Tobacco Documents Online

Tobacco Control 1993 ; 2: 103-113 103 ORIGINAL ARTICLES Tobacco industry response to a risk assessment of environmental tobacco smoke Institute for Health Policy Studies, School of Medicine and Division of Clinical Pharmacy, School of Pharmacy, University of California, San Francisco, California, USA LA Bero Division of Cardiology, Department of Medicine and Institute for Health Policy Studies, School of Medicine, Universitv of California, San Francisco, California, USA SA Glantz C;c+rresl+undence to : Lisa A Bero. Institute for Health Policy Studies, L'niversit}• of C:ahii rnia, San Francisco, 1388 Sutter Street, 11th Floor. San Franc sco, CA 44104. L:SA Lisa A Bero, Stanton A Glantz Abstract Objective - To analyse the tobacco industry's response to the US Environ- mental Protection Agency (EPA) draft risk assessment of environmental tobacco smoke (ETS) and compare it to the response of the EPA's Science Advisory Board. To examine the peer review status of the literature cited to support criticisms of the risk assessment. Method - A content analysis of all sub- missions received by the EPA, analysis of all literature cited in the submissions and classification by topic and whether or not the literature was peer reviewed. Results - Sixty-four percent (69/107) of submissions received by the EPA claimed that the conclusions of the draft were invalid and, of these, 71 % (49/69) were submitted by tobacco industry-affiliated individuals. The specific arguments pre- sented by commentors affiliated with the industry differed qualitatively from those of the Science Advisory Board and were supported by the selective citation of non- peer reviewed literature. Compared to tobacco industry-affiliated reviewers, the risk assessment cited more references on ETS (32 % vs. 15 %) and fewer non- refereed publications (27 % vs. 37 %) (p < 0.001). Conclusion -The majority of comments that were critical of the risk assessment were submitted by tobacco industry- affiliated reviewers. The critical com- ments differed in content from the Sci- ence Advisory Board Report and were more often (compared to the draft EPA report) supported by non-peer reviewed references. (Tobacco Control 1993; 2: 103-113, Introduction Public policies to regulate exposure to en- vironmental hazards are generally based on formal risk assessments prepared through a process open to scrutiny and comment from the ptiblic and industries that might be affected by regulation of their product. It is sometimes difficult to determine whether or not an industry's response to a regulatory policy is a valuable scientific contribution to the risk assessment or mere advocacv of the industry's economic interest. The tobacco industry has been subject to increasing restrictions of its product because of growing evidence that cigarette smoke is an air pollutant.'-' The industry's response has been multi-faceted, consisting of advertising, political advocacy, funding research, and "smokers' rights" campaigns."0 Despite these efforts by the industry, there has been an increasing aware- ness of the health effects of tobacco, including the effects of tobacco smoke on non-smokers. Restrictions on where people can smoke have led to decreases in the amoun"t of tobacco people consume."-13 Recently, the public and the tobacco in- dustry have been offered the opportunity to respond with scientific comments to a risk assessment of environmental tobacco smoke (ETS) conducted by the US Environmental Protection Agency (EPA) entitled "Health effects of passive smoking: assessment of lung cancer in adults and respiratory disorders in children" ("the draft risk assessment")." The draft risk assessment concluded that ETS causes lung cancer in adults and respiratory problems in children. The draft had significant policy implications since its approval led to a decision to classify ETS as a Group A (known human) carcinogen and has accelerated the trend towards smoking restrictions in work- places and public places. The draft was open to technical comment from June 25 to October 1, 1990. The draft was also reviewed bv the EPA's Science Advisory Board (SAB) to provide an outside, expert assessment of its scientific accuracy. A summary of the public comments was provided to the SAB. The EPA revised and updated the document following the public comment period and SAB review. This study compares the content of the comments submitted by persons affiliated with the tobacco industrr_ to the comments by the SAB. The onlv study to evaluate the quality of scientific data presented by the tobacco in- dustry in response to a government report on ETS was conducted by the New Zealand Department of Health." Its analysis of the New Zealand Tobacco Institute's submission in response to the New Zealand Department of Health's report on ETS," found that the Tobacco Institute submitted an incomplete, selective and distorted analysis of the scientific data. Specifically, the New Zealand Depart- ment of Health concluded that the submission : reviewed less than 40 "•„ of the refereed scien- tific literature published since the US Surgeon General's report on ETS; cited mainly un- refereed research for post-1986 studies; ignored scientists' defence of their work,

104 presenting criticisms of the work as if they had not been answered; ignored most articles on child health and ETS; accorded small studies as much merit as large studies; and denied the validity of pooling results of existing studies (meta-analvsis) to obtain sufficient numbers for anal_vsis." This study examines all submissions received by the EPA during its public com- ment period to determine if submissions from the US tobacco industry presented a pattern of argument and citation similar to that presented by the New Zealand Tobacco Institute. We tested the hypothesis that comments which opposed the conclusions of the draft risk assessment were no more likely to be submitted by individuals affiliated with the tobacco industrv than others. We also compared the editorial review policies for the citations in critical comments to the policies for citations in the draft risk assessment to assess the scientific rigor of these two sets of citations. Methods DATA SOURCES All submissions received during the public comment period on the draft risk assessment " Health effects of passive smoking: assessment of lung cancer in adults and respiratory disorders in children" (Docket number EPA/600/6-90/006A), and the SAB Report were obtained from the EPA. CHARACTERISATION OF SUBMISSIONS RECEIVED DURING THE PUBLIC COMMENT PERIOD Each submission was classified as either sup- porting, neutral to, or opposing the EPA's tentative conclusion that ETS causes lung cancer in adults and respiratory problems in children. A total of 107 submissions were received by the EPA. (Due to space limitations, these are unable to be fully referenced in this paper. Complete references are available through the National Auxiliary Publications Service. *) Of these, 64 °;, (69/ 107) were critical of the draft risk assessment While 31 `i„ (33/107) supported the conclusion of the risk assessment and 1",~, (2/107) were neutral (eg, pointed out a typographical error) and were * See NAPS document no. 05016 for 16 