Tobacco Documents Online

THE UNIVERSITY OF ALABAMA THE GRADUATE SCHOOL AI'PUED STATISTICS PROGRAM September 24, 1990 Project Officer for Environmental Tobacco Smoke Technical Information Staff Office of Health and Environmental Assessment (RD-689 U.S. Environmental Protection Agency 401 M Street, SW. Washington, DC 20460 Dear Project Officer: This letter is in response to the request for public comments announced in the Notices of the Federal Register, June 25, 1990. My comments relate to the publication entitled "Health Effects of Passive Smoking; Assessment of Lung Cancer in Adults and Respiratory Disorders in Children; External Review Draft" prepared by the Environmental Protection Agency. This publication is apparently the basis for the external review draft of "Environmental Tobacco Smoke: A Guide to Workplace Smoking Policies," which also invited public comment. As a professional statistician and author of many books and articles on statistical theory and methods, I have a special interest in the proper uses of statistical techniques in real data situations, especially those that affect public policy. My current position is Thomas D. Russell Professor of Applied Statistics in the Graduate School of the University of Alabama and Chairman of the University's Applied Statistics Program. I received my Ph.D. in statistics from Virginia Polytechnic Institute and State University in 1963. My bachelor's and master's degrees, both in mathematics, were earned at Duke University. I was elected a Fellow of the American Statistical Association in 1972. In 1973, I was a senior Fulbright-Hays Scholar at the Indian Statistical Institute. In 1985 I received the Burlington- 75 6IDGOOO /(205) 348-6085 / BOX 870226 / TUSCALOOSA. ALA6AMA 35487-0226

2 Northern Award for Faculty Scholarship. Some of my pertinent professional activities include serving on the Board of Directors of the American Statistical Association, serving as an editorial collaborator on several statistical journals, and reviewing grant proposals for the National Science Foundation. A copy of my complete curriculum vitae is attached as Appendix B. I have reviewed in great detail 24 recent publications on the subject of lung cancer risk from passive smoking, particularly the studies concerning nonsmoking wives of men who are smokers. The references to these studies are listed in Appendix A. On the basis of my examination of this literature, I have some problems with the conclusions stated in the EPA document "Health Effects of Passive Smoking: Assessment of Lung Cancer Risk in Adults and Respiratory Disorders in Children," May 1990 [hereinafter referred to as EPA (1990)]. These problems fall into the following categories, each of which is discussed below. I. Definition of Exposure to E.T.S. (Environmental Tobacco Smoke) II. Sources of Bias in Studies III. Reliance on Studies from Foreign Countries IV. Little Attention Paid to Confidence Limits Given in Individual Studies V. The Meta-Analysis Approach I. Definition of Exposure to E.T.S. Most of the studies define the cases with exposure to E.T.S. as a nonsmoking female married to a smoking male and the controls as a nonsmoking female married to a nonsmoking male. These definitions are convenient but have little to do with actual exposure and nonexposure to E.T.S. 1. A smoking male may smoke 20 cigarettes a day, but all at work and none at home, or vice versa. When a smoking male does smoke at home, is his wife present or absent? In the same room? 2. A nonsmoking female may be exposed to E.T.S. through other residents and/or guests in the home or colleagues in the workplace, irrespective of whether her husband smokes.

