Tobacco Documents Online

B851 MW47 7 P612 i3EtdS AArAa-I a Sponsored Symposia "° "" M`U ""U` on Environmental Tobacco Smoke Lisa A. Bero, PhD; Alison Gaibraith; Drummond Rennie, MD Objective. To test the hypothesis that symposia on environmental tobacco smoke (ETS) are more likely to present unbalanced data and be authored by to- bacco industry-affiliated individuals than journal articles on ETS. To compare the publication records and affiliations of authors of symposia with the authors of sci- entific consensus documents on ETS. Design.-Content analysis of articles; computerized literature searches of English-language publications (except for one symposium) supplemented with ad- ditional sources. Participants (Articles).-All 297 symposium articles on ETS and a random sample of 100 journal articles on ETS published between January 1, 1965, and March 31,1993; the 1986 Surgeon General's report on ETS; and the 1986 National Research Council's report on ETS. Main Outcome Measures.-For cach article, regardless of whether it had a methods section, agreement with the tobacco industry position that ETS is not harmful; topic; funding source(s); affiliation(s) of author; and publication records of authors. Results.-Of the symposium articles 41 % were reviews, compared with 10% of journal articles. A total of 83% of original symposium articles and 100% of journal articles contained methods sections (P=.0001). Symposium articles were more likely to agree with the tobacco industry position (46% vs 20%), less likely to as- sess the health effects of ETS (22% vs 49%), less likely to disclose their source of funding (22% vs 60%), and more likely to be written by tobacco industry-affiliated authors (35% vs 6%) than journal articles (P=.0001). Symposium authors published a lower proportion of peer-reviewed articles (71 % vs 81 %) (P=.0001) and were more likely to be affiliated with the tobacco industry (50% vs 0%) than consensus document authors (P=.0004). Conclusions.-Symposium articles on ETS differ from journal articles and con- sensus documents in ways that suggest that symposia are not balanced. (JAMA. 1994;271:612617) A NUMBER of scientif c consensus docu- ments and reviews indicate that envi- ronmental tobacco smoke (ETS) is a cause of lung cancer, heart disease, and respi- ratory disorders.1-5 The tobacco industry claims that the data on ETS are meth- odologically flawed~-" and has identified multiple strategies, one of which is the publication of scientific research, to at- tempt to influence public opinion, tobacco control legislation, and legal testimony and to refute studies published in the medical literature 9,'o,"zl To suggest sub- stantial support for their position that ETS is not harmful, the tobacco industry From the Institute for Health Policy Studies, School of Medicine (Drs Bero and Rennie and Ms Galbraith), and Division of Clinical Pharmacy, SchoohofPharmacy (Dr Bero), University of California-San Francisco. Ms Gal- braith is currently a medical student at the University of Rochester (NY). Reprint requests to Institute for Health Policy Stud- ies, Universiryof California-San Francisco, 1388 Sutter St, 11th Floor, San Francisco, CA 94109 (Dr Bero). frequently cites industry-funded, non- peer-reviewed publications, such as sym- posium proceedings.l'x This study ex- amines symposia on ETS to determine whether they should be considered equivalent to other sources of scientific information on ETS. This study was de- signed to examine the extent of tobacco industry sponsorship of symposia on ETS; to see whether there was a balanced pre- sentation of the data on ETS in such symposia compared with articles on ETS in the peer-reviewed literature; to test whether these symposia represented the work of tobacco industry-sponsored sci- entists relatively more than did articles on ETS in the peer-reviewed journals; and to compare the symposium authors and their sponsorship with the authois of two scientific consensus documents that concluded that ETS exposure is harmful to health.2~ In short, this study gauges the number, sponsorship, author- ship, and quality of symposia on ETS. METHODS Identification and Description of Symposia on ETS Symposium was defined as a collec- tion of articles from a conference that was edited for publication.' This study focuses on the scientific data on ETS and included conferences on indoor air quality only if they focused mainly on ETS. We included epidemiologic, ani- mal toxicology, and basic science stud- ies on the health effects of ETS, as well as studies on measuring exposure to ETS. The following databases were searched electronically to identify sym- posia published between January 1,1965, and March 31,1993: MEDLINE, CATA- LOG, DIALOG, CONFERENCE PA- PERS INDEX, TOXLINE, and IN- TERNATIONAL GUIDE TO PERI- ODICALS. Symposia were selected us- ing the key words "environmental tobacco smoke," "tobacco smoke pollu- tion," "passive smoking," "involuntary smoking," and "symposia," "proceed- ings," and "conference paper." Staff at the Office on Smoking and Health Tech- nical Information Center, Centers for Disease Control and Prevention, assisted with identifying the symposia. Two sym- posia were identified from Tobacco In- stitute publications and press releases. Each symposium was classified as a journal publication (the symposium was published as part of, or as a supplement to, a medical/scientific journal) or as a book. We also determined whether the symposium explicitly acknowledged fund- ing from the tobacco industry or affili- ates. Sometimes the sponsor of a publi- cation, such as Healthy Buildings Inter- national, Fairfax, Va, could not be clearly identified as having ties with the tobacco industry.2" For organizations such as this, additional written documentation was analyzed to determine tobacco industry af6liations.'~' Comparison of Content of Articles To test the hypothesis that symposium publications on ETS differ in content from articles on ETS appearing in medical/ scientific journals, we selected a random sample of articles from the scientific lit- erature on ETS and compared them with the symposium ai-ticles. We tested the following four hypotheses: 4 o et al 612 JAMA, February 23, 1994-Vo1 271, No. 8 Sponsored Sympc C flp YF I GI~T s S D~ AMER ~{EDICAL a IL

predisposition to smoke rather to the be- havior itself.13-A Smoking initiation by birth cohort has been estimated by smoking prevalence in 20- to 24-year-olds using the same NHIS data source as in this study.l',is These prevalence data indicate that be- tween 1965 and 1987, smoking initiation remained stable in young women and that smoking prevalence decreased in young men.16 Thus, since prevalence was measured several years after most ini- tiation occurred, increased initiation in girls aged 12 to 17 years was offset by the decrease in initiation in young women aged 18 years and older. We found no evidence that initiation increased among males or among young women aged 18 years or older during this 1967 to 1973 or 1974 period. In women, as in men of the same age, the initiation rate actually declined. It is pos- sible that along with young women who were college bound, those who reached age 18 years without starting to smoke had achieved a degree of maturity that made them less susceptible to the ciga- rette advertising messages of these early women's campaigns. References L Davis R. Current trends in cigarette advertising and marketing. N,Engl J Med. 1987;316:725-732. 2. Raftery J. Advertising and smolring: a smolder- ing debate. Br J Addict. 1989;84:1241-1246. 3. Centers for Disease ControL Cigarette adver- tising-United States, 1988. MMWR Morb Mortal Wkly Rep. 1990;39261-265. 4. Pierce JP, Gilpin E, Burns DM, et al. Does to- bacco advertising targetyoung people to start smok- ing- evidence from California. JAMA.1991;266:3154- 3158. 