Tobacco Documents Online

I I ~ I I An Alternative Explanation for the Apparent Elevated Relative Mortality and Morbidity Risks of Spouses and Other Family Members of Smokers Associated with Exposure to Environmental Tobacco Smoke T.D. Sterling, Professor W.L. Rosenbaum, Senior Research Associate J.J. Weinkam, Professor Faculty of Applied Sciences, School of Computing Science Simon Fraser University Burnaby, British Columbia, Canada, V5A 1S6 May 18, 1994

SUNIlVIAR.Y I I ~ I I I I I In its proposed air quality standard in the workplace, OSHA (1994) faces the dilemma of a lack of evidence for elevated risk associated with actual workplace exposure to ETS. OSHA briefly discusses and attempts to resolve its' dilemma in 6ection IV. (Preliminary Quantitative Risk Assessment). OSHA concludes that "risk estimates calculated from studies of the general population or of selected subgroups, such as non-smoking wives of smoking husbands, are relevant to the working non-smoking population.". OSHA then defends extrapolating from assumed or observed risks of female spouses of smokers to all persons in the workplace by concluding that "it is the exposure to environmental tobacco smoke and not the environment in which that exposure occurs that is the important factor." (OSHA, 1994:15994). This set of assumptions does not justify OSHA's extrapolation from risks associated with having a smoking spouse to risks associated with ETS exposure in the workplace because there is an alternative explanation for the apparent elevated lung cancer risk found in some spousal studies and the failure to find such elevated lung cancer risks in studies of workplace ETS exposure. In this review we show that there is a strong possibility that factors other than ETS were acting in those spousal studies that compared lung cancer and other disease risks of non-smoking female spouses of smoking males with the same diseases among non-smoking female spouses of non-smoking males. Our analysis details an alternative explanation for elevation of relative risks observed in some spousal studies: The presence of a smoking spouse or parent within the household demonstrably corresponds to a greater likelihood that the household belongs to a I

2 I I I lower socio-economic stratum and that its' members are subjected to paraoccupational exposure than if no smoking spouse or parent is present. (Paraoccupational exposure is defined as an exposure to a substance outside the ... ...... ... ... occupational setting in which workers are exposed to that substance.) Insofar as service or industrial workers are much more likely to smoke than persons in other occupations and in higher socio-economic strata, observed differences in risk of morbidity or mortality ascribed to ETS on the basis of the comparison of households with or without smokers may be partly or entirely due to differences in socio- economic strata and paraoccupational exposure. The major supporting evidence are based: 1. On the substantial confounding of occupation and economic factors wit smoking patterns; 2. On the link between paternal and spousal occupation and disease; 3. On the possible exposure to carcinogens brought home by parents and spouses from the workplace; 4. On the inverse relation between socio-economic levels and mortality; I I I 5. On the fact that elevated lung cancer risks is reported in non-smoking • females living with smoking males but not generally for non-smoking males living with smoking females; 6. On the lack of evidence for an elevated lung cancer risk associated with EfiS exposure in the workplace; and 7. On the observation of "healthier" lifestyles of wives married to non- smokers. I

3 I I I I I Important factors besides paraoccupational exposure rooted in socio-economic differences between households with or without smokers are discussed as well. Our investigation not only of possible ETS effects but also on effects socio- economicc and paraoccupational variables concludes that it is the peculiar setting of spousal studies and related socio-economic factors involved in househol stu ies that make for an apparent elevated disease risk in spousal studies. There is no justification for extrapolating from assumed or observed risk levels in spousal studies to calculate risks associated with ETS exposure in the workplace. ~

