TABLE &-Comparison of smoking habit data obtained during life and after death Smoking habit data obtained at death Smokmg habit data obtained dunng life, 1971 Number NW?r Formerly smoked smoked Smoked Smoked at some time Never smoker 12 8 2 2 0 Ex-smoker 26 2 1.5 2 7 Smoker 76 1 12 33 30 SOURCE Berry et al. (19851. and for those who have died (which would include most individuals with lung cancer> by questioning next of kin or checking hospital records. Berry and colleagues (1985) examined the comparability of these data sources in a prospective evaluation of asbestos workers in which smoking data were accumulated both at the start of the study period (i.e., prospectively) and at the time of death from lung cancer (i.e., retrospectively). A comparison of the smoking status obtained by the two methods for the same individuals is shown in Table 5. In general, there was good agreement between the two methods, but both methods identified as never smokers individuals who were classified as smokers by the other method. No data were presented to allow determination of which method was more accurate. The random misclassification of smoking status, of itself, should not introduce spurious associations for the population as a whole, or for the smokers in the population (Greenland 19801. However, when the question being asked is whether a risk exists in the absence of smoking and synergism between smoking and the occupational exposure is present, the misclassification of even small numbers of exposed smokers as nonsmokers can lead to the conclusion of increased risk of lung cancer due to an occupational exposure in the absence of cigarette smoking. The potential for misclassification exists and is of greatest concern when decisions are being made on small numbers of cases. The second caveat that may need to be applied in the examination of the effects of occupational exposure among people who have never smoked is the potential effect of involuntary exposure to cigarette smoke. A number of studies have shown increased lung cancer risks in the nonsmoking wives of smokers, raising the question of a carcinogenic risk due to environmental tobacco smoke exposure (IARC, in press). If these studies can be extrapolated to the workplace, then the potential exists for environmental tobacco smoke in the worksite to act as an occupational carcinogen, 126 particularly in those occupations in which there is a high prevalence of active smoking among workers. The considerations raised by examination of smokers with work- place exposures are somewhat different from those raised by examination of nonsmokers. Comparisons of smokers with and without an occupational exposure r. `I- pnqiire careful attention to the correlations among age, duration of exposure, and smoking dose. Age adjustment of the death rates in the exposed group and the control population is generally accepted as more useful than simply compar- ing the mean age of the two populations, because of the rapid rise in lung cancer death rates in the older age groups. It is less widely understood that age adjustment does not eliminate the effects of differences in the age distributions of smokers between the two populations. The smoking-related risk of developing lung cancer occurs disproportionately in older smokers compared with younger smokers. Therefore, in two populations with similar prevalences of smoking, but with different age distributions of that smoking prevalence, the population with the higher prevalence of smoking in the older age group will have the higher number of lung cancer deaths. This difference in number of lung cancers will persist after an age adjustment using the age distributions of the entire popula- tion (smoker and nonsmoker). Therefore, in considering the differ- ences between occupationally exposed smokers and smokers who are not exposed, the lung cancer deaths should be adjusted for age on the basis of the age distribution of the smokers in the two populations rather than the age distribution of the entire population. Several attempts have been made to combine the strengths of large population-based measurements with the detailed measure- ments of smoking status available in cohort studies. Hammond and colleagues (1979) used the American Cancer Society (ACS) study of 1 million men and women to develop a control group for a study of asbestos insulation workers. From the ACS study population, they extracted a group of more than 73,000 men who were white, not a farmer, had no more than high school education; did have a history of occupational exposure to dust, fumes, vapors, gases, chemicals, or radiation; and were alive at the time of the initiation of followup of the insulators. From this control group, they were able to develop age-specific and smoking-specific expected lung cancer death rates for comparison with the observed death rates in the insulation workers. There was a difference in the time period of followup between these two studies; therefore, the expected lung cancer death rates were adjusted upward on the basis of differences in the national lung cancer death rates during the years of differential followup. This approach allowed the expected rates to be calculated from a large enough population to provide stable rates in a number of separate age and smoking categories. The control group and the 127 exposed populations were also matched for a number of those characteristics that raise questions about the comparability of national death rate data to populations of employed workers. A somewhat different approach to the same problem was taken by Berry and colleagues (1985). They used data from a prospective mortality study of British physicians by smoking status (Doll and Peto 1978, 1981) to develop factors that related the risks of smokers, nonsmokers, and ex-smokers separately to the risk in the entire population of physicians. They calculated the expected number of deaths for the exposed workers in each smoking category, using national death rate data, and multiplied this expected number of deaths by the smoking factor to get a smoking-specific expected number of deaths for each category of exposed workers. They also adjusted the number of expected deaths for differences in g-graphic location by multiplying the expected deaths by the ratio of the local lung cancer SMR to the national lung cancer SMR. This approach is obviously quite sensitive to the method by which the smoking- specific factors are developed, and it is not clear that one set of factors can be applied to all ages. When an explicit control population is being used, the differences in smoking behavior can be