Tobacco Documents Online

I I I I I I I I I I I I I I I I I J+~tEO 5r~r~S UNITED STATES ENVIRONMENTAL PROTECTION AGENCY n v WASWNGTON, D.C. i 20460 ~ ~ ~~r~[ ppptE~ Indoor Air Quality Proposed Rule OFFICE OF RESEARCH AND DEVELOPMENT Post-Post Hearing Comments . Docket No. H-122, Room N-2625 OSHA Docket Office U.S. Department of Labor February 9,_ 1996 200 Constitution Ave. N.W. ^ ^ . ..~ ~.._...,~ _ .~ Washington D.C. 20210 ~v.:..~. ... . . .~. ~ Dear Sir or Madam: As my additional comments on the environmental tobacco smoke (ETS) portion of the proposed Indoor Air Quality (IAQ) Rule, I wish to comment on a portion of a comment to OSHA by witness Dr. Stanton Glantz, da'.:ed January 23, 1996. I refer to his reference to the recently released report of the Congressional Research Service (CRS) on "Environmental Tobacco Smoke and Lung Cancer Risk" (Redhead, S. and Roweberg, R.; CRS, Library of Congress, Nov. 14, 1995). I am familiar with that report since I was asked for review comments on an earlier draft and participated in a workshop review in June, 1995. As an EPA reviewer, I and my EPA colleague, Ms. Jennifer Jinot, provided eighteen pages of official EPA comments and concerns to that draft through our office director, Mr. Michael Callahan (see attached letter from Michael Callahan to Dr. David Gushee, Congressional Research Service, September 29, 1995). While the final report did incorporate some of these comments, there were many which it failed to acknowledge. Ms. Jinot and I have also provided these additional concerns on the final report to Dr. Dan Mulhollan, Director of the CRS, in a January 31, 1996 letter from Mr. Callahan (attached). These two letters reflect my comments on the CRS report. I also enclose a "fact sheet" prepared by Ms. Jinot titled "Comments on the CRS Report, ETS and Lung Cancer Risk" which highlights some of EPA's major concern. I also wish to make a few comments on the comments to OSHA by Dr. Michael Ogden, R.J. Reynolds Tobacco Company, September 1, 1995, Comments on Smoking-Status Misclassification. His Appendix A is a multi-authored manuscript submitted for publication titled "National Incidence of Smoking and Misclassification Among the U.S. ~ cs ~ Married Female Population". It is a large study, and potentially ca 1%3 worthwhile in providing information on smoker-status m misclassification rates. However, since the subjects were picked ~ w * by mall-intercept and the interviews were actually conducted by a 4 ao 1 Printed on Recycled Paper I

Reynolds' employed Ohio market research firm rather than epidemiologist interviewers, the results may be suspect. Reynolds ran smaller studies (not in the references) with similar recruitment designs in the past which were submitted to EPA. They also have a large 16 city ETS personal exposure study, in conjunction with Oak-Ridge scientists, which estimates smoker- status misclassification rates as a subtopic. All these studies use, I think, the same market research firm to conduct the interviews. As a general observation these studies all have higher misclassification rates than those using other personnel. In my opinion, the study was not done as a basic research design; it was intended to estimate smoker-status misclassification rates in order to produce bias estimates for downward adjustment of the relative risk estimates in the U.S. spousal ETS - lung cancer studies among female never-smokers. This is exactly what is done in the discussion section in its use on EPA's meta-analysis (pgs.- 20-21). While the study can be presented independent of the arguments for bias, the authors make the bias arguments and application to the ETS-lung cancer studies and the EPA analysis a central part of their paper (see statement top of page 20 - "the misclassification bias (is) of greatest interest since it contributes the largest bias to observed risk estimates" ). This leads me to question the age selection of this sample. Considerina the large mean•age differences in these misclassification rate studies (about 40 in this study vs. the mid sixties in the lung cancer case-control studies), the generalization of both its smoker-status misclassification rates and concordance rates to the U.S. ETS lung cancer studies would be highly suspect. While the EPA report criticized by the authors also used studies whose rates were less than ideal, those were the best available at the time. Considering the known purpose of this study and the apparently careful and tried design, it is puzzling why no attempt was made to sample the proper age distribution. In their abstract (pg. 2) the authors claim that "these data indicate that misclassification bias alone is likely to explain any lung cancer risk elevation observed in the U.S. epidemiology of ETS exposure among nonsmoking women." This is an overstatement based on their analysis in the discussion section of pooled average relative risk estimates presented in the EPA report. Considering the large non-differential exposure misclassification downward bias, especially between the un- and lower-exposed groups, most of the increased risk observable is in the highest exposed groups, and Dr. Ogden should have acknowledged this. These small smoker-status misclassification bias estimates cannot explain the increased risks seen consistently in the highest exposure groups in either nearly all these U.S. studies or in the many non-U.S. studies. Furthermore, while the EPA used 1.09 as the smoker-status misclassification rate for the U.S. current smokers categorized as never smokers, bias estimates were calculated separately for the individual studies. Thus, I believe his Figure 4, should have reflected this. 2 I I I I I i I I I I I I I I I I 111) a . IM na I