- INTRODUCTION I I I I I I I I In its proposed air quality standard, OSHA (1994) faces a dilemma which never is directly discussed. That dilemma is the lack of evidence for elevated risk + associated with actual workplace exposure to ETS. There are 10 studies which calculate 13 relative lung cancer risks associated with workplace exposure to ETS , for males, females and all workers. Of these 13 relative risks, 3 are of borderline statistical significance. These are Fontham's relative lung cancer risk for females of 1.34 (confidence interval 1.03 - 1.73); Kabat and Wynder's relative lung cancer risk for males of 3.27 (confidence interval 1.01 - 10.60); and Wu-Williams relative lung cancer risk for females of 1.20 (confidence interval 1.00 - 1.40). The findings of all available studies combined would ordinarily be dismissed as not demonstrating a relationship between workplace exposure to ETS and lung disease. Yet OSHA is pressed to set a standard for regulating tobacco smoke exposures at the workplace. OSHA resolves this dilemma by concluding that "risk estimates calculated from studies of the general population or of selected subgroups, such as non-smoking wives of smoking husbands, are relevant to the working non-smoking population". _ _ OSHA. defends that extrapolation from female spouses of smokers to all persons in the workplace by concluding that "it is the exposure to environmental tobacco smoke and not the environment in which that exposure occurs that is the important factor. " (OSHA, 1994: 15994). .. OSHA does not further refer to its dilemma but instead, deals with it by assuming that the lung cancer relative risk observed by Fontham et al, (1991) of 1.34 is representative of the risk for workplace exposures and uses a spousal study,

5 I I I I I I I I I Helsing et al, (1988) to estimate the relative heart disease risk for workplace exposure -- a risk for which no direct workplace investigations exist at this time. A conclusion by OSHA that a risk exists in the workplace based on evidence ~ that is not related to workplace exposure and, in fact, is in conflict with negative results of workplace exposure studies, would have far reaching implications for , setting standards altogether. It is therefore important to investigate whether or not an explanation exists that would link the negative results of workplace ETS exposure to the apparent observed elevation in risk of family members of smokers, especially of non-smoking spouses. The alternative to OSHA's reasoning is that the environment in which the exposure occurred in so called spousal ETS studies ig an important factor. In this review we show that there is a strong possibility that factors other than ETS were acting in those spousal studies that compared lung cancer and related disease of ~ non-smoking female spouses of smoking males with the same diseases among non- smoking female spouses of non-smoking males. We believe that there is an alternative explanation for these observed elevation in relative risk, namely: The presence of a smoking spouse or parent within a household corresponds to greater likelihood that the household belongs to a lower socio-economic stratum and that its members are subjected to paraoccupational exposurei than when a smoking spouse or parent is not present. Insofar as service and industrial workers are much more likely•to smoke than those in other occupation and in higher socio-economic IParaoccupational exposure is defined as an exposure to a substance outside the occupational setting in which workers are exposed to that substance. It usually is an exposure of workers' family materials brought home on the hair, skin, and clothing ofworkers. ~

6 strata, observed differences in risk of morbidity or mortality ascribed to ETS on the basis of a comparison of households with and without smokers may be partly or entirely due to differences in socio-economic. strata and in paraoccupational exposures. MAJOR SUPPORTING EVIDENCE2 I I I I I I I I I I There is persuasive, albeit only indirect, evidence that an elevated disease risk in homes of smokers (primarily in homes of male smokers) is due at least in part to socioeconomic and especially occupation-related factors. This evidence rests on six sets of interrelated observations: 1. On The Extent of Confounding of Occupation and Economic Factors with Smoking Patterns . Sterling (1976, 1990) and Weinkam (1987) have shown a pattern between occupation of members of the household and their smoking behavior. Two of the findings by Sterling and Weinkam are especially relevant: (a) Nonsmoking females whose husbands smoke are about 40% more likely to live in a blue collar household and about 50 % less likely to live in a professional household than are nonsmoking females whose husbands do not smoke. (b) Infants whose mothers smoke are 25 % more likely to live in blue collar households and one third less likely to live in professional households than infants whose mothers do not smoke (Sterling, 1990). 2 An alternative explanation for the apparent association with ETS exposure of non-smoking _ spouses is discussed in a lengthy paper presently uader review by the Epidemiologic an Environmental Journal. I

