Tobacco Documents Online

518 Journal of Health Politics, Policy and Law before funding. The primary purpose of special projects, according to tobacco industry lawyers, was to develop scientific data that could be used to defend tobacco companies against litigation (Saroldn 1988; Bero et al. 1995). In addition, the special projects allowed the industry to cul-. tivate relationships with scientists who could later be called on to testify . on the industry's behalf (Bero et al. 1995). The Council for Tobacco Research's special projects were often designed specifically to divert attention from tobacco as a cause of disease (Bero et a1.1995). For exam- ple, studies were funded to show that poor nutrition, occupation, or genetic predisposition could cause the same diseases attributed to smok- ing. In addition, the study designs of the special projects were sometimes altered so they would be more likely to produce results that would sup- port the tobacco industry position (Bero et al. 1995). Given general evidence suggesting that industry sponsorship can influ- ence the results or conclusions of sponsored research and specific evi- dence of previous efforts by the tobacco industry to control the direction of the research it funds, we hypothesized that research funded by CIAR might be affected by conflict of interest. The tobacco industry argues that exposure to ETS is too minimal to cause any real damage, and that epi- demiologic studies of passive smoking are flawed and do not prove that ETS is harmful (Tobacco Institute 1986). The industry has an obvious interest in producing scientific data to support these arguments, because policy makers and juries would be more likely to accept them if they were based on hard data. In fact, a confidential report prepared by the Roper Organization (1978: A7) for the Tobacco Institute noted that the industry's best strategy for countering public concern about passive smoking was to fund scienti8c research. The report stated: "The strate- gic and long run antidote to the passive smoking issue is, as we see it, developing and widely publicizing clear-cut, credible, medical evidence that passive smoking is not harmful to the non-smoker's health" Thus the purpose of our study was to analyze the content, quality, and use of research funded by CIAR to determine whether it is funding truly independent research or whether it is financing studies that appear to be motivated by tobacco industry interests. We hypothesized that, although CIAR claims to be objective and independent, the tobacco industry might smoking has not been provedd harmful to health. be funding research through CIAR to support its position that passive

522 Journal of Health Politics, Policy and Law 0 studied an aspect of tobacco, tobacco smoke, or a tobacco-specific sub- stance such as nicotine; or as non-ETS related if it studied general indoor ai= quality and did not specifically mention tobacco. Then we subclassified the topics of the projects into the categories of health effects, exposure, or other. A health effects study evaluated the effects of exposure to a substance, including studies conducted in cells, animals, or humans. An exposure study measured exposure to a sub- stance but did not evaluate the effects of that exposure; these included studies of exposure markers sucb as cotinine (a nicotine metabolite). Studies that did not evaluate either health effects or exposure were clas- sified in the other category. One project, for example, studied whether tobacco smoke is perceived primarily through sight or smell. Table I summarizes our findings on the topics of projects funded by CIAR. Almost two-thirds of CIAR's special-reviewed projects were related to ETS, compared with 30 percent of its peer-reviewed projects. In addition, most special-reviewed projects studied exposure, rather than health effects. Thesefindings suggest that research related to ETS, par- ticularly exposure research, is a high priority for the tobacco industry executives who select CIAR's special-reviewed projects. One of the tobacco industry's main arguments regarding passive smoking is that per- sons are not exposed to sufficient levels of ETS to cause any serious dam- age (Tobacco Institute 1986). It is therefore possible that the tobacco industry is funding research through CIAR's special-reviewed projects to develop data it can use to support this claim. In addition, from a legal per- spective, a study measuring exposure to ETS would be much less dam- aging to the industry than one showing that EfS causes disease; thus it is "safer" for the industry to sponsor exposure research. We also examined the affiliations of CIAR-funded principal investiga- tors (PIs). Based on the information provided by CIAR, we coded the affiliation of each PI as either academic, private sector, or other. They were classified as academic if their stated affiliation was a university or college and as private sector if they were affiliated with a private-sector business, consulting firm, or contracting agency. All other affiliations were classified as other. None of the PIs had more than one affiliation listed. Table 2 summarizes our findings regarding the affiliations of CIAR- funded PIs. Nearly all of the PIspn peer-reviewed projects had academic affiliations, whereas almost 40 percent of the PIs on special-reviewed projects were affiliated with private-sector organizations. Two special- reviewed projects were awarded to the president of Healthy Buildings ® W s.

® 532 'Journal of Health Politics, Policy end Law. ® can use to influence policy. The motivation behind CIAR's peer-reviewed projects is less clear. It is possible-that the industry is using them to deflect attention away from ETS as an indoor air pollutant, because most of them are -not related to ETS: . In addition, it is possible 'that peer- reviewed projects are being funded to gain good publicity and improve the industry's credibility. Influencing Policy Our analysis of the use of CIAR-funded research suggests that the tobacco industry is using CIAR to develop scientific data it can cite in legislative settings to influence policy related to ETS. For example, data from two of CIAR's special-reviewed projects were presented at OSHA hearings on its proposed indoor. airi quality standard, and data from a third special-reviewed project were presented at a congressional hearing related to the proposed ban on smoking on commercial aircraft. In all three cases, these data'were used to s~pport the tobacco industry position that smoking.should not be regulate4. Although the tobacco industry applears to be funding research through CIAR's special-reviewed projects i an attempt to influence policy on ETS, the impact of this strategy ma~ be minimized by the lack of peer review associated with these projects. For example, the Oak Ridge National Laboratory study shows hQw lack of peer review appears to have led to a poor study design with al high potential for bias. Participants were selected for the study by a marf eting research firm with ties to the industry, and laboratory tests were c~nducted by scientists employed by R. J. Reynolds. Given these sorts of c~mpromises in study design, policy makers may be unwilling to accept the conclusions of the study, which support the tobacco indust ry position that workplace smoking should not be regulated. In addition, the congressional su~ ommittee report on the HBI study suggests that data from one of CYARRrs special-reviewed projects appear to have been dramatically altered. It }s not clear from the subcommittee report whether HBI acted independently or with the approval of its tobacco-industry sponsors. However~' it is likely that policy makers will examine testimony and research resuQts presented by HBI with a partic- ularly critical eye in the future Is r4