pages of supplementary material. This material con- tains the names and addresses of authors of all comments submitted to the EPA. When refer- ences are made to the comments throughout the text, * indicates that the specific references to individual comments are given in the NAPS document. Order from NAPS c/o Microfiche Publications, P.O. Box 3513, Grand Central Station, New York, NY 10163-3513, USA. Remit with your order, not under separate cover, 57.75 (US funds on a US bank only) for photocopies or 54.00 for microfiche. Outside the US and Canada, add postage of 54.50 for photocopies or $1.75 for microfiche. Institutions and organisations may order by purchase order; however, there is a billing and handling charge for this service of 515, plus any applicable postage. Bera Glanzz excluded from any further analyses. Three comments were missing from the materials received from EPA. We determined whether or not the author of a comment was affiliated with the tobacco industry. An affiliation as defined in this study means that the reviewer had a financial or other interest in the tobacco industry. An affiliation does not mean that the reviewer was biased or supported the position of the tobacco industry that ETS is not harmful to health. For example, a tobacco industry-affiliated (TIA) reviewer could have submitted a comment to the EPA which supported the conclusions of the draft. A reviewer was categorised as being affiliated if he or she, 1) disclosed in the written comment that he or she was a paid consultant to the tobacco industry, 2) received grant funding from the tobacco industry, 3) appeared at least twice at tobacco industry- sponsored symposia, 4) testified as an expert witness in support of the tobacco industry during legal proceedings, or 5) was on the Board of Directors of a tobacco company or the Tobacco Institute. Sources used to determine tobacco industry affiliations were, 1) the sub- missions themselves, 2) curriculum vitae accompanying the submissions, 3) proceedings of tobacco industry-sponsored symposia, and 4) transcripts from legal proceedings. PATTERNS OF ARGUMENT PRESENTED BY TIA REVIEWERS AND THE SAB A content analysis of each comment was made to place the criticisms of the draft into one of six categories: 1) study selection, 2) epi- demiologic methods, 3) statistical analysis, 4) bias and confounding, 5) exposure estimates, and 6) other. We compared the comments by TIA reviewers to the comments presented by the SAB. We determined the peer-review status of the literature used to support the arguments of the TIA reviewers. We examined the concordance between the criticisms of TIA reviewers and the criticisms of the SAB. For example, if both the TIA reviewers and the SAB commented that the risk assessment analysis omitted studies, but the TIA reviewers suggested including more epidemio- logical studies which did not find an association between ETS and ill health, while the SAB suggested including more animal toxicology studies which demonstrated an association between active smoking and lung cancer, we concluded that the TIA reviewers and the SAB were not concordant. For each category of criticism, the TIA reviewers' argument, the peer-review status of the literature used to support the argument, and the SAB's con- clusion regarding the argument, if any, is described. COMPARISON OF PEER REVIEw' STATUS AND TOPICS OF CITATIONS IN THE RISK ASSESSMENT TO THOSE OF CITATIONS IN SUBMISSIONS CRITICAL OF THE RISK ASSESSMENT We tested the hypothesis that there was no difference in the proportion of peer-reviewed

Tobacco industry response ta ETS risk assessment 105 literature cited in the EPA risk assessment and the proportion cited in the critical comments. The citations were classified according to their published peer-review policies into five cate- gories of peer-reviewed literature and 16 categories of non-peer-reviewed literature** (see Appendix 1). Multiple citations were not counted. We also compared the literature cited in the draft risk assessment to that cited by the critical reviewers with regard to focus on the health effects of ETS. For this analysis, we classified each citation into one or more of six categories: I) children (health effects of ETS and/or other variables in children), 2) con- founding variables (health effects of variables other than ETS, eg, car exhaust, radon), 3) ETS (health effects), 4) exposure (ETS ex- posure measurement or critiques of measure- ment), 5) statistics (description or critique-of statistical techniques or study design, such as meta-analysis), and 6) other (eg, mechanisms of carcinogenesis, risk assessment methods in general). I STATISTICS All comparisons were done with chi-square analvsis of contingency tables. ** We used two resources to search for the literature cited: 1) the University of California Periodicals Database and 2) the Technical Information Center, Office on Smoking and Health, Centers for Disease Control and Pre- vention, Atlanta, Georgia. The Periodicals Data- base consists of 799136 periodical titles repre- senting 1194200 holdings from the University of California, the California State Library, the California State Universities, Stanford Univer- sity, L:niversitv of Southern California, the Getty Center for the History of Art and the Humanities, the Center for Research Libraries, and the California Academy of Sciences. The Technical Information Center compiles the annual Biblio- graphy on Smoking and Health, keeps a com- puterised database which includes journal articles and other materials related to tobacco, and produces the annual Report of the Surgeon General on the health consequences of smoking. Staff at the Technical Information Center are available to do specialised literature searches. Results CHARACTERISATION OF SUBMISSIONS RECEIVED DURING THE PUBLIC COMMENT PERIOD Seventy-one percent of critical comments came from TIA authors, while 100 °,o of supportive comments were submitted by individuals with no disclosed affiliations to the tobacco industry (table 1). The proportion of TIA reviewers who submitted critical comments (100°.•0, 49/49) was significantly greater than the pro- portion of non-TIA reviewers who submitted critical comments (38 %, 20/53) (p = 0.001, Chi-square analysis). Of the 49 reviewers who satisfied our criteria for tobacco industry affiliation, 33 (67%) disclosed on their sub- mission to the EPA that they were writing their response at the request of the tobacco industry. PATTERNS OF ARGUMENT PRESENTED BY TIA REVIEWERS AND THE SAB Table 2 lists the criticisms presented by TIA reviewers and whether or not the SAB made the same criticisms. Overall, the SAB was in concordance with 4 out of 50 criticisms presented by TIA reviewers and was not in concordance with 27 criticisms made by TIA reviewers. The SAB did not mention 19 criticisms made by TIA reviewers. The most frequent criticisms and the literature cited to support them are described below. Criticisms of study selection Twenty TIA reviewers stated that the litera- ture database used for the draft risk assessment was incomplete and excluded negative studies*. Submissions from TIA reviewers cited unsupported opinion, such as letters-to- the-editor, editorials, or discussions at con- ference proceedings, as evidence that should have been considered by the EPA rather than peer-reviewed publications. For example, the Tobacco Institute's submission cited papers from symposia 32 times, one unpublished article, one letter-to-the-editor, one editorial, seven peer-reviewed journal articles, and one unidentified citation*. Thirteen TIA reviewers recommended that two negative studies be included in the analysis regardless of their source, quality, or country Table 1.