7 I I I I I Differences in the prevalence of smoking among various occupations are indeed striking. Table 1 shows the 20 occupations with the highest smoking prevalence and Table 2 shows the 20 occupations with the lowest smoking prevalence among white males in 1970. While these prevalences were observed in 1970, the last year the ~ National Health Interview Survey collected detailed information about smokers in the total probability'sample, they give insight into the relative levels of smoking by occupation that are relevant today - even though the absolute number of smokers has declined for all occupations. Prevalence of tobacco use among the 20 highest ranking occupations ranges from 72% to 38%. On the other hand, prevalence of smoking among the lowest ranking occupations ranges from 30% to 7%. Tables 1 and 2 also show that there are no occupations with high smoking prevalence that also might be expected to incur relatively low exposure to toxic or carcinogenic substances on the job. All of the 20 occupations with high smoking prevalence expose workers to various types of toxic products in many instances to substances that may be brought home on the worker's person. The reverse is true for the lowest levels of smoking. With the possible exception of chemical engineers and farmers, all occupations with low smoking prevalence involve work in clean environments and most individuals in these occupations belong to higher socio%conomic strata. All 20 occupations with the highest prevalence of current smokers are blue collar occupations. None of the occupations with the lowest smoking prevalence are blue collar._ Weinkam and Sterling (1987) showed that by 1980, the percent of current smokers had decreased and the percent of former or never, smokers had increased

8 I I I I I I I I I I I I I I ~ within each occupation. However, despite the decline among current smokers and increase among former and never smokers, the basic pattern of distribution of occupation within smoking categories remains substantially unchanged from 1970 to 1980. Also, changes in smoking habits for cohorts moving from younger to older ages from 1970 to 1980 are similar for all occupations. I The 1970 National Health Interview Survey also provides information which is especially relevant to comparison of households with or without male smokers. Information in the 1970 NHIS makes it possible to compare the percent of blue collar and professional occupations in households with different patterns of smoking of husbands and wives (i.e. where both the husband and wife smoke, where only the husband smokes, only the wife smokes, and neither smoke). Because spousal studies of the effect of ETS compare disease incidence of nonsmoking wives married to smoking males with nonsmoking wives married to nonsmokers, these two categories are especially important. We see from Figure 1 (based on the population of employed males of ages 20-64) that the percentage of blue collar households was 48.1% where the husband smokes and the wife doesn't, but only 35.1% where neither spouses smoke-- a ratio of 1.37. The opposite tendency is seen for professional occupations. For employed males only, 7.1% of households with nonsmoking wives and smoking husbands were of a professional background while 17.3% were of that background if neither the husband nor the wife smokes or a ratio of 0.40. It is obvious that a comparison of households with a nonsmoking wife and a smoking husband with households where neither the wife nor husband smokes compares two groups of households which differ quite significantly. These differences also affect the likelihood that either the nonsmoking wife or other

9 I I I members of the family may be paraoccupationally exposed to materials brought home from the workplace by industrially employed members of the family. That likelihood is greater if the husband is a smoker. The 1970 NHIS also made possible comparison of the blue collar and professional backgrounds of husbands of smoking and nonsmoking mothers. Figure ~ 2 summarizes that information and is based on wives of employed males ages 20-64. This figure shows that for households in which the mother of an infant was a smoker, 51.4% were of blue collar background whereas in households where the ~ mother did not smoke 41.8% were of blue collar background. Again, the relative frequencies are the reverse for professional households with respect to smoking and ' nonsmoking mothers of infants. In families of employed males ages 20 - 64, the ~ percent of professional households is 10.2 if the mother smokes versus 14.9 if the mother does not smoke. I I I I I Again, these data show that a comparison of households in which the mother of an infant smokes as opposed to households in which the mother of an infant does not smoke also compares households with a greater prevalence of blue collar backgrounds and a smaller prevalence of professional backgrounds. Since smoking is distributed unequally among different categories of individuals, any comparison of infants based on differences in mother's smoking habits must adjust for the differences in socioleconomic backgrounds of the infants who are being compared and of the possibilities of their paraoccupational exposures. ~ 2. On The Link Between Paternal and Spousal Occupation and Disease Epidemiologic studies of parental occupation and childhood cancer date back ~ to 1974 when Fabia and Thuy reported a two-fold excess risk of cancer mortality N among children less than 5 years of age whose fathers had jobs classified as 4 ~ CJI ~ ~ Q? ~ ~