3 3. E.T.S. is present in the atmosphere at all times, but in different amounts at different locations and times. 4. Since there is no such thing as nonexposure to E.T.S. for persons not confined inside a perfectly controlled environment, some attempt must be made to measure and quantify the actual amount of exposure by both cases and controls before groups can be formed. Further, attention must be paid to exposure over time, as children, in dormitories or residence halls, with former spouses or live-ins, etc. II. Sources of Bias in Studies Many of the studies do offer caveats about possible sources of bias. Some of these are: 1. Misclassification of smokers as nonsmokers, and vice versa. 2. Small data sets in many cases. 3. Assumptions of normal distributions and/or large sample sizes to carry out the statistical tests. 4. Possible relevant factors other than E.T.S. are completely ignored in most studies. Some of these factors are heredity, diet, culture, environment at home and work, ventilation, type of home cooking, type of heating and cooling at home and work, pollutants at home and work, modes of transportation, etc. This means that each study is predicated under the a priori assumption that smoking is the only possible factor. Kabat et al.(1984) suggest that adenocarcinoma might be estrogen related, but no one seems to have paid any attention to this or other matters that make females different from males. 5. Not only is smoking the only factor considered, but also no attention is paid to different kinds of smoking, different kinds of cigarettes, or different kinds of tobacco. 6. There is, especially in relation to smoking, a publication bias. Studies that find no significant results and therefore cannot make

4 headlines are frequently found not "suitable" for publication, for one reason or another. III. Reliance on Studies From Foreign Countries Of the 24 papers I studied, nine relate to persons in the U.S. and 15 relate to persons outside the U.S., including Hong Kong, China, Japan, Sweden, Greece, England and Scotland. Of the 19 papers listed in Table 3-5 on pages 3-15 to 3-17 of EPA (1990), seven relate to U.S. persons and 12 to persons outside the U.S. For the papers listed in Table 3-6 on page 3-23, five relate to the U.S. and eight to foreign countries. 1. Since none of the other possible relevant factors, like those listed in 11.4 is considered in most of these studies, the subjects are different and extrapolation to U.S. persons is not possible, especially when the purpose is to make workplace policies for the U.S. 2. Tobacco and types of cigarettes are not the same for all countries in the world. See 11.5. IV. Little Attention Paid to Confidence Limits in Individual Studies The-EPA (1990) report seems to stress the risk ratios computed in the individual studies, rather than the confidence limits. Only by looking at confidence limits can we take into account margins of error. 1. Of the nine papers I studied relating to persons in the U.S., almost all of the computed confidence intervals for odds ratio or risk ratio include one. A confidence interval that includes one implies no significant additional risk to nonsmoking females exposed to E.T.S. through a smoking husband. The majority of the foreign studies also give confidence intervals that include one. 2. For the seven papers listed in Table 3-5 on pages 3-15 to 3-17 of EPA (1990) for U.S. females, all seven confidence intervals include one. In fact, only four out of the 19 given confidence intervals do not include one, and

5 these four are all for studies in foreign countries, namely, China, Hong Kong (two), and Greece. In Table 3-6 on page 3-23 of EPA (1990), all of the five U.S. studies have confidence intervals that include one. V.__The Meta-Analysis Approach Meta-analysis is appropriate only for situations in which every factor is comparable, e.g. selection of cases and controls, diagnosis, classifications, areas, external factors, etc. 1. The studies combined for the EPA (1990) meta- analysis do not meet any of the criteria that justify meta-analysis. It is a perfect example of combining apples, oranges, lemons and even some prunes. 2. If a meta-analysis is to be done, it should be carried out using only comparable studies based on U.S. females as cases and controls. Given the magnitudes of the confidence intervals for odds and risk ratios reported in these individual studies, I suspect that the overall risk ratio confidence interval will include one and therefore not indicate a significant risk of lung cancer for U.S.-females exposed to E.T.S. through a smoking spouse. Conclusions The conclusion in EPA (1990) is not warranted by either the individual studies or the meta-analysis study that was reported. More and better quality studies need to be carried out in the U.S. before any appropriate conclusions about E.T.S. can be made and any meta-analysis should be done properly. I appreciate the opportunity to express my comments on this important issue. Sincerely, Jean D. Gibbons Russell Professor of Applied Statistics and Chairman, Applied Statistics Program