5. Albright CL, Altman DF, Slater MD, Maccoby N. Cigarette advertisements in magazines: evidence for a differential focus on women's and youth maga- zines. Health Educ Q. 1988;15:225-233. 6. Ernster VL. Mixed messages for women: a so- cial history of cigarette smoking and advertising. N Y State J Med. 1985;85:335-340. 7. Warner KE, Goldenhar LM. Targeting of ciga- rette advertising in US magazines, 1959-86. To- bacco Control. 1992;125-30. 8. Jones KE. Women's Brands; Cigarette Adver- tising Explicitly Directed Toward Women. Cam- bridge, Mass: Harvard University Press; 1987. 9. Whiteside T. Selling Deathr Cigarette Advertis- ing and Public Health. New York, NY: Liveright; 1971. 10. Federal'Irade Commission. Federal Trade Com- mission Report to Congress for 1989: Pursuant to the Federal Cigarette Labeling and Advertising Act. Washington, DC: Federal Trade Commission; 1991. 11. Roberts DF, Maccoby N. Effects of mass com- munication. In: Lindzey G, Aronson E, eds. Hand- book of Social Psychology. 3rd ed. Reading, Mass: Addison Wesley; 1984:539-598. 12. Pierce JP, Fiore MC, Novotny TE, Hatzian- dreu E, Davis R. Trends in cigarette consumption We also found evidence for an earlier decline in initiation in young men aged 18 through 20 years between 1944 and 1948, the immediate years after World War II. In 1944, more than one quarter of all 18-year-old nonsmoking men started to smoke, which may reflect the fact that cigarettes were freely avail- able to GIs during World War II.1-1 By 1948, the initiation rate for this age group had decreased by one third and coin- cided with the demobilization of troops; without the military, free or heavily dis- counted cigarettes were unavailable. The health consequences of smoking are well documented, as is the addic- tiveness of the behavior that prevents many smokers from being able to suc- cessfully quit before these health con- sequences manifest."," In this study, we have demonstrated that tobacco ad- vertising has a temporal and specific relationship to smoking uptake in girls younger than the legal age to purchase cigarettes. Our findings add to the evi- dence that tobacco advertising plays an important role in encouraging young people to begin this lifelong ad- diction before they are old enough to in the United States: educational differences are increasing. JAMA. 1989;261:56-60. 13. Macaskill P, Pierce JP, Simpson JM, Lyle DM. Mass media-led antismoking campaign can reduce the education gap in quitting behavior. Am J Pub- lic Health. 1992;82:96-98. 14. US Dept of Health and Human Services. Re- ducing the Health Consequences of Smoking: 25 Years of Progress: A Report of the Surgeon Gen- eral. Rockville, Md: US Dept of Health and Human Services; 1989. US Dept of Health and Human Ser- vices publication (CDC) 89-8411. 15. Fiore MF, Novotny TE, Pierce JP, Hatzian- dreu E, Patel K, Davis R. Trends in cigarette smok- ing in the United States: the changing influence of gender and race. JAMA. 1989;261:49-55. 16. Pierce JP, Fiore MC, Novotny TE, Hatzian- dreu E, Davis R. Trends in cigarette smoking in the United States: projections to the year 2000. JAMA. 1989;261:61-65. 17. Pierce JP, Naquin M, Gilpin E, Giovino G, Mills S, Marcus S. Smoking initiation in the United States: a role for worksite and college smoking bans. JNatl Can.cer Inst. 1991;83:1009-1013. 18. Harris JE. Cigarette smoking among succes- sive birth cohorts of men and women in the United States during 1900-80. J Natl Cancer Inst. 1983; 71:473-479. 19. The National Health Interview Survey Design, 1978-84, and Procedures, 1975-1983. Hyattsville, Md: US Dept of Health and Human Services;1985. US Dept of Health and Human Services publication (PHS) 85-1320. 20. US Dept of Health and Human Services. Sample Design and Estimation Procedures for the National Health Interview Survey,1985-1994.. Hyattsville, Md: National Center for Health Statistics; 1989. . fully appreciate its long-term health risks. If this is true, the popular and suc- cessful Joe Camel cigarette advertising campaign may pose a considerable prob- lem for the public health movement. This campaign has been demonstrated to have a substantial impact on minors.''ztA In- deed, the impact has been so great that another tobacco company, Philip Mor- ris, has adopted some of the same ad- vertising strategies (ie, redemption of proof-of-purchase codes from cigarette packages for clothing and other items, some of which are marked with ciga- rette brand identification) in its recently revitalized campaign for its major brands, including Virginia Slims (Plain Dealer. March 6,1993;B9). The prudent public health approach to prevent an- other increase in initiation amongyoung people is urgent action to extend the ban on tobacco advertising to cover all forms of advertising and promotion 44,30 This study was supported by grant 1RT245 from the University of California Tobacco-Related Dis- ease Research Program, Oakland. 2L Larsen RJ. An Introduction to Mathematical Statistics and Its Applications. 2nd ed. Englewood Cliffs, NJ: Prentice-Hall International Inc; 1986. 22. Chambers JM, Hastie TJ, eds. Statistical Mod- els. South Pacific Grove, Calif: Wadsworth and Brooks/Cole Advanced Books and Software; 1992: 314. 23. Leventhal H, Cleary PD. The smoking prob- lem: a review of research and theory in behavior risk modification. Psychol Bull. 1980;88:3701105. 24. Pierce JP, Farkas A, Evans N, et al. Tobacco Use in California 1992: A Focus on Preventing Uptake in Adolescents. Sacramento: California Dept of Health Services; 1993. 25. Kennet L. GI.• The American Soldier in World War II. New York, NY: Charles Scribner's Sons; 1987:94. 26. US Dept of Health and Human Services. The Health Consequences of Smoking: Nicotine Addic- tion: A Report of the Surgeon General. Rockville, Md: US Dept of Health and Human Services; 1988. US Dept of Health and Human Services publication (CDC) 88-8406. 27. DiFranz.a JR, Richards JW Jr, Paulman PM, et al. RJR Nabisco's cartoon camel promotes Camel cigarettes to children. JA.NIA. 1991;266:3149-3153. 28. Fischer PM, Schwartz MP, Richards JW Jr, Goldstein A0, Rojas TH. Brand logo recognition by children aged 3 to 6 years: Mickey Mouse and Old Joe the cameL JAMA. 1991;26Fi:3145-3148. 29. Tye JB, Warner KE, Glantz SA. Tobacco ad- vertising and consumption: evidence of a causal relationship. J Public Health Policy. 1987;8:492- 508. 30. Warner KE. Effects of the antismoking cam- paign: an update. Am J Public Health. 1989;79: 144-151. JAMA, February 23, 1994-Vol 271, No. 8 Smoking by Adolescent Girls-Pierce et al 611

