ray may be normal, or rounded nodules, less than 10 mm in
diameter, may be present, predominantly in the upper lobes. These
findings characterize the simple form of coal workers’ pneumoco-
niosis. In spite of the presence of roentgenographic and pathologic
abnormalities, only subtle abnormalities of small airways function
are demonstrable in simple coal workers’ pneumoconiosis (Morgan
1984a).

In certain chronic occupational lung diseases, parenchymal lung
injury may be accompanied by evidence of restriction alone; in
others, variable combinations of restriction and obstruction may
occur. Relevant examples of these two types of processes are
asbestosis (Seaton 1984a) and the complicated forms of coal workers’
pneumoconiosis and silicosis (Morgan 1984a; Seaton 1984b).

Although asbestos exposure is associated with fibrosis of the
respiratory bronchioles, the injury often progresses and involves the
alveolar interstitium with the development of parenchymal fibrosis
(Seaton 1984a). The clinical consequences of this parenchymal injury
are cough and dyspnea. Other changes found in asbestosis include
crackles, clubbing, and basilar, irregular, linear opacities on chest x
ray. Pulmonary function testing shows only a restrictive pattern
with reduced FVC, normal FEV/FVC%, and decreased TLC.

In contrast, the complicated forms of silicosis and coal workers’
pneumoconiosis may be accompanied by obstruction in addition to
restriction. In both disorders, large masses of dust and fibrosis
replace the normal lung parenchyma and reduce FVC and TLC.
Obstruction may also be present, presumably because of increased
airways resistance and parenchymal abnormalities. Dyspnea is
generally a prominent symptom.

Thus, for some occupational agents, the associated lung injury at
specific anatomic loci resembles that from cigarette smoking. Large
airway irritation, regardless of the exposure, is accompanied by
abnormalities of the mucous glands and mucus hypersecretion.
Small airways may be affected by occupational agents, and a pattern
of injury distinct from that found in cigarette smokers has been
described (Churg et al. 1985). However, the parenchymal abnormali-
ties of advanced pneumoconiosis can be readily distinguished from
emphysema associated with cigarette smoking.

Methods for Evaluating the Effects of Occupational Exposures
on the Lungs

Workers exposed to occupational agents that cause chronic lung
disease may be examined for diagnostic reasons, for surveillance, or
for research. Regardless of the purpose of the evaluation, the same
assessment techniques are generally used: history of respiratory
symptoms, physical examination of the chest and extremities,

151
spirometry and other physiological tests, and chest x ray (American
Thoracic Society 1982b; Boehlecke 1984; Townsend and Belk 1984).
These techniques and their sensitivity to the effects of cigarette
smoking are described below.

History of Respiratory Symptoms

Symptoms of lung disease are nonspecific; the most prevalent are
cough, phlegm production, wheezing, and breathlessness or dyspnea
(Gandevia 1981). Although a physician may take a conventional
history to evaluate these symptoms, standardized questionnaires are
generally used for surveillance and research purposes.

In the 1950s, the British Medical Research Council developed a
standardized respiratory symptoms questionnaire for studies of the
epidemiology of chronic bronchitis and chronic obstructive lung
disease (Samet 1978; Florey and Leeder 1982). In 1968, this question-
naire was adopted for use in the United States by a committee of the
American Thoracic Society (1969). Three years later, the National
Heart and Lung Institute made available a version that had been
modified to improve its suitability for the United States (US DHEW
1971). In 1978, the American Thoracic Society published a further
revised respiratory symptoms questionnaire (Ferris 1978). The
Medical Research Council questionnaire or one of these modified
versions has been used in most studies of chronic lung disease in the
workplace. All include a series of questions related to cough, phlegm,
wheezing, and dyspnea.

The Medical Research Council questionnaire was originally devel-
oped for investigating the etiology of chronic bronchitis and airflow
obstruction (Fletcher et al. 1959; Samet 1978). The questionnaire was
designed, in part, to test one of the prevailing hypotheses about
airflow obstruction: that mucus hypersecretion predisposed repeated
lower respiratory tract infections and consequent airflow obstruction
(Fletcher et al. 1959, 1976). Accordingly, the cough and phlegm
questions were worded to be sensitive to the earliest phases of mucus
hypersecretion, a condition largely attributable to cigarette smoking
(US DHHS 1984). The questions may be less satisfactory for cough
and sputum associated with other exposures, particularly if those
other exposures produce a pattern of symptoms different from those
due to cigarette smoking, such as nocturna! cough or episodic cough.
Further, their sensitivity to cigarette-associated mucus hypersecre-
tion may hinder separation of an occupational exposure’s effect on
the occurrence of cough and phlegm from that of cigarette smoking.
The dyspnea and wheeze questions probably do not share this
sensitivity.

In population surveys, cigarette smoking is the major determinant
of the prevalence of cough and phlegm (US DHHS 1984). This
association has been confirmed in occupational groups as well as in

152
population samples (Gandevia 1981; US DHHS 1984; Petersen and
Castellan 1984). Wheezing is also associated with cigarette smoking
(Mueller et al. 1971; Samet et al. 1982; Schenker et al. 1982).
Dyspnea has multiple determinants that interact in a complex
fashion; cigarette smoking and smoking-related impairment of lung
function contribu.e to the occurrence of dyspnea (Wasserman and
Whipp 1975; Cotes 1979; Killian and Jones 1984).

Chest X Ray

The pneumoconioses are associated with characteristic radio-
graphic abnormalities, although the chest film may be normal in the
presence of biopsy-proven disease (Epler, McLoud et a]. 1978). A
conventional clinical interpretation is usually sufficient for estab-
lishing the presence of pneumoconiosis. Preferably, however, the
chest x ray should be coded according to the classification estab-
lished by the International Labour Office (ILO) (1980). This system,
originally published in 1950, categorizes the types of abnormalities
on the chest x ray by shape and size, and provides a grading (the
profusion) for describing the density of small opacities. The opacities
classified as small are grouped as rounded or irregular. If the
opacities are less than 1 cm in diameter, they are called small; if
equal to or greater than 1 cm, they are called large.

The effects of cigarette smoking on chest x-ray findings have been
examined, using both conventional interpretations and readings in
the ILO system. Human autopsy evidence and animal exposure
studies show that cigarette smoking leads to abnormalities in the
airways and parenchyma that might produce radiographic abnor-
malities (US DHEW 1979b; Weiss 1984). However, these changes are
subtle in comparison with the pathological findings in the pneumo-
conioses. Cigarette smoking is associated with modest amounts of
interstitial fibrosis in the lungs, in addition to airways abnormalities
and emphysema (US DHEW 1979b; Weiss 1984). For example,
Auerbach and colleagues (1974) examined lung sections from 1,443
men and 388 women deceased between 1963 and 1970, and found
more fibrosis in smokers than in nonsmokers and a dose-response
relationship between the degree of fibrosis and the amount smoked.
The small airways of cigarette smokers, even at young ages, display
inflammation with edema of the bronchiolar walls, smooth muscle
hypertrophy, and goblet cell metaplasia (US DHHS 1984). These
changes may underlie, at least in part, the pattern of increased lung
markings in smokers described anecdotally by clinicians, but are
unlikely to be confused with the more extensive fibrosis found in
moderate or advanced pneumoconiotic lung disease.

Comparisons of chest x-ray findings generally show a higher
frequency of abnormalities, interpreted as representing interstitial
fibrosis, in smokers than in nonsmokers. These investigations have

153
been based on chest films from both the general population and
specific occupational groups. Weiss (1967, 1969) reviewed chest films
from two samples of adults—participants in a tuberculosis screening
program and hospital employees. In both groups, he identified a
pattern of increased lung markings, termed diffuse pulmonary
fibrosis, more often in smokers, and showed that the prevalence of
this finding increased with the amount and duration of smoking.
These studies have been criticized because the films were 70 mm
photofluorograms taken for screening purposes and not full sized
(Kilburn 1981). Further, the films were not read directly according to
the ILO classification. In another study that did not use the ILO
system, Carilli and colleagues (1973) showed that radiologists could
generally distinguish smoking women from nonsmoking women by
the presence of linear and nodular fibrotic changes in the smokers.
Epstein and colleagues (1984) read the chest x rays of 200 hospital-
ized patients according to the ILO classification. Twenty-two patients
with at least category 1/0 profusion and no documented dust
exposure or other explanation for nodular densities were identified,
10 of whom had not smoked cigarettes. Because this study included
only hospitalized people, the results may not be generalizable to
working populations.