\'umher and percenrage of total critical and supportive contments strbmitted to the EPA according to tobacco irrdustrn• affrharron of the conrntentor So ac<• nl tornnrrnt .\'ttmher •'.,., o1 crurcJ( emtnnents 1n= 69: .\'umber of supporurd comments ' n= 33 Affiliated with tobacco tndustn ' Consultant no untversirv affihattun; 24 (351 0 (0; Consultant universin• affiliated, - 17 (24i 0 (0; Tobacco industry 8 t11; 0 f0/ Total affiliated 49 (71 0 (0; Not aA-iliated with tobacco industr}Consultant no university affiliation; - 0 (0, 2 (6/ University afaiated 12 (20, 7 (19; Government health agency 0 (0~ 9 (29) Phvsioan 0 (0r 3 (10; Other pmate citizen 6 (8; 3 (10) Professional organ sauon 1 (1; 4 (13 Other private nonprofit organisation 1 '1> 5 (13, Total not affiliated 20 (29. 33 (1(K31

106 Brro, Glanta Table 2 Arguments presented bY tobacco industry-affiliated (TIA) reviewers* Argument h'o- of inentions in TIA comments TIA conmtentt & SAB report concordant STUDY SELECTION Studies inappropriately included or excluded • Negative studies were excluded 20 no • Studies were excluded due to publication bias 15 not mentioned • Only US studies should have been included 14 no • Studies of males should have been included 3 not mentioned • Study results were reported selectively 2 not menttoned Criticisms of Hirayama study • Criticisms of the study remain unanswered 13 not mentioned • There was a reporting bias 4 not mentioned • The data are not representative 3 not mentioned EPIDEMIOLOGICAL METHODS The criteria for scientific proof have not been met 3 not mentioned Criticisms of inethodolog y • Recall bias is a problem 4 not mentioned • Use of questionnaires is problematic 3 not mentioned Criticisms of epidemiology • Epidemiology should not be relied upon 9 no • Epidemiology criticised in general 7 not mentioned • Results are meaningless when strength of association is low 5 no No animal data exist to support an association between ETS and lung 14 no cancer STATISTICAL ANALYSIS Statistics used in individual studies were inappropriate 9 no Criticisms of ineta-analvsis • Cannot be used when studies ha ve inconsistent results 16 no • Cannot be used when individual studies have flaws 16 not mentioned • Cannot be used with nonhomogeneous studies 11 not mentioned • Meta-analysis was used inappropriately 8 no • Should have used weight of evidence instead 6 no • Method criticised in general 4 no • Wald method criticised 2 not mentioned BIAS AND CONFOUNDERS The problem of misclassification was not dealt with adequately 12 no Criticisms of diagnostic data • All diagnoses were not confirmed by histopathologv 8 yes • Lung cancers detected were of inconsistent histological type 7 not mentioned • Childhood diagnoses have different meanings I not mentioned Confounding variables were not adequately controlled • Confounders mentioned in general 19 no • Environmental 9 no • Socioeconomic 8 no . CulturalrLifestv_le 6 no • Diet 4 no . Genetic 4 no • Viral infections 3 no EXPOSURE ESTIMATES Criticisms of exposure estimates • Cotinine!nicotine are not valid markers 16 no • Spousal smoking is not a good estimate of ETS exposure 13 no • No valid marker for ETS exposure exists 7 no • The composition of ETS is not known 7 no • Background exposure estimates are invalid 6 yes • ETS is a minor indoor air pollutant 5 not mentioned • The distribution of ETS is not known 3 not mentioned • The nose and lung defend the body from ETS 2 not mentioned ADDITIONAL CRITICISMS Btologtcal plausibilitv •.\tatnstream smoke and ETS differ 10 no • Exposure to ETS is too low to have health effects 9 not mentioned • General arguments against plausibility 9 no • Mainstream smoke does not cause cancer 2 no Criticisms of dose response estimate • No dose-response relationship exists 9 no • Linear extrapolation to low doses is invalid 6 no • The rdative potency approach is not valid 2 ees The EPA did not follow its own guidelines for risk assessment 5 yes * TIA comments are the comments submitted by tobacco industry-affiliated reviewers. Concordance between TIA comments and the SAB report was rated as "yes" if the two presented exactly the same argument, "no" if they presented opposing arguments, or "not menttoned" if the SAB report did not make the argument put forth by TIA reviewers. of origin," " or argued that the risk assessment should be based onlv on studies conducted in the US.* The SAB believed that data from all studies should be utilised in evaluating whether exposure to ETS increases the risk of lung cancer.'4 The SAB did identifv some deficiencies in the literature reviewed in the draft report. In contrast to the TIA reviewers, the SAB suggested inclusion of peer-reviewed pub- lications rather than non-peer reviewed litera- ture. The SAB also suggested the inclusion of positive studies which found an association between ETS and health effects rather than negative studies. The SAB concluded that the draft EPA report failed to consider all studies on the toxicology of ETS, the evidence on active smoking and lung cancer, and the effect of ETS on respiratory disease in children, but stated that including these additional studies would strengthen, not weaken, the conclusion that ETS is a Group A carcinogen.'y

Tabacco induury response to ETS risk assessment Seventeen TIA reviewers* suggested that certain studies be removed from the EPA's analysis, especially Hiravama's original 1981 study linking ETS to lung cancer.20 To support this suggestion, four letters disputing the conclusions of Hirayama's paper linking involuntary smoking to cancer were cited.2'-2' Onlv the EPA risk assessment and one TIA reviewer* cited Hirayama's published response to these criticisms.=' Two TIA reviewers* claimed that the EPA failed to consider the criticisms of the Hirayama study that have been published since 1986. These criticisms consist entirely of articles or panel discussions from sponsored symposia."37 The SAB did not criticise the EPA's use of the Hirayama study.19 Criticisms of epidemiologic methods Nineteen TIA reviewers argued that the criteria for scientific proof had not been met by the risk assessment.* They denounced the reliance on epidemiological studies, criticising epidemiological survey methods and arguing that epidemiology is a poor and imperfect science. If references were cited to support these arguments, they were usually general reviews of the discipline. More often, TIA reviewers stated opinion, rather than refer- encing a critique of methodology.*** The SAB approved the use of epidemiologic studies in the EPA risk assessment, but stated that additional chapters addressing the physics and chemistry of ETS, its relation to mainstream smoke, and the exposure of various populations to ETS would strengthen the conclusions of the draft." The draft was criticised by 14 TIA reviewers for considering epidemiologic data rather than animal toxicology data on the premise that toxicology is a more definitive science than epidemiology.