------
of smoking risks. Also, Doll and Peto's procedure would include
all interactions in the category of cancers attributed to smoking.
Doll and Peto's actual numerical estimate of the fraction
of cancers attributable to air pollution appears to be derived
from the study of Pike et al. (1975) and the more informal
review by Cederlof et al. (1978), both of which led to
the conclusion that atmospheric pollution, in conjunc-
tion with cigarette smoke, might have contributed
to some 10% of all cases of lung cancer in big cities
(and so to a few percent of lung cancer in the country
as a whole, i.e., about 1% of all cancer).... These
crude estimates provide the best basis for the forma-
tion of policy.
Doll and Peto did not review the other studies listed in Table IV-1
in this report, and did not consider the point made in Chapter IV,
that extrapolation from data on persons exposed to high concentra-
tions of products of incomplete combustion, using BaP as an
index, yields estimates only of the fraction of lung cancers
associated with these components of air pollution, and not
with other components.
In summary, Doll and Peto's conclusions about air pollution
were informal and do not appear to be based on a critical review
of the limited literature which they cited.
Shy and Struba (1982) presented another review of scientific
evidence on the association between air pollution and cancer.
They recognized the existence of four of the "converging lines
of evidence" that have been reviewed in this report: the unex-
plained urban factor, the known carcinogenic effects of combustion
products in workers occupationally exposed to high concentrations,
G-4
-------
the geographic correlations between lung cancer rates and some
indices of air pollution, and the presence of carcinogenic sub-
stances in ambient air. However, they concluded:
In spite of these converging lines of evidence, we
will argue in this section that firm conclusions
about air pollution and lung cancer are simply not
warranted by the current state of knowledge. Serious
deficiencies exist in making even qualitative esti-
mates of persons exposed or not exposed to atmospheric
carcinogens. Analytic (individual risk) studies of
air pollution as a human carcinogen have not yet
been reported, and none of the epidemiologic studies
allows one to make a direct link between lung cancer
incidence and exposure to air pollution. The support-
ing arguments for this judgment will be given as we
review the epidemiologic evidence in the following
parts of this section.
Although Shy and Struba cited more studies of the associa-
tion between air pollution and cancer rates than Doll and Peto,
they nevertheless listed only a limited number of papers, and
did not cite the studies that we regard as individually most
persuasive (e.g., Haenszel and Taeuber 1964, Dean et al. 1978,
Hammond and Garfinkel 1980). They dismissed studies of urban/rural
differentials with the following incorrect statement:
Thus far, none of the studies provide even qualitative
estimates of personal exposure to ambient air pollu-
tion, and all lack any quantitative data whatsoever
on carcinogenic levels in the ambient air.
As noted in the text, they dismissed as "extremely low" a calcu-
lated risk from ambient concentrations of BaP that actually
falls within the range of other estimates (see Table IV-1).
Although Shy and Struba's critical approach to the studies
they cited is appropriate, their standards of proof seem unreason-
ably high:
G-5
-------
It would seem essential, in future epidemiologic
studies, to identify cohorts exposed to specific
classes of suspected atmospheric carcinogens, such
as formaldehyde in particle board, plastic vapors,
indoor cigarette smoke, classes of solvents in closed
environments, motor vehicle diesel exhaust, and
so on. Many of these exposure situations may be
best studied in an occupational setting, but the
characterization of chemical species and dose will
be difficult in any environment. General population-
based studies do not promise satisfactory results,
owing to the heterogeneity of exposure and lack
of individual data on confounding factors in most
such studies.
...The proposed approach for advancing our knowledge
in this area is to define individual exposure to
specific sources of atmospheric carcinogens, to
attempt to characterize this exposure in terms of
specific organic chemical classes of compounds, and
to use these exposure characterizations as a basis
for well-designed analytic epidemiologic studies. It
is hoped that this approach will yield more testable
and refutable hypotheses than have been developed
to date.
Their insistence on rigorous, analytic (apparently prospective
and long-term) studies reflects a reluctance to consider the
weight of evidence provided by the large body of literature on
this subject, much of which they did not cite.
G-6
-------
APPENDIX H
DATA ON SMOKING HABITS IN NORTHEASTERN ENGLAND
Tables H-l to H-4 summarize data on three characteristics
of smoking habits (age at starting to smoke, depth of inhalation,
and proportion of filter cigarettes), stratified by age, sex,
and location of residence. These data were derived from a
survey in northeastern England and were originally published
as Tables H17, HIS, H21, and H24 in Dean et al. (1978).