The results of investigations involving occupational groups do not
show strong effects of cigarette smoking on the profusion of small
opacities. Glover and colleagues (1980) read the chest films of slate
workers and a nonexposed control group according to the 1971 ILO
classification. In the controls, small irregular opacities were not seen
in nonsmokers, but were present in 2 percent of current and former
smokers. Investigators from the National Institute for Occupational
Safety and Health interpreted chest x rays of 1,422 blue-collar
workers whose present and past employment should not have
involved exposure to respiratory hazards (Castellan et al. 1984). Only
three workers had at least category 1/0 profusion, two with small
rounded opacities and one with small irregular opacities. Sixty-one
percent of the subjects were current or former smokers. However,
the mean age of subjects in this study was only 33.9 years,
substantially lower than the age at which pneumoconiosis or
significant cigarette-related airflow obstruction would generally be
manifest if exposure began at about age 20. In a much smaller study
of similar design, Cordier and colleagues (1984) identified small
opacities in only 1 person in a control group of 48 office workers, 31
percent of whom smoked.

Studies of workers exposed to hazardous agents show that
cigarette smoking may modify the pattern of radiographic abnormal-
ity. In coal workers, small rounded opacities predominate in the
simple phase of coal workers’ pneumoconiosis, but irregular opaci-
ties may also be present (Amandus et al. 1976; Cockcroft et al. 1982,

154
1983). The irregular opacities are associated with cigarette smoking
and with reductions of FEV,, FVC, and diffusing capacity (Cockcroft
et al. 1982). In autopsy specimens obtained from coal workers in the
United Kingdom, Ruckley and colleagues (1984) demonstrated that
emphysema was present in 90 percent of the lungs with small
irregular opacities, but in only 60 percent with small rounded
opacities alone. Dick and colleagues (1983) examined the radiographs
of a stratified random sample of miners from 10 British coal mines
and concluded that smoking did not influence the prevalence of
rounded opacities. Smokers had a greater prevalence of irregular
opacities, but after adjusting for the effects of differences in age and
dust exposure, these results were not statistically significant.

Studies of other occupationally exposed groups also demonstrate
that cigarette smoking may affect the pattern and extent of
radiographic abnormality. In granite workers, Theriault and col-
leagues (1974) found that rounded opacities were related to an
estimate of lifetime dust exposure, whereas small irregular opacities
were more strongly related to smoking. In workers exposed to
manmade vitreous fibers, the prevalence of small opacities was
determined not only by estimated exposure but also by smoking
habits (Weill et al. 1983). Using multiple logistic regression, Peters
and colleagues (1984) showed that cigarette smoking, but not
particulate exposure, predicted the occurrence of linear opacities in
silicon carbide workers. In asbestos workers, the predominance of
evidence indicates that cigarette smoking acts independently and
additively with asbestos to create radiographic abnormalities (Weiss
1984).

The findings of these studies of occupationally exposed and
nonexposed individuals indicate that cigarette smoking may affect
chest x-ray readings. Cigarette smoking alone is occasionally associ-
ated with definite abnormalities classified in the ILO system.
Smoking may also affect the radiographic pattern and independently
increase the prevalence of abnormality. In addition, the threshold
for detection of an abnormality on chest x ray may be exceeded more
frequently or at an earlier age in workers who smoke than in
workers who do not smoke.

Physiological Assessment

An evaluation of workers for diagnosis and surveillance may
include auscultation of the chest, for breath sound quality and
intensity and for the presence of adventitious sounds including
crackles, and examination of the fingernails for evidence of clubbing.
Crackles, also referred to as rales or crepitations, are discontinuous,
interrupted sounds thought to arise from the sudden opening of
small airways or from the bubbling of air through secretions in
larger airways (Loudon and Murphy 1984). Fine crackles may be

155
heard in people with diffuse interstitial fibrosis. For example, Epler,
Carrington, and colleagues (1978) reported that fine crackles were
present in 60 and 65 percent of subjects with biopsy-proven and
clinically diagnosed asbestosis, respectively. Some definitions of
asbestosis incorporate the presence of crackles as a diagnostic
criterion (Murphy et al. 1978). Because crackles may be heard in
asbestosis and other occupational lung diseases, auscultation has
been advocated as a surveillance technique for monitoring workers
exposed to asbestos and other agents (Loudon and Murphy 1984;
Murphy et al. 1984).

Few studies have addressed the effects of cigarette smoking on
auscultatory findings, however. Epler, Carrington, and colleagues
(1978) reported the results of a conventional clinical auscultation of
patients with various interstitial disorders or with chronic obstruc-
tive lung disease, which is largely attributable to cigarette smoking.
Fine crackles, characteristic of asbestosis, were heard in only 10 to
12 percent of the latter group, though coarse crackles were more
common in those with chronic bronchitis. Two studies of asbestos
workers suggest that cigarette smoking may independently increase
the frequency of crackles. To quantify the separate effects of asbestos
exposure and cigarette smoking on the prevalence of bilateral fine
crackles, Samet and colleagues (1979) analyzed data from 409 survey
subjects, using multiple logistic regression. Statistically significant
effects of both smoking and asbestos exposure were found. In the
other study (Murphy et al. 1984), a technician examined each subject
with a standardized approach and a summary crackles score was
calculated. Multivariate analysis suggested that cigarette smoking
was associated with the lower abnormality levels of this score. The
consistent findings of these two investigations seem plausible in view
of the effects of cigarette smoking on the small airways, the site
where fine crackles are presumed to originate (Loudon and Murphy
1984). In 590 employed men not exposed to respiratory hazards,
crackles were heard predominantly in the older smokers (Gandevia
1981). This finding further supports a relationship between cigarette
smoking and the presence of crackles.

Clubbing refers to a change in the configuration of the nail beds,
which can be best quantitated by the hyponychial angle (Regan et al.
1967). It has many causes and is a nonspecific manifestation of
advanced chronic respiratory diseases, lung cancer, and other
disorders (Shneerson 1981). Because clubbing may be occasionally
found with COLD, its presence may be related to cigarette smoking
as well as to occupational lung disease. Samet and colleagues (1979)
found that cigarette smoking and occupational exposure to asbestos
were independent determinants of the prevalence of clubbing in four
different populations of asbestos workers.

156
Findings on clinical examination, like respiratory symptoms, are
nonspecific, and a conventional physical examination alone is an
insensitive method for diagnosing chronic occupational lung dis-
eases. However, the presence of fine crackles, in the setting of an
appropriate exposure, should alert the clinician to the possibility of
pneumoconiosis, even if the chest x ray is unremarkable. Clubbing,
when attributable to a chronic pulmonary process, is generally a
marker for more advanced disease. Diseases associated with ciga-
rette smoking may be accompanied by crackles or clubbing.

Evaluation of pulmonary function in occupationally exposed
individuals, whether for diagnostic or research purposes, should
include spirometry, which measures FVC, FEV,, and maximal
expiratory flow rates (Ferris 1978; American Thoracic Society
1982b). The effects of smoking on spirometric parameters are
discussed elsewhere in this chapter. The diffusing capacity for
carbon monoxide may also be measured; it is a sensitive test that
may detect early abnormalities in chronic occupational lung diseases
(Weinberger et al. 1980). As with FVC, FEV,, and other spirometric
measures, cigarette smoking habits must be considered in interpret-
ing the level of diffusing capacity, which is reduced by smoking-
related lung disease (particularly emphysema) as weil as by occupa-
tional lung disease (Make et al. 1982; Miller et al. 1983). FVC can be
reduced either by restrictive lung diseases, such as asbestosis, or by
COLD; therefore, TLC should be measured with a physiological or
radiological method in order to establish the presence of a restrictive
disorder. In evaluating subjects for occupational asthma, nonspecific
bronchial reactivity may be assessed with pharmacologic agents,
such as methacholine, or with cold air inhalation (Brooks 1982).
Some studies indicate that nonspecific bronchial reactivity is in-
creased in cigarette smokers (Kabiraj et al. 1982; Gerrard et al.
1980), though others do not (Kennedy et al. 1984; Wanner et al.
1985).

