Tobacco Documents Online

s06s C ats on Coagressional Research Service Assessment of the ; Sealth Risks of Environmental Tobacco Sasoke I I I I I I I I I I I I I I I office on Smoking and Health May 12, 1994 quite mialea ing and the authors leave themselves in no position to make any conclusi e statements about the relationship between ETS exposure and The Con ressional Research Service, in a report to Congress, attempted to assess th health risks of environmental tobacco smoke (ETS). The report's "re iew" of the scientific evidence relating to the health effects of ETS is fl wed. The report concludes that the link between passive smoking and isease is uncertain, and goes to great lengths to attempt to point out scientific flaws in the epidemiologic studies. However, the authors of t4e report ignore basic principles of epidemiology as well as the nse ~ody of research on the subject of passive smoking. I The CRS~report can hardly be considered a review or an analysis of the epidemioloqi evidence relating to the health effects of ETS exposure. There are hu dreds of epidemiologic studies of the relationship between ETS and health o4tcomes, yet the CRS report directly cites only two of these studies. In tact, the report cites as many articles from Consumers' Research as it does from the epidemiologic literature. Despite its failure to actually ieview or analyze the evidence scientifically, the report draws sweeping condlusiona that are not supported by the data. The presentation of the evidedce is so incomplete and so selective that the •review" is health. The aut~lors are both economists, and are apparently not qualified to conduct a va3Jid epidemioloqic analysis. in testimony before the Unitcd States Senatd, they themselves admitted that they have no expertise in assessing tb mechanisme of disease causation: "Please note that we are trainad as e onomists and our area of expertise relates to economic analysia aad the associated areas of statistical inference and quantificati n of effects for purposes of cost-benefit analysis and related economic pol ies. We do not have technical expertise in the physiologicall and biologic 1 transmission mechanisms of disease causing agents." The EpAlreport was written and reviewed by many reputable scientists trained in th~e field of epidemiology, who do have expertise in the phyaiologieale and biological tranestisaion m.chaniams of disease causing agents, and a therefore a credible assessment of the health risks of ETS exposure. Ba q publicipolicy decisions on the conclusions of two economists, tIIer than on the consensus of most reputable epidemiolog3sts, would be a se ious mistake. The foll`owing are the most important flaws in the CRS "review" of the health risks of environmental tobacco smoke: 1. On page 45, the report questions whether there is a dose-response relationship between tobacco smoke exposure and adverse health effects over the wide rangb of E'rs levels to which nonsmokers are exposed. The opidemiologicl data in this regard, however, are absolutely clear. Risk of lung cancer h6s consistently been shown to rise as ETS exposure levels increase. ~ 2. On page 45, the report suggests that there may be a threshold level txs below which t~ere is no biological damage from ETS exposure. The existence ~ ~ .~

of a thresho d for tobacco smoke in lung cancer is theoretically possible, but there are no empirical data to support the threshold concept. Indeed, the data tha do exist clearly favor the non-threshold hypothesis. virtually ev ey one of the thousands of studies evaluating lung cancer among smoker shows that risks increase in proportion to level of smoking, with demonstlable increases in risk even among the very lightest smokers. 3_ On page d~', the report suggests that the relationship between tobacco smoke exposu~te and health effects at the level of passive smoke exposure i:' such that th# effects would be minor. The CRS report refers to the 1989 Surgeon GeneTal's report citation (page 44) of a cohort study of lung cancer and sking among about 34,000 British physicians. This study , estimated th t risk of lung cancer rose in proportion to the square of th e number of ci~arettes smoked per day. The CRS states that this implies that the relati.on is "strongly nonlinear". This statement is questionable, sincq~ the departur~ from linearity with the quadratic model may not be great. J/ More importa tly, the CRS report ignores data presented on page 45 of th ~e same Surgeon;General's report. There a graph is presented showing a clearly linear relatlon between the number of cigarettes sawked per day and lung cancer risk.;The data on page 45 derive from a much larger cohort study of , nearly 250,0Q0 American veterans. 1 4. On page 47, the report criticizes the weighting of studies by the . variance of the risk estimates. This is a very rational and well accepte d technique inlmeta-analysis and statistical analyses in general. The report claims that itudiea with large sample size would +end to find more significant ffecta for passive stiwking. Statistically, this is true only if there rea ly is a true effect of passive smoking. If there were no true effect, then large studies would have an equal probability of detecting a significant increase or'decrease in lung cancer risk associated with ETS ' exposure. ~ 5. On pages 4~7 and 48, the report criticizes the use of a one-tailed test of significar~ce (approximately equivalent to using a 90% confidence ' interval to tleat the chance that a relative risk differs from 1.0), even though it of~ers no plausible reason to believe that exposure to ETS may decrease lung cancer risk. Although 95% confidence limits are commonly used, 90% coqidence intervals are often used in health studies, and are fact standardd in epidemiologic studies of agents such as radiation, which like tobacco, is a strong lung carcinogen. in putting a largQ emphasis on this rather nor part of the analysis, the report ignores the importance of other fac rs used by epidemiologists in causal inference, such as . consistancy association, strength of association, temporality of th i1e association, i ose-response relationship and biologic plausibility. 6. on page 48, the report suggests that recall bias may invalidate the , conclusion of+ an adverse health effect of ETS exposure. This ignores the fact that the~e is a consistent association between $TS exposure and disease even n cohort studies, in which recall bias cannot occur, because ~ exposure is a eertained before disease has occurred. 7. on page 48, the report suggests that confounding factors may explain the observed asso4ation between ETS exposure and lung cancer. However, it does, not suggest aisingle factor that could plausibly explain the association. Specifically,iit does not point out any factor that is a strong independent risk factor fPr lung cancer that is also strongly related to ETS exposure. Moreover, th.,consietency of results across a wide variety of study designs' and populations argues against confounding as an explanation of the consistent sthdy findings. r•.~~,' I 4" Co r,3 ~ ra U1 4 Q '