controlled through the use of a statistical model for lung cancer risk in the population. Models may include a variety of measures of cigarette smoking dosage and duration, and the mortality experienced by the exposed population can be exam- ined by using the risk model developed in the control population. This approach allows the confounding due to smoking to be adjusted through the use of terms for intensity and duration of exposure. Comparisons Using Internal Control Populations The use of an internal control group drawn from the same workforce as the exposed population, but not exposed to the agent of interest, may produce a control group that is more closely matched to the exposed population than the total US. population would be (Breslow et al. 1983; Pasternack and Shore 1976; Redmond and Breslin 1975). Working populations tend to have a lower overall mortality than the U.S. population of the same ages (McMichael 1976; Enterline 1975; Fox and Collier 1976; Shindell et al. 1978; Vinni and Hakama 19801, at least in part because workers with illness tend to drop out of the working population. This lower mortality has been called the healthy worker effect and is one of the reasons the selection of an internal control population may be more appropriate than using SMRs for evaluating occupational exposure risks. External control groups, selected from populations geographi- cally or demographically similar to the exposed population, may also provide a population more similar to the exposed workers than the general U.S. population. 128 That the smoking behaviors of the exposed group and the control population are comparable must still be established. The selection of a control population based on its similarity in one variable (such as worksite) does not allow the assumption of comparability on other variables (such as smoking behaviors). It is possible for a control population to deviate from national measures of smoking behavior in one direction and for the exposed population to deviate in the opposite direction; thus it is important to actually examine the comparability of the smoking behaviors in the exposed group and the control population even when an internal control population is used. The absence of an external control group means that the entire population has some exposure. Potential confounding of cumulative occupational exposure by cumulative smoking exposure can be reduced by stratification of the two exposures in question. The risk with increasing exposure to an occupational agent can then be examined within each strata of smoking exposure. Stratification of smoking by intensity only (cigarettes per day) would lead to a residual confounding of smoking and cumulative dust exposure, owing to the importance of duration of smoking for lung cancer risk and the association of age with both duration of smoking and cumulative dust exposure. The reduction of residual confounding should also guide the selection of the number of strata selected for smoking and the occupational exposure. The larger the risk due to smoking in relation to the risk due to the occupational exposure, the larger the number of smoking strata needed to control the confounding. The use of too few strata may result in the residual confounding producing the appearance of a dose-response relationship with the occupational exposure. A second method of controlling the confounding of occupational exposure by smoking behaviors is through the use of modeling techniques. By using a multiple logistic regression, a model of the smoking variables that contribute to lung cancer risk can be developed. The model should include measures of intensity and duration as well as a factor for cessation. Other factors that may contribute to the model are type of cigarette smoked, use of pipes or cigars, and age of initiation (as separate from duration). Once the model is established for smoking variables, a term or terms for the occupational exposure can be added to the risk prediction equation and tested to see whether the term improves the fit of the model to the observed data. Case-control analyses can also be applied in the absence of an external control group by examining the distribution of exposures in cases of lung cancer and in a control group selected from the sample population of workers, but who have not died of lung cancer. Confounding due to cigarette smoking can then be controlled by 129 stratification (Liddell et al. 1984) or by modeling (Whittemore and McMillan 1983; Pathak et al., in press). This approach is particularly useful when a case-control analysis can be nested within an ongoing study of a cohort of workers. In this setting, the smoking habits of the workforce are often known prior to the development of lung cancer, eliminating the potential for biased recall of smoking habits by the lung cancer patients (or their survivors) compared with the controls. Examination of Occupational Exposures When Smoking Habits Are Not Known In many occupational settings the smoking habits of the workforce are either unavailable or incompletely ascertained. In these cases, the death rates for these workers are compared with rates for a control population or with national mortality data (to generate an SMR). The potential for smoking pattern differences to influence the SMR is then evaluated by calculating the maximal distortion that would be produced, assuming that the exposed population had a very high smoking prevalence. The calculations used are similar to those used in generating Tables 2 and 3. As discussed earlier, extremes of differences in smoking prevalence and dosage could be expected to generate SMRs in excess of 200, and differences in age distribution and type of cigarette smoked may increase this number even more. Once an outer limit for smoking-related distortions of the SMR is estimated, it becomes the value that must be below (outside) the confidence interval surrounding the actual SMR for the exposed population in order to exclude a potential smoking effect. This approach may be useful in settings where smoking data are unobtainable, but should not be used as a substitute for collecting smoking information. When the mortality in a control population is compared with the mortality of an exposed population in the absence of smoking data, the potential for differences between the smoking habits of the two populations may be larger than the differences when using SMRs. The control group and the exposed population may deviate in opposite directions from the mean smoking behaviors represented in the SMR, and correspondingly, the differences in cancer outcome may also be magnified. One method of adjusting for differences in smoking patterns between populations when smoking data are not available, or would be too costly to obtain, is to survey a random sample of the two populations for smoking behavior. The limitation of this technique is that the sample size needed to obtain estimates of usable precision is large and may approximate the size of the two populations com- bined. 