• that a larger proportion of sympo- sium than journal articles would be re- views rather than original research ar- ticles, • that a larger proportion of sympo- sium than journal articles would sup- port the tobacco industry's position that ETS is not harmful, • that a larger proportion of sympo- sium than journal articles would focus on data that do not directly assess the effects of ETS on health, • and that symposium articles would be more likely than journal articles to contain research sponsored by the to- bacco industry and to be authored by individuals affiliated with the tobacco industry. The proportions of characteristics of symposium and journal articles were compared using the ?~ statistic. We also tested the secondary hypothesis that the proportion of characteristics of sympo- sia that explicitly acknowledged tobacco industry funding would differ from those of symposia that did not acknowledge such funding. A MEDLINE search identified 583 articles on ETS published in medical/ scientific journals between January 1, 1965, and March 31, 1993. Random num- bers were generated by computer and used to select 100 articles for compari- son with symposium articles. Editori- als, articles from "throwaway" journals, and articles from supplements were ex- cluded. At an a value of.05 and a[3 value of.10, a maximum sample size of 72 j our- nal articles and 72 symposium articles was required to test for significant dif- ferences in proportions of articles?' Therefore, our actual sample size of 297 symposium articles and 100 medical/sci- entific journal articles was large enough to test all of our hypotheses at 90% power. The descriptive characteristics of each article were coded independently by two reviewers. One reviewer was blinded to the authors, the source of the article, and the funding of the article. Before discrepancies were resolved, the blinded and nonblinded reviewers agreed on their assessments 81% of the time for the symposium articles and 86% of the time for the journal articles. Discrep- ancies between reviewers were resolved by discussion of the blinded article 2s The nonblinded reviewer coded the source of the article, the funding source of the article, and the affiliation(s) of the author(s) of the article. Both reviewers determined whether the article pre- sented original research and had a meth- ods section and assessed the topic of the article and the article's agreement with the tobacco industry's position on ETS. The content of each article was coded according to a standard procedure10-s1 into mutually exclusive categories de- scribed below. (Detailed definitions of the coding procedures and examples are available through the National Auxil- iary Publications Service.) Articles Coded as Original or Review.-Articles that appeared to present new findings were coded as origi- nal. Meta-analyses that contained new calculations of relative risks were counted as original articles rather than reviews. Literature reviews without new analyses were coded as reviews. Articles Coded as Having or Not Having a Methods Section.-Any de- scription under a subheading entitled "Methods" was counted as a methods section. If an article contained a descrip- tion of methods not under a subheading, this also counted as a methods section. Methods referenced in other papers or methods of even only a few sentences were counted as methods sections. Articles Coded as Agreed, Disagreed, or Neutral in Relation to the Tobacco Industry Position That ETS Is Not Harmful.-Each article was read to de- termine if its conclusion supported the tobacco industry's position that "there is no persuasive evidence that cigarette smoke in the air, or ETS, poses any significant risk to the health of nonsmok- ers."8 The position of the tobacco indus- try on the health effects of ETS has been summarized in a 1986 Tobacco In- stitute document as follows8: •"exposure to ETS has not been shown to cause lung cancer in nonsmokers,"'(Pl1) •"[ETS] has not been shown to im- pair the respiratory or cardiovascular health of nonsmoking adults or children, or to exacerbate preexisting disease in these groups,"&~) • "evidence relating ETS to health effects is scanty, contradictory and of- ten fundamentally flawed,"&P'f •"no marker has yet been found for chronic rather than very recent [ETS] exposures,"&0'o •"the existence of so many co- and confounding variables will, of course, con- tinue to present formidable obstacles to the study of the health effects, if any, of ETS."8(a44) If an article agreed with at least one of the five points listed above, it was coded as agreeing. If an article's con- clusion was contrary to any of the five points presented above, it was coded as not agreeing. If an article made no con- clusions regarding the tobacco indus- try's position or made some conclusions that supported the tobacco industry's position and some that did not, we coded the article as neutral. Only the conclu- sions of the article, not the actual re- sults, were assessed. Topics of Articles Coded as Health Effects, Exposure, Methodology, Con- founders, or Other.-To deteimine whether the articles represented balanced and comprehensive coverage of data on the health effects of ETS, we grouped the articles by scientific topic into one of five mutually exclusive categories: (1) health effects of ETS (ie, epidemiologic or ani- mal studies on the effect of ETS on lung cancer, pulmonary function in adults, car- diovascular effects, and child health; in vitro studies of ETS carcinogenicity); (2) exposure to ETS (ie, measured amount of exposure to ETS in different settings, critiques of methods for measuring ex- posure, chemical composition of ETS); (3) critiques of methodology (ie, criticisms of meta-analyses, discussions of misclassi- fication, discussions of biases and pitfalls of epidemiologic studies; (4) confounders (ie, articles that studied factors other than ETS that could contribute to lung cancer, cardiac disease, or respiratory effects in children [ie, diet, lifestyle, bird owner- ship, occupational exposures, or radon]); and (5) other (ie, articles on government procedures for risk assessment, basic sci- ence studies on the mechanism of cancer, articles on the annoyance and irritation of ETS, but not its health effects). Sources of Funding Coded as Tobacco Industry, Government, Other, Multiple, or Not Disclosed.-For each article, we recorded the funding source(s) that were acknowledged in the article. If an article had multiple funding sources, each was coded. An article was classified as funded by the tobacco industry if it stated that the research was supported by a tobacco company, a cigarette manufacturer, the Tobacco Institute, the Smokeless Tobacco Council, the Center for Indoor Air Re- search, or other tobacco industry-spon- sored groups (ie, the Australian Tobacco Council). The article was coded as having a government funding source if the work was supported by local, state, or national US or foreign governments. Additional funding sources were coded as other. Affiliations of Authors Coded as Industry Affiliated or Not Industry Affiliated.-Using a previously pub- lished procedure,l" we determined whether there were any known affilia- tions between the authors of the articles and the tobacco industry. Briefly, an au- thor was characterized as affiliated if he or she (1) disclosed tobacco industry funding in the article, (2) had received grant funding from the tobacco industry as stated in other publications, (3) had testified as an expert witness in support of the tobacco industry during legal pro- ceedings, (4) was an employee of the tobacco industry, or (5) was on the board of directors of a tobacco company or the Tobacco Institute. JAMA, February 23, 1994-Vol 271, No. 8 Sponsored Symposia on Environmentai Tobacco Smoke-Bero et al 613