APPENDIX A REFERENCES Akiba, S., et al., "Passive Smoking and Lung Cancer Among Japanese Women," Cancer Research 46: 4804-4807, 1986. Brownson, R.C., et al., "Risk Factors for Adenocarcinoma of the Lung," American Journal of Epidemioloay 125(1): 25-34, 1987. Buffler, P.A., et al., "The Causes of Lung Cancer in Texas," Lung Cancer: Causes and Prevention, eds. M. Mizzell and P. Correa (New York: Verlag Chimie International, 1984): 83-99. Chan, W.C. and S.C. Fung, "Lung Cancer in Non-Smokers in Hong Kong," Cancer Campaign Vol. 6, Cancer Epidemioloay, ed. E. Grundmann (Stuttgart: Gustav Fischer Verlag, 1982): 199-202. Correa, P., et al., "Passive Smoking and Lung Cancer," The Lancet II: 595-597, 1983. Gao, Y.-T., et al., "Lung Cancer Among Chinese Women," International Journal of Cancer 40: 604-609, 1987. Garfinkel, L., "Time Trends in Lung Cancer Mortality Among Nonsmokers and a Note on Passive Smoking," Journal of the National Cancer Institute 66: 1061-1066, 1981. Garfinkel, L., et al., "Involuntary Smoking and Lung Cancer: A Case-Control Study," Journal of the National Cancer Institute 75(3): 463-469, 1985. Geng, G.-Y., et al., "On the Relationship Between Smoking and Female Lung Cancer" Smoking and Health 1987, eds.-M. Aoki, et al. (Amsterdam: Excerpta Medica, 1988): 483-486. Gillis, C.R., et al., "The Effect of Environmental Tobacco Smoke in Two Urban Communities in the West of Scotland," ETS-Environmental Tobacco Smoke. Report_for a Workshop__on__Effects and Extiosure ve s, eds. R. Rylander, et al., European Journal of Resniratorv Diseases, Supplement 133(65): 121-126, 1984. Hirayama, T., "Non-Smoking Wives of Heavy Smokers Have a Higher Risk of Lung Cancer: A-Study from Japan," British Medical Journal I, 282: 183-185, 1981. Humble, C.G., et al., "Marriage to a Smoker and Lung Cancer Risk," American Journal of Public'Health 77(5): 598-602, 1987. Inoue, R. and T. Hirayama, "Passive Smoking and Lung Cancer in Women" Smoking and Health 1987, eds. M. Aoki, et al. (Amsterdam: Excerpta Medica, 1988): 283-285.

Kabat, G.C. and E.L. Wynder, "Lung Cancer in Nonsmokers," Cancer 53(5): 1214-1221, 1984. Koo, L.C., et al., "Measurements of Passive Smoking and Estimates of Lung Cancer Risk Among Non-Smoking Chinese Females," International Journal of Cancer 39: 162-169, 1987. Lam, T.H., et al., "Smoking, Passive Smoking and Histological Types in Lung Cancer in Hong Kong Chinese Women," British Journal of Cancer 56(5): 673-678, 1987. Lee, P.N., et al., "Relationship of Passive Smoking to Risk of Lung Cancer and Other Smoking-Associated Diseases," British Journal of Cancer 54: 97-105, 1986. Pershagen, G., et al., "Passive Smoking and Lung Cancer in Swedish Women," American Journal of Epidemiology 125(1): 17-24, 1987. Shimizu, H., et al., "A Case-Control Study of Lung Cancer in Nonsmoking Women," Tohoku Journal of Experimental Medicine 154: 389-397, 1988. Svensson, C., et al., "Smoking and Passive Smoking in Relation to Lung Cancer in Women," Acta Oncologica 28(5): 623-629, 1989. Trichopoulos, D., et al., "Lung Cancer and Passive Smoking," International Journal of Cancer 27(1): 1-4, 1981. Trichopoulos, D., et al., "Lung Cancer and Passive Smoking: Conclusions of Greek Study," The Lancet II: 677-678, 1983. Varela, L.R., Assessment of the Association Between Passive Smoking and Lung Cancer, dissertation submitted to Yale University, 1987. Wu, A.H., et al., "Smoking and Other Risk Factors for Lung Cancer in Women," Journal of the National Cancer Institute 74(4): 747- 751, 1985. -