130 An additional method of examining the effects of unknown differences in smoking habits on the rates of one smoking-related cancer is to look at the rates of other smoking-related cancers in the same population. The various smoking-associated cancers do not all have the same incidence rates, rate of change in incidence with time, ethnic distribution, cure rate, or age distribution. These differences make cross-comparison between rates of these cancers as a measure of differences in smoking patterns between populations a complex and uncertain exercise at best. This kind of comparison may be useful as a point of discussion, but probably offers little in the way of an estimate of the differences between populations in their smoking behavior. Summary and Conclusions 1. Cigarette smoking and occupational exposures may interact biologically, within a given statistical model and in their public health consequences. The demonstration of an interaction at one of these levels does not always characterize the nature of the interaction at the other levels. 2. Information on smoking behaviors should be collected as part of the health screening of all workers and made a part of their permanent exposure record. 3. Examination of the smoking behavior of an exposed population should include measures of smoking prevalence, smoking dose, and duration of smoking. 4. Differences in age of onset of exposure to cigarette smoke and occupational exposures should be considered when evaluating studies of occupational exposure, particularly when the ex- posed population is relatively young or the exposure is of relatively recent onset. 131 References AMERICAN CANCER SOCIETY. Cancer statistics, 1985. CA-A Cancer Journal for Clinicians 35(1):19-35, January-February 1985. ARCHER, V.E., GILLIAM, J.D., WAGONER, J.K. Respiratory disease mortality among uranium miners. Annals of the New York Academy of Sciences 271:280-293, 1976. ARMITAGE, P., DOLL, R. Stochastic models for carcinogenesis. In: Neyman, J. (ed.). Contributions to Biology and Problems of Medicine. Proceedings of the Fourth Berkeley Symposium on Mathematical Statistics and Probability, Vol. 4. Berkeley, University of California Press, 1961, pp. 19-38. AXELSON, 0. Aspects on confounding in occupational health epidemiology. (letter). Scandinavian Journal of Work, Environment and Health 4(1):98-102, March 1978. BERRY, G., NEWHOUSE, M.L., ANTONIS, P. Combined effect of asbestos and smoking on mortality from lung cancer and mesothelioma in factory workers. British Journal of Industrial Medicine 42(1):12-18, January 1985. BLAIR, A., HOAR, SK., WALRATH, J. Comparison of crude and smoking adjusted standardized mortality ratios. Journal of Occupational Medicine 27(12):881484, December 1985. BLOT, W.J., DAY, N.E. Synergism and interaction: Are they equivalent? American Journal of Epidemiology 110(1):99-100, July 1979. BRESLOW, N. Multivariate cohort analysis. In: Selection, FoZZow-up, and Analysis in Prospective Studies: A Workshop. National Cancer Institute Monograph No. 67. U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Institute, NIH Pub. No. 85-2713, 1985, pp. 149-156. BRESLOW, N.E., DAY, N.E. The Analysis of Case-Control Studies. Statistical Methods in Cancer Research, Vol. 1. IARC Scientific Pub. No. 32. Lyon, International Agency for Research on Cancer, 1980,338 pp. BRESLOW, N.E., LUBIN, J.H., MAREK, P., LANGHOLZ, B. Multiplicative models and cohort analysis. Journal of the American Statistical Association 78(381):1-12, March 1983. BRESLOW, N.E., STORER, B.E. General relative risk functions for case-control studies. American Journal ofEpidemiology 122(1):149-162,1985. DAY, N.E., BROWN, CC. Multistage models and primary prevention of cancer. Journal of the National Cancer Institute 64(4):977-989, April 1980. DOLL, R., PETO, R. Cigarette smoking and bronchial carcinoma: Dose and time relationships among regular smokers and lifelong nonsmokers. Journal of Epidemiology and Community Health 32(4):303-313, December 1978. DOLL, R., PETO, R. The causes of cancer: Quantitative estimates of avoidable risks of cancer in the United States today. Journal of the National Cancer Institute 66(6):1191-1308, June 1981. ENTERLINE, P.E. What do we expect from an occupational cohort? Not uniformly true for each cause of death. Journal of Occupational Medicine 17(2):127-128,1975. FARBER, E. Chemical carcinogenesis: A biologic perspective. American Journal of Pathology 106(2!:271-296, February 1982. FOX, A.J., COLLIER, P.F. Low mortality rates in industrial cohort studies due to selection for work and survival in the industry. British Journal of Preuentiue and Social Medicine 30(4):225-230, December 1976. GELBOIN, H.V.. TS'O, P.O.P. (eds.). Enuironment, Chemistry, and Metabolism. Polycyclic Hydrocarbons and Cancer, Vol. 1. New York, Academic Press, 1978a. GELBOIN, H.V., TS'O, P.O.P. (eds.). Molecular and Cell Biology. Polycyclic Hydrocarbons and Cancer, Vol. 2. New York, Academic Press, 1978b, 397 pp. GREENLAND, S. Limitations of the logistic analysis of epidemiologic data. American JournaZ OfEpidemiology 110(6):693-698, December 1979. 132 GREENLAND, S. The effect of misclassification in the presence of covariates. American Journal ofEpidemiology 112(41:564-569, October 1980. GREENLAND, S. Tests for interaction in epidemiologic studies: A review and a study of power. Statistics in Medicine 2(21:243-251, AprilJune 1983. HAMMOND, E.C. Smoking in relation to the death rates of one million men and women. In: Haenszel, W. (ed.). Epidemiologic& Approaches to the Study of Cancer and Other Chronic Diseases. National Cancer Institute Monograph No. 19. U.S. Department of Health, Education, and Welfare, Public Health Service, National Institutes of Health, National Cancer Institute, January 1966, pp. 127-204. HAMMOND, E.C., SELIKOFF, I.J., SEIDMAN, H. Asbestos exposure, cigarette smoking and death rates. Annals of the -Vew York Academy of Sciences 330:473- 490,1979. HOGAN, M.D., KUPPER, L.L.. MOST, B.M., HASEMAN, J.K. Alternatives to Rothman's approach for assessing synergism Ior antagonism) in cohort studies. American Journal of Epidemiology 108(11:60-6?. July 1978. INTERNATIONAL AGENCY FOR RESEARCH ON CANCER. IARC Monograph No. 