Exercise testing is one of the methods used to assess the degree of
impairment resulting from a chronic occupational lung disease
(American Thoracic Society 1982a). Exercise testing has been used to
characterize the pathophysiology of chronic occupational lung
diseases, but is rarely used for establishing clinical diagnoses or for
epidemiological studies (Wiedemann et al. 1984) and is not discussed
further in this chapter. Cigarette smoking can impair exercise
performance through a variety of mechanisms (Cotes 1979).

157
Quantification of Effects of Smoking and Occupation in
Populations

Concepts of Interaction

Interaction has been defined as “the interdependent operation of
two or more causes to produce an effect” (Last 1983, p. 51).
Epidemiclogists may also apply the term “effect modification” to
variation in the magnitude of an exposure’s effect as the level of
another exposure changes (Last 1983). Synergism refers to an
increased effect of the exposures when both are present, and
antagonism refers to a reduced effect (Last 1983). Statistical model-
ing techniques are generally used to test for the presence and
direction of interaction. The most widely applied statistical tech-
niques measure interaction on either an additive or a multiplicative
scale (Rothman et al. 1980; Kleinbaum et al. 1982). Ideally, the
choice of a model should be based on a specific biological formulation
of disease pathogenesis; most often, however, the underlying biologi-
cal mechanisms are not well established and largely statistical
considerations govern the selection of an analytical model.

The results of such models must be interpreted not only statistical-
ly but also in biological and public health contexts (Rothman et al.
1980). Rothman and colleagues (1980) argued that biological models
should be explicitly described; in their view, the labeling of mecha-
nisms as synergistic or independent does not advance the under-
standing of disease etiology. They broadly described two categories of
mechanisms: those with the multiple etiological factors acting
interchangeably at the same step and those with the factors acting at
different steps. The corresponding statistical models are the additive
and the multiplicative, respectively. These authors and others (Blot
and Day 1979; Saracci 1980; Kleinbaum et al. 1982) have concluded
that, from the public health viewpoint, departure from additivity
represents interaction.

Both advancing the understanding of disease etiology and the need
for protecting public health provide a compelling rationale for
assessing interaction between cigarette smoking and workplace
exposures. Cigarette smoking may interact with a particular expo-
sure through diverse mechanisms that range from behavioral to
molecular levels (Table 4). The 1979 Report of the Surgeon General
(US DHEW 1979b) partially addressed different forms of interaction
between smoking and occupational exposures; other plausible hy-
potheses concerning interaction between cigarette smoking and
occupational agents can also be postulated. The interactions listed in
Table 4 are intended to be illustrative and not exhaustive.

Some consequences of cigarette smoking might lead to a reduction
of the dose of an inhaled agent. In comparison with nonsmokers,
current and former smokers have higher rates of absenteeism from

158
TABLE 4.—Some potential interactions between cigarette

smoking and occupational exposures in the
pathogenesis of chronic occupational lung

diseases

 

Source of interaction

Potentia] consequence

 

Increased absenteeism by smokers from work

Selection of more fit nonsmokers into
aerobically demanding jobs

Contaminated cigarettes act as a vector

Workplace chemicals are metabolized
to toxic or more toxic agents by
cigarettes

Increased tracheobronchial deposition of
particulates in smokers and people
with chronic bronchitis

Reduced mucociliary transport in smokers

Reduced alveolar clearance of
particulates in smokers

Increased numbers of polymorphonuclear

Reduced inhaled dose in
smokers

Reduced inhaled dose in
smokers

Increased exposure of smokers

Increased exposure of smokers

Differing regional lung doses
in smokers and nonsmokers

Increased dose in smokers

Increased dose in smokers

Increased lung injury in smokers

leukocytes and other inflammatory
cells in lungs of smokers

 

work (US DHEW 1979b). Because cigarette smoking and cigarette-
related cardiorespiratory diseases are associated with reduced aero-
bic capacity, nonsmokers may tend to perform the more strenuous
tasks in the workplace. The higher ventilatory requirements of such
jobs might increase the amount of dust or other agents inhaled;
smokers would be spared to the extent that they are selected for
more sedentary jobs. The excess mucus production of chronic
bronchitis might protect against soluble agents through the in-
creased absorptive capacity of the mucus.

Tobacco products might serve as vectors for the transformation of
workplace chemicals into more harmful agents. For example,
smokers are placed at increased risk for polymer fume fever through
contamination of their cigarettes by fluorocarbons; toxic products
are generated by the cigarette’s heat and are inhaled by the smokers.
Reduced pulmonary defenses in smokers might also increase the
effects of occupational agents. The mucociliary apparatus of the
airways removes particles and absorbed gases by physical transport
(Wanner 1977; Lippmann et al. 1980). Both cilia and mucus are
affected by tobacco smoke, and direct measurements of mucociliary

159
transport in animals and in humans confirm that long-term smoking
impairs particle clearance (Wanner 1977; Lippmann et al. 1980; US
DHHS 1984). Cohen and colleagues (1979) have demonstrated
impaired alveolar clearance of particulates in smokers, as well. A
plausible, though not established, consequence of reduced clearance
is the increased pulmonary residence time of harmful agents and an
increased dust burden in the lungs. Finally, alterations of lung cell
populations and the presence of inflammation in smokers might
amplify the effects of inhaled occupational agents. Inflammatory
cells are thought to have a central role in lung injury caused by
occupational agents (Campbell and Senior 1981; Bitterman et al.
1981). The lungs of smokers yield markedly increased numbers of
macrophages and neutrophils in bronchoalveolar lavage fluid in
comparison with the lungs of nonsmokers (US DHHS 1984). Thus,
synergism between cigarette smoking and an occupational agent
could reflect a greater release of enzymes and other toxic products
from the large numbers of inflammatory cells that have been
recruited into the lung by cigarette smoke.

Study Design

Several epidemiological study designs are used to assess the
independent and interactive effects of smoking and occupational
exposures in human populations. The cross-sectional study, or
survey, is the most widely used approach, primarily because of its
feasibility and low cost. Most surveys involve data collection from
samples defined by employment status or union membership. In a
cohort study, exposed and nonexposed people are followed over time
and monitored for the development of disease. Large-scale cohort
investigations of workers exposed to asbestos, silica, and coal dust
have been carried out. The case-control design involves the identifi-
cation of cases with the disease of interest and a control series of
people without the disease who would be potentially selected as cases
if they were to develop the disease. The exposure histories of the
cases and controls are ascertained and compared. This design has
been used infrequently for studying chronic occupational lung
diseases.

As a minimum, when cigarette smoking and a single occupational
agent are of interest, the study should provide estimates of their
independent effects and of the combined effect. This minimum is
suggested because the impairment observed in a particular popula-
tion reflects the consequences not only of the occupational agent but
also of all other damaging environmental exposures. Of these,
cigarette smoking is by far the most important and the most readily
assessed. Cross-sectional, case-control, and cohort designs meet this
requirement if the cigarette smoking practices and exposure histo-
ries of the subjects can be accurately determined.

160
Assessment of Exposures
Cigarette Smoking

The American Thoracic Society (Ferris 1978) has recommended
that a cigarette smoking questionnaire include smoking status
(never, current, or previous), age started smoking, age stopped
smoking (for former smokers), current and usual amount smoked,
and depth of inhalation. Questions concerned with brand and extent
of filter cigarette smoking are optional, but should be used when
possible to address research questions related to types of cigarettes
smoked. The recommended items provide several measures of
exposure to cigarette smoke for data analysis: usual amount smoked,
duration of smoking, and cumulative consumption. The items
related to cigarette smoking status can be used to stratify a study
population into current, former, and never smokers.