Industry-Funded Research and Conflict of Interest: An Analysis of Research Sponsored by the Tobacco Industry through the Center for Indoor Air Research Deborah E. Barrrnes and Lisa A. Bero University of California, San Francisco Abstract The Center for Indoor Air Research (CIAR) was created by three United States tobacco companies itt 1988. Its stated mission is to fund high-quality, objective research related to indoor air, including studies of environmental tobacco smoke (ETS). Because CIAR is financed by the tobacco industry and funds research related to tobacco, it fosters an inherent conflict of interest. We consider whether this conflict of interest has affected the content, quality, or use of research funded by CIAR. We hypothesize that the tobacco industry might be using CL4R to develop sci- entific data to support its position that ETS is not harmful to health. CIAR funds two types of projects: "Peer-reviewed" projects are awarded after peer review by a group of scientists, whereas "special-reviewed" projects are awarded directly by tobacco industry executives. CIAR's special-reviewed projects are more likely than its peer- reviewed projects to be related to ETS, to support the tobacco industry position, and to be used by the industry to argue tbat smoking should not be regulated in public places. Our findings suggest that the tobacco industry is funding special-reviewed projects through CIAR to develop scientific data that it can use in legislative and legal settings. The industry may be financing peer-reviewed projects through CIAR to enhance its credibility, to provide good publiciry, and to divert attention from ETS as an indoor air pollutant. CIAR's stated mission of funding high-quality, objective research has been compromised by conflict of interest, and at least some of CIAR's projects are being used to promote the tobacco industry's agenda. 'this study was fundedghrough the Cigarette and Tobacco Surtax Fund of the State of Califor- nia through the TbbaocR-Related Disease Research Progntm of the University of California (awards 2KT0078 and 4RT0035). We thank all of our friends and cotlea;uea at the Institute for Health Policy Studies who provided Invaluable insight as we prepared this article. Presented in pan at the American Public Health Association meeting in Washinatoa. DC, 2 October 1994. JournaJ offfealrh Politics. Policy anQ law. Vol. 21. No. 3, Fall 1996. Copyright O 1996 by Duke University Prees.

:l• 542 Joumal of Health Politics, Policy and law. White, L. G•'.'n.d. Ethical Coruideroti,oru of Accepting Financial Support from the Tobacco lnduatry.. New York: American Council on Science and Health. Witt, M. D., and L: O. Gostin. 1994. Conflict of Interest Dilemmas in Biomedical Research:'Journal of the Americon Medical Asaociation 271(7):547-551. Wolinsky, H. 1985. When Researchers Accept Funding from the Tobacco Industry, Do Ethics Go up in Smoke? New York State Journal of Medicine 85(7):451-455. ®

520 Journal of Health Politics, Policy and Law group of individuals with reputations for expertise and scientific lead- ership in the disciplines relevant to indoor air research. ... After the SAB establishes the research agenda, the Center announces to "the sci- entific community at large that research applications in response to the agenda are being accepted. The review of proposals and their selection for funding is accomplished in a scientifically rigorous and objective manner. Applications are reviewed first for scientific quality by the applicant's peers selected from the group listed in Appendix B [of the RFA). The SAB, in turn, reviews the applications and peer evalua- tions, and develops recommendations on the selection of applieations- Studies recommended by the SAB are subject to final approval by the Board of Directors. (Center for Indoor Air Research 1994: 3) The RFAs also state that investigators funded by CIAR are encouraged to publish their findings in the peer-reviewed literature (Center for Indoor Air Research 1994). CIAR's review process is similar to the grant- approval processes used by many foundations, although all of CIAR's projects are funded as contracts rather than as grants. Although CiAR's publications state that its projects are funded through the peer-review process previously-described, we found that some projects receive a special-review process. CIAR publishes a brochure that lists the projects it has funded to date (Center for Indoor Air Research 1992b). They are classified as "research," "applied;' or "other;" but the differences ecrtbparr ries are not explained. Through inter- views with staff members e were informed that research proj- ects tion n c investigations funded through the peer- iew process described in CIAR's publications; applied and other projects are more goal-oriented studies by the board of directors without receiving peer revie (Marquardt 1993 ). ' s we noted, CIAR's board of directors consists pn atives of the charter tobacco companies. None of CIAR's publications mention that some of its projects are awarded through a special-review process. We tried to confirm the information about CIAR!s special-review process by writing a letter to the executive director of CIAR, Max Eisen- berg. We explained in detail our understanding of the special-review process used for applied and other projects and asked for confirmation or clarification of the process. In response, we received only a note stating "For your information;' along with a packet of CIAR's informational brochures.. However, one recipient of an applied project, Roger A. Jenk- ins (1995), of Oak Ridge National Laboratory, testified at length regard- ing ing app leal R.] fort wh. CIy I pro Co lec to , inf- Re pef fro l CI, mc try usE pul Co Wi wr ulz re: E1 tbl de uK