38. Lyon, France, International Agency for Research on Cancer, in press. KAHN, H.A. The Dorn study of smoking and mortality among US. veterans: Report on eight and one-half years of observation. In: Haenszel, W. (ed.) Epidemiological Approaches to the Study of Cancer and Other Chronic Diseases. National Cancer Institute Monograph No. 19. U.S. Department of Health, Education, and Welfare, Public Health Service, National Institutes of Health, National Cancer Institute, January 1966, pp. 1-125. KLEINBAUM, D.G., KUPPER, L.L., MORGENSTERN, H. Epidemiologic Research: Principles and Quantitatir:e Methods. Belmont. California, Lifetime Learning Publications, 1982. KUPPER, L.L., HOGAN, M.D. Interaction in epidemiologic studies. American Journal of Epidemiology 108(6):447-453, December 1978. LAST, J.M. (ed.1. A Dictionary of Epidemiology. New York, Oxford University Press, 1983. LIDDELL, F.D.K., THOMAS, D.C., GIBBS, G.W., MCDONALD, J.C. Fibre exposure and mortality from pneumoconiosis, respiratory and abdominal malignancies in chrysotile production in Quebec 192675. Annals of the Academy of Medicine, Singapore 13(2, Supp.):340-344, April 1984. LUNDIN, F.E., Jr., WAGONER, J.K., ARCHER, V.E. Radon Daughter Exposure and Respiratory Cancer, Quantitative and Temporal Aspects. Joint Monograph No. 1. U.S. Department of Health, Education, and Welfare, Public Health Service, National Institute for Occupational Safety and Health and National Institute of Environmental Health Sciences, 1971, 176 pp. MCKAY, F.W., HANSON, M.R., MILLER, R.W. (eds.1. Cancer Mortality in the United States, 1950-1977. National Cancer Institute Monograph No. 59. U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Institute, NIH Pub. No. 82-2435. 1982. McMICHAEL, A.J. Standardized mortality ratios and the "healthy worker effect": Scratching beneath the surface. Journal of Occupational Medicine 18(31:165-168, March 1976. MOSSMAN, B., LIGHT, W., WEI, E. Asbestos: Mechanisms of toxicity and carcinoge- nicity in the respiratory tract. Annual ReLjieu- of Pharmacology and Toxicology 23595615, 1983. NATUSCH, D.F.S., WALLACE, J.R., EVANS, CA., Jr. Toxic trace elements: Preferential concentration in respirable particles. Sczence 183(4121):202-204, January 18,1974. NICHOLSON, W.J., PERKEL, G., SELIKOFF, I.J. Occupational exposure to asbestos: Population at risk and projected mortality-198&2030. American Journal of Industrial Medicine 3(4):259-311. 1982. 133 PASTERNACK, B.S., SHORE, R.E. Statistical methods for assessing risk following exposure to environmental carcinogens. In: Whittemore, A. (ed.). Environmental Health: Quantifatitle Methods. Philadelphia, SIAM Institute for Mathematics and Society, 1976, pp. 49-69. PATHAK. D.R., SAMET, J.M., HUMBLE, C.G., SKIPPER, B.J. Determinants of lung cancer risk in cigarette smokers in New Mexico, in press. REDMOND, C.K., BRESLIN, P.P. Comparison of methods for assessing occupational hazards. Journal of Occupational Medicine 17(51:313-317, May 1975. ROTHMAN, K.J. Induction and latent periods. American Journal of Epidemiology 114(2):253-259, August 1981. ROTHMAN. K.J. Synergy and antagonism in cause-effect relationships. American Journal of Epidemiology 99(6): 385-388, June 1974. ROTHMAN, K.J. Occam's razor pares the choice among statistical models. American Journal of Epidemiology 108(5):347-349, November 1978. ROTHMAN. K.J., BOICE, J.D. Epidemiologic Analysis With a Programmable Calculator. U.S. Department of Health, Education, and Welfare, Public Health Service, National Institutes of Health, NIH Pub. No. 79-1649, June 1979. ROTHMAN, K.J., GREENLAND, S., WALKER, A.M. Concepts of interaction. American Journal of Epidemiology 112(4):467-470,1980. SAMET, J.M., LERCHEN, M.L. Proportion of lung cancer caused by occupation: A critical review. In: Gee, J.B.L., Morgan, W.K.C., Brooks, SM. (eds.). Occupational Lung Disease. New York, Raven Press, 1984, pp. 55-67. SARACCI, R. Interaction and synergism. American Journal of Epidemiology 112(4):465466,October 1980. SCHLESSELMAN, J.J.. STOLLEY, P.D. Case-Control Studies; Design, Conduct, Analysis. New York, Oxford University Press, 1982. SELIKOFF, I.J., LEE, D.H.K. Asbestos and Disease New York, Academic Press, 1978, 549 pp. SHINDELL, S., WEISBERG, R.F.. GIEFER, E.E. The "healthy worker effect": Fact or artifact? Journal of Occupational Medicine 20(12):807-811, December 1978. SHOPLAND, D.R., BROWN, CD. Analysis of Cigarette Smoking Behavior by Birth Cohort. Paper presented at the Interagency Committee on Smoking and Health meeting, Washington, D.C., October 1,1985, 12 pp. SIEMIATYCKI, J.. THOMAS, DC. Biological models and statistical interactions: An example from multistage carcinogenesis. International Journal of Epidemiology lOt4):383-387, December 1981. SIMANI, AS'., INOVE, S., HOGG, J.C. Penetration of the respiratory epithelium of guinea pigs following exposure to cigarette smoke. Laboratory Investigation 31(11:75-81, July 1974. US. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE. Smoking and Health: A Report of the Surgeon General. U.S. Department of Health, Education, and Welfare, Public Health Service, Office of the Assistant Secretary for Health, Office on Smoking and Health, DHEW Pub. No. (PHS)79-50066, 1979, 1,136 pp. U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES. The Health Conse- quences of Smoking: Chronic Obstructiue Lung Disease. A Report of the Surgeon General. U.S. Department of Health and Human Services, Public Health Service, Office of the Assistant Secretary for Health, Office on Smoking and Health, DHEW Pub. No. IPHS) 84-50205,1984,545 pp. U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES. The Health Conse- quences of Smoking: The Changing Cigarette. A Report of the Surgeon General. U.S. Department of Health and Human Services, Public Health Service, Office of the Assistant Secretary for Health, Office on Smoking and Health, DHEW Pub. No. lPHS)Sl-50156,1981,252 pp. 134 U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES. The Health Conse- quences of Smoking: Cancer. A Report of the Surgeon General. Department of Health and Human Services, Public Health Service, Office of the Assistant Secretary for Health, Office on Smoking and Health, DHHS Pub. No. tPHSX32- 50179,1982,322 pp. VINNI, K., HAKAMA, M. Healthy worker effect in the total Finnish population. British Journal of Zndustrral Medicine 37t2):18G184, May 1980. WALKER, A.M., ROTHMAN. K.J. Model of varying parametric form in case-referent studies. American Journal of Epidemio1og.v 115(11:129-137, 1982. WALTER, SD., HOLFORD, T.R. Additive, multiplicative, and other models for disease risks. American Journal of Epidemiology 108t5~:341-346, November 1978. WEINSTEIN, I.B. Chemical carcinogenesis. In: Calabresi. P., Schein, P.S., Rosenberg, S.A. (eds.). Medical Oncology: Basic Principles and Clinical Management of Cancer. New York, Macmillan Publishing Company. 