These simple measures of exposure to cigarette smoking strongly
predict the risk of both age-specific overall mortality and COLD
mortality (US DHEW 1979b; US DHHS 1984). In the major
prospective cohort studies, dose-response relationships between
amount smoked and age-specific mortality have been demonstrated;
the findings have been similar for duration of smoking (US DHEW
1979b). Associations with self-reported depth of inhalation have been
less consistent. Indices of pulmonary morbidity also vary with
measures of cigarette smoke exposure (US DHHS 1984). The
consistency of these findings for morbidity and mortality emphasizes
the importance of collecting information on the parameters of
cigarette smoking in epidemiological investigations.

Self-reported data may underestimate true cigarette consumption;
however, the degree of bias has not been shown to vary with
occupational status. For the United States and other countries,
estimates of nationwide consumption based on survey data are
generally lower than consumption figures calculated with informa-
tion from manufacturers and government agencies (Todd 1978;
Warner 1978). In the Multiple Risk Factor Intervention Trial
(MRFIT), validation of smoking with serum thiocyanate measure-
ments documented underreporting of smoking, which was greater in
the group randomized to special intervention (Neaton et al. 1981;
Ockene et al. 1982). This finding implies that bias in reported
smoking may vary with the context in which the information is
collected. Workers exposed to agents associated with lung disease
might report their smoking habits differently from unexposed
workers; both more and less accurate reporting by the exposed
population can be postulated.

161
Occupational Exposures

For clinical and research purposes, exposure to occupational
agents should be documented and both duration and concentration
estimated, when possible. The techniques used to establish exposure,
duration, and concentration are diverse, and are not considered in
detail here. Comprehensive reviews and books about them have been
published (Hammad et al. 1981; Dodgson 1984; Cralley and Cralley
1979). The methods include self-report, use of industry, occupation,
or job title as a surrogate for exposure, area sampling, personal
dosimetry, and biological markers.

Data Analysis

In an epidemiological investigation of a population at risk for
chronic occupational lung disease, information concerning work-
place exposures and cigarette smoking is collected and appropriate
health outcome measures, such as the chest radiograph and spirome-
try, are assessed. Data analysis is directed at characterizing associa-
tions between risk factors and disease and at the modifiers of these
associations; in studies of chronic occupational lung disease, ciga-
rette smoking and exposure to the occupational agent are the
primary risk factors to be considered. Data analysis with epidemio-
logical methods can provide estimates of the independent effects of
smoking and the occupational agent and test for interaction between
them (Kleinbaum et al. 1982). These techniques, some quite complex,
are not described here, but approaches for assessing interaction are
considered.

Analysis of data related to a chronic occupational lung disease,
regardless of the study design, must address the potential confound-
ing and effect modification, or interaction, resulting from cigarette
smoking. Confounding refers to the bias introduced when the effects
of one factor are not separated from those of another. In studies of
chronic occupational lung diseases, confounding may occur when
estimates of exposure to the occupational agent are associated with
cigarette smoking. For example, in a study of asbestos workers,
confounding would be present if the more heavily exposed individu-
als were also heavy smokers. Comparisons of blue-collar workers
with white-collar employees may be confounded because the former
are more often smokers.

Confounding can be controlled at the design phase or at analysis
by either stratified or multivariate techniques (Kleinbaum et al.
1982). Options in study design include restriction of participants to
smokers or to nonsmokers alone and matching of occupationally
exposed and nonexposed subjects for smoking habits. At analysis,
whether stratified or multivariate, biologically appropriate and valid
measures of cigarette smoking are needed. More simplistic variables,

162
such as categorical indicators designating never and ever smokers,
may not be satisfactory, and their use may only partially control
confounding. In particular, measures of cumulative consumption
seem most appropriate for the lung function changes of COLD
(Burrows et al. 1977; US DHHS 1984). However, errors in the
measurement of smoking may reintroduce confounding and appar-
ent effect modification (Kleinbaum et al. 1982).

Simple generalizations cannot be offered concerning the potential
magnitude of bias that uncontrolled confounding by cigarette
smoking can produce. The bias will depend on the strength of the
association between the occupational exposure and cigarette smok-
ing and on the magnitude of smoking’s effects in the population.
However, because there is a high prevalence of smoking in the
workforce and smoking has a strong association with lung function
impairment, it should not be dismissed as a confounder merely
because some particular level of effect is found for an occupational
exposure. Further, the attainment of statistical significance for the
effect of an occupational exposure does not exclude confounding.

Either stratified or multivariate statistical techniques can be used
to test for interaction. In the first approach, variation in the effects
of one factor (e.g., an occupational agent) is examined across strata
defined by the second factor (e.g., cigarette smoking). More often,
multivariate regression models, either linear or logistic, are used to
test for interaction (Kleinbaum et al. 1982). In linear regression
models, the dependent variable is a continuous measure, such as
FEV, in the logistic model, the dependent variable is the occurrence
or nonoccurrence of a discrete outcome, such as the presence of
crackles. In both types of models, the independent variables may
include terms for the individual exposures and cross-product terms
to test for interaction. The regression coefficients estimate the
effects of the exposures on the dependent variables. For example,
models developed for an asbestos-exposed study population might
include a variable for cumulative asbestos exposure, a variable for
cumulative cigarette consumption, and a variable created by multi-
plying the two. Statistically, the null hypothesis of no interaction is
tested by the cross-product term. Failure to reject this null hypothe-
sis indicates that the data are consistent with the two factors acting
independently. However, interpretation of such analyses must
consider the scale on which interaction is measured; linear models
assess departure from additivity, whereas logistic models test
departure from a multiplicative interaction (Kleinbaum et al. 1982).
The coefficient for the cross-product term specifies the direction and
magnitude of the effect of interaction, at various levels of the two
interacting factors.

The limitations posed by sample size must also be considered in
interpreting the results of modeling. In studies of occupational

163
groups, the number of subjects is most often determined by the size
of the workforce and by feasibility considerations, and rarely on the
basis of more formal sample size calculations with statistical
methods. The statistical power of tests for interaction tends to be low
(Greenland 1983), and potentially important interactions may not
attain conventional levels of statistical significance without a
sufficiently large population.

Analysis of epidemiological data can also provide estimates of the
effects of exposure at the individual level and at the group level
(Kleinbaum et al. 1982). Measures of association between exposure
and disease estimate the excess risk incurred by exposed individuals.
Measures of impact combine measures of association with the
prevalence of exposure and estimate the contribution of specific
exposures to the disease burden in a population. The most widely
used is the population attributable risk or etiologic fraction. These
measures can be used to gauge the relative importance of cigarette
smoking and occupational agents.

Specific Investigation Issues
Population Selection

The most widely employed design for investigating occupational
lung disease, the cross-sectional study or survey, may be biased when
subjects are selected from the active workforce. The individuals
examined at any particular time in a cross-sectional study may be
regarded as survivors from the entire population that entered the
particular workplace. Individuals with illness tend to leave the
workforce, whereas healthy individuals tend to remain. This bias,
often called the healthy worker effect, must be considered in both
longitudinal and cross-sectional designs (Fox and Collier 1976; Wen
et al. 1983). The implications for surveys of occupational lung disease
are evident and have been widely discussed (McDonald 1981; Field
1981; Lebowitz 1981). If only employed workers are considered and
individuals with occupational lung disease leave the workforce, the
measures of association will underestimate the true effect of
exposure. In fact, the leaving of employment by people who are ill
has been demonstrated in several industries (Fox and Collier 1976;
Musk et al. 1977; McDonald 1981; Soutar and Maclaren 1982; Eisen
et al. 1983). The resulting bias should be evaluated by examining
retirees and others who have left.

The role of cigarette smoking in determining the magnitude of the
healthy worker effect has not been fully evaluated. Overall mortality
ratios for cigarette smokers are greater below age 65 (US DHEW
1979b), and cardiovascular diseases, respiratory diseases, and lung
cancer generally contribute prominently to the reduced all-cause
mortality of the healthy worker effect (Fox and Collier 1976; Wen et
al. 1983). Thus, cigarette smokers would be anticipated to leave the

164
workforce prematurely more often than nonsmokers. A recent study
of Vermont granite workers provides data that conflict with this
hypothesis, however. Eisen and colleagues (1983) compared men who
remained in the industry during a 5-year followup period with those
who terminated. The rate of FEV, loss was greater in those who left
the industry, but their cumulative cigarette consumption was not
significantly greater than that of those who stayed. These data do
illustrate the selection bias that results from differential termina-
tion of employment, contingent on the development of disease.