* References to tobacco industry-sponsored symposia, a progress re- port, and one publication which could not be located were the only citations used >;o support the claim that no animal inhalation studies have reported adverse effects of ETS.'"a' The final report of a project funded by the Council for Tobacco Research (a tobacco industrv- funded group) which concluded that active smoking does not cause cancer in rodents was also used to argue that adverse effects of ETS were not biologically plausible." In contrast, the EPA risk assessment referenced peer- reviewed journal articles which stated that tobacco smoke inhalation induces carcinomas *** For example, one reviewer stated, "Epi- demiological studies have more difficulty than experimental studies in achieving a level of significance strong enough to be considered proof".* "In an epidemiological investigation we are not studying facts, but only responses to questions".* One reviewer preferred to attack epidemiologists themselves, rather than the disci- pline: "Epidemiologists are often biased in favor of finding positive results and of giving causal interpretations to such findings" and "reluctant to accept criticisms (by, for example, a stat- istician) about their findings".* 107 in hamsters.'6•"' The SAB did not identify the lack of animal toxicology data as a flaw in the draft risk assessment. The SAB pointed out that the carcinogenicity of tobacco smoke is not based on long-term animal studies, which are negative, but on epidemiologic studies of lung cancer and smoking in humans.'y Criticisms of statistical analyses TIA reviewers criticised the statistical methods used in individual papers that were evaluated for the draft risk assessment. Nine submissions suggested that one statistical test be replaced by another, but the test in question was never central to the argument in the EPA risk assessment.* The criticism of 34 TIA reviewers focused on the risk assessment's choice of meta-analysis to combine the results of several epidemiological studies to obtain a single risk estimate.* Literature cited to sup- port the argument against meta-analysis con- sisted of general review articles or editorials on meta-analysis, which neither supported nor refuted specifically the studies on ETS, and letters-to-the-editor (without citing the responses to the letters) critiquing meta- analysis and the effects of ETS on children.°"•'9 The SAB report concluded that meta-analysis "is an appropriate tool to summarize the epidemiological studies investigating the risk of ETS"." Five TIA reviewers argued that the strength of association of ETS with adverse health effects calculated from the meta-analysis was too low to warrant classification as a Group A carcinogen.* However, the EPA risk assess- ment established a lower bound of 1.26 for a relative risk of 1.41." The SAB emphasised that a low risk from a hazard to which a large number of people are exposed can lead to a significant number of preventable deaths.19 The problems of bias and confounding The EPA risk assessment addressed the methodological problem of misclassification bias (eg, coding smokers as exposed non- smokers) and made a downward adjustment of the relative risk estimates to correct for misclassification. Twelve TIA reviewers either ignored or found fault with the EPA's ad- justment and cited literature that shows that misclassification increases the estimated rela- tive risk, while failing to cite literature which shows that misclassification can also lower it.* Eight TIA reviewers suggested methods from an unpublished paper, letters-to-the-editor, and tobacco industry-sponsored symposia to adjust for misclassification, instead of the method used by the EPA risk assessment, which was based on a peer-reviewed paper and used previously by the National Research Council in its evaluation of ETS.= 31'S°* The SAB concluded that misclassification was con- sidered in detail in the draft risk assessment and that appropriate corrections were made.1° Sixteen TIA reviewers suggested that diag- noses of lung cancer might have been over- estimated because not all diagnoses were

109 confirmed by histology.* The SAB acknow- ledged this potential bias, but suggested that diagnoses of lung cancer might also be under- estimated if an actual lung cancer was in- correctly classified as some other diagnosis.14 A criticism levied by 36 TIA reviewers was that the draft failed to consider adequately potential confounding variables that could cause lung cancer in adults and respiratory disorders in children.* Nineteen TIA reviewers mentioned confounders in general, while others mentioned specific confounders described in table 2. The majority (91 °,~, 237/260) of the papers cited on confounders were peer reviewed and included publications describing associations between lung cancer and occupation, diet, or pet bird ownership. The SAB stated that "Important potential confounders of the ETS-lung cancer relation- ship were addressed in the report.... The potential main confounders included in these adjusted analyses were age and surrogates for confounding factors, including education and social class. As for other potential confounders of the ETS-lung cancer relationship, including occupation, radon exposure, and diet, there is no way to evaluate their importance as con- founders or to adjust for them, since virtually none of the studies contain information on them "." Criticisnt of ETS exposure nteasurements Thirtv-six TIA reviewers refuted ETS ex- posure estimates by claiming that ETS cannot be reliably measured, that spousal smoking (v,,hich was used to define exposure groups in most epidemiological studies) does not ac- curately indicate ETS exposure, that cotinine and nicotine are not reliable markers for ETS exposure, and that ETS distribution in the body is not understood (table 2).* Criticism of spousal smoking estimates was based primarily on symposia and references to papers on questionnaire methodology.'' The EPA risk assessment cited a peer-reviewed journal article that validated self-reported smoking behaviour data, while a TIA reviewer claimed that the risk assessment should have cited a chapter in a book which used a non-validated questionnaire to determine that universiry students underreported their smoking.'" TIA reviewers also argued that spousal smoking estimates were irrelevant because exposure at home could not be extrapolated to exposure at other sites. They attacked the work of Repace and Lowrey, published in a peer- reviewed journal, which estimated the number of deaths in the US due to ETS exposure, and failed to cite Repace and Lowrey's earlier peer- reviewed work and responses to criticism.5`~" The-TIA reviewers listed "numerous scientific articles" criticizing this study, but in fact, the "articles" consist of two symposia, one letter- to-the-editor, two editorials and one peer- reviewed paper.""" Citations from industry- sponsored symposia and the National Research Council report on ETS were also used to support the reviewers' contention that there is no reliable measure for ETS exposure.t '"-'' Befp, Gl.lllt.