1985, pp. 63-81. WHI'ITEMORE, AS., McMILLAN, A. Lung cancer mortality among U.S. uranium miners: A reappraisal. Journal of the National Cancer Institute 71(3):489-499, September 1983. 135 CHAPTER 4 EVALUATION OF CHRONIC LUNG DISEASE IN THE WORKPLACE 157-964 0 - 86 - c CONTENTS Introduction Chronic Lung Diseases Sources of Information Occurrence of Chronic Lung Diseases Patterns of Lung Injury Injury From Cigarette Smoke Injury From Occupational Exposures Methods for Evaluating the Effects of Occupational Expo- sures on the Lungs History of Respiratory Symptoms Chest X Ray Physiological Assessment Quantification of Effects of Smoking and Occupation in Populations Concepts of Interaction Study Design Assessment of Exposures Cigarette Smoking Occupational Exposures Data Analysis Specific Investigation Issues Population Selection External Control Populations Colinearity of Aging, Cigarette Smoking, and Oc- cupational Exposure Effects Quantification of Effects in Individuals Summary and Conclusions References 139 Introduction Exposure to harmful agents in the workplace is, and will probably continue to be, an important and avoidable cause of both acute and chronic lung diseases. The major chronic lung diseases associated with workplace exposures can be classified as the pneumoconioses (fibrotic diseases of the lung parenchyma secondary to dust inhala- tion), industrial bronchitis and other processes involving the lung's airways, and occupational asthma. Some of these diseases were recognized long before cigarette smoking became prevalent. During the 16th century, Agricola and Paracelsus described diseases of miners (Hunter 1978); early in the 18th century, Ramazzini (1940) reported further on the respiratory problems of miners and noted that the lungs of stonecutters were full of sand. Occupational lung disease in coal miners was recognized during the 1800s (Morgan 1984a). In the 20th century, many chronic lung diseases caused by workplace exposures have been studied intensively using epidemio- logical, physiological, and clinical approaches. The resulting data have been essential for developing the standards that govern workplace exposures and for evaluating worker safety. In this century, however, assessment of the effects of occupational agents on the lung has been made difficult by the widespread smoking of cigarettes. This behavior has been particularly prevalent among those at high risk for occupational lung diseases-men employed in blue-collar jobs (US DHEW 1979b). The degree of pulmonary impairment in any individual represents the summation of the effects of all harmful environmental factors, including cigarette smoking, occupational agents, and other expo- sures. Cigarette smoking, in the absence of other exposures, causes chronic bronchitis (cough and mucous hypersecretion), airway abnor- malities, and emphysema (abnormal dilation of the distal airspaces with destruction of alveolar walls); together, the last two disease processes underlie the expiratory flow limitation found in chronic obstructive lung disease (COLD) (US DHHS 1984). Cigarette smoking may potentiate the effects of some occupational agents on the lung. This potentiation may occur through an effect of cigarette smoke on the mechanism of lung injury that results from a given occupational exposure, or it may result from a mechanism of lung injury due to cigarette smoke that is independent of the mechanism of occupation- al injury but produces a level of combined lung damage capable of potentiating the level of disability or the level of abnormality detected by pulmonary function tests, x rays, or symptoms. The term "synergism" is used in this chapter to refer to an effect of combined exposure to cigarette smoke and occupational agents that results in a level of abnormality (by whatever measure being used) that is significantly greater than the sum of the levels of abnormality 141 produced by the agents separately. Such interactions are of impor- tance not only for researchers but also for the exposed workers and their employers. Synergism between cigarette smoking and occupa- tional agents may, at the individual level, markedly raise the risk of developing disease and, at the group level, greatly increase the burden of occupational disease in the workforce. Thus, in evaluating the effects of workplace exposures on the lung, consideration must be given not only to the independent effects of cigarette smoking and of the agent of interest but also to the possible interaction of these factors. This chapter describes the techniques used to evaluate chronic lung disease in the workplace and addresses the methodological issues raised by cigarette smoking. The focus of the chapter is largely confined to the chronic, fixed lung injuries that result from these exposures rather than the acute reversible responses that character- ize occupational asthma. This focus was adopted in the interest of clarity and brevity and does not suggest that the issues related to the evaluation of occupational asthma are either unimportant or unrelated to cigarette smoking. Emphasis is placed on methodologi- cal problems; specific exposures are reviewed in other chapters of this Report. Chronic Lung Diseases Sources of Information Although cigarette smoking is the predominant cause of preventa- ble morbidity and mortality from respiratory diseases in the United States (US DHHS 1984), occupational exposures also produce sub- stantial disease. Because the occurrence of nonmalignant respiratory diseases is not directly monitored, its frequency must be estimated from diverse information sources such as the National Center for Health Statistics, the U.S. Bureau of Labor Statistics, the Social Security Administration, and epidemiologic surveys. The extent to which chronic lung diseases are ascertained by these sources is difficult to establish, but coverage is probably not comprehensive. Vital statistics enumerate the numbers of deaths from specific causes. Chronic conditions, such as respiratory diseases, may be listed on the death certificate, but remain uncoded unless they led directly to death. For example, Rank and Bal (1984) reviewed death certificates and found that in comparison with its frequency as an underlying cause of death, emphysema was listed nearly twice as often as an uncoded "other" condition. Vital statistics data cannot readily be used for addressing questions related to the pulmonary effects of cigarette smoking and occupational exposures. Cigarette smoking is not included on the death certificate, and only usual 142 TABLE l.