Table 1.-Sponsored Symposia on Environmental Tobacco Smoke Table 2.-Descriptive Character'istics ofArticles on Symposia That Explicitly Acknowledge Tobacco Industry Support Environmental Tobacco Smoke by Source of Article 1. Assessing Low-Risk Agents for Lung Cancer. Methodological Aspects (proceedings of an intemational symposium held in Clearwater, Fla, October 10-14, 1989). Int J Epidemiol. 1990;19(suppt 1). Source of Article, No. (%)* Sponsored by the Center for Indoor Air Research, Linthicum, Md* 2. Ecobichon DJ, Wu JM, eds. Environmental Tobacco Smoke: Proceedings of the International Symposium Symposium Journal at McGill University. Lexington, Mass: DC Heath & Co; 1990. (n=297 (n=100 Sponsored by "a tobacco industry grant and by grants and other support° from cosponsors (including Characteristic Articles) Articles) Healthy Buildings, International, Fairfax, Va) 3. ETS-Environmental Tobacco Smoke (report from a Workshop on Effects and Exposure Levels at the University of Geneva [Switzerland], 1983). EurJ Respir Dis. 1983;65(suppt 133):7-145. Agrees with tobacco Industry position No 38(13) 65(65) Sponsored by The Tobacco Institute, Washington, DC Yes 137(46) 20(20) 4. Environmental Tobacco Smoke Effects on the Non-Smoker (report from a workshop, Bermuda, 1974). S d J R i i Neutral 122(41) 15(15) can esp r D s. 1974;(suppl 91):7-90. d b F bri d T b R i N h t l S S it l d x'=106.24, df=2 ponsore y a ques e a eun es, euc a e , ac w zer an P--.0001 5. Symposia That Do Not Explicitly Acknowledge Tobacco Industry Support Bieva CJ, Courtois Y, and Govaerts M, eds. The Present and Future of Indoor Air Quality: Proceedings of the Brussels [Belgium] Conference (February 14-16, 1989). New York, NY: Elsevier Science Publishing Cc; 1989. 6. Fustinoni 0, ed. Air Quality in Closed Places: Symposium (December 6-7, 1988, San Carlos de Bariloche, Argentina). Buenos Aires, Argentina, Academia Nacional de Ciencias de Buenos Aires, Argentina, 1988 7. Perry R, Kirk PWW, eds. Indoor and Ambient Air Qualityt (proceedings of the Imperial College London [England] symposium, 1988). London, England: Selper Ltd; 1988. 8. Kasuga H, ed. Indoor Air Quality (papers from the Conference on Indoor Air Quality, Tokyo, Japan, November 4-6, 1987). New York, NY: Springer-Verlag; 1990. 9. Environmental Tobacco Smoke and Passive Smoking [Helsinki, Finland]. Mutat Res. 1989;222:1-140. 10. Proceedings of the lntemational Experimental Toxicology Symposium on Passive Smoking [Essen, Germany]. Toxicol Letf. 1987;35:1-168. 11. Medical Perspectives on Passive Smoking [Vienna, Austria].t Prev Med. 1984;13:557-746. *The Center for lndoor Air Research (CIAR) is a tobacco industry-sponsored group supported by charter members Philip Morris, USA, RJ Reynolds Tobacco Company, and Lorillard Corporation.m tAlthough these symposia did not acknowledge funding from the tobacco industry, they were organized by individuals who are affiliated with the industry (written communication from F. Schmidt to D. Simpson, Action on Smoking and Health, London, England, January 12, 1988) .~ Six of the symposia on environmental tobacco smoke have been pubBshed as special issues of medical joumals, but five out of the six most recent symposia have been published independently as books. Comparison of Authors of Symposium Articles With Authors of the Surgeon General's Report and the National Research Council Report on ETS In 1986 two scientific consensus docu- ments, the Surgeon General's report The Health Consequences of Involuntary Smoking2 and the National Research Council's report Environmental Tobacco Smoke: Measuring Exposures and As- sessing Health Effects," concluded that ETS exposure is harmful to health. The two reports were based on a review of the literature available at the time. Pre- sumably, the scientists who wrote these reports were selected because they were believed to be experts on ETS and to- bacco-related disease. Since the authors who appear frequently in tobacco indus- try-sponsored symposia are often cited by the tobacco industry as experts on ETS research,',1,11 we tested the hypoth- esis that the symposium authors and the authors of the consensus documents would publish similar proportions of articles on their ETS-related area of research. The publication records of the 14 au- thors who appeared four or more times in symposia, the 16 authors of the 1986 Surgeon General's report on ETS, and the 11 authors of the National Research Council's report on ETS were obtained from searches of the MEDLINE and CATALOG databases for all of the au- thors' publications between January 1, 1965, and March 31, 1993. There was no overlap between authors who frequently appeared in symposia and consensus re- port authors. Two people were authors of both the Surgeon General's report and the National Research Council's re- port. We counted the number of publi- cations of each author in the area of expertise relevant to his or her work on ETS. For example, if a symposium au- thor presented an article critiquing the statistical analyses used in epidemiologic studies of ETS, then the author's pub- lications related to statistics were counted, even if all the articles were not on ETS. Articles were not counted if they had nothing to do with the topic of the symposium article or the consensus reports. We categorized the relevant publications as journal articles or not. Publications that were not considered journal articles were typically not peer- reviewed and included letters to the edi- tor, editorials, published invited lectures, symposium presentations, book chap- ters, and articles appearing in throw- away medical journals ~,37 The X1 sta- tistic was used to compare the propor- tions of peer-reviewed and non-peer- reviewed publications between the symposium authors and consensus re- port authors. RESULTS Identification and Description of Symposia on ETS We identified 11 symposia on the topic of ETS; four of these explicitly acknowl- edged tobacco industry funding (Table 1)." Although seven symposia did not acknowledge industry sponsorship, at least two of these (numbers 7 and 11, Topic of article Health effects 65(22) 49(49) Exposure 89(30) 26(26) Methodology 16(5) 3(3) Confounder 85(29) 7(7) Other 42(14) 15(15) X~°-35.58, df=4 P=.0001 Funding for article Tobacco Industry 14(5) 6(6) Government 38(13) 28(28) Other 11 (4) 11 (11) Multiple sourcest 3(1) 15(15) Not disclosed 231 (78) 40(40) Affiliations of authorst Tobacco industry Non-tobacco industry )~665.77, df=4 P=.0001 100 (35) 6(6) 183(65) 94(94) X1--31.77, df=1 P=.0001 *Percentages may not total 100 due to rounding. tThe three symposium articles that acknowledged multiple sources of funding were funded by a govem- ment source and another source. Among the 15 medical journal articles with multiple funding sources, 13 were funded by a government source plus another source, one was funded by a government source plus the to- bacco industry, and one was funded by a govemment source, another source, and the tobacco industry. tFourteen symposium authors were excluded from the analysis because their affiliations were unknown. For affiliations, n=283 for symposium authors. Table 1) were organized by individuals who are affiliated with the tobacco in- dustry (written communication from F. Schmidt to D. Simpson, Action on Smok- ing and Health, London, England, Janu- ary 12, 1988).11 We made comparisons between (1) all symposium articles and journal articles and (2) the four sympo- sia that explicitly acknowledged fund- ing and the seven symposia that did not explicitly acknowledge funding. Comparison of Content of Articles Types of Articles.-Fifty-nine percent (175/297) of symposium articles presented original research and 41% (122J297) were review articles, compared with 90% (90/ 100) of journal articles that presented original research (>e=32.56, df=1, P=.0001). Seventy-two percent (33/46) of the ar- ticles published in symposia that acknowl- edged tobacco industry sponsorship were review articles. Of the articles that pre- sented original research, 17% (29/175) of symposium articles did not contain a meth- ods section, while all of the journal ar- ticles (90/90) contained a methods section (X2=16.75, df=1, P=.0001). Agreement of Articles With the Tobacco Industry Position That ETS Is 614 JAMA, February 23, i994-Vol 271, No. 8 Sponsored Symposia on Environmental Tobacco Srnoke-Bero et al