e to envirotunen- rsons who do not :ause lung cancer !ase (Glantz and "eases the risk for ) and various res- rotection Agency harmful effects of e National Acad- xal (U.S. Depart- )2, the U.S. Envi- -i Group A human 92). The growing is led to increased ionsmokers from :wed journal con- :d [indoors) with ?). This study was avhich was created nission is to fund icluding studies of ~ that CIAR is an : that CIAR's proj- .ve manner" based and that this peer rch. . . is recom- 1994: 3). ctly related to its nay influence the .991; Blank 1992; In particular, the nvestigator to pro- ). For example, the . study design that zsor could encour- ie final publication -stigators may feel i findings that are :illman et al. 1991). Barnes and Bero  industry-Funded Research 517 In particular, investigators may fear that future funding will be denied if they publish unfavorable data (Hillman et al. 1991). Several studies on the effects of industry sponsorship indicate that these concerns about conflict of interest are justified. One showed thai research funded by the chemical industry is more likely than govem- ment-funded research to conclude that occupational exposure to cbemi- cal agents is not harmful (Swaen and Meijers 1988). Another study found that research sponsored by the pharmaceutical industry is more likely than research funded through other sources to favor the new drug being evaluated (Davidson 1986). Similarly, a third study showed that research sponsored by pharmaceutical companies almost always concludes that the sponsor's drug is equivalent or superior to comparison drugs, even when the data do not completely support this conclusion (Rochon et al. 1994). These studies provide compelling evidence that industry funding may influence the type of research condueted and the conclusions drawn from the data. • History has also shown that, when scientific findings are particularly damaging, industry may try to conceal, manipulate, or deny the findings. The asbestos industry, for example, funded external contract research on the health effects of asbestos from the 1930s to the 1950s (Lilienfeld 1991). According to internal memos, these studies suggested that expo- sure to asbestos could cause asbestosis and lung cancer. However, the study results were suppressed, and the industry denied that asbestos was hazardous to health (Lilienfeld 1991). Similarly, the lead industry has denied or minimized evidence suggesting that lead in paint is a common cause of disease in children (Rabin 1989). The tobacco industry has a history of financing scientific research that it claims is independent but is actually designed to fulfill the industry's needs (Bero et al. 1995). In 1954, U.S. tobacco companies created the Tobacco Industry Research Committee, which was later renamed the Council for Tobacco Research-U.S.A., Inc. The industry stated publicly that the council's purpose was to fund independent scientific research on the health effects of smoking and that, to ensure the'objecdvity of the research, projects would be selected based on peer review by an inde- pendent science advisory board (Council for Tobacco Research 1992). However, internal tobacco industry.documents reveal the existence of a "special projects" division within the Council for Tobacco Research (Freedman and Cohen 1993; Bero et al. 1995). Special projects were funded based on the recommendations of tobacco industry lawyers, rather than the science advisory board, and were not peer reviewed

.rd some members r their projects. quality of research : the quality of the the proposals sub- ity of publications 'unded research by urent Contents for ided two lists that I resulted from its , 1993c). Ten more ng were identified a total of seventy- •ticles (15 percent) 'alifornia, Califor- ore were not avail- I five of these arti- stitutional reports, 3 percent), both of :hored by a PI on a 92). Therefore we ilted from CIAR- next section. sociated with pub- ms have been asso- shed data; Rochon a (Bero et al. 1992) .larly, articles pub- iiewed, have been ; (Bero et al. 1992, nerally have been .viewers (Chalmers based on industry- ts (Davidson 1986; al. 1994; Cho and Barnes and Bero  lndustry-Funded Research 525 To assess factors associated with publication quality, we coded the peer-review status, type of publication, and source(s) of funding acknowl- edged for the articles that resulted from CIAR-funded projects.. An article was coded as peer reviewed if the journal in which it bad been published stated it was peer reviewed, published a list of reviewers, or required man- uscripts to be submitted for review before publication; otherwise, it was coded as non-peer reviewed. Type of publication was coded as symposium article, original research article, review, or other based on how the article was described in the table of contents. All articles published in symposia were classified as symposium articles, regardless of whether they were original research articles or reviews. Source(s) of funding acknowledged were coded as tobacco if the stated sponsor(s) was a tobacco company, the Tobacco Institute, CIAR, or the Council for Tobacco Research; as multi- ple if both CIAR and a nontobacco source of funding were acknowledged; as other if an organization other than CIAR was acknowledged; and as none if no source of funding was acknowledged. We also evaluated the conclu'sions.of the articles to determine whether those resulting from CIAR-funded research tend to support the tobacc industry position that ETS is not harmful to health. An article was clas sified as pro-industry if it concluded that there is no evidence that ETS is associated with adverse health effects, that the evidence is inconclu- sive, that confounding variables are responsible for health effects attrib- uted to ETS, that studies on ETS are.statistically flawed, or that the level of exposure to ETS is not high enough to cause health problems. In addi- tion, articles that made policy recorrimendations consistent with tobacco industry proposals, such as accommodating both smokers and nonsmok- ers in restaurants or workplaces, were classified as pro-industry. An arti- cle was considered neutral if it did not discuss the health effects of ETS. An article was considered anti-industry if it concluded that ETS is asso- ciated with adverse health effects or if it made policy recommendations consistent with tobacco-control efforts. These categories have been used previously (Bero and Glantz 1993; Bero et al. 1994) and are based on the written statements of the tobacco industry (Tobacco Institute 1986). High-quality publications were considered those that were published in peer-reviewed journals, those that presented original research findings and had not been published in symposia, and those that acknowledged a source of funding other than the tobacco industry. We used the chi-square statistic to test the hypothesis that articles resulting from peer-reviewed projects would be more likely than articles resulting from special- reviewed projects to be associated with these high-quality characteristics.