-Number of deaths in selected categories of the International Classification of Diseases (ICD), for three time periods, United States Year (classification) Cause of death 1960 (ICDI 1970 (ICDI 1980 (ICDI COLD Chronic bronchitis 2.287 15021 5,014 (491) 3,269 (491) Emphysema 9.253 (527.11 22,721 (4921 13.677 (4921 Chronic airways obstruction n.e.c. ' - 4,444 1519.31 34,743 (496) Occupational disorders Coal workers' pneumoconiosls 810 1523.1) 1,160 (515.1) 982 mlo) Asbestosis 21 1523.2) 26 (515.2~ 101 will Silicosis 550 1523.01 355 (515.0) 207 (502) Other inorganic dusts 13 (516.01 8 (5031 Other dusts 62 (524) 7 (516.1) 3 (504) Unspecified 210 (523.3) 281 (505) Conditions due to chemical fumes/vapors 5 !516.21 43 (506) Chronic interstitial pneumonia 3,973 (525) 3,351 (517) 202 (516.31 ' Not elsewhere claasifwzd SOURCE, US DHEW (19631; National Center for Health Statista (1974), unpublished data (1980). occupation and industry are noted. Further, the occupational information is not routinely coded by States (Kaminski et al. 1981). Cause of death is coded according to the International Classifica- tion of Diseases, currently in its ninth revision (WHO 1977). For the chronic respiratory diseases, separate categories cover the obstruc- tive disorders, major pneumoconioses, and other interstitial diseases (Table 1). As the International Classification of Diseases has been modified from the seventh through the ninth revisions, major changes in the coding of chronic respiratory diseases have been made. The categories for occupational lung diseases have been expanded and their titles have been made more specific. With the eighth revision (US DHEW 1968), a category (519.3) was added for the diagnosis of chronic obstructive lung disease (COLD). These changes must be considered in examining time trends of mortality. For example, after the introduction of a category for COLD, the number of deaths assigned to this code increased and deaths attributed to emphysema decreased (Table 11. 143 Estimates of disease occurrence based on vital statistics must be interpreted with caution. Some causes of death may be underreport- ed, and mortality rates may not directly reflect incidence. The mortality rate for a particular disease approximates the incidence rate as the case-fatality rate approaches unity (Kleinbaum et al. 1982). Competing causes of death will also influence the relationship between incidence and mortality (Kleinbaum et al. 1982). For example, Berry (1981a) examined the mortality of 665 men certified as having asbestosis by medical boards in England and Wales. Of the 283 deaths, 39 percent were from lung cancer, 9 percent were from mesothelioma, and only 20 percent were from asbestosis. The distribution of competing causes of death should be different in smokers and nonsmokers; thus, even for non-smoking-related occu- pational lung diseases the relationship between incidence and mortality may vary with smoking practices. For several respiratory diseases, vital statistics underestimate mortality. For COLD, Mitchell and colleagues (1971) compared cause of death, as reported on the death certificate, with clinical and autopsy-derived diagnoses. In 211 subjects who died of COLD, as determined by autopsy, another cause of death was listed on the death certificate for 51. For asbestosis, Hammond and colleagues (1979) used "the best available medical information" and identified 160 deaths from this pneumoconiosis in a cohort study of asbestos workers. Only 76 were similarly classified by the death certificate statement of cause of death. State workmen's compensation claims are another source of information about the occurrence of occupational lung diseases. However, most workmen's compensation claims involve acute prob- lems (Whorton 1983) and may more accurately measure conditions associated with irritant gas or vapor inhalation than with the pneumoconioses. Under the Occupational Safety and Health Act, selected employ- ers are required to maintain records of occupational injury and illness (US House of Representatives 1984). In an annual survey, the Bureau of Labor Statistics collects and reports the injury and illness data. During 1982, 2,000 reports for dust diseases of the lungs and 8,800 for respiratory conditions due to toxic agents were filed, but more specific diagnoses were unavailable (US DOL 1984). In the introduction to the 1982 survey, it was acknowledged that "to the extent that occupational illnesses are unrecognized and therefore unreported, the survey estimates understate their occurrence" (US DOL 1984, p. 3). On a national level, the Social Security Administration operates a compensation program for people who have been disabled for at least 5 months (US DHHS 1983). People receiving compensation for chronic lung diseases must meet this criterion as well as stringent 144 requirements for the extent of impairment on lung function testing (US DHEW 1979a). Data from the Social Security Administration probably underestimate the prevalence of most chronic lung dis- eases. For example, Epler and colleagues (1980) showed that approximately 9 percent of a series of clinically diagnosed patients with pneumoconiosis met the Social Security disability criteria. Epidemiological surveys offer the most accurate estimates of disease frequency, though the surveyed populations are generally limited to employed workers and disease frequency may therefore be underestimated. Estimates of disease frequency from a particular survey should be generalized cautiously. Nonrandom selection of occupational groups for study as well as the nonrandom enrollment of workers within a particular workforce may introduce bias. Occurrence of Chronic Lung Diseases Although the available data sources have limitations, they can be used to document the relative frequencies of cigarette-related and occupation-related chronic lung diseases. Most indicate that the diseases associated with cigarette smoking are much more common in the general population than those resulting from occupational exposures. In recent years, mortality from COLD has steadily increased; the number of deaths rose from 32,179 in 1970 to 51,889 in 1980 (Table 1). The 1984 Surgeon General's Report, The Health Consequences of Smoking: Chronic ObstructiveLung Disease (US DHHS 19841, offered the estimate that 60,000 people would die from COLD during that year. Examination of COLD mortality rates for smokers and nonsmokers suggests that 85 to 90 percent of