Eisen and colleagues (1983, 1984) have explored other sources of
bias in respiratory disease surveys. In the granite workers’ study,
men whose spirometric testing repeatedly failed to meet criteria for
acceptability had a more rapid decline of FEV, than those with a
better performance. This finding suggests that the exclusion of
subjects whose lung function testing is judged unacceptable may
introduce bias toward the null.

External Control Populations

When subjects are selected for an epidemiological investigation, a
population, not exposed to the agent of interest but similar in other
respects to those who are, may not be available for comparison
purposes. In this circumstance, an investigator may consider only
the exposed subjects and evaluate the dose-response relationships if
the necessary data are available, or identify an external population
as controls. If the latter approach is used, the control population
must be comparable to the exposed group on potential confounding
factors such as age, sex, race, and cigarette smoking. At times,
appropriate external populations may not be readily identified.

Nevertheless, external control populations are frequently used. In
mortality studies, the use of general population rates for calculation
of “expected” deaths assumes that the general public is the control
group. Frequently, lung function levels in exposed people are
compared with those predicted from tests performed on “normal”
populations, most often asymptomatic nonsmokers without respira-
tory disease (Clausen 1982). Recently, Peterson and Castellan (1984)
reported the prevalence of chest symptoms, as measured with a
modified Medical Research Council questionnaire, in 1,372 blue-
collar workers employed in plants considered to be free of respira-
tory hazards. The data are illustrative of the effects of smoking on
the prevalence of major respiratory symptoms; even in this young,
employed population, all of the symptoms examined were more
common in current and former smokers. The authors provided
smoking-specific prediction equations and suggested that these data
can be used for comparative purposes.

165
Colinearity of Aging, Cigarette Smoking, and Occupational
Exposure Effects

From approximately age 25, measures of ventilatory function
gradually and progressively decline. In nonsmokers, the rate of loss
is approximately 20 to 30 mL annually for FEV, and FVC (US
DHHS 1984). The decline in FEV, with age may not be a linear
function with a constant decline each year, but rather, the absolute
rate of annual decline may vary with age. In addition, the rate of
decline in lung function with age derived from cross-sectional studies
may be an overestimate of the actual rate of decline because of
possible differences in lung function among different birth cohorts in
cross-sectional studies. Some cigarette smokers lose function at much
more rapid rates and ultimately develop COLD, unless they stop
smoking (US DHHS 1984). Presumably, a similar insidious excess
loss of function antedates the appearance of clinically evident
chronic occupational lung disease.

This simultaneous contribution of aging, smoking, and occupation-
al exposure to lung function loss represents a formidable analytical
problem. Further complicating its solution is the temporal colineari-
ty or correlation of these three independent factors; age, cumulative
smoking, and cumulative exposure all increase with the passage of
time. Failure to address this colinearity may lead to confounding and
to an incorrect assessment of the effect of exposure.

This problem is most often addressed by using external standard
populations to control for aging and, at times, cigarette smoking, or
by multiple regression modeling (Berry 1981b). In the first approach,
expected lung function levels in the exposed workers are calculated
with prediction equations developed in other populations; sex, age,
race, and cigarette smoking habits may all be considered in the
calculations. For example, Beck and colleagues (1984) conducted a
cross-sectional survey of cotton textile workers in Columbia, South
Carolina. Spirometric test results for the cotton workers were
compared with the expected values calculated from survey data
collected in two towns in Connecticut and one town in South
Carolina. For each cotton worker, an expected value was predicted
on the basis of sex, age, height, and weight, with regression
equations derived from asymptomatic nonsmokers in the control
communities. Deviations from the expected value were then exam-
ined within the strata defined by smoking. This approach is effective
when appropriate external populations are available. Prediction
equations developed for clinical purposes are frequently used,
primarily owing to availability; investigators should, however,
consider the comparability of the exposed workers with the ‘nor-
mal” population from which the prediction equations were derived.

Multiple regression techniques permit a simultaneous examina-
tion of the effects of age, exposure, and smoking, as well as their

166
interactions, on lung function measures. Comprehensive treatments
of these methods have been published (Draper and Smith 1966;
Kleinbaum and Kupper 1978), and only their use for lung function
data is considered here. With this approach, the lung function
measures are the dependent variables, and age, smoking, and
exposure are the independent variables in a model of this form:
Y=a+B,X,+B,X%,+B;X;+ ... BiXi+e; where Y is a lung function
parameter, a is a constant term, X, through Xi are the independent
variables and B, through Bi are their regression coefficients, and ¢ is
a term for error. The regression coefficients describe the change in Y
per unit change in a particular Xi, with all other independent
variables held constant. An estimated regression equation is general-
ly obtained by the least squares criterion.

Most standard statistical packages for computers include this
technique, and it can be readily applied to a data set. However, the
results of such modeling may be misleading, and the plausibility of
such models should be assessed by careful examination of the raw
data and residuals and by other formal means. In addition, model
development should be guided by biological rather than primarily
statistical considerations; that is, the investigator should specify the
regression mode] in the most appropriate fashion biologically, rather
than rely on statistical procedures for variable selection. Colinearity
of the age, smoking, and exposure effects may limit the multiple
regression approach. High correlation in a data set between any two
of these factors may prevent assessment of their independent effects.

Quantification of Effects in Individuals

Properly designed epidemiological investigations can provide
essential information about the occurrence of chronic occupational
lung diseases in populations. They can establish that an occupational
exposure is hazardous, quantify the risk associated with exposure,
describe the agent’s contribution to the disease burden in the
population, and document the consequences of reducing the expo-
sures. For an individual, epidemiologically derived estimates of
relative risk generally indicate the excess risk incurred by virtue of
exposure to a particular agent, as compared with nonexposure. But
such a measure of relative risk cannot be interpreted directly as a
quantitative indicator of the chance that a particular individual’s
exposure to the agent was responsible for the occurrence of the
disease concerned. Statements concerning causality in an individual
case are particularly difficult when the disease of interest has
multiple causes and interactions among them are of potential
importance.

Judgments concerning the causation of disease in specific individu-
als are frequently necessary, however, for deciding claims made

167
through workmen’s compensation, the courts, or other mechanisms
(Hoffman 1984; Hadler 1984). Legal proof of causation hinges on a
finding that the exposure more likely than not caused the disease
(Danner and Sagall 1977; Hoffman 1984). Allocation of probability of
causation when multiple agents interact is particularly problematic
(Cox 1984), but frequently necessary. In particular, the evaluation of
impairment in cigarette smokers exposed to harmful occupational
agents requires judgment concerning the independent and combined
effects of all exposures.

Accepted methods for accomplishing this quantification have not
yet been developed. Enterline (1983) considered the problem for two
agents that interact in a multiplicative fashion. Cox (1984) has
suggested an approach that covers the situation of joint and
interacting causes. Algorithms have been proposed for specific
diseases, such as asbestosis (Mitchell et al. 1985), and for specific
agents, such as radiation (NRC 1984). However, these approaches
have only recently been proposed and their applicability remains to
be established.