.rtise and scientific lead- • research. . . . After the ter announces to the sci- ations in response to the osals and their selection , rigorous and objective ;cientific quality by the d in Appendix B [of the rtions and peer evalua- election of applications. to final approval by the search 1994: 3) y CIAR are encouraged ature (Center for Indoor s similar to the grant- although all of CIAR's •ants. :s projects are funded bed, we found that some ublishes a brochure that or Indoor Air Research ied;' or "other;" but the plained. Through inter- med that research proj- ided through the peer- )ns: applied and other firectly by the board of rdt 1993). As we noted, ' representatives of the blications mention that 31-review process. CIAR's special-review )r of CIAIZ, Max Eisen- of of the special-review ked for confirmation or ived only a note stating ' CIAR's informational project, Roger A. Jenk- stified at length regard- i Barnes and Bero  Industry-Funded Research 52t ing the funding process used for his project. He stated that CIAR had approached him with a proposal for the project, and that be and his col- leagues had developed the study methodology with input from CIAR, R. J. Reynolds, and Bellomy Research (a marketing research firm). There- , fore, we believe that the information we received by telephone interview,............-......... which characterized applied and other projects as being awarded by CIAR's board of directors without undergoing peer review, was accurate. Because CIAR's applied and other projects receive a special-review process, and because we hypothesized that they might be similar to the Council for Tobacco Research's special projects, we refer to them col- lectively as "special-reviewed" projects throughout this article. We refer to CIAR's "research" projects as "peer-reviewed" projects. Based on information in CIAR's tax forms, which we obtained from the Internal Revenue Servim, we determined that CIAR awarded $11,209,388 for peer-reviewed projects and $4,022,723 for special-reviewed projects from 1989 to 1993.1 , For our analysis of the content, quality, and use of research funded by CIAR, we hypothesized that CIAR's special-reviewed projects would be more likely than its peer-reviewed projects to, support the tobacco indus- try position, and that the results of special-reviewed studies would be used by the tobacco industry to lobby against regulation of smoking in public places. Content of Research Funded by CIAR . We conducted a content analysis of CIAR-funded research to determine what projects and which investigators are being funded. We were partic- ularly interested in determining what percentage of CIAR-funded research is related to ETS. As we noted previously, research related to ETS has a high potential for being biased. by con8ict of interest, because the tobacco industry could use the results in legal and legislative settings. CIAR provided information regarding projects it had funded from Jan- uary 1988 to October 1993, including the title of each project, the name and affiliation of each principal investigator, the date the project was originally funded, and a one-paragraph description of the purpose of the project (Center for Indoor Air Research 1992b, 1993a). Based on the description, we classified the topic of each project as ETS related if it i. Figures do not include awards made in 199o beanse tax forms were oot avallabe from the Internal Revenue Service.

research to argue tobacco industry istry continue to ;ree that the case iewed projects to ipecial-reviewed i has little to lose ise most of them directly affect its :urs because the dly sound, unbi- :ss of how those ,ther hand, has a is best spent on can overtly or in various ways d Gostin 1994). t it believes will consciously or eir findings in a gators who rely ible to industry I be imposed to .1 (Hillman et al. hren (1994) has ded through an is try and inves- unded research :, which should entific freedom indings. tprofit corpora- nd the tobacco Barnes and Bero ' Industry•Funded Research 535 industry. In addition, all of CIAR's projects are funded through contracts, rather than grants, and CIAR's RFAs explicitly state that funded investi- gators are encouraged to publish their findings in peer-reviewed journals. CIAR has also taken additional steps to mini*niie the role of its sponsors in the funding process. In particular, it has given the primary responsi- bility of project selection to an independent group of scientists who serve on its science advisory board. Our analysis suggests that the projects funded through peer review by the advisory board are of high quality and have not been unduly influenced by conflict of interest. However, our analysis also suggests that Chren's proposal does not necessarily minimize conflict of interest in industry-sponsored research. In particular, we found that the tobacco industry developed a system for circumventing CIAR's peer-review process to fund projects that would serve its needs. The studies funded through CIAR's special-reviewed projects appear to be those that had the highest potential to produce data that could be either harmful or beneficial to the industry, because most of them were related to ETS. It seems likely that the tobacco industry funded these studies through its special-review process to retain control of selection of the principal investigators and study design. If other industries were to create organizations similar to CIAR for funding external research, it seems possible that they might be similarly tempted to fund particularly sensitive studies through a special process. implications for Disclosure Our study raises several important issues'related to disclosure. None of CIAR's publications disclosed that one-third of its projects are funded through a special-review process. In fact, CIAR's most recent publication of supported research no longer separates its projects into the categories of research, applied, and other (Center for Indoor Air.Research 1995). If we had received this new brochure, we probably would not have thought to ask whether all of CIAR's projects are funded based on the peer- review process described in its publications. CIAR should clearly describe the funding process for its special-reviewed projects in all of its publications. In addition, special-reviewed projects should be clearly labeled in its publications to indicate that they were selected by tobacco industry executives rather than through a process of peer review. Similar disclosure statements should be made in all publications resulting from CIA.R-funded research. Journal editors and policy makers have increasingly called for researchers to.disclose any potential conflicts 9024/031

534 Joumal of H i In addition, the ind that "more rese4 memo stated, "It spend their dolla against smoking i : The tobacco in enhance CIAR's projects will be m by funding peer-r are not related to il Implications for Whenever an indu business, a confli researcher has a r ased research and findings might aff responsibility to research that prod As many auth covertly influence (Hillman et ai.:19 Industry may be ~ produce desirable unconsciously des way that is likely t heavily on indust pressure (Hillman Several authors minimize the pote 1991; Blank 1992; proposed that ind independent third tigators. In additio should be awarde specifically state regarding study de CIAR fulfills th tion that acts as a Ith Potitics, Policy and Law . try often uses the fact that it funds research to argue is needed:' For example, a 1978 tobacco industry extremely important that the industry continue to n research to:ahow that we don't agree that the case losed" (Bloch 1994). try may`be using CIAR's peer-reviewed projects to •dibility, ao that the results of its special-reviewed likely to be accepted. The industry has little to lose ewed projects through CIAR, because most of them S. [I ustry Funding In General y funds scientific research that will directly affect its of interest arises. This conflict occurs because the " nsibility to conduct methodologically sound, unbi- ~ disseminate the findings, regardless of bow those the industry. The industry, on the other hand, has a ximize its profits, and its money is best spent on favorable data. ~have noted, industry sponsorship can overtly or ~ conduct and publication of research in various ways ; Blank 1992; Chren 1994; Witt and Gostin 1994). e likely to fund research studies that it believes will ata. In addition, investigators may consciously or i their research studies or present their findings in a e favorabld-to their sponsors. Investigators who rely unding may be particularly susceptible to industry al. 1991). ve suggested that guidelines should be ioposed -to ~ for bias. in industry-funded research (Hillman et al. ~ren 1994; %'jtt and Gostin 1994). Chren (1994) has Iry-funded research should be awarded through an ty to- irI ize contact between industry and inves- Chren suggested that all industry-funded research t ngh contracts, rather than grants, which should investig$tors will retain complete scientific freedom Fn, data "ysis, and publication of findings. iteria proposed by Chren. It is a nonprofit corpora- *rmediary between investigators and the tobacco 9