COLD deaths in the United States can be attributed to cigarette smoking (US DHHS 1984). As described in the 1984 Surgeon General's Report, numerous surveys provide estimates of the prevalence of COLD (US DHHS 1984). Representative recent data have been collected in Tucson, Arizona, in six other U.S. cities, and nationwide in the National Health Interview Survey (NHIS). Lebowitz and colleagues (1975) sampled 3,805 subjects in Tucson from 1972 through 1974. In men over 44 years of age, physician-diagnosed chronic bronchitis and emphysema were reported to be 10.2 and 13.3 percent, respectively. In women over 44 years of age, the percentage with chronic bronchitis was 9.0 percent and with emphysema, 4.3 percent. From 1974 through 1977, Ferris and colleagues (1978) surveyed 7,909 men and women in six U.S. cities; 5 percent of the men and 1.9 percent of the women had airway obstruction, defined as a ratio of forced expiratory volume in 1 second (FEV,) to forced vital capacity (FVC) less than or equal to 60 percent. The 1970 NHIS included about 116,000 persons in a nationwide sample (NCHS 1974). Individuals 19 145 years of age and older were asked whether they or other family members not present at the time of the interview had bronchitis or emphysema during the previous 12 months. On the basis of this survey, 3.4 million Americans over 45 years of age were projected as having chronic bronchitis or emphysema. In contrast, data from the Social Security Administration, not included in the 1984 Surgeon General's Report, showed only 20,246 new claimants for COLD receiving disability benefits in 1979 (US DHHS 1983). The available data sources also probably do not comprehensively document the nationwide occurrence of occupational lung diseases. The number of deaths recorded as due to several occupational lung diseases was stable from 1960 to 1980 (Table l), but it is unlikely that these death certificate data provide accurate estimates of the actual prevalence or severity of these disease processes in the U.S. population, owing to the inaccurate reporting of these diseases as cause of death. The Social Security Administration is also an ongoing source of information. In 1977, 820 persons were granted disability for pneumoconiosis; in 1979, the number had decreased to 389 (US DHHS 1983). Data from the 1970 NHIS provide an estimate of the prevalence of work-related chronic lung diseases across the Nation (NCHS 1974). Participants were queried concerning dust in the lungs, silicosis, or pneumoconiosis during the previous 12 months; their responses were used to estimate that 126,000 people nationwide had these conditions. Numerous workforces in the United States and elsewhere have been surveyed to establish the prevalence of occupational and nonoccupational lung diseases. Representative recent surveys of workers in the United States are presented in Table 2, showing the prevalence of disease and of cigarette smoking. Various disease indicators were considered in these studies. Chronic bronchitis was diagnosed on the basis of persistent cough and phlegm as ascertained by questionnaire. For the pneumoconioses, the presence of disease was based on the presence of radiographic abnormality. Of note is the high prevalence of coal workers' pneumoconiosis reported by Morgan and colleagues (1973). A different group of readers subse- quently reinterpreted the chest films reported in the Morgan and colleagues study and found a prevalence of only 12 percent; this lower prevalence suggests overinterpretation on the initial reading (Morring and Attfield 1984). Regardless of the occupational group, cigarette smoking is com- mon, even in workforces exposed to acknowledged respiratory hazards (Tabit: 2). At the time the selected surveys of these workers were conducted, 1966 to 1977 for the asbestos workers (Weiss and Theodos 1978; Samet et al. 1979) and 1981 for the uranium miners (Samet et al. 19841, knowledge of the hazards of these occupations was widely disseminated and information concerning interaction 146 TABLE 2.-Prevalence of cigarette smoking and occupational lung disease in selected survey populations Study, location. years of study Study population Prevalence of smokmg Prevalence of disease Iper 100) cper 1M)l Kibelstis et al 119731, c s.. 1969 Morgan et al 119731, U.S.. 1969 Weiss and Theodos (19781, U.S.. 1975 Samet et al 11979J. U.S.. 196G1977 Theriault et al (19741, U.S., 1971 &met et al 11984), U.S.. 1981 Merchant et al. (1973), U.S., 197M971 Do Pica et al. (19771. U.S., 1974 Gruchow et al. I1981 I, U.S. 8.555 coal miners 9,076 coal m,ners 88 workers from two asbestos manufacturing phIItS 409 workers from two asbestos manufacturing plants and two shtpyards 792 granite shed workers 192 uranium miners 787 male, 473 Smokers 51 Ex-smokers 24 Nonsmokers 25 Smokers 56 Ex-smokers 23 Nonsmokers 21 Smokers 4666 Ex-smokers 27-42 Nonsmokers 10-18 Exposed workers Smokers 61 Ex-smokers 25 Nonsmokers 13 Smokers 43 Exsmokers 39 Nonsmokers 19 M W female cotton textile Smokers 63 44 mill workers Ex-smokers 16 6 Nonsmokers 21 50 300 grain workers Smokers 59 Chronic bronchitis Ex-smokers 22 Smokers 42 Nonsmokers 19 Nonsmokers 30 1,510 farm workers Smokers 15 Ex-smokers 27 Nonsmokers 57 Chronic bronchitis Smokers 3~3 Ex-smokers 30 Nonsmokers 25 Airway obstruction ' Smokers 18 Ex-smokers 14 Nonsmokers 6 Coal workers' pneumoconiosis Simple 30.0 Complicated 2.5 X-ray profusion >l/O 20 X-ray profusion 2110 12-31 2211 513 X-ray profusion l/O 21 210 7 3/o 2 X-ray profusion 2110 8.0 Chronic cough/phlegm < 10 years mining 52 10-19 years mining 46 220 years mining 59 Byssinosls M W Current smokers 25 19 Never smokers 18 15 Farmers lung disease 0.5 147 with cigarette smoking was accumulating. Nevertheless, a large proportion of the participants in these surveys smoked cigarettes. The findings from these surveys with regard to the prevalence of smoking are supported by larger data sets collected from population samples (Friedman et al. 1973; Sterling and Weinkam 1976). Friedman and colleagues (1973) questioned 70,289 participants in the Kaiser-Permanente Multiphasic Health Checkups program and found a higher proportion of smokers in those reporting occupational exposure to asbestos, silica, or fumes. Similarly, Sterling and Weinkam (1976) examined smoking patterns by employment status in data from the 1970 NHIS and found the prevalence of smoking to be highest among blue-collar workers. Association between occupa- tional group and cigarette smoking practices is addressed in detail elsewhere in this Report. Thus, in research and clinical care related to chronic occupational lung diseases, consideration