H-l
-------
TABLE H-l
DISTRIBUTION OF AGE AT STARTING TO SMOKE
BY AREA AND SEX IN THE LIVING POPULATION, 1973
Eston
Male
Female
Stockton
Male
Female
Rural Districts
Male
Female
Number
Age at
<15
15-19
20-24
25+
Smokers
35+ 7,230
starting to smoke
18.3
43.0
12.6
5.9
2.1
7,570
7.6
21.6
10.5
10.3
1.4
18,370
14.5
41.5
11.5
8.2
5.4
20,460
4.7
23.1
8.9
13.2
1.9
15,380
11.7
36.3
10.3
5.7
6.9
16,510
2.2
17.8
8.3
8.7
1.7
unclassified
Never smokers 18.1
48.5
19.0
48.2
29.1
61.4
H-2
-------
TABLE H-2
DISTRIBUTION OF AGE AT STARTING TO SMOKE
BY AREA AND SEX IN THE LIVING POPULATION, 1973
Eston
Male
(%)
Number 35-44
Age at starting
<15
15-19
20-24
25+
Smokers,
unclassified
Never smokers
2,270
to smoke
15.5
45.0
14.0
1.6
1.6
22.5
Number 45-54 2,070
Age at starting
<15
15-19
20-24
25+
Smokers ,
to smoke
14.9
50.4
10.7
6.6
1.7
Female
(%)
1,980
11.7
37.9
15.2
4.1
0.7
30.3
2,080
12.4
25.5
11.7
8.8
2.9
Stockton
Male
(%)
4,950
9.4
47.2
11.0
' 4.7
8.7
18.9
5,680
11.8
45.7
9.4
7.1
5.5
Female
(%)
5,220
8.1
30.4
8.9
12.6
1.5
38.5
5,420
5.3
31.8
13.6
7.6
1.5
Rural Districts
Male
(%)
4,750
8.6
39.5
11.1
5.6
4.3
30.9
3,950
13.4
37.3
12.7
4.2
7.7
Female
(%)
4,460
1.2
27.2
10.1
8.9
1.2
51.5
3,920
5.4
25.9
10.9
6.1
2.7
unclassified
Never smokers 15.7
38.7
20.5
40.2
24.6
49.0
H-3
-------
TABLE H-2 (continued)
Eston
Number 55-64
Age at starting
<15
15-19
20-24
25+
Smokers,
unclassified
Never smokers
Number 65+
Age at starting
<15
15-19
20-24
25+
Smokers,
Male
(%)
1,910
to smoke
18.6
43.3
15.5
8.2
3.1
11.3
980
to smoke
28.4
27.0
9.5
9.5
2.7
Female
(%)
1,720
2.3
11.5
13.8
19.5
2.3
50.6
1,790
0.9
4.3
0.9
12.9
0.0
Stockton
Hale
(%)
4,150
20.0
30.0
11.1
12.2
3.3
23.3
3,590
20.5
38.6
15.7
10.8
2.4
Female
(%)
4,330
5.2
17.7
6.3
22.9
2.1
45.8
5,490
0.0
9.8
5.7
12.3
2.5
Rural Districts
Male
(%)
3,420
10.8
35.1
7.2
7.2
8.1
31.5
3,260
14.8
31.5
9.3
6.5
8.3
Female
(%)
3,770
1.6
12.6
8.7
10.2
1.6
65.4
4,360
0.6
4.4
3.8
9.5
1.3
unclassified
Never smokers 23.0
81.0
12.0
69.7
29.6
80.4
H-4
-------
TABLE H-3
DISTRIBUTION OF DEPTH OF INHALATION BY DISTRICT AND SEX
IN THE LIVING POPULATION, 1973
Eston
Male
Female
Stockton
Hale
Female
Rural Districts
Male
Female
Number 35+
Inhalation
A lot
7,230
category
36.8
A fair amount 17.6
A little
None
Smokers,
15.7
10.0
1.9
7,570
17.5
10.3
14.0
9.5
0.2
18,370
29.0
19.7
12.9
12.9
6.6
20,460
12.2
11.8
13.8
12.8
1.2
15,380
22.8
15.5
13.4
14.5
4.8
16,510
9.0
11.6
10.0
6.8
1.2
unclassified
Never smokers 18.1
48.5
19.0
48.2
29.1
61.4
H-5
-------
TABLE H-4
PROPORTION OF MANUFACTURED-CIGARETTE SMOKERS
WHO SMOKE FILTER CIGARETTES—BY AREA, SEX AND PERIOD
FOR WHICH SMOKING HABITS REPORTED
Eston Stockton Rural Districts
Male Female Hale Female Hale Female
Filter Smokers (%) (%) (%) (%) (%) (%)
Current 60.5 83.6 68.6 86.9 74.8 88.0
3-5 years ago 52.4 75.5 61.9 76.4 69.4 83.6
6-10 years ago 33.3 51.3 38.4 63.4 52.7 71.2
>10 years ago 9.9 23.6 18.2 35.4 30.8 45.8
U.S. Envfronmentaf Protection Agency
Region V, Library
230 South Dec,,:p-i "''oet
Chicago, mine's 6 '
H-6
------- |