-. . After considering the limitations of ETS exposure measurements, the SAB concluded that ETS exposure can be estimated and that "spousal smoking status seems to be a reason- able method of identifying people with greater, versus lesser, ETS exposure ". `4 Additional criticisn:s Twenty-three TIA reviewers suggested that adverse health effects of ETS are not bio- logically plausible for a variety of reasons, including that actual exposure to ETS is too low to have any biological effect, that main- stream smoke and ETS differ in chemical composition, and that the specific cancer- causing agent in mainstream smoke is un- known and might not be contained in ETS.* The SAB concluded that the chemical com- position and particulate sizes of mainstream smoke and ETS were sufficiently similar to "directly imply carcinogenicity of ETS".1" The TIA reviewers' claims were again based primarily on papers published in symposia, while the SAB conclusion was based on peer- reviewed literature.'3`Thirteen TIA reviewers argued that no dose-response relationship for the health effects of ETS exists and that extrapolation of a dose-response effect is invalid.* TIA reviewers argued against the well-accepted theory that the lower limit of exposure necess- ary for carcino ;enicity is not known. The SAB commented that "determination of the dose- response effect of ETS based on the epi- demiologic studies of Hirayama would be most valuable, "but did not say whether or not a dose-response relationship existed.'9 One criticism of the EPA risk assessment did not assess the science upon which the as- sessment based its conclusion, but the govern- mental procedure used to assess risk. Seven TIA reviewers argued that the risk assessment did not follow the EPA's own risk assessment guidelines.'' * The SAB stated that the risk assessment guidelines, which were designed to address a single chemical compound, were not directly applicable to ETS. They concluded that the risk assessment of ETS was "fully consistent with the risk assessments that [the EPA has] done for many other carcinogens "." CRITICISMS BY NON-TIA REVIEwERS Thirty-three of the non-TIA reviewers com- mented that the draft was a thorough and balanced review of the scientific literature on ETS and that the conclusions were justified by the data presented. Reviewers not affiliated with the tobacco industrv did criticize some aspects of the draft. They suggested that the EPA should include an additional section on the effect of ETS on coronary artery disease, that the report should establish the magnitude of risk of ETS relative to other environmental hazards, that the evidence linking childhood ETS exposure and ETS was stronger than suggested in the report, that the EPA's defin- ition of bias should be clarified, and that the report should include additional discussion of ,

Tobacco industry response to ETS risk assessment socioeconomic status as a confounding factor. Non-TIA reviewers suggested that a peer- reviewed paper which found an association between ETS and disease be added to the risk assessment.76 The SAB agreed that the evi- dence on health effects of ETS in children was stronger than suggested in the report and suggested the addition of several peer- reviewed references on children. The SAB also agreed that the definition of bias could be clarified.'y - COMPARISON OF PEER REVIEW STATUS AND TOPICS OF CITATIONS IN THE RISK ASSESSMENT TO THOSE OF CITATIONS IN SUBMISSIONS CRITICAL OF THE RISK ASSESSMENT Overall, submissions critical of the risk as- sessment cited four times as manv individual. references (n = 1620) as the draft risk as- sessment (n = 391). Only 3% of the citations could not be located. The proportion of peer- reviewed material cited by the EPA risk assessment differed from that cited by critical reviewers (73 °11, vs. 63 °,o, p = 0.0002) (table 3). The pattern of topics of citations in the risk assessment differed from citations in the criti- cal comments (p = 0.0001, table 4). The risk assessment cited a greater proportion of papers examining the health effects of ETS, whereas the critical comments cited a larger pro- portion of papers on potential confounding variables. The papers on confounding vari- ables did not assess the health effects of ETS or control for ETS exposure. The risk assessment 109 and critical comments cited similar pro- portions of papers on respiratory health in children, ETS exposure measurements, stat- istical methods, and other topics. Discussion The strategy of criticizing the methodology of studies that have a potential impact on one's economic or political position is common to all industries." The strategy calls for framing one's reactions as a seemingly objective, scien- tific evaluation of the methodology used in the study.'' The tobacco industry's specific ap- proach in response to the EPA risk assessment was to criticize the draft's methodology by selectively citing the scientific literature that supported its position, rather than citing all the relevant literature. The TIA reviewers often cited unrefereed literature, such as symposia, editorials, and letters-to-the-editor, and they cited this material as though it were peer reviewed. The large number of critical com- ments on the draft risk assessment suggested on the surface that it was seriously flawed and that the evidence regarding the health effects of ETS remains controversial. However, the majority of the critical comments were sub- mitted by reviewers affiliated with the tobacco industry. The pattern of argument and citation displayed in the TIA submissions was similar to that used by the Tobacco Institute of New Zealand in its rebuttal to the New Zealand Department of Health report on ETS.t' The Table 3Number and percentage of total citations in critical contments and EPA risk assessment by peer review status t 1 _ Number ("a) cited in critical comments - Number (';,,) cited in draft risk assessment Tvpe of re(erence (n - 1620) (n=391) Not peer reviewed Abstract 13 (I) 1 (0) Book 95 (6) 18 (5) Dissertation 4 (0) 2 (0) Editorial 35 '2; 4 (1) Environmental ioumal 9 (1) 3 (1) Government report 14 (1) 3 (1) Letter-to-the-editor 63 t4, 11 (3) Medtcalisoentnfic ioumal 3 (0) 0 (0) News 9 l) 0 (0) Other 74 ~5 13 (3; Submitted publication Q !1 ) 1 (0) Symposium proceestng 220 (14, 39 (10) Tobacco trade publication 14 I 1 (01 L'nknoun 40 (3) 9 (2r Total not peer reviewed 602 !37, 105 (27) Peer reviewed Revieu in review journal 26 (2, 3 (1) Consensus report 53 :3, 17 (4) Law journal 1 '0 0 (0, Environment )ournal 61 (4) 16 (4) Medtcal; setentific )oumal 877 (54 , 250 (64) Total peer reviewed 1018 '63, 286 (73) Table 4 - \'wnber and percentaRe of total citations in critical comrnettts and EPA risk assessment br topic. ,Pcrcentagcs do not add up to 100 because topics cc'ere not mutually exclusi2'e) .Vumb<r ,.. ) of cttattons Number ("„) of citations in Topic of in critical comments draft risk assessment ntatton n = 1620. (n=391t ETS 235 C15• 123 (32i Confuunders - - 260 (16 12 (3) Children 259 -!l6, 95 (24, Exposure 218 14 53 (14) Statistics 134 (8r 21 (5) Other 766 '47 248 (63+