Some guidance can be found, however, in the pattern of physiologi-
cal abnormality. For example, the impairment in a smoker with
asbestosis, but with no evidence of airflow obstruction, can be
attributed mostly to the pneumoconiosis. Correspondingly, the
presence of airflow obstruction and an increased TLC in an asbestos
worker who smokes and who has a normal chest x ray suggests that
the impairment is largely attributable to cigarette smoking. The
problem is more complicated in those situations where reduced
expiratory airflow is present and TLC is decreased or in those
pneumoconioses where reductions in the rate of expiratory airflow
are part of the pattern of the pneumoconiosis. For example,
reductions of FEV,, FVC, and FEV,/FVC may all be found in
complicated silicosis and coal workers’ pneumoconiosis, and these
patterns are similar to those found in cigarette smokers. Emphyse-
ma decreases lung elastic recoil, whereas some pneumoconioses, such
as asbestosis, increase it. These competing effects may result in a
TLC that is increased, normal, or reduced in a smoker with COLD
and pneumoconiosis, depending on which effect predominates. Thus,
smokers with COLD and pneumoconiosis display diverse patterns of
lung function abnormality. Evidence of airflow obstruction on
spirometry may be accompanied by a reduced, normal, or increased
TLC, and the diffusing capacity for carbon monoxide will generally
be reduced regardless of the cause of the injury. In this setting, the
diagnosis of pneumoconiosis can often be established from the chest
x ray findings, but responsibility for impairment cannot readily be
divided between COLD and pneumoconiosis. For chronic occupation-
al lung diseases associated with airflow obstruction, even diagnosis
may be difficult in an individual cigarette smoker.

168
A second method of separating the relative effects of two agents in
a combined exposure is to use the known dose-response relationships
for the agents. This approach is most useful when exposure to one
agent has been slight in comparison with the exposure to the second
agent. Difficulty arises when an individual has been exposed to
biologically equivalent doses of both agents or when exposure to one
of the agents cannot accurately be assessed.

Summary and Conclusions

During the 20th century, cigarette smoking has become prevalent
among workers at risk for occupational lung disease. By itself,
smoking causes pulmonary impairment; among people exposed to
harmful occupational agents, the interactive effects of smoking may
increase the number of individuals developing clinically significant
impairment. For both clinicians and researchers, cigarette smoking
by workers poses difficult and important challenges.

1. Existing resources for monitoring the occurrence of occupation-
al lung diseases are not comprehensive and do not include
information on cigarette smoking. Other approaches, such as
registries, might offer more accurate data and facilitate
research related to occupational lung diseases. Because of the
variability in diagnostic criteria for chronic lung disease, in
studies on occupational lung diseases emphasis should be
placed on measures of physiological change, roentgenographic
abnormality, and other objective measures.

2. Further studies that correlate lung function with histopatholo-
gy should be carried out in occupationally exposed smokers and
nonsmokers.

3. The effects of cigarette smoking on the chest x ray should be
clarified. In particular, the sensitivity of the ILO classification
to smoking-related changes should be further evaluated in
healthy populations.

4. To determine if smoking is reported with bias by occupational-
ly exposed workers, self-reported histories should be compared
with biological markers of smoking in appropriate populations.

5. Mechanisms through which specific occupational agents and
cigarette smoking might interact should be systematically
considered. Both laboratory and epidemiological approaches
should be used to evaluate such interactions.

6. Statistical methods for evaluating interaction require further
development. In particular, the biological implications of
conventional modeling approaches should be explored. Fur-
ther, the limitations posed by sample size for examining
independent and interactive effects should be evaluated. The

169
157-964 0 - 86-7
consequences of misclassification by exposure estimates and of
the colinearity of exposure variables should also be addressed.
7.The role of cigarette smoking in the ‘healthy worker effect”
requires further evaluation.
8. Approaches for apportioning the impairment in a specific
individual between occupational causes and cigarette smoking
should be developed and validated.

170
References

AMANDUS, H.E., LAPP, N.L., JACOBSON, G., REGER, R.B. Significance of
irregular small opacities in radiographs of coalminers in the USA. British Journal
of Industrial Medicine 33(1):13-17, February 1976.

AMERICAN THORACIC SOCIETY. Standards for Epidemiologic Surveys in Chronic
Respiratory Disease. New York, National Tuberculosis and Respiratory Disegse
Association, 1969, 32 pp.

AMERICAN THORACIC SOCIETY. Evaluation of impairment/disability secondary
to respiratory disease. American Review of Respiratory Disease 126(5):945-951,
November 1982a.

AMERICAN THORACIC SOCIETY. Surveillance for respiratory hazards in the
occupational setting. American Review of Respiratory Disease 126(5):952-956,
November 1982b.

AUERBACH, O., GARFINKEL, L., HAMMOND, E.C. Relation of smoking and age to
findings in lung parenchyma: A microscopic study. Chest 65(1):29-35, January
1974.

BECK, G.J., MAUNDER, L.R., SCHACHTER, E.N. Cotton dust and smoking effects on
lung function in cotton textile workers. American Journal of Epidemiology
119(1):33-43, January 1984.

BERRY, G. Mortality of workers certified by pneumoconiosis medicai panels as
having asbestosis. British Journal of Industrial Medicine 38(2):130-137, May
198 1a.

BERRY, G. Statistical analyses. In: Weill, H., Turner-Warwick, M. (eds.). Occupational
Lung Diseases: Research Approaches and Methods. Lung Biology in Health and
Disease, Vol. 18. New York, Marcel Dekker, 1981b, pp. 427-463.

BITTERMAN, P.B., RENNARD, S.L, CRYSTAL, R.G. Environmental lung disease
and the interstitium. Clinics in Chest Medicine 2(3):393-412, September 1981.

BLOT, W.J., DAY, N.E. Synergism and interaction: Are they equivalent? (letter).
American Journal of Epidemiology 110(1):99-100, July 1979.

BOEHLECKE, B. Medical monitoring of lung disease in the workplace. In: Gee, J.B.L.
(ed.). Occupational Lung Disease. New York, Churchill Livingstone, 1984, pp. 225-
240.

BOUSHY, S.F., ABOUMRAD, M.H., NORTH, L.B., HELGASON, A.H. Lung recoil
pressure, airway resistance, and forced flows related to morphologic emphysema.
American Review of Respiratory Disease 104(4):551-561, October 1971.

BROOKS, S.M. The evaluation of occupational airways disease in the laboratory and
workplace. Journal of Allergy and Clinical Immunology 70(1):56-66, July 1982.

BURROWS, B., KNUDSON, RJ., CLINE, M.G., LEBOWITZ, M.D. Quantitative
relationships between cigarette smoking and ventilatory function. American
Review of Respiratory Disease 115(2):195-205, February 1977.

CAMPBELL, E.J., SENIOR, R.M. Cell injury and repair. Clinics in Chest Medicine
2(3):357-375, September 1981.

CARILLI, A.D., KOTZEN, L.M., FISCHER, M.J. The chest roentgenogram in smoking
females. American Review of Respiratory Disease 107(1):133-136, January 1973.

CASTELLAN, R.M., SANDERSON, W.T., PETERSEN, M.R. Prevalence of radio-
graphic appearance of pneumoconiosis in an unexposed blue collar population.
(abstract). American Review of Respiratory Disease 1294):A155, April 1984.

CHURG, A., WRIGHT, J.L., WIGGS, B., PARE, P.D., LAZAR, N. Small airways
disease and mineral dust exposure: Prevalence, structure, and function. American
Review of Respiratory Disease 131(1):139-143, January 1985.

CLAUSEN, J.L. Prediction of normal values. In: Clausen, J.L. (ed.). Pulmonary
Function Testing Guidelines and Controversies: Equipment, Methods, and Normal
Values. New York, Academic Press, 1982, pp. 49-59.

COCKCROFT, A., BERRY, G., COTES, J.E., LYONS, J.P. Shape of small opacities and
lung function in coalworkers. Thorax 37(10):765-769, October 1982.

171
COCKCROFT, A., LYONS, J.P., ANDERSSON, N., SAUNDERS, M.J. Prevalence and
relation to underground exposure of radiological irregular opacities in South
Wales coal workers with pneumoconiosis. British Journal of Industrial Medicine
40(2):169-172, May 1983.

COHEN, D., ARAL, S.F., BRAIN, J.D. Smoking impairs long-term dust clearance from
the lung. Science 204(4392):514-517, May 4, 1979.

CORDIER, S., THERIAULT, G.. PROVENCHER, S. Radiographic changes in a group
of chrysotile miners and millers exposed to low asbestos dust concentrations.
British Journal of Industrial Medicine 41(3):384-388, August 1984.