® i 528 Journal of Health Politia, Pot'rty and larr We examined these sources to identify CIAR-funded researchers who bad either testified orally or submitted written statements. When avail- able, we read both written statements and transcripts of oral testimony to determine whether they discussed CIAR-funded research, whether the researcher's relationship with the .tobacco industry was disclosed, and whether the testimony supported the. tobacco industry position on ETS (as we previously defined for concliisions of articles). We found that 28 percent (five of eighteen) of PIs on special-reviewed projects had submitted statements related to ETS, compared with 3 per- cent (one of thirty-four) of PIs on peer-reviewed projects (Fisher's exact test, p=.0154). One PI on :a special-reviewed project had testified on five separate occasions, whereas all others had testified only once. All of the statements submitted by CIAR-funded PIs supported the tobacco industry position that ETS is not harmful to health. The PI who had testified five times consistently argued that smoking can be accom- modated indoors through adequate ventilation; two PIs argued that evi- dence is insufficient to conclude that ETS causes disease; one maintained that there is a lack of biological plausibility to conclude that ETS is harmful, and that confounding variables could explain the statistical association between ETS and disease;:one argued that ETS is not a major cause of sick building syndrome and poor air quality; and one argued that workplace exposure to ETS is very low (implying that the exposure is too minimal to cause disease). f : Data from at least three oif CIAR's special-reviewed projects have been presented at hearings related.to ETS. Two special-reviewed projects were discussed at OSHA'si hearings tegarding its proposed indoor air quality standard, which recommends restricting or banning smoking in workplaces nationwide, whereas the third was discussed at hearings related to banning smoking on comttiercial airline flights. In all three cases, the data from CIAR's' special-reViewed projects were used to sup- port the tobacco industry position. One of the OSHA statements was submitted by Roger A. Jenkins, who discussed preliminary data from 'a special-reviewed project entitled "Determination of Human ExpQsure to, Environmental Tobacco Smoke" According to CIAR's tax records, Oak Ridge National Laboratory received $797,892 in 1993 to conduct this study. Jenkins disclosed in his statement-that the project had- been fuailed by CIAR; however, be stated orally that he was not appearing as awadvocate for or against OSHA's proposed standard, but "as a:scientist that's been conducting a study and Qoi7/osi

a tobacco-specific sub- :studied general indoor :o. s into the categories of Is study evaluated the iies conducted in cells, •ed exposure to a sub- posure; these included i nicotine metabolite). or exposure were clas- mple, studied whether ht or smell. of projects funded by viewed projects were teer-reviewed projects. i exposure, rather than :h related to ETS, par- r the tobacco industry projects. One of the ve smoldng is that per- :ause any serious dam- sible that the tobacco 31-reviewed projects to !idon, from a legal per- tld be much less darn- ltlSes disease; thus it is rch. ed principal investiga- CIAR, we coded the sector, or other. They :)n was a university or I with a private-sector All other affiliations •e than one affiliation affiliations of CIAR- projects had academic ; on special-reviewed zations.'Iwo special- of Healthy Buildings Table I Topics of Projects Funded by CIAR, 1988 to 1993 Topic ETS Health Effects ETS Exposure Other ETS Total ETS Non-ETS Health Effects Non-ETS Exposure Other Non-ETS Total Non-ETS r of Peer- d Projects d =40) Number of Reviewed (n =1 Special- Projects 9) 6 2 6 8 0 ; 2 1Zlo096) 12 (63%) 14 0 13 . 6 1 1 28 0096) 7 (37%) Comparison of total ETS vs. total non-ETS for peer-reviewed vs. special-reviewed projects: Fisher's Exact'itiat, D +. .02. Table 2 Affiliations of Principal Investigators (Pis) funded by CIAR' Percentage of PIs on Percentage of PIs on Peer-Reviewed Special-Reviewed Projects Projects Affiliation (n = 34) (n = 18) Academic 94 ! : 56 Private Sector 3 . 39 Other 3 6 Chi square =12.26. df = 2, p.0.002. +'IWo PIa received funding for both peer-reviewed projects and special-reviewed projects. They were cl.ssitied as PIs on special-reviewed projects throughout our analyses. However, onr results would not have been different if we had classified them as PIs on peeFreviewed projects, or if we had placed them in a separate category. International (HBI), Gray Robertson, a long-time tobacco industry con- sultant who has testified frequently on the industry's behalf (Levin 1993; U.S. House 1994a). It is therefore possible.that special-reviewed projects tend to be awarded to PIs in the private sectbr because some of them may be sympathetic to the tobacco industry position. Alternatively, academic investigators may be wary of accepting filnd,ing for special-reviewed proj- ects because they have not been subject to a formal peer-review process. We also found that three of CIAR's sp4cial-reviewed projects were awarded to PIs who are cturently meinb~ of CIAR's science advisory board. This sort of financial relationship beiween the advisory board and the tobacco industry could raise concern atoout the board's ultimate inde- Numbe Reviewe Reviewe (n tn 1~ J cn J J tn 00