I I I I I I I I I I I I I I I I I I Since Dr. Ogden wants to use these misclassification rates to estimate misclassification bias in ETS-lung cancer studies, I feel they should have clarified the difference between rates and bias in order to specify conditions for use of these rates in both the case-control and cohort studies. For example, in case-control studies differential misclassification rates between cases and controls could create very different biases. The large Fontham (1994) U.S. case-control study, specifically designed for minimizing smoker-status misclassification rates actually found lower rates in controls than in cases and both were.lower than in this present study. In cohort studies the differential misclassification rates between the exposed and unexposed are important. This is a point I thought was interesting in Dr. Ogden's paper - potential differential smoker-status misclassification rates among exposed vs. nonexposed nonsmokers. However, with such small numbers (Table 8), the conclusion on page 2 that "substantial differential misclassification was found between exposed and unexposed populations" is not justified. This is a merely an interesting suggestion. Perhaps additional analysis to determine whether these."deceivers" were also deceiving about their husbands' smoking would tell us more. Perhaps their answers about their own and their husband's smoking habits were just crossed or misreported. As a further consideration of the Ogden et al study's intent to shed doubt on the studies showing increased lung cancer risk, I point out that this same study can just as easily provide estimates of exposure misclassification bias, a known downward bias in ETS- lung cancer studies. All that would be needed would be an analysis comparing cotinine levels by ETS exposure group (smoking vs. nonsmoking husbands) among never smokers by amount of spousal smoking (Fig. 3 presents the results for self-reported current smokers). Yet exposure misclassification bias is never even mentioned as a possible source of bias. Salivary cotinine levels by ETS exposure group for self reported never-smokers could have been presented analogous to Fig. 3. While the same claims of inapplicability to the lung cancer studies due to different age distributions would still be pertinent, at least it would have provided a less biased approach. I also enclose copies of two recent published papers I coauthored on ETS rebutting critics of EPA's risk assessment. These are papers that appeared in Risk Analysis and Environmetrics. I have also asked an EPA colleague of mine, Dr. James Repace, for an analysis of several comments to the OSHA docket which may impact my IAQ hearing testimony and previous submissions to the OSHA docket. I enclose his comments to me, which includes 1) a 29 page memorandum to me, dated February 9, 1996; 2) an editorial by Dr. Moghissi from Environmental Internationa ; 3) a letter to the Editor of Environmental International from Repace and Lowry; 4) a letter to the British Medical Journal from Repace; 5) a letter to the British Medical Journal from Martin Jarvis, and 6) a letter to 3 I

I Risk Analysis from Repace and Lowry. - Thank you for your attention to my comments. Sincerely,/ , even Bayard, Statistician U.S. EPA I Attachments: (11) 1 ~ ~ ~ I I I ~ a ~ 4 i0D ~ on c3 w fla I I ^Ph.D. I I I I I I I