110 comments from TIA reviewers and citations used to support them were in sharp contrast to the comments provided by the SAB, a group of independent experts who also reviewed the draft. The SAB pointed out areas in which the risk assessment could be improved, but its criticisms tended to be more balanced than the TIA reviewers. For example, TIA reviewers stated that misclassification and diagnostic bias could only increase the estimated relative risk, while the SAB noted that these biases could serve to increase or decrease the relative risk. Recently, several of the arguments presented by TIA reviewers in response to the risk assessment have been rebutted.'' These in- clude the arguments that ETS is far less important than other indoor air pollutants; assessment of non-smokers' exposure to ETS is inadequate for risk assessment; markers for ETS in body fluids are not reliable indicators of ETS exposure; non-smokers' doses of ETS are too small to result in disease; smoking by a spouse is not a good surrogate for T:TS exposure; most causes of lung cancer are unknown but include radon, diet, and genetics; most studies of passive smoking show a high correlation with adenocarcinoma and there is no correlation between adenocarcinoma and active smoking; and misclassification of smokers as non-smokers can entirely account for the observed risk of passive smoking. The industry strategy of focusing criticism on individual studies, rather than the evidence as a whole, has been noted previously.''~a`"' Bv identifying individual studies that should have been included or excluded in the risk assessment, the TIA reviewers attempted to show that the analysis was unbalanced. How- ever, they often cited unsupported opinion, such as letters-to-the-editor, editorials, or discussions at meeting proceedings, as evi- dence that ETS is not harmful. TIA reviewers sometimes misrepresented the findings of in- dividual studies. Several reviewers claimed that the risk assessment failed to consider a doctoral dissertation that found no effect of passive smoking on non-smokers."` When an analysis of data from this study was published in the peer-reviewed literature, it actually showed a moderate association between ETS and lung cancer." Criticism of cotinine as a marker for ETS exposure centered around citation of the National Research Council report and the work of various authors who called for more research on biochemical markers.- The TIA reviewers misrepresented literature describing how methods of ETS exposure measurement can be incrementally improved by using it to suggest that currently available measures of ETS are invalid and should not be used at all. • TIA reviewers attempted to create an un- attainable "standard of scientific proof" by failing to accept epidemiological models of causality and rejection of the null hypothesis. This tactic is identical to one that the industry has used for decades in contending that there is a continuing scientific "controversy" as to whether or not active smoking harms smokers." Beso. Glanrz A frequent criticism of the draft risk as- sessment was that it ignored the literature on potential confounding factors that can con- tribute to lung cancer or respiratory problems in children. Although most of the literature cited to support the argument on confounders was peer reviewed, the papers cited did not usually include ETS exposure as a variable and, therefore, were not directly relevant to the EPA analysis. (For example, a paper might examine the association between socio- economic class and lung cancer, but would not include ETS exposure as a variable in the same study.) We found that the draft risk assessment cited the available literature that assessed the effects of both ETS and confounders. Since the potential number of confounders is un- limited in a human study, it is important to identify the major factors, as the risk as- sessment did, and determine their impact on the reported outcome. The EPA decided to conduct the risk assessment without waiting for data on the relative contributions to disease of every possible confounder related to ETS, and the SAB report concurred with this decision. The analysis of the scientific literature used to support specific arguments of the TIA reviewers reveals that the reviewers based their comments on non-peer-reviewed literature to a greater degree than did the EPA risk assessment. Competitive peer review appears to be associated with scientific rigor.R' In addition, a published review of the literature on ETS which rated the quality of studies according to epidemiological criteria, found that the majority of papers which ranked low on the criteria were derived from non-peer- reviewed sources."6 The citation of non-peer- reviewed literature by reviewers affiliated with the tobacco industry suggests that the industry was attempting to refute the scientific evidence published in peer-reviewed publications with research or opinion of inferior quality. In addition, the non-peer-reviewed material cited by the critical reviewers was not acknowledged as such and criticisms included in letters-to- the-editor were cited without citing investi- gators' responses to the letters. The tobacco industry appears to be pro- ducing publications which can be used to support its contention that ETS is not harmful to health, including pseudo- or quasi-scientific booklets, pamphlets, editorials, and letters-to- the-editor refuting the harmful effects of ETS." In addition, four of the symposia cited in the critical comments explicitly stated that they were sponsored, at least in part, by the tobacco industry."'-y" This material was heavily cited by TIA reviewers. The majority of respondents who supported the risk assessment were not paid for their critiques (two were consultants), while the majority of critical commentators had some financial connection to the tobacco industry in the form of grant funding, honoraria, payment for expert testimony, or employment by the industry. As financial support can present a potential conflict of interest, the source of funding for comments should be disclosed and

tobacco indusrry response to ETS risk assessment considered as they are reviewed. In the current study, only 67 °,~ of the affiliations with the tobacco industry were disclosed. One limi- tation of this study is that we relied only on written sources to identify affiliations with the tobacco industry; therefore, we might have underestimated the number of commentors who received funding from the industry.**** The tobacco industry has used the com- ments as part of its public relations campaign to refute the dangers of ETS. The tobacco industry prepared excerpts of criticisms of the risk assessment and released them at a press conference in 1990.