COSIO, M., GHEZZO, H., HOGG, J.C., CORBIN, R., LOVELAND, M., DOSMAN, J.,
MACKLEM, P.T. The relations between structural changes in small airways and
pulmonary-function tests. New Engiand Journal of Medicine 298(23):1277-1281,
June 8, 1978.

COTES, J.E. Lung Function Assessment and Application in Medicine. 4th ed. Oxford,
Blackwell Scientific Publications, 1979, 605 pp.

COX, L.A., dr. Probability of causation and the attributable proportion of risk. Risk
Analysis 4(3):221-230, 1984.

CRALLEY, L.J., CRALLEY, L.V. Theory and Rationale of Industrial Hygiene Practice.
Patty’s Industrial Hygiene and Toxicology, Vol. 3. New York, John Wiley and
Sons, 1979, 752 pp.

DANNER, D., SAGALL, E.L. Medicolegal causation: A source of professional
misunderstanding. American Journal of Law and Medicine 3(3):303-308, Fall 1977.

DICK, J.A., JACOBSEN, M., GAULD, S., PERN, P.O. The significance of irregular
opacities in the chest radiographs of British coal miners. Proceedings of the Sixth
International Pneumoconiosis Conference, 1983. Vol. 1. Geneva, International
Labour Office, pp. 283-299.

DODGSON, J. The measurement of dusts and fumes. In: Morgan, W.K.C., Seaton, A.
(eds.). Occupational Lung Diseases. 2d ed. Philadelphia, W.B. Saunders Co., 1984,
pp. 212-238.

DO PICO, G.A., REDDAN, W., FLAHERTY, D., TSIATIS, A., PETERS, M.E., RAO, P.,
RANKIN, J. Respiratory abnormalities among grain handlers: A clinical, physio-
logic, and immunologic study. American Review of Respiratory Disease 115(6):915--
927, June 1977.

DOSMAN, J.A., COTTON, D.J., GRAHAM, B.L., LI, K.Y.R., FROH, F., BARNETT.
G.D. Chronic bronchitis and decreased forced expiratory flow rates in lifetime
nonsmoking grain workers. American Review of Respiratory Disease 121(1):11-16,
January 1980.

DOUGLAS, A.N., LAMB, D., RUCKLEY, V.A. Bronchial gland dimensions in coal
miners: Influence of smoking and dust exposure. Thorax 37(10):760-764, October
1982.

DRAPER. N.R., SMITH, H. Applied Regression Analysis. New York, John Wiley and -
Sons, 1966, 407 pp.

EDWARDS, C., MACARTNEY, J., ROOKE, G., WARD, F. The pathology of the lung
in byssinotics. Thorax 30(6):612-623, December 1975.

EISEN, E.A., ROBINS, J.M., GREAVES, LA., WEGMAN, D.H. Selection effects of
repeatability criteria applied to lung spirometry. American Journal of Epidemiolo-
gy 125):734-742, November 1984.

EISEN, E.A., WEGMAN, D.H., LOUIS, TA. Effects of selection in a prospective study
of forced expiratory volume in Vermont granite workers. American Review of -
Respiratory Disease 128( 4):587-591, October 1983.

ENTERLINE, P.E. Sorting out multiple causal factors in individual cases. In: Chiazze,
L., Jr., Lundin, F.E., Watkins, D. (eds.). Methods and Issues in Occupational and
Environmental Epidemiology. Ann Arbor, Ann Arbor Science Publishers, 1983, pp.
177-182.

172
EPLER, G.R., CARRINGTON, C.B.. GAENSLER, E.A. Crackles (rales) in the
interstitial pulmonary diseases. Chest 73(3):333-339, March 1978.

EPLER, G.R., McLOUD, T.C., GAENSLER, E.A., MIKUS, J.P., CARRINGTON, C.B.
Normal chest roentgenograms in chronic diffuse infiltrative lung disease. New
England Journal of Medicine 298(17):934-939, April 27, 1978.

EPLER, G.R., SABER. F.A., GAENSLER, E.A. Determination of severe impairment
(disability) in interstitial lung disease. American Review of Respiratory Disease
121(4):647-659, April 1980.

EPSTEIN, D.M., MILLER. W.T., BRESNITZ, E.A., LEVINE, M.S., GEFTER, W.B.
Application of ILO classification to a population without industrial exposure:
Findings to be differentiated from pneumoconiosis. American Journal of Roentge-
nology 142(1):53-58, January 1984.

FERRIS, B.G. Epidemiology standardization project. American Review of Respiratory
Disease 118(6):1-120, December 1978.

FIELD, G.B. Worker surveys. In: Weill, H., Turner-Warwick, M. (eds.). Occupational
Lung Diseases: Research Approaches and Methods. Lung Biology in Health and
Disease, Vol. 18. New York, Marcel Dekker, 1981, pp. 405-426.

FLETCHER, C.M., ELMES, P.C., FAIRBAIRN, A.S., WOOD, C.H. The significance of
respiratory symptoms and the diagnosis of chronic bronchitis in a working
population. British Medical Journal 2(5147).257-266, August 29, 1959.

FLETCHER, C.M., PETO, R.. TINKER, C., SPEIZER, F.E. The Natural History of
Chronic Bronchitis and Emphysema: An Eight-Year Study of Early Chronic
Obstructive Lung Disease in Working Men in London. Oxford, Oxford University
Press, 1976, 272 pp.

FLOREY, C. du V., LEEDER, S.R. Methods for Cohort Studies of Chronic Airflow
Limitation. WHO Regional Publications, European Series No. 12. Copenhagen,
World Health Organization, Regional Office for Europe, 1982, 140 pp.

FOX, AJ., COLLIER, P.F. Low mortality rates in industrial cohort studies due to
selection for work and survival in the industry. British Journal of Preventive and
Social Medicine 30(4):225—230, December 1976.

FRIEDMAN, G.D., SIEGELAUB, A.B., SELTZER, C.C. Cigarette smoking and
exposure to occupational hazards. American Journal of Epidemiology 98(3):175-
183, September 1973.

GANDEVIA, B.H. Clinical techniques. In: Weill, H., Turner-Warwick, M. (eds.).
Occupational Lung Diseases: Research Approaches and Methods. Lung Biology in
Health and Disease, Vol. 18. New York, Marcel Dekker, 1981, pp. 11-33.

GERRARD, J.W., COCKCROFT, D.W., MINK, J.T., COTTON, D.J., POONAWALA,
R., DOSMAN, J.A. Increased nonspecific bronchial reactivity in cigarette smokers
with normal lung function. American Review of Respiratory Disease 122(4):577-
581, October 1980.

GLOVER, J.R., BEVAN, C., COTES, J.E., ELWOOD, P.C., HODGES, N.G., KELL,
R.L., LOWE, C.R., McCDERMOTT, M., OLDHAM, P.D. Effects of exposure to slate
dust in North Wales. British Journal of Industrial Medicine 37(2):152-162, May
1980.

GREENLAND, S. Tests for interaction in epidemiologic studies: A review and a study
of power. Statistics in Medicine 2(2):243-251, April-June 1983.

GRUCHOW, H.W., HOFFMANN, R.G., MARX, JJ., Jr, EMANUEL, D.A., RIMM,
A.A. Precipitating antibodies to farmer’s lung antigens in a Wisconsin farming
population. American Review of Respiratory Disease 124(4):411-415, October 1981.

HADLER, N.M. Occupational illness: The issue of causality. Journal of Occupational
Medicine 26(8):587-593, August 1984.

HAMMAD, Y., CORN, M., DHARMARAJAN, V. Environmental characterization. In:
Weill, H., Turner-Warwick, M. (eds.). Occupational Lung Diseases: Research
Approaches and Methods. Lung Biology in Health and Disease, Vol. 18. New York,
Marcel Dekker, 1981. pp. 291-371.

178
HAMMOND, E.C., SELIKOFF, 14J., SEIDMAN, H. Asbestos exposure, cigarette
smoking and death rates. Annals of the New York Academy of Sciences 330:473-
490, December 14, 1979.

HANKINSON, J.L., REGER, R.B., MORGAN, W.K.C. Maximal expiratory flows in
coal miners. American Review of Respiratory Disease 116(2):175-180, August 1977.