Not Harmful.-Table 2 shows that a larger proportion of symposium articles than journal articles agreed with the to- bacco industry's position. Neutral articles typically included methodologies for mea- suring ETS exposure or assessments of the effect of potential confounding fac- tors (such as radon, wood smoke, or bird ownership) on health, but did not assess the effects of ETS in the same study. Topics of Articles.-The distribution of topics of articles on ETS differed sig- nificantly between the symposium ar- ticles and the journal articles (Table 2). The journals contained a larger propor- tion of articles that assessed the health effects of ETS, including epidemiologic studies and animal studies. The sympo- sia contained a larger proportion of ar- ticles that assessed the effects of factors that could confound an association be- tween ETS and tobacco-related diseases (eg, cooking smoke, radon, and diet). So urces of Funding for Articles.-The funding of the research in each article was not acknowledged for 78% of sym- posium articles and 40% of journal ar- ticles (Table 2). A larger proportion of journal articles than symposium articles acknowledged funding from government sources, multiple sources, or other sources. Eight journal articles acknowl- edged research funding from the tobacco industry (two were funded by other sources as well). Of these, six of the ar- ticles did not support the tobacco indus- try's position that ETS is harmless, one supported the industry's position, and one was neutral. Affiliations of Authors of Articles.- Table 2 shows that a higher proportion of symposium authors than journal ar- ticle authors were affiliated with the tobacco industry. The types of affilia- tions of the authors are shown in Table 3. None of the three journal article au- thors affiliated with the tobacco indus- try acknowledged tobacco industry fund- ing for the research. All three presented data that were neutral regarding the tobacco industry's position on ETS. The symposia also contained 38 com- mentaries, such as panel discussions, prefaces, and introductions. The tone for the symposium was often suggested by the commentaries. For example, two of the commentaries in the symposia stated the following: The flames of the ETS controversy are fed by real or perceived concerns, distrust of studies published in the literature, frustra- tions over the inconclusiveness or ambiguity of many of the studies and the biases that many people appear to have brought to the study of ETS.40 ' . [T]he published data, when critically analyzed and examined, are inconsistent with the notion that ETS is a health hazard °1 Table 3.-Affliations of Authors of Articles on Envi- ronmentai Tobacco Smoke by Source of Article* Source of Article, No. (%) Affiliation Symposium (n=297 Articles) Journal (n=100 Articles) Unknown affiliation 14(5) 0(0) Private consultant 49(17) 2(2) Unive rsity-aff liated consultant 35(12) 1 (1) Tobacco industry 16(6) 3(3) Total Industry-Affiliated 100(34) 6(6) University-affiliated 129(46) 66(61) Private nonprofit 6(2) 0 (0) Government health agency 30(11) 15(15) Professional organization 1 (0.35) 0 (0) Other 17(6) 13(13) Total Non-tndustry- Affiliated 183(62) 94(94) *The total number of industry-affiliated authors and total number of non-industry-affiliated authors were com- pared between symposia and journals (X2--31.77, 0f--1, P=.0001). Authors with unknown affiliations were ex- cluded from the X2 analysis. Percentages may not total 100 due to rounding. Comparison of Symposia That Acknowledged Funding From the Tobacco Industry With Symposia That Did Not Acknowledge Funding We compared the characteristics of ar- ticles from the symposia that explicitly acknowledged tobacco industry funding with those of articles from the symposia that did not acknowledge funding. As de- scribed above, at least two symposia in- cluded with the group that did not ac- knowledge funding were probably sup- ported by the tobacco industry. Table 4 shows that a larger proportion of articles from symposia that acknowledged indus- try funding agreed with the tobacco in- dustry position that ETS is not harmful than articles from symposia that did not acknowledge funding. Articles from sym- posia that did not acknowledge funding were more likely to focus on confounding factors than articles from industry-spon- sored symposia. The sources of funding for individual articles and the affiliations of the authors did not differ between sym- posia that acknowledged funding and those that did not (Tables 4 and 5). Comparison of Authors of Symposium Articles With Authors of the Surgeon General's Report and the National Research Council's Report on ETS Table 6 shows that the symposium authors published a lower proportion of peer-reviewed journal articles than the authors of the consensus documents. The ratio of peer-reviewed to non-peer-re- viewed publications for symposium au- thors was 2.5, compared with 4.3 for the consensus report authors. Table 6 also shows that the proportion of authors with tobacco industry affiliations differs between the consensus report authors Table 4.-Descript}ve Characteristics of Sympo- sium Articles on Environmental Tobacco Smoke by Acknowledged Source of Funding for the Symposium Type of Symposium, No. (%) haracteristic Tobacco Industry- Funded (n=46 Articles) Non-Tobacco Industry- Funded (n=251 Articles) Agrees with tobacco industry position No Yes Neutral 2(4) 33(72) 11 (24) 36(14) 104(41) 111 (44) X=14.60, dh2 P=.0006 Topic of article Health effects Exposure Methodology Confounder Other 15((35) 1(2) 6(13) 8(17) 74(29) 15(6) 79(31) 34(14) X2-10.33, df=4 P_-.04 Funding for articie Tobacco industry 0(0) 14(6) Government 6(13) 32(13) Other 1(2) 10(4) Multiple sources 1(2) 2 (1) Not disclosed 38(83) 193(77) Affiliations of authors* Tobacco industry Non-tobacco industry x1-3.81, df--4 P=.43 17(39) 83(35) 27(61) 156(65) xZ0.25, dt-1 n6.62 *Two authors from tobacco industry-funded sympo- sia and 12 authors from non-tobacco industry-funded symposia were excluded because theiraffiiiations were unknown. For affiliations, n=44 for authors of tobacco industry-funded symposia and n=239 for authors of non-tobacco industry-funded symposia. Percentages may not total exactly 100 due to rounding. and the symposium authors. While 85% (23/27) of the consensus document au- thors were affiliated with academic in- stitutions, 36% (5/14) of the authors who frequently wrote symposium articles were from universities. In contrast, 43% (6/14) of the symposium authors were consultants to the tobacco industry and one worked for the tobacco industry, while none of the authors of the con- sensus reports were in similar positions. COMMENT The content of symposia on ETS dif- fers from the content of journal articles on ETS in ways that suggest that sym- posia are not balanced and that they present the tobacco industry position on ETS. Symposia consist, in large part, of review articles that reach conclusions that are contrary to independent scientific con- sensus documents and meta-analyses pub- lished in journals?4'S9 Symposium ar- ticles minimized the potential effects of ETS on health by claiming that ETS ex- posure is impossible to measure, or if it can be measured, levels of exposure are not high enough to produce physiological effects. The journal articles acknowledged the difficulties of accurately measuring ETS exposure but did not conclude that JAMA, February 23, 1994-Vol 271, No. 8 Sponsored Symposia on Environmental Tobacco Smoke-Bero et ai 615