® 08/09/98 ® 14:44 8 0 0 n 910 716 2186 ® t 526 Journal of Hea We also an,6lyzed ' whether articles res th Politics, Policy and Law e stated conclusions of the articles to determine lting from special-reviewed projects would be more likely than t'hose res; Table 3 s min • " of CIAR's roject-r, ing from sp~ecial-re quality char~cteristi reviewed p addition, 3 published lications. Fi ects tended funding. Table 3 al ects were m port the tob ting from peer-reviewed projects to be pro-industry. es our findings regarding the quality and conclusions lated publications. As hypothesized, articles result- iewed projects tended to be associated with poor- s. Almost one-half of articles resulting from special- jects were published in non-peer-reviewed journals. In percen were •published in symposia and 20 percent were ~ book s ctions, institutional reports, and other types of pub- ally, articles associated with CIAR's special-reviewed proj- o ackn~wledge the tobacco industry as the sole source of l o show that articles based on CIAR's special-reviewed proj- re likely than articles from its peer-reviewed projects to sup- o indistry position. These findings support evidence from previous studies su esting that industry funding is associated with pro- industry out,6omesavidson• 1986; Swaen and Meijers 1988; Lexchin 1993; Rochon et al. 1994). In addition, these findings provide further sup- port for oui hypot esis that the tobacco industry may be funding research thrqugh C R's special-reviewed projects to produce data that GO are favorable to its osition. Use of CIAR-Fundeb Research The final phapse of ou study was an ysis of the use of research funded by CIAR. We hypo esized that, alth ugh CIAR-funded research has led to publicatioits both upporting and ing the tobacco industry posi- tion, the industry vbould selectivel cite the studies that support its claims. In addition,, ve hypothesized that Pl:s on special-reviewed proj- ects would be more 1~keIy than PIs on r-reviewed projects to have tes- tified on the utdustryf s behalf. To de e how CIAR-fuaded research was being u);ed, w analyzed tes ' ony presented by CIAR-funded researchers t~t fed hearings relate to ETS. We used three sou s to identify -funded researchers who have testified or st#bmi statements on ' ues related to ETS: the list of per- sons who tesfified ore the Occupa 'onal Safety and Health Adminis- tration (OSHA) rega ding its propose indoor air quality standard (U.S. Department bf Occu ational Safety nd Health Administration 1994); the list of persons w, o submitted tec 'cal oomments to the U.S. Envi- i I ~ ® 015/031 T: Ft Pc T) So Ac ro) w1 an ide mt an. Pn inF bu: to 19! Hc Hc me

© ® ® I ® 516 Journal of Health Politics, Policy and Law Scientific evidence now strongly suggests that exposure to environmen- tal tobacco smoke (ETS) is harmful to the health of persons who do not smoke. Studies have shown that passive smoking can cause lung cancer (Hirayama 1981; Fontham et al. 1994) and heart disease (Glantz and Parmley 1991; Steenland 1992) in adults and that it increases the risk for sudden infant death syndrome (DiFranza and Lew 1995) and various res- piratory disorders in children (U.S. Environmental Protection Agency 1992). In 1986, two scientific consensus reports on the harmful effects of exposure to ETS were published independently by the National Acad- emy of Sciences (1986) and by the U.S. Surgeon General (U.S. Depart- ment of Health and Human Services 1986). And in 1992, the U.S. Envi- ronmental Protection Agency classified ETS as a known Group A human carcinogen (U.S. Environmental Protection Agency 1992). The growing awareness of the harmful effects of passive smoking has led to increased restrictions on smoking in public places to protect nonsmokers from exposure to ETS. . In contrast, a study published in 1992 in a peer-reviewed journal con- cluded that "acceptable air quality can be maintained (indoors] with moderate amounts of smoking" (Turner et al. 1992: 19). This study was funded by the Center for Indoor Air Research (CIAR), which was created by U.S. tobacco companies in 1988. CIAR's stated mission is to fund high-quality, objective research related to indoor air, including studies of ETS. Both CIAR and the tobacco industry maintain that CIAR is an independent organization. In particular, they emphasize that CIAR's proj- ects are funded in a "scientifically rigorous and objective manner" based on peer review by a group of independent scientists and that this peer review process "ensures that only high quality research ... is recom- mended for funding" (Center for Indoor Air Research 1994: 3). Whenever an industry funds research that is directly related to its product, there is concern that conflict of interest may influence the research in some way (Bond 1991; Hillman et al. 1991; Blank 1992: Rothman 1993; Chren 1994; Witt and Gostin 1994). In particular, the sponsor might apply overt or covert pressure on the investigator to pro- duce results that will be favorable (Hillman et a1.1991). For example, the sponsor could recommend that the investigator use a study design that would be more likely to favor its product. Or the sponsor could encour- age a researcher to emphasize certain conclusions in the final publication of the data. Even without any external pressure, investigators may feel consciously or subconsciously compelled to publish findings that are pleasing, or at least not damaging, to their sponsors (Hillman et al. 1991). 19 M ®

M rojects, according to ic data that could be . (Sarokin 1988; Bero d the industry to cul- x called on to testify Council for Tobacco specifically to divert t al. 1995). For exam- ition, occupation, or s attributed to smok- jects were sometimes :sults that would sup- sponsorship can influ- irch and specific evi- i control the direction arch funded by CIAR o industry argues that 9amage, and that epi- and do not prove that .ustry has an obvious :e arguments, because ) accept them if they sport prepared by the istitute noted that the ancern about passive rt stated: "The strate- issue is, as we see it, ble, medical evidence oker's health." e content, quality, and ther it is funding truly idies that appear to be hesized that, although obacco industry might position that passive Barnes and Bero  Ina C1AR's Peer-Reviewed and Special-Reviewed Projects CIAR has released several Request for App that describe its organizational structure, m (Center for Indoor Air Research 1989, 1991, these RFAs, CIAR is an independent, nonA 1988 (Center for Indoor Air Research 1994) rate members: charter, regular, and asso~ "those corporations that established the Cen ing the majority of funding;" whereas regulI "those persons or corporations that are im research but were not involved in the estab: ter for Indoor Air Research 1994: 3). Chat represented on the board of directors bu (Center for Indoor Air Research 1994). Although these types of memberships ar RFAs, the names of the members are listed M original charter members were Philip Tobacco Company, and Lorillard Corpo Research 1989); Svenska Tobaks A.B., a S pany, was added to the list of charter mem Air Research 1994). CIAR has two reguls and Mead Paper, its associate members it aging companies. Thus CIAR was create tobacco companies, and its board of direl tobacco industry representatives. C1AR's mission, as stated in its 1989- ate a focal point organization of the high and foster quality, objective research in i ronmental tobacco smoke, and to effectivi ings to the broad $cientific comrnunity" 1989: 1). This mission statement was includes a specific reference to ETS ( 1992a). CTAR states that its mission is ft entific and technical research related to i The peer-review process that CIAR ui is described in its RFA brochures as foll The research agenda of the Center fo lated by the Science Advisory Boa -Funded Research 519 bations (RFAs) brochures 6on; and funding process ~92a, 1994). According to W corporation formed in t has three types of corpo- ite. Charter members are r and are currently provid- and associate members are ested in indoor air quality zment of the Center" (Cen- - and regular members are ssociate members are not learly described in CLAR's ly in the appendix. CIAR's 's U.S.A., R. J. Reynolds on (Center for Indoor Air dish domestic tobacco com- in 1994 (Center for Indoor ~ embers: Hoechst Celanese de several paper and pack- nd nd is primarily financed by rs is composed primarily of ~ t research agenda, is "to cre- scientific caliber to sponsor oor air issues including envi-. communicate research find- ter for Indoor Air Research en ified in 1992 and no longer iter for Indoor Air Research lled primarily by funding sci- loor air. to select projects for funding s: idoor Air Research is formu- (SAB), a multi-disciplinary ® !: i