HOFFMAN, R.E. The use of epidemiologic data in the courts. American Journal of
Epidemiology 12Q2):190-202, August 1984.

HUNTER, D. The Diseases of Occupations. 6th ed. London, Hodder and Stoughton,
1978, 1,257 pp.

INTERNATIONAL LABOUR OFFICE. International Classification of Radiographs of
Pneumoconioses. Occupational Safety and Health Series, No. 22. (rev.). Geneva,
International Labour Office, 1980.

IRWIG, L.M., ROCKS, P. Lung function and respiratory symptoms in silicotic and
nonsilicotic gold miners. American Review of Respiratory Disease 117(3):429-435,
March 1978.

KABIRAJ, M.U., SIMONSSON, B.G., GROTH, S., BJORKLUND, A., BULOW, K.,
LINDELL, S.-E. Bronchial reactivity, smoking, and alpha,-antitrypsin: A popula-
tion-based study of middle-aged men. American Review of Respiratory Disease
126(5):864-869, November 1982.

KAMINSKI, R., BROCKERT, J., SESTITO, J., FRAZIER, T. Occupational information
on death certificates: A survey of state practices. American Journal of Public
Health 71(5):525-526, May 1981.

KENNEDY, S.M., ELWOOD, R.K., WIGGS, BJ.R., PARE, P.D., HOGG, J.C. Increased
airway mucosal permeability of smokers: Relationship to airway reactivity.
American Review of Respiratory Disease 129(1):143-148, January 1984.

KIBELSTIS, J.A., MORGAN, E.J., REGER, R., LAPP, N.L., SEATON, A., MORGAN,
W.K.C. Prevalence of bronchitis and airway obstruction in American bituminous
coal miners. American Review of Respiratory Disease 108(4):886-893, October 1973.

KILBURN, K.H. Cigarette smoking does not produce or enhance the radiologic
appearance of pulmonary fibrosis. American Journal of Industrial Medicine
2(4):305--308, 1981.

KILLIAN, K.J., JONES, N.L. The use of exercise testing and other methods in the
investigation of dyspnea. Clinics in Chest Medicine 5(1):99-108, March 1984.

KLEINBAUM, D.G., KUPPER, L.L. Applied Regression Analysis and Other Multi-
variable Methods. North Scituate, Massachusetts, Duxbury Press, 1978, 556 pp.

KLEINBAUM, D.G., KUPPER, L.L., MORGENSTERN, H. Epidemiologic Research:
Principles and Quantitative Methods. Belmont, California, Lifetime Learning
Publications, 1982, 529 pp.

LAST, J.M. (ed.). A Dictionary of Epidemiology. New York, Oxford University Press,
1983.

LEBOWITZ, M.D. Epidemiological recognition of occupational pulmonary diseases.
Clinics in Chest Medicine 2(3):305-316, September 1981.

LEBOWITZ, M.O., KNUDSON, RJ., BURROWS, B. Tucson epidemiologic study of
obstructive lung diseases: I. Methodology and prevalence of disease. American
Journal of Epidemiology 102(2):137-152, August 1975.

LIPPMANN, M., YEATES, D.B., ALBERT, R.E. Deposition, retention, and clearance
of inhaled particles. British Journal of Industrial Medicine 37(4):337-362, Novem-
ber 1980.

LOUDON, R., MURPHY, R.L.H., Jr. Lung sounds. American Review of Respiratory
Disease 130(4):663-673, October 1984.

MAKE, B., MILLER, A., EPLER, G., GEE, J.B.L. Single breath diffusing capacity in
the industrial setting. Chest 82(3):351-356, September 1982.

McDONALD, J.C. Epidemiology. In: Weil}, H., Turner-Warwick, M. (eds.). Occupation-
al Lung Diseases: Research Approaches and Methods. Lung Biology in Health and
Disease, Vol. 18. New York, Marcel Dekker, 1981, pp. 373-403.

174
MERCHANT, J.A., KILBURN, K.H., O'FALLON, W.M., HAMILTON, J.D., LUMS-
DEN, J.C. Byssinosis and chronic bronchitis among cotton textile workers. Annals
of Internal Medicine 76(3):423-433, March 1972.

MERCHANT, J.A., LUMSDEN, J.C., KILBURN, K.H., O’FALLON, W.M., UJDA,
J.R., GERMINO, V.H., Jr.. HAMILTON, J.D. An industrial study of the biological
effects of cotton dust and cigarette smoke exposure. Journal of Occupational
Medicine 15(3).212-221, March 1973.

MILLER, A., THORNTON, J.C., WARSHAW, R., ANDERSON, H., TEIRSTEIN, A.S.,
SELIKOFF, IJ. Single breath diffusing capacity in a representative sample of the
population of Michigan, a large industrial state: Predicted values, lower limits of
normal, and frequencies of abnormality by smoking history. American Review of
Respiratory Disease 127(3):270-277, March 1983.

MITCHELL, R.S., CHASE, G.R., KOTIN, P. Evaluation for compensation of asbestos-
exposed individuals: I. Detection and quantification of asbestos-related nonmalig-
nant impairment. Journal of Occupational Medicine 27(2):95-109, February 1985.

MITCHELL, R.S., MAISEL, J.C., DART, G.A., SILVERS, G.W. The accuracy of the
death certificate in reporting cause of death in adults, with special reference to
chronic bronchitis and emphysema. American Review of Respiratory Disease
104(6):844-850, December 1971.

MORGAN, W.K.C. Industrial bronchitis. British Journal of Industrial Medicine
35(4):285-291, November 1978.

MORGAN, W.K.C. Coal workers’ pneumoconiosis. In: Morgan, W.K.C., Seaton, A.
(eds.). Occupational Lung Diseases. 2d ed. Philadelphia, W.B. Saunders Co., 1984a,
pp. 377-448.

MORGAN, W.K.C. Industrial bronchitis and other nonspecific conditions affecting the
airways. In: Morgan, W.K.C., Seaton, A. (eds.). Occupational Lung Diseases. 2d ed.
Philadelphia, W.B. Saunders Co., 1984b, pp. 521-540.

MORGAN, W.K.C., BURGESS, D.B., JACOBSON, G., O'BRIEN, RJ., PENDER-
GRASS, E.P., REGER, R.B.,, SHOUB, E.P. The prevalence of coal workers’
pneumoconiosis in U.S. coal miners. Archives of Environmental Health 27(4):221-
226, October 1973.

MORRING, K.L., ATTFIELD, M. Dust Concentrations and Prevalence of Pneumoco-
niosis in the United States and Great Britain: A Re-Examination and Comparison
of Data Prior to 1970. Paper presented at the Coal Mine Dust Conference,
Morgantown, West Virginia, October 8-10, 1984.

MUELLER, R.E., KEBLE, D.L., PLUMMER, J., WALKER, S.H. The prevalence of
chronic bronchitis, chronic airway obstruction, and respiratory symptoms ina
Colorado city. American Review of Respiratory Disease 103(2):209-228, February
1971.

MURPHY, R.L.H., Jr., GAENSLER, E.A., FERRIS, B.G., FITZGERALD, M., SOLLI-
DAY, N., MORRISEY, W. Diagnosis of “asbestosis”: Observations from a longitudi-
nal survey of shipyard pipe coverers. American Journal of Medicine 65(3):488-498,
September 1978.

MURPHY, R.L.H., Jr, GAENSLER, E.A., HOLFORD, S.K., DEL BONO, E.A.,
EPLER, G. Crackles in the early detection of asbestosis. American Review of
Respiratory Disease 129(3):375-379, March 1984.

MUSK, A.W., PETERS, J.M., WEGMAN, D.H. Lung function in fire fighters: II. A five
year follow-up of retirees. American Journal of Public Health 67(7):630-633, July
1977.

NATIONAL CENTER FOR HEALTH STATISTICS. Mortality. Vital Statistics of the
United States, 1970, Vol. 2, Part A. U.S. Department of Health, Education, and
Welfare, Public Health Service, National Center for Health Statistics, DHEW Pub.
No. (HRA)75-1101, 1974, 648 pp.

175