I I I I I I I I I I I I I I I I I 8. In their estimony, the authors of the CRS report suggest that the presence of consistent dose-response relationship in the epidemiologic studies shou d not have increased confidence in the conclusion of a causal relationship between ETS and lung cancer. This statement is in direct conflict wit one of the most basic principles of epidemiology. 9. In their ~estimony, the authors suggest that the findings of two recent studies, pub iished too late to be included in the EPA review, cast doubt on the relation hip between ETS and lung cancer. on the contrary, in combination hese studies are entirely consistent with the body of epidemiologic evidence on the issue, and they add additional support for the conclusi n that ETS increases lung cancer risk. The authors of the CRS report ignor.d the published conclusions of both of these studies. In the 4rownson study, the authors found an elevated lung cancer risk for adults w th more than 40 pack-years of exposure to ETS from household members, and,concluded that "exposure to high levels of environmental tobacco smoke in adulthood increases the risk of lung cancer in nonsmokers."I In the tockwell study, the authors found that the risk of lung cancer more than do led for women with more than 40 pack-years of household ETS exposure and conclude that "long-term exposure to environmental tobacco smoke increades the risk of lung cancer in women who have never smoked." 10. The Aith~rs attempt to determine the risk of passive smoking by extrapolatin from urinary cotinine levels in active compared to passive smokers. The conclude that the average exposure is 0.1 cigarette equivalents er day. Bowever, using this same approach, Guerin et al. have shown that 0.I1 cigarettd equivalents of exposure to nicotine in inhaled air implies expo ure to 60 cigarette equivalents of pyridine, and 56 cigarette equivalents f ammonia. The concentrations of many of the carcinogenic components o tobacco smoke are much higher in sidestream than in mainstream s ke. But it is exposure to these components, and not to nicotine, that will determine the lung cancer risk among exposed nonsmokers. I I

I VALIDITY OF THE CRS REVIEW OF THE EPA REPORT Q. The Congressional Research Service recently completed a review of the EPA report that classified environmental tobacco smoke as a Group A "known human" carcinogen. This review concludes that the link between passive smoking and disease is uncertain and that the EPA report's review of the scientific evidence relating to the health effects of ETS is flawed. What is your reaction to this review? A. There are a number of points that need to be addressed surrounding this issue: 1. The EPA report was written and reviewed by many reputable scientists trained in the field of epidemiology, who have expertise in the physiological and biological transmission mechanisms of disease causing agents, and is a therefore a credible assessment of the health risks of ETS exposure. 2. The CRS report, on the other hand, is a review written by two economists not trained in the principles of epidemiology, and is not a credible assessment of the health risks of ETS exposure. 3. The CRS report is not a complete review or analysis of the epidemiological evidence relating to the health effects of ETS exposure. There are hundreds of epidemiologic studies of the relationship between ETS and health outcomes, yet the CRS report direct_y cites only two of these studies. This report draws sweeping conclusions that are not supported by the data. This incomplete and misleading presentation of the evidence makes it impossible for this report to make any conclusive or credible statements about the relationship between ETS exposure and health. 4. CDC has recognized ETS as a significant public health issue since the 1986 Surgeon General's report, "Health Consequences of Involuntary Smoking." As recently as 1991, CDC published the conclusion that ETS is a potential occupational carcinogen, and recommended that exposures be reduced to the lowest feasible concentration. 5. The Centers for Disease Control and Prevention (CDC) has and continues to strongly support the findings and methodology of the EPA report. I I I I I I I I I I I I I I I I I