°' These excerpts have been circulated to local legislators where smoking restrictions were being considered in California. In addition, letters-to-the-editor and editorials in local newspapers where smoking restrictions were being considered have cited the TIA reviewers comments as if they represented the EPA's position.y'•1° Cne letter stated that, according to EP,1 Docket No. 600-6-90-006A, "over 50 scientists speak out against the overblown health issue of ETS "." None of this press material acknow- ledges that the majority of the comments came from sources affiliated to the tobacco industry, or that these statements did not represent EPA policy. On 22 July 1992, the SAB approved the final draft of the EPA risk assessment of the health effects of ETS." The final draft identifies ETS as a Group A carcinogen that causes 3000 lung cancer deaths annuallv_ in the US. The draft also concludes that ETS causes increased respiratory symptoms, middle ear disease, and reduced lung capacity in children. Although the EPA was not required by law to respond to the public submissions described in this paper, a reading of the final document suggests that all comments were carefully considered in revising the draft." By responding to the tobacco industry criticisms, as well as to the criticisms of other reviewers and the SAB, the EPA scientific staff produced a stronger docu- ment. However, the involvement of the tobacco industrv in the critical comments and the literature cited to support the comments detracted from the scientific value of their criticism. The price to the public has been that the sheer volume of industry documents re- quiring consideration probably slowed down release of the final report and, perhaps, reduced the resources that EPA devoted to other work. Tobacco industry activities might also have drawn public attention to the report and thereby increased public interest in and aware- ness of the dangers of ETS. As the regulation of ETS exposure continues `* A letter sent to the EPA on 8 July 1992 supports our supposition that we underestimated the number of critical reviewers who received funding from the tobacco industry. The critical commentators who sent the letter acknowledged that "we inadvertently neglected to clearly in- dicate that we are both consultants to the Tobacco Institute and that our submission was made at the request and expense of the Tobacco Institute" (Philip Witorsch, .c•ritten communication, 8 July 1992). II1 to be an issue, the tobacco industry can be expected to persist in using the tactics outlined in this paper at the local, national, and international level to refute the independent scientific findings regarding the health effects of ETS. Policymakers concerned with limiting the effects of ETS exposure should be in- formed about these tactics in order to evaluate fairly the conclusions of scientific consensus documents on ETS. This study was supported by funds provided by the Cigarette and Tobacco Sunax Fund of the Statc of California through the Tobacco-Related Disease Research Program of the Um- versitv of California under award 2KT0072 (LAB) and award IRT 520 (SAG). ~'e would like to thank George Alexeef, Charles Becker, Alison Galbraith, Thomas Novrotnc, Drummond Rennie, and Donald Shopland for their comments on this manuscript and Adrienne Brazelton for administrative help. Appendix Literature that was peer reviewed was classified as: 1) Consensus report - a report compiled by a panel of scientific experts and peer reviewed before publication; - - 2) Environntental journal article - an article from a journal which focussed on environmental issues, but not issues related to health, and had a published peer-review policy (eg, Environ Int, Environ Sci Technol); 3) Law journal article - an article from a law journal that had a published peer-review policy; 4) Medical/scientiftc journal article - an article from a medical or scientific journal that had a published peer-review policy; and, 5) Reviet:, article -an article from a medical or scientific journal which did not present original research, but reviewed the literature on a topic. Review articles were further classified as derived either from peer-reviewed scientific or medical journals (eg, Appl Occupat Environ Hyg, N Engl ,7 Med) or from peer-reviewed specialty journals which publish only review articles (eg, Nutr Rev, Semin Respir Med, Compr Ther). Literature that was not peer reviewed was classified as: 1) Abstract - an abstract from a conference pro- ceeding; 2) Advertisentent - an advertisement from a maga- zine or newspaper; 3) Book - an entire book, text book, book chapter, or monograph; , 4) Dissertaticnr-a doctoral dissertation; 5) Editorial - an opinion piece containing no data which was published in a newspaper, magazine, medical journal, etc.; 6) Enz•ironnrental journul article - an article from a journal which focussed on environmental issues, but not issues related to health, and did not have a published peer-review policy (eg, Environ Technol Lctt,,3' Environ Econ Afan); 7) Government Report - a report issued by a gov- ernment agency, but not prepared by a scientific panel (eg, "Risk Assessment in the Federal Government: Managing the Process ", National Research Council, 1983;; 8) Law journal article - an article from a law journal that did not have a published peer- review policy; 9) Letter-to-the-Editor- a letter published in peer- reviewed medical, scientific, or environmental journals, but not subjected to the same type of competitive peer-review as articles published in the journals; 10) Afedical/scientific journal article-an article I

112 from a medical or scientific journal that did not have a published peer-review policy; 11) News clipping-a clipping from a magazine, newspaper, or a press release; 12) Radio transcript - a published transcript from an interview broadcast on radio; 13) Submitted publication - a manuscript which had been submitted to a journal for review, but had not yet been accepted; 14) Symposium - a paper which was presented at a meeting and published in either a medical journal or book. The proceeding did not contain a published peer-review policy; 15) Tobacco trade publication - article in a trade magazine published by the tobacco industry; and, 16) Unknown - if the citation could not be located in the University of California library system or with the aid of the Technical Information Center, Office on Smoking and Health, Centers for Disease Control and Prevention, Atlanta, Georgia, USA. 1 International Agency for Research on Cancer (IARC Monograph). Evaluation of carcinogenic risk of chemicals to humans: tobacco smoking, Vol. 38. Lyon: IARC 1986: 163-314. 2 National Research Council. Environmental tobacco smoke: measuring exposures and assessing health effects. Washing- ton, DC: National Academy Press, 1986. 3 US Department of Health and Human Services. The health consequences of involuntary smoking : a report of the Surgeon General. Washington, DC: DHHS, 1986. (DHHS Pub- lication No (CDC) 87-8398.) 4 Glantz SA, Parmlev W\L^. Passive smoking and heart disease - epidemiology, physiology and biochemistry. Circulation 1991 ; 83: 1-12. 5 Steenland K. Passive smoking and the risk of heart disease. ,j< Am ttfed Assoc 1992; 267: 94-9. 6 Marshall E. Tobacco science wars. Science 1987; 236: 250-1. 7 Marshall E. Tobacco industry does slow burn over EPA advisor. Science 1991 ; 250: 203. 8 Emster V, Bums D. A rebuttal to the tobacco industry's paper, "Cigarette smoke and the nonsmoker." 51 Public Health Polrcv 1984; 5(3): 368-i5. 9 Warner KE, Goldenhar LM, McLaughlin CG. Cigarette advertising and magazine coverage of the hazards of smoking. N Engl3 Sfed 1992; 326: 305-9. 10 Davis RM. Current trends in cigarette advertising and marketing. \' Engl3' .tfed 1987; 316: 725-32. 