539 Journal of Health Politics, Policy and Law . 1994. Sponsored Symposia on Environmental Tobacco Smoke. Journal of the Ameriean Medical Association 271(8):612-617. Bero, L. A., and S. A. Glantz.1993: Tobacco Industry Response to a Risk Assessment of Bnvironmental Tobacco Smoke.Tobacco Connvl 2(2):103-113. • Blank, L H. 1992. Industry-Funded Dermatologic Research within Academia in the United States: Fiscal and Ethical Considerations. Journal of lm+estigative Derma- tology 98(3):265-268. Bloch, M.1994: Tobacco Industry Funding of Biomedical Research. Tobacco Control 3(4):297-298. Bond, 0. 0. 1991. Ethical Issues Relating to the Conduct and Interpretation of Epi- demiologic Research in Private Industry. Journal of Clinical Epidemiology 44(Suppl 1):29S-34S. Center for Indoor Air Research. ;1989.1989-90 Research Agenda: Request for Appli- c+ations. Linthicum, MD: Center for Indoor Air Research. 1991. 1991 Research Agenda: Request for Applications. Linthicum, MD: Center for Indoor Air Research. 1992a. 1992-93 Research Agenda: Request for Applications. Linthicum, MD: Center for Indoor Air Reacareh. 1992b: Supported Studies. Linthicum, MD: Center for Indoor Air Research. 1993a..List of projects funded by CIAR in 1993, provided by V. Christine Marquardt. I:inthicum, MD: Center forindoor Air Research, 13 October. 1993b. List of publications resulting from CIAR-funded research, provided by V. Christine Marquardt. Linthicum, MD: Center for Indoor Air Research, 13 October. 1993c. List of publications i•esulting from CIAR-funded research, provided by Paula Raimondo, Research Librarian. Linthicum, MD: Center for Indoor Air Research, 16:Novembet: 1994. 1994 Research Agenda: Request for Applications. Linthicum, MD: Center for Intlnor Air Research. •• 1995. Supported Studies and Peiblications. Linthicum, MD: Center for Indoor Air Research. Chalmers, T. C., C. S. Frsnk, and 0. Reitman. 1990. Minimizing the Three Stages of Publication Bias. Journal of dusAmerican:Medical A.rsociatton 263(10):1392-1395. Cho, M. K, and L A. Bero. Forthcoming. The Quality of Drug Studies Published in Symposium Proceedings. Chren, M. 1994. Independent Investigators and For-Profit Companies: Guidelines for Biomedical Scientists Consideaing Funding by Industry.Archives ofDermarology 130:432-437. Council for Tbbaceo Research. 1992. Report of the Council for Tobacco Research- USA, Inc. Washingto0. DC: Councii for'Ibbaeco Research. Davidson, R. 1986. Source of Funding and Outcome of Clinical Trials. Journal of General Inteinal Medicrne l i l55- 08. DiFranra, J. R.; and R. A. Lew. 1995. Effects of Maternal Cigarette Smoking on Pregnancy Complications and•Sndden Infant Death Syndrome. Journal of Family Practice 40(V85-•394. i

1 ® ® 524 Journal of Health PolPtics, Policy and Law pendence. It also suggests that the industry may reward some members of the science advisory board by providing funding for their projects. f Quality of Research Funded by CIAR " The next phase of 'our study was an evaluation of the quality of research funded through CIAR. We could not directly evaluate the quality of the projects themselves, because we did not have aocess to the proposals sub- mitted to CIAR.. As a surrogate, we evaluated the quality of publications that have resulted from C1AR-funded research. We identified articles that had resulted from CIAR-funded research by contacting CIAR and by searching MEDLINE and Current Contents for articles coauthored by CIAR-funded PIs. CIAR provided two lists that collectively identified sixty-one publications that had resulted from its funded projects (Center for Indoor Air Research 1993b, 1993c). Ten more articles that acknovtiledged CIAR as a source of funding were identified through MEDLINE and Current Contents, giving us a total of seventy- one possible articles for inclusion. Eleven of these articles (15 percent) were not contained within any of the University of California, Califor- nia State, or Stanford University libraries and therefore were not avail- able through our interlibrary loan system. We obtained five of these arti- cles by contacting the authors; another four were institutional reports, which we purchased. We could not obtain two articles (3 percent), both of which were published in symposia, and had been coauthored by a PI on a special-reviewed project (Hedge 1991; Hedge et al. 1992). Therefore we assessed the quality of sixty-nine articles that resulted from CIAR- funded research based on the criteria described in the next section. Several studies identified various factors that are associated with pub- lication quality. FQr example, peer-reviewed publications have been asso- ciated with higher quality (Barnes and Bero, unpublished data; Rochon 1994) and a more balanced presentation of information (Bero et al. 1992) compared with non-peer-reviewed publications. Similarly, articles pub- lished in symposia, which typically are not peer reviewed, have been associated with poor quality and unbalanced findings (Bero et al. 1992, 1994; Rochon 1994). In addition, review articles generally have been associated with biases based on the afi'iliations of the reviewers (Chalmers et al. 1990). Studies have also shown that publications based on industry- funded research tend to draw pro-industry conclusions (Davidson 1986; Swaen and Meijers 1988; Lexchini 1993; Rochon et al. 1994; Cho and Bero forthcoming). I ®