Table 5.-Aff liations of Authors of Symposium Ar- ticles on Environmental Tobacco Smoke by Ac- knowledged Source of Funding forthe Symposium* Type of Symposium, No. ( k) Tobacco Industry- Funded (n=46 Articles) Non-Tobacco Industry- Funded (n=251 Articles) Unknown 2(4) 12(5) Private consultant 5(11) 44(18) University-aff liated consultant 12(27) 23(10) Tobacco industry o (o) 16(7) Total Industry- Affiliated 17(39) 83(35) University-affiliated 20(45) 109(46) Private nonprofit 0(0) 6(3) Government health agency 4(9) 26(11) Professional organization o(o) 1(0.4) Other 3(7) 14(6) Total Non-Industry- Affiliated 27(61) 156(65) *The total number of industry-affiliated authors and total number of non-industry-affiliated authors were com- pared between symposia that acknowledged tobacco industry funding and those that did not (xz--0.25, df=1, P=.62). Authors with unknown affiliations were excluded from the analysis. these methodological issues invalidate all studies on the health effects of ETS. Sym- posium articles contained a greater em- phasis on potential confounding factors than journal articles. Most of the sympo- sium articles, however, exclude ETS from comparative analysis of other health risk factors such as exposure to radon, cook- ing fuel, or automobile emission. For ex- ample, most articles discussing sick build- ing syndrome did not mention ETS as a component of indoor air. The scientific reliability of the journal articles or consensus documents should be considered superior to that of sym- posia. Symposium articles are not from the regular issues of peer-reviewed j our- nals and peer review can be a proxy for quality.41 Furthermore, only 83% of the original research articles in symposia con- tained a methods section. Although re- view articles frequently lack methods, an original article without methods would be unlikely to withstand peer review. Publication bias against negative stud- ies in the peer-reviewed medical litera- ture could be responsible for the differ- ence in content between the symposium and journal articles.11-'4 Although publi- cation bias does exist,43,44 negative stud- ies on ETS are published in the peer- reviewed literature ' The analysis of com- bined individual studies results in the con- clusion that ETS is harmful.L4,1 Poor quality or the failure of authors to submit negative studies to medical journals could also explain the lack of negative studies in the peer-reviewed literature 4s,41 Our data suggest that industry control over publication, rather than industry funding for research, is likely to influence the presentation of findings. Tobacco in- Table'6-Publication Records and Affiliations of Symposia Authors Compared With Authors of Consensus Reports on ETS* Most Frequent Authors of Symposia SG and NRC Authors, Reports, No. (%) No. (%) (n=14) (n=27) Publication type Peer-reviewed 369(71) 1303 (81) Non-peer-reviewed 149(29) 302(19) X2=23.17, d/=1 Affiliations Tobacco industry Non-tobacco industry F'=.0001 7(50) 0(0) 7(50) 27(100) X--16.28, df=1 P=.oo04 *ETS indicates environmental tobacco smoke; SG, the US Surgeon General's report, The Health Conse- quences of Involuntary Smoking: A Report of the Sur- geon General2; and NRC, the National Research Coun- cil report, Environmental Tobacco Smoke: Measuring Exposures and Assessing Health Effects.23 dustry-funded researchers have argued that their source of funding does not in- troduce bias into their work,41 and our study was not designed to suggest that research by industry-funded scientists is biased. All but one of the eight peer-re- viewed journal articles that acknowledged tobacco industry support presented data contrary to the tobacco industry's posi- tion that ETS is not harmful. An addi- tional industry-supported study, which was not among those randomly selected for this study, has also found an associa- tion between ETS exposure and ill health.49 The influence of industry fund- ing for research could not be examined in great detail because a much larger pro- portion of symposium articles (78%) than journal articles (40%) failed to disclose their source of funding. While these data suggest that industry-funded authors who publish in the peer-reviewed medical lit- erature do not consistently present data that support the industry's position, the same might not be true for researchers who publish in symposia. The ratio of to- bacco industry-affiliated researchers to non-industry-affiliated researchers is greater in symposia than in journal ar- ticles, and symposia are more likely to agree with the tobacco industry position than journal articles. Researchers who have been employed by the tobacco in- dusuy have recently stated that their attempts to publish work describing an association between active smoking and lung cancer were suppressed andlmis- represented by the tobacco industry.50 The finding that data and reviews sug- gesting that ETS is not a health hazard are published in non-peer-reviewed lit- erature has substantial implications for public policies regarding smoking restric- tions. In June 1993, the US Supreme Court ruled that federal judges must en- sure that scientific evidence and testi- mony admitted in trials are reliable- Whether the data have been subjected to peer review and publication is one criterion that judges are expected to con- sider.si-" This ruling is directly relevant to the lawsuit that the tobacco industry has recently filed against the Environ- mental Protection Agency (EPA) claim- ing that the EPA used inaccurate and incomplete data for its risk assessment of ETS.6 As the court hears this case, it will have to decide whether non-peer- reviewed, tobacco industry-sponsored re- search, such as that presented in sym- posium proceedings, is admissible as evi- dence. Sponsored symposium proceedings in- fluence public policy because they are often presented in a misleading fashion, as if they are equivalent to peer-re- viewed journal articles, as if they are balanced reviews of the scientific litera- ture, and as if they are not affiliated with the tobacco industry. Symposia on ETS have been used to attempt to re- fute peer-reviewed journal articles and the EPA's risk assessment.l'= Sympo- sium proceedings have also been cited in tobacco industry public relations ma- terials'~~-11 and in the lay press.11X The tobacco industry has cited symposia as "the consensus of scientific experts"* and a gathering "of leading experts from around the world"" who disagree with the published literature on ETS. The publication and dissemination of symposia on ETS appear to be part of the tobacco industry's highly politicized cam- paign to mislead the public regarding the health risks of tobacco.ts,a9,5'{° The first tobacco industry-sponsored symposium occurred in 1974, well before most jour- nal articles on ETS were published. This suggests that the tobacco industry was concerned about adverse health effects of ETS before the medical community had conducted much research on the topic. By 1978, the tobacco industry had iden- tified the passive smoking issue as a threat to its product and had begun planning a strategy to refute scientific findings that indicate an association between passive smoking and ill health. A study conducted by the Roper Organization for the To- bacco Institute concluded: [ETS] is the most dangerous development to the viability of the tobacco industry that has yet occurred.... The strategic and long run antidote to the passive smoking issue is, as we see it, developing and widely publicizing the clear-cut, credible medical evidence that passive smoking is not harmful to the non- smoker's health.1e The publication, dissemination, and ci- tation of symposia on ETS are some of the means by which the tobacco industry has administered the "antidote" to data on the adverse effects of passive smoking. 616 JAMA, February 23, 1994-Vol 271, No. 8 Sponsored Symposia on Environmental Tobacco Srnoke=Bero et al