must be given not only to occupational exposures but also to cigarette smoking. The remainder of this chapter describes the general patterns of lung injury by cigarette smoking and occupational exposures and the methods used for evaluating workers who are exposed to both. Patterns of Lung Injury The sites of lung injury caused by cigarette smoke and occupation- al agents may be broadly categorized as the large airways, the small airways, and the parenchyma. The effects of cigarette smoke on these sites are summarized in Table 3. A comparison of injury patterns from cigarette smoke and from selected, but representative, occupational exposures follows. Injury From Cigarette Smoke The pattern of lung injury associated with cigarette smoking has been comprehensively described elsewhere (US DHHS 1984). In the large airways, cigarette smoke causes an increase in mucous gland size and in goblet cell number. These changes result in increased mucus production and the associated symptom of chronic bronchitis. Large airway injury may contribute to airflow obstruction, but the peripheral airways are the predominant site of the increased airflow resistance in COLD (US DHHS 1984). Changes in the small airways are one of the earliest manifesta- tions of cigarette smoking. Niewoehner and colleagues (1974) exam- ined the lungs of 20 smokers and 19 nonsmokers who died suddenly at a mean age of 25 years. A pattern of small airways injury, termed "respiratory bronchiolitis," was readily identified, even in these young smokers. Clusters of brown pigmented macrophages were found in the respiratory bronchioles, which also displayed increased 148 TABLE 3.-Pathologic changes and manifestations of lung injury by cigarette smoke Large airways Small airways Parenchyma Pathologic changes Manifestations Mucous gland hyper- Goblet cell metaplasia. Emphysema, plasia. inflammation inflammation and minimal and edema, ' bronchial fibrosis of the interstital smooth muscle respiratory bronchiole fibrosis Symptoms Cough, phlegm Physical signs None Cough, phlegm Crackles Dyspnea Diminished breath SOUdS X ray None Pulmonary function ? j FEV, testing ? Linear opacities ? Linear opacities FEV,, 1 FEV,%, i FEV,, 1 FEV,%, : TLC, 1 RV. J DLCO `TLC, TRV. LDLCO Accelerated annual Accelerated annual decline of FEV, decline of FEV, numbers of inflammatory cells and denuded epithelium. To charac- terize the physiological consequences of small airways injury associ- ated with smoking cigarettes, Cosio and colleagues (1978) correlated small airways morphology with lung function in 36 patients under- going thoracotomy for a localized lesion. With increasing cumulative consumption, both inflammation and fibrosis of the respiratory bronchioles increased. Furthermore, airflow obstruction, as mea- sured by the ratio of FEV, to FVC or by the maximum midexpirato- ry flow rate (FEFzMNI, progressively decreased and residual volume increased with the amount smoked. Physiological measures of airflow obstruction correlated with the severity of small airways abnormalities. The major parenchymal injury associated with cigarette smoking is emphysema: "abnormal dilation of air spaces distal to the terminal bronchioles accompanied by destruction of air space walls" (US DHHS 1984, p. 119). Emphysema and small airways injury contribute to the physiological impairment found in COLD; in individual patients with COLD, either may be predominant, but both are probably important in most (US DHHS 1984). By itself, emphyse- ma is accompanied by spirometric evidence of airflow obstruction, increased lung compliance, and increased total lung capacity (TLC) and residual volume (RV). The diffusing capacity for carbon monox- ide varies inversely with the extent of emphysema (Park et al. 1970; Cotes 1979). Emphysema is also associated with abnormalities of gas exchange. 149 Cigarette smoking, through its effects on the small airways and lung parenchyma, produces the clinical syndrome of expiratory flow limitation with dyspnea. The chronic airflow obstruction found in COLD develops progressively and insidiously in most cases through a sustained excessive decline of ventilatory function (US DHHS 1984). In COLD, spirometry shows reduced FEV, and a reduced FEV, to FVC ratio; FVC may also be diminished. The airflow obstruction is accompanied by increases in RV and TLC (Boushy et al. 1971; Cotes 1979). Injury From Occupational Exposures For occupational exposures in the absence of cigarette smoking, the patterns of lung injury vary among the agents, presumably on the basis of differences in their physical and chemical properties. Although the clinical and physiological manifestations of occupa- tional lung injury may be distinct from those of cigarette smoking, overlap occurs for some exposures. As with cigarette smoke, chronic irritation of the large airways by dusts and gases is associated with mucous gland enlargement and mucus hypersecretion (Morgan 1978, 1984b). This pattern of injury has been well documented clinically and pathologically for coal and cotton dust (Douglas et al. 1982; Edwards et al. 1975; Kibelstis et al. 1973; Merchant et al. 1972). Gold miners and grain workers also develop chronic bronchitis attributable to occupational dust expo- sure (Irwig and Rocks 1978; Dosman et al. 1980). Industrial bronchitis may be associated with airflow obstruction. Hankinson and colleagues (1977) studied approximately 9,000 coal miners from 1973 to 1974. Among the nonsmoking miners with dust- induced bronchitis, decreased airflow at high lung volumes was demonstrated, a finding suggestive of changes in the larger airways. Abnormalities of the small airways seem to be one of the earliest responses to mineral dust exposure (Churg et al. 1985). In a recent study of hard-rock miners and people employed in the asbestos, construction, and shipyard industries, Churg and colleagues (1985) showed that the abnormalities of the respiratory bronchioles associ- ated with mineral dust are accompanied by airflow abnormalities. The lesions consisted of fibrosis and pigmentation in the small airways and were considered by these researchers to represent a nonspecific response to dust. Involvement of the small airways has also been demonstrated in workers with specific exposures. For example, the coal macule is characterized by the deposition of alveolar macrophages loaded with coal dust in the respiratory bronchioles (Morgan 1984a). Subsequent- ly, the involved respiratory bronchioles dilate, a change termed "focal emphysema" (Morgan 1984a). At this stage, individuals usually are asymptomatic and have no physical findings. The chest x 150