540 Jou~al of Health Politics, Policy and Law Marquardtj V. C., and staff scientist. 1993. Telephone interview regarding funding process for CIAIt projects. 13 October. .National Acadetay of Sciences. 1986. Enirironmental Tobacco Smoke: Measuring Exposurles and Assesstng Health Jrffects. Washington, DC: National Academy Press. ' Parish, L C.. J: A. Witkowsld, and L. E. Milh'han.1991. Conflict of Interest and Sci- entific Publications: International Journal of Dennatology 30(4):250-251. Rabin, R. 1989. Warnings Unheeded: A History of Child Lead Poisoning. American Journal of Public Health 79(12):1668-1674. Relman, A: S. - 1984. Dealing with Conflicts of Interest. New England Journal of Med- icine 31()(18):1182-1183. . 1990. New "Information for Authors"-and Readers. New England Journal of Medicnd 323:56. Rochon, P. X994.: Evaluating the Quality of Articles Published in Journal Supplements Compared with the Quality of Those Published in the Parent Journal. Journal of the Ameiiean Medical Association 272(2):108-113. Rocbon, P. A.; J. H. Qunvitz, R. W. Simms, P. R. Fortin, D. T. Felson, K. L. Minaker, and T. 6. Chalmers..I994. A Study of Manufacturer-Suppotted Trials, of Non- steroidal Anti=inflammatory Drugs in the Treatment of Arthritis. Archives ofInter- nal Mediciiu'154:157-163. Rodwin, k A.1989. Physicians' Conflicts of Interest: The Limitations of Disclosure. New En~landJournal ofMedlcine 321(20):1405-1408. Roper Otg nization. 1978. A Study of Public Attitudes towanda Cigarette Smoking and the To cco Industry in 1978. Stotm, CT: Roper Organization. Rothman, J. 1993. Conflict of Interest: The New McCarthyisrn in Science. Journal of the Ainerican MsdicalAssoeiation 269(21):2782-2784. Sarokin, H; L: 1988. Opinion, Cipollone v Liggett Group et al. 683 Federal Supple- ment 1487, NJ District Court,1988. . Southgate, hvl. T: 1987. Conflict of Interest and the Peer Review Process. Journal of the Ametiean'Medieal Association 258(10):1375. Steeniand, ~C. 1992. Passive Smoking and Risk of Heart Disease. Journal of the Amer- tcan Me;Iical Alaociation 267:94-99. Stolberg, S. 1994. R*arcbers Try to Separate Smoking Fact from Fiction. Los Ange- les Timlis, 28 May, ~. A 1. Swaen, G., knd J: Meijfrs.:1988. Influence of Design Chanacteristica on the Outcome of Retr ve CobodStudies. British Journal of Industtial Medictne 45:624-629. Taking Mo ey from the Devil (Editorial].1985. British Medical Journal 291(6511): 1743-1744. i Tobacco Ittstitute. 19d6. Tobacco Smoke and the Non-smoker: Scientific Integrity at the Crossroaats. W' ington, DC: Tobacco Institute. . Ciscr 1988. Decades of Initiatives by a Responsible Cigarette Indus- try. Wasbington, : Tobacco Institute. Todd, J. S. 1944: Leiter to Robert S. Blacklow, M.D., president and dean of North- eastern hio Univetai6es College of Medicine, 8 Novembec ~ I .~ .

0 0 i.. . 530 Journal of Health Politics, Policy shd Law ® 0 Research 1992b). Hedge's oral testimo:ny was not available for our analy- sis. However, in his written statement, Hedge (1994:2) noted that his research showed "environmental toba6co smoke plays a very minor role, if any, in eliciting sick building syn ~, ome c.omplaints." In addition, 'he stated that OSHA's proposal to requi~e separate ventilation systems in offices, that allow smoking was "unnecessarily restrictive" (Hedge 1994: 3). Thus his statement also supports the tobacco industry position that OSHA should not regulate smoking in the workplace. A third CIAR special-reviewed project was discussed at a congres- sional bearing related to banning smoking on commercial aircraft (U.S. House 1989). This study was conductea by Torbjorn Malmfors, of Malm- fors Consulting Inc., and it measured-ltvels of ETS on aircraft. Larry C. Holcoinb, of Holcomb Environmental Services, presented preliminary findings of the study and stated that be( acted as a consultant to CIAR on the project. Holcomb also noted that ; te had been asked to submit his statement by the Tobacco Institute. According to Holcomb, "the results of the CIAR study confirm that ETS levels in general are very low in pas- senger cabins" (U.S. House 1989: 369), The study was later published in a peer-reviewed journal, and it concluded that exposure to ETS on air- craft is probably not responsible for health problems in either passengers or flight attendants (Malmfors et a1. 1989). - These findings strongly suggest thai the tobacco industry is funding research through CIAR to develop scientific data that it can use in leg- islative forums to support its position. .: Concern regarding the quality of dgta resulting from CIAR special- reviewed projects has also been raised f'U.S. House 1994a; Barnes et al. 1995). A congressional subcommittee rdcently concluded that 25 percent of the data from one of CIAR's special-ieviewed projects may have been falsified or fabricated (U.S. House 1991a). The study was conducted by Gray Robertson, of HBI, and it involved an analysis of indoor air qual- ity, including measures of ETS, in 585joffice buildings. The study was published in a peer-reviewed journal er et al. 1992) and reported that ETS levels in rooms used for "ligb' smoking were similar to levels in nonsmoking rooms. The article concl.~ ded that "with good ventilation, acceptable air quality.can be maintai#ed with moderate amounts of smoking" (Turner et al. 1992: 19). I The U.S. Congress's Subcommittee n Health and the Environment (U.S. House 1994a) conducted an investi.gation of the HBI study in which it obtained that company's raw data and compared them with data sub- mitted in an interim report to CrAR. The:~ubcommittee's analysis showed W

.. i . ' of interest, particularly-fniancial ones (Relman 1984; Southgate 1987; International Committee of Medical Journal'Editors 1988; Lundberg and Flanagin 1989