In summary, we urge policymakers to consider the type and quality of the scientific literature on which they must base policy regarding ETS exposure. To assist laypersons in their evaluation of the research on ETS, we recommend full disclosure on all publications of financial support from the tobacco in- dustry for research or publication and full disclosure of the peer review pro- cess, if any, to which the data were sub- jected. The editor's note that prefaced a symposium published in Cardiovascu- lar Drugs and Therapy is a good ex- ample of disclosure sl The editor's com- ment stated that the symposium did not emphasize a single drug product, was not funded by a single pharmaceutical company, described the peer review process for the symposium issue, and listed the names of the peer reviewers. Data from symposia should be reviewed skeptically since their funding sources may not be acknowledged, they have not been peer-reviewed, and tr ,-y may not present a balanced overview of the scientific literature. This study was supported by funds provided by the Cigarette Tax and Tobacco Surtax Fund of the state of California through the Tobacco-Related Disease Research Program of the University of California under award 2KT0072 to L.A.B. See NAPS document No. 05092 for 4 pages of supplementary material. Order from NAPS cJo Mi- crofiche Publications, Po Box 3513, Grand Central Station, New York, NY 10163-3513. Remit in ad- vance, in US funds only, $7.75 for photocopies or $4 for microfiche. There is a $15 invoicing charge on all orders filled before payment. Outside the United States and in some parts of Canada, add postage of $4.50 for the first 20 pages and $1 for each 10 pages of material thereafter, or $1.50 for the first micro- fiche and 50¢ for each microfiche thereafter. We thank Mina Silberberg, PhD, for translating a symposium; the staff at the Office on Smoking or Health, Centers for Disease Control and Preven- tion, for technical help; Steve Sizemore, PhD, for programming; Phillip Lollar for administrative as- sistance; and participants in the Institute for Health Policy Studies Writing Seminar and Stan- ton Glantz, PhD, for useful comments. References 1. US Environmental Protection Agency. Respi- ratory Health Effects of Passive Smoking: Lung Cancer and Other Disorders. Washington, DC: In- door Air Division, Office of Atmospheric and In- door Air Programs, Office of Air and Radiation, US Environmental Protection Agency; 1992. 2. The Health Consequenees of Involuntary Smok- ing: A Report of the Surgeon General. Washington, DC: US Dept of Health and Human Seivices;1986: 332. DHHS publication CDC 87-8398. 3. Evaluation of Carcinogenic Risks of Chemicals to Humans: Tobacco Smoking. Lyon, France: In- ternational Agency for Research on Cancer; 1986; 38:163-314. 4. Glantz SA, Parmley WW. Passive smoking and heart disease. Circulation. 1991;83:1-12. 5. Steenland K. Passive smoking and risk of heart disease. JAMA. 1992;267:94-99. 6. Flue-Cured Tobacco Cooperative Stabilization Corporation and the Council for Burley Tobacco, Inc, and Universal Leaf Tobacco Company, Inc, and Philip Morris, Inc, and R. J. Reynolds To- bacco Company and Gallins Vending Company v United States Environmental Protection Agency and Carol Browner. 1993. 7. Cigarette Smoke and the Nonsmoker. Washing- ton, DC: Tobacco Institute; 1984. 8. Tobacco Smoke and the Non-Snwker: Scientifie Integrity at the Crossroads. Washington, DC: To- bacco Institute; 1986. 9. Tobacco Institute, Washington, DC, mailing to the Tobacco Action Network, 1990. 10. Environmental Tobacco Smoke and Health: The Consezsus. Washing-ton, DC: Tobacco Insti- tute; undated. 11. Sterling TD. Publication decisions and their possible effects on inferences drawn from tests of significanceor vice versa. JAm Stat Assoc. 1959; 54:30-34. . 12. Vandenbroucke JP. Passive smoking and lung cancer: a publication bias? BMJ. 1988;296:391-392. 13. Hedges LV. Modeling publication selection ef- fects in random effects models in meta-analysis. Presented at meeting of the I nstitute of Mathemati- cal Statistics; 1990; Uppsala, Sweden. 14. Armitage AK. Environmental tobacco smoke and coronary heart disease. J Smoking Related Dis. 1993;4(1):27-36. 15. Repace JL, Lowrey AH. Issues and answers concerning passive smoking in the workplace. To- bacco Control. 1992;1(3):208-219. 16. A Study ofPublic Attitudes Towards Cigarette Smoking and the Tobacco Industry in 1978. May 1978 ed. Roper Organization; 1978;1. 17. Bero LA, Glantz SG. Tobacco industry response to a risk assessmentofenvironmentaltobacco smoke. Tobacco Control. 1993;2(2):103-113. 18. FTC Staff Report on the Cigarette Advertising Investigation by ML Meyers, C Iscoe, C Jennings, W Lenox, E Minsky, and A Sacks. Washington, DC: Federal Trade Commission; 1981. 19. Glantz SA. The tobacco industry response to scientific evidence on passive smoking. In: Pro- ceedings of the Fifth World Conference on Tobacco and Health; 1983; Perth, Australia287-292. 20. Rylander R, Peterson Y, Lebowitz MD, eds; Samet JM, guest ed. Assessing low-risk agents for lung cancer. Int J Epidemiol. 1990;19(suppl 1):S3- S87. 21. Cipillone v Liggett Group I. Transcript of pro- ceedings ofAntonio Cipillone, individually and as an Executor of the Estate of Rose D. Cipillone, Plaintiff v Liggett Group, Ine, a Delaware Corpo- ration; Philip Morris, Inc, a Virginia Corporation; and Loew's Theatres, Ine, a New York Corpora- tion, defendants. Civil aetion No. 83-2864(SA).1988: TPLR 3.431-3.468. Litigation documents 3.6. 22. Chapman S, Borland R, Hill D, Owen N, Wood- ward S. Why the tobacco industry fears the passive smoking issue. Int J Health Services. 1990;20(3): 417-427. 23. National Research Council. Environmental To- bacco Smoke: Measuring Exposures and Assess- ing Health Effects. Washington, DC: National Acad- emy Press; 1986. 24. Bero LA, Galbraith A, Rennie D. The publica- tion of sponsored symposiums in medical journals. N Engl J Med. 1992;327:1135-1140. 25. Ecobichon D, Wu J. Environmental tobacco smoke. In: Proceedings of the International Sym- posium at McGill University, November 3-4, 1989. Lexington, Mass: DC Heath & Co; 1990:1-388. 26. Taylor ST, Mintz M. Indoor air consultants act as tobacco industry duckblind. Health Lett. October 1992:1-4. 27. Clean-Air Companies Have Ties to Tobacco In- dustry [transcript of 'NBC Nightly News,' June 10, 1992]. National Broadcasting Co Inc; 1992:1-3. 28. Hulley SB, Cummings SR, eds. Designing Clini- cal Research. Baltimore, Md: Williams. & Wilkins; 1988:247. 29. Chalmers TC, Smith HJ, Blackburn B, et al. A method for assessing the quality of a randomized control trial. Controlled Clin Trials. 1981;2:31-49. 30. Krippendorff K. Content Analysis. Beverly Hills, Calif: Sage Publications Inc; 1980;5:189. In: Kline FG, ed. The Sage CommText Series, No. 62. 31. Weber RP. Basic Content Analysis. 2nd ed. New- bury Park, Cali#': Sara Miller McCune, Sage Publl- cations Inc;199A:96. In: Lewis-Beck, MS, ed. Series: Quantitative Applications in the Social Sciences, No. 49. 32. Smokers'Rights in the TYorkplace: An Employee Guide. Washington, DC: Tobacco Institute; not dated. 33. Smoking Restrictions: The Hidden Threat to Public Health. Washington, DC: Tobacco Institute; not dated. 34. S?nokinginthelPorkplace:SomeConsiderations. Washington, DC: Tobacco Institute; not dated. 35. Scientists at international conference dispute 'passive smoking.' S»wkers Advocate. June 1990:2. 36. Smoke signals: scientists dispute data. Philip Morris Magazine. September-October 1990:41. 37. Rennie D, Bero LA. Throw it away Sam: the controlled circulation journals. CBE Views. 1990; 13(2):31-35. 38. Request for Applications: 1989-1990 Research Agenda. Linthicum, Md: Center for Indoor Air Re- search; 1989. 39. Timmins N. Row over passive smoking study. The Independent [London, England]. Mareh24,1988. 40. Ecobichon DJ. Opening remarks. In: Ecobichon DJ, Wu JM, eds. Environmental Tobacco Smoke. Lexington, Mass: Lexington Books; 1990:xxi-xxii. 41. Wu JM. Summary and concluding remarks. In: Ecobichon DJ, Wu JM, eds. Environmental To- bacco Smoke. Lexington, Mass: Lexington Books; 1990:367-376. 42. Rennie D, ed. Guarding the guardians: research on editorial peer review. JAMA. 1990;263(theme issue):1317-1444. 