I I I I I I I I I I I I I I I I AREN'T THESE CRITICISMS WORTH CONSIDERING? Q. How can you simply dismiss the findings of the CRS review as a flawed review written by two economists not trained in the principles of epidemiology? Aren't their criticisms worth considering? A. No. The authors of this review are both economists not epidemiologists. In testimony before the United States Senate, they themselves admitted that they have no expertise in assessing the mechanisms of disease causation. In their written testimony, the CRS economists made the following statement: Please note that we are trained as economists and our area of expertise relates to economic analysis and the associated areas of statistical inference and quantification of effects for purposes of cost-benefit analysis and related economic policies. We do not have technical expertise in the physiological and biological transmission mechanisms of disease causing agents. This lack of technical expertise is reflected by the fact that the authors of the CRS report ignored basic principles of epidemiology as well as the immense body of research on the subject of passive smoking. I believe that public health policy decisions should be based on the sound scientific conclusions of trained epidemiologists, not the conclusions of two economists who lack training and technical expertise in the field of epidemiology.

RECENT ETS STUDIES CITED IN THE CRS REVIEW In their testimony, the authors suggest that the findings of two recent studies, published too late to be included in the EPA review, cast doubt on the relationship between ETS and lung cancer. Are you familiar with these studies? Are there suggestions accurate? A. Yes, we are very familiar with the studies in question and no the suggestions contained in the CRS review are not accurate. On the contrary, both of these studies are entirely consistent with the body of epidemiologic evidence on the issue, and they add additional support for the conclusion that ETS increases lung cancer risk. The authors of the CRS report ignored the published conclusions of both of these studies. In the Brownson study, the authors found an elevated lung cancer risk for adults with more than 40 pack-years of exposure to ETS from household members, and concluded that "exposure to high levels of environmental tobacco smoke in adulthood increases the risk of lung cancer in nonsmokers." In the Stockwell study, the authors found that the risk of lung cancer more than doubled for women with more than 40 pack-years of household ETS exposure and conclude that "long-term exposure to environmental tobacco smoke increases the risk of lung cancer in women who have never smoked." I I I 1, I I I I I I I I I I I I

I I 4- I I I I I I I I I I I I I I I I A. METHODOLOGICAL CRITICISMS The CRS review levies a number of criticisms against EPA for the research methods used in their study. Specifically, the CRS review criticizes the EPA's weighting of studies by the variance of the risk estimates, the use of a one-tailed test of significance, and the fact that the studies with larger sample sizes would tend to find more significant effects for passive smoking. How would you respond to these criticisms? There are very reasonable explanations for each of these criticisms. 1. Weighting studies by the variance of the risk estimates is a very rational and well-accepted technique in meta-analyses, and statistical analyses in general. 2. The use of 90% confidence intervals was very appropriate for this study because there was no reason to believe that ETS would DECREASE the risk of lung cancer. Further, the use of a 90-percent confidence interval is a minor part of this analysis. By placing a large emphasis on this, the report ignores the importance of other factors used by epidemiologists in causal inference, such as consistency of association, strength of association, temporality of the association, dose-response relationship and biologic plausibility. 3. Statistically, studies with larger sample sizes will tend to find more significant effects for passive smoking only if there really is a true effect of passive smoking. If there were no true effect, then large studies would have an equal probability of detecting a significant increase or decrease in lung cancer risk associated with ETS exposure.

RECALL BIAS Q. The report suggests that recall bias may invalidate the conclusion of an adverse health effect of ETS exposure. Do you agree with this criticism? A. No. This criticism ignores the fact that there is a consistent association between ETS exposure and disease even in cohort studies, in which recall bias cannot occur, because exposure is ascertained before disease has occurred. I I I I I I I I I I I I I I I Q .~ ~ ~ CO C~ CA I 00 I- I

I I I I I I I I I I I I I I I CRS RISKS ESTIMATES The CRS review attempted to determine the risk of passive smoking by extrapolating from urinary cotinine levels. What is your analysis of this portion of the review? A. The CRS report concludes that the average ETS exposure is .1 cigarette equivalents per day. Using this same approach, Guerin et al. have shown that .1 cigarette equivalent of nicotine exposure implies exposure to 60 cigarette equivalents of pyridine and 56 cigarette equivalents of ammonia. The concentrations of many of the carcinogenic components of tobacco smoke are much higher in sidestream than in mainstream smoke. But it is exposure to these components, and not to nicotine, that will determine the lung cancer risk among exposed nonsmokers. I