11 Becker D.M• Conner HF, Waranch HR, Stillman F, Pennington L, Lees PS et al. The impact of a total ban on smoking in the Johns Hopkins Children's Center.I Am .4fed Assoc 1989; 262(6): 799-802. 12 Atarshall E. Involuntary smokers face health risks. Science 1986; 234: 1066-7. 13 Stillman FA, Becker DM. Swank RT• Hantula D, Moses H. Glantz S et al. Ending smoking at the Johns Hopkins medical institutions: An evaluauon of smoking pre- t•alence and indoor air pollutionJ Am .Sfed Assoc 1990; 264 12i: 1565-9, 14 US Environmental Protection Agency. Health effects of passive srnnkt,tg: assessnrenf of lung canter nt adults and resptrafarrduordcrs in children. Washington D.C.: Indoor Air Division. Otizee of Atmospheric and Indoor Air Programs. Office of Air and Radiation• USEPA, 1990. (EPA ~ 600 16-90 006A.. 15 New Zealand Depanment of Health, prepared by J. Re- inken. Through the smokescreen - A crttrquc of enrrr- rntntental tabacco snrok<: A recvev oI the literature bt• the Tobacco Institute o! .\~•t: Zealand Wellington: NZDH, 1990. 16 New Zealand Department of Health, Creatnrg snmkefree em•tronments. Welltngton: NZDH, 1988. l i Shimizu H, Monshtta \t. Mtzuno K, Masuda T• Ogura Y• Santo M ct al. A case control study of lung cancer in nonsmoking women. Tohoku .7 E.rp Afed 1988; 154: 389-96. 18 Sobue T, Suzuki T. Nakayama N, Inubushi T, Matsuda :.1, Doi 0 ct al. Association of indoor air pollution and passive smoking with lung cancer in Osaka. ,fap f Cancer Clrn 1990: 36: 329-33. 19 US Environmental Protection Agencc. Report of the Indoor Air Quality an,l Total Hurnan Exposure Commtttee : r 7•teaof the Office o , f Research and Dez elopment Draft Report, •'Health effects of passive smoktng: assessment of lung ca,tc<r nr adults and respsratnrdisorders m children." \L"ashtngton, DC: EPA. 1991. !EPA/600!6-90/006A.t 20 Hiravama T. Non-smoking wives of heavy smokers have a higher risk of lung cancer: A study from Japan. BalJ 1981 ; 282: 183-85. 21 .\Sacdonald EJ. BJf7 1981; 283: 915, 22 Komegap HR, Katitenbaum MA. B.tf.f 1981; 283: 914. 23 Harns JE. DuMouchel 1X'H. B.SfI 1981 : 283: 914. 24 Mantel N. B.Sf•7 1981; 283: 914. Bero, Glmn_- ' 25 Hirayama T. BM•7 1981; 283: 916. 26 Letzel H, Uberla K. Meta-analcsis on passive smoking and lung cancer. In: Kasuga H. ed. Indoor air qualtt.•. New York: Springer-Verlag, 1990: 317-22. 27 Letzel H et a1. Meta-analvsis on passive smoking and lung cancer; effects of study selection and misclassification of exposure. In: Perry R, Kirk PW, eds. Indoor and ambient air qualitv. London: Selper Ltd, 1988: 293-302. 28 Bacon-Shone J. The use of statistics to evaluate the level and effect of indoor air pollutton. In: Lunau F, Reynolds GL, eds. Indoor air a,uahtr and ventilation. London: Selper Ltd, 1990: 11-4. 29 Uberla K, Ahlbom W. Passive smoking and lung cancer: A reanalysis of Hirayama's data. In: Kasuga H, ed, Indoor air qualitv. New York: Spnnger-Verlag, 1990: 333-41. 30 Layard M, Viren J. Assessing the quality of a Japanese cohort study. In: Bieva et al, eds. Present and future of indoor air qualttr. New York : Elsevier Science Publishing Co., 1989: 177-50. 31 Layard M. Environmental tobacco smoke and cancer: The epidemiologic evidence. In: Ecobichon D, Wu J, eds. Ent•ironmental tobacco smoke: proceedrngt of the inter- national meeting at AfcGil/ L'mvcrnt-v 1989. Massa- chusetts: Lexington Books, 1990: 99-115. 32 Mantel N. What is the epidemiologic evidence for a passive smoking-lung cancer association? In: Kasuga H, ed. Indoor air quality. New York: Springer-Verlag, 1990: 341-7. 33 Panel discussion on lung and other cancers. In: Ecobiochon D, Wu J, eds. Environmental tobacco smoke : proceedings of the international svmposru n at McGill University 1989. Massachusetts: Lexington Books, 1990: 117-36. 34 Kilpatrick SJ, Viren J. Age as a modifying factor in the association between lung cancer in non-smoking women and their husbands' smoking status. In: Perry R, Kirk PW, eds. Indoor and ambient air quality. London: Selper Ltd, 1988: 195-202. 35 Kilpatrick SJ. Model specification effects in ETS/nutrition research. In: Kasuga H, cd. Indoor air quality. New York: Springer Verlag, 1990: 256-71. 36 Kilpatrick SJ. An example of extra-poison variation suggesting an under-specific model. In: Bicva et a!. eds. Present and future of indoor air qualite• New York: Elsevier Science Publishing Co., 1989: 83-90. 37 Ahlbom W, Uberla K. Passive smoking and lung cancer: A reanalysis of Hirayama's data. In: Perry R, Kirk PW, eds. Indoor and ambient air qual]ty. London: Selper, Ltd., 1989: 169-78. 38 Mohtashamipur E, ed. Proceedings of the international experimental toxicologp symposium on passive smoke. Toxicol Lett 1987: 35. 39 Adlkofer F, Scherer G, Wenzel-Hartung R. Exposure of hamsters and rats to sidestream smoke of cigarettes: Preliminary results ofa 90 day inhalation study. In: Perry K, Kirk PW, eds. Indoor and ambient air quality. London: Selper, Ltd., 1988: 252-8. 40 Adlkofer F, Scherer G, et al. Exposure to ETS and its biological effects. In: Bieva CJ, ed. Present and future of indoor air qua6.v. New York, Elsevier Science Publishing Co., 1989: 186-96. 41 Haley N, Adams JD, Alsofon J, Hoffman D. Uptake of sidestream smoke by Syrian golden hamsters. Toxicol Letr 1987; 35: 83-8. 42 Haley N. Sidestream smoke uptake by Syrian golden hamsters in an inhalanon btoassay. Indoor air 87, proc int conf indoor air qualitv and climate 1987; 2: 68-73. 43 Haley N. Progress report Am Health Foundation. Wash- ington, DC: I;SDHHS• 1988: 71-81. (USDHHS Publication l:o'PHS grant C-%29580-08.) 44 Dontenwill W. Ararctmutelforschung 1971 ; 21 : 142-3. 45 Henry C, Kouri R. Final report : chronic exposure of nuce to cigarette srnoke. New York: Field, Rich and Assoc., 1984. 46 Dontenwill W, Chavalier HJ, Harke HP, Lafrenz U, Reckzeh G• Schneider B. Investigations on the effeets of chronic cigarette smoke inhalation in Syrian golden hamsters. 3 Nail Cancer Inst 1973; 51: 1781-1832. 47 Dontenwil W, Chevalier HJ. Harke HP, Klimisch HJ, Kuhnigle C, Reckzeh Ci er al. Studies on the effect on chronic cigarette smoke inhalation in Syrian golden hamsters and the importance of vitarriin A on mor- phologrca] alterations after smoke expusure, Z ICrebsfarsch 1977: 89: 145. - 48 Eysenck HJ. An exerase in mega-silliness. Am Psycho 1978; (May:: 517 49 Mantel N. Does passn•e smoking stunt the growth of children? hrt 7 7:'prde,nrol 1986; I.5' 3t: 427-8. 50 Wald NJ. Nanchahal K, Thompson SG, Cuckle HS. Does breathing other people's smoke cause lung cancer? Bh13 1986; 293: 121 7-?'. 51 Lee PN. Does breathing other people's tobacco smoke cause lung cancer? B,tf_7 1986; 293: 1503-t. 52 Lee PN. Passtve smoking and lung cancer association: A result of bias. Hum Brol 1987; 6: 517-24, 53 Lee PAn alternative explanauon for the increased risk of lung cancer in nonsmokers, In: Pcrry R, Kirk PW, eds. Indoor and ambrent air qualur. London: Selper Ltd, 1988:149-58. 54 Lee P. Misclassificatron as a factor in passive smoking. Lancet 1986; 2: 867. 55 Lee PN. lfuclastrJ catrnn of smoking habtts and passive nnoktng. New York: Springer-Verlag, 1988. 56 Lee PN. Lung cancer and passive smoktng: Association an artefact due to mrsclassificatton of smoking habits. Toxicol Lett 1987; 35: 157-62.