43. Dickersin I6. The existence of publication bias and risk factors for its occurrence. JAMA. 1990; 263:1385-1389. 44. EasterbrookPJ, BerlinJ, Copalan R, Matthews DR. Publication bias in clinical research. Lancet. 1991;337:867-872. 45. Bero LA, Rennie D. Publication bias and policy. Im Proceedings of the Second International Con- ference on Peer Review in Biomedical Publication; September 9-11, 1993; Chicago, 111:17. 46. HunterJE, Schmidt FL. Methods ofMeta-Analy- sis: Correcting Error and Bias in Research Find- ings. Newbury Park, Ca1it Sage Publications; 1990. 47. Rennie D, Flanagin A. Publication bias. JAMA. 1992;267:411-412. 48. Sterling TD. Intimidation of C7°R-flmded saen- tists d aim ed. Am J Pub l fc Health.1992;82:1292-1293. 49. Spitzer WO, Lawrence V, Dales R, et al. Links between passive smoking and disease: a best evi- dence synthesis. Clin Invest Med. 1990;13:17-42. 50. Freedman AM, Cohen LP. Smoke and mirrors: how cigarette makers keep health question 'open' year after year. Wall Street Journal. February 11, 1993:A1, A6, A7. 51. Angier N. Court ruling on scientific evidence: a just burden. New York Times. June 30, 1993:A8. 52. Greenhouse L. Justices put judges in charge of deciding reliability of scientific evidence. New York- Times. June 29, 1993:A9. 53. Daubert v Merrell Dow Pharmaceuticals Inc. Decision ofthe Supreme Court ofthe United States. Washington, DC: 1993. No. 92-102. 54. Peterson J. `Passive smoking' risk unproven. Wisconsin State J. August 6, 1990:B4. 55. Sullum J. Smoke and mirrors. Reason. Febru- ary 1991:28-33. 56. Passive Smoking: An Active Controversy. Win- ston-Salem, NC: RJ Reynolds Tobacco Co; 1985. 57. Slade J. An analysis of RJ Reynolds' position paper on the health effects of smoking. Tobacco Products Litigation Reporter. 1986:1.8 TPLR 5.97- 22 TPLR 521. 58. Catanoso J. Man who once helped now criticizes Reynolds. News Record. September 26, 1992:A4. 59. Strum C. Judge cites possible fraud in tobacco research. New York Times. February 8, 1992:A1. 60. Durbin RJ. The tobacco industry strategy: new subject, same tactics. Tobacco Control. 1993;2(2): 8-9. Editorial. 61. Review policy for supplement. Caxdiovasc Drugs Ther. 1993;7(suppl 3):493. JAMA, February 23, 1994-Vol 271, No. 8 Sponsored Symposia on Environmental Tobacco Smoke--Bero et al 617

Brief Reports . The Problem and Sociocultural Context of Single-Cigarette Sales Elizabeth A. Klonoff, PhD; Jan M. Fritz, PhD; Hope Landrine, PhD; Richard W. Riddle; Laurie Tully-Payne Objective: To provide data on the sale of single cigarettes to adults and minors and to examine the sociocultural context in which these sales occur. Design.-A naturalistic observation study using repeated measures. Two hun- dred six stores in a convenience sample were visited by one minor and one adult who each attempted to purchase a single cigarette. Main Outcome Measures.-Single-cigarette sales to an adult and/or minor. Data were obtained on type of store and, if a sale occurred, on the price, brand, and packaging of the cigarette. The ethnic composition of the neighborhood surround- ing each store was determined. Results.-One hundred one (49.1 %) of the stores sold single cigarettes. Singles were sold significantly more often to minors than to adults, and when both could make a purchase, minors paid more for these singles than did adults. Singles were least likely to be sold in white neighborhoods, more likely to be sold in integrated neighborhoods, and most likely to be sold in minority neighborhoods. Minors were able to purchase single cigarettes during 34.4% of the visits to white neighborhoods but could do so during 71.2% of the visits to minority neighborhoods; adults were able to make similar purchases during 24% of the visits to white neighborhoods and 37.3% of the visits to minority neighborhoods. Conclusions.-The illegal sale of single cigarettes involves complex sociocul- tural factors heretofore unexamined. An understanding of such factors may be useful in planning merchant education programs and drafting policy to control ille- gal sales. (JAMA. 1994;271:618-620) A VARIETY of interventions have been designed to prevent and decrease to- bacco use by minors and adults. One strategy involves primary prevention programs directed at youth.i-' A second strategy has been the enactment of vari- ous laws and policies designed to regu- late or restrict smoking, tobacco adver- tising, and access to tobacco. One com- mon policy intervention has been the passage of laws to prevent minors' ac- cess to tobacco products; 45 states and the District of Columbia have such laws, From the Behavioral Health Institute, California State University-San Bernardino (Dr Klonoff, Mr Riddle, and Ms Tully-Payne), the School of Planning, University of Cincinnati (Ohio) (Dr Fritz), and Public Health Founda- tion Enterprises, City of Industry, Calif (Dr Landrine). Reprint requests to Behavioral Health Institute, Cali- fornia State University, 5500 University Pkwy, San Ber- nardino, CA 92407 (Dr Klonoff). the majority of which establish a mini- mum age for sale 8 Despite the wide- spread existence of these laws, enforce- ment has been minimal. A survey by the Department of Health and Human Ser- vices9 found only 32 instances of enforce- ment, despite the fact that an estimated 1 billion packs of cigarettes are sold each year to persons less than 18 years ofage.10 California enacted novel legislation in October 1991 banning the sale of single cigarettes. This legislation bans the sale of cigarettes to anyone in anything other than a sealed and properly labeled pack- age. Violators are guilty of an infraction and fined from $100 to $250. Single-cig- erette sales are also in violation of a federal law prohibiting the sale of ciga- rettes without a warning label stating the known health risks of tobacco. The cost of cigarettes has increased in recent years,ll particularly in Califor- nia, where Proposition 99 added a 25- cent tax per pack of cigarettes to fund tobacco control interventions and re- search. Although increased cost may not deter committed smokers, it may result in a decrease in smoking among minors and economica.lly disadvantaged casual smokers,12 who may not have sufficient funds to buy a pack of cigarettes. Thus, minors and the economically disadvan- taged might resort to purchasing one or more single cigarettes at a fraction of the cost of an entire pack. The avail- ability of these inexpensive, accessible tobacco products may facilitate experi- mentation in adolescents" and may per- mit intermittent smoking by adults who otherwise might quit because of the high cost of a pack of cigarettes.1z A small study of 35 stores in Nash- ville, Tenn, found that 25~'o sold single cigarettes." No additional data on single- cigarette sales in California or any other state have been reported. We do not know how frequently single cigarettes were sold before the ban in California, how frequently they are being sold since the ban, if they are sold more often to minors than to adults, or if other factors such as the ethnic composition of the community affect sales. Given that packs of cigarettes are sold to minors despite the illegality of these sales,"i$ we sus- pect that single cigarettes are being simi- larly sold to adults and minors in Cali- fornia despite the ban. The purpose of this study was to obtain data on the illegal sale of single cigarettes to adults and minors and to assess the extent to which the sociocultural context of the sale affects the frequency of the sales. Methods A convenience sample of 206 retail outlets located in middle-class and lower- middle class metropolitan areas of San Bernardino and Riverside counties, Cali- fornia, served as the targets for this 618 JAMA, February 23, 1994-Vol 271, No. 8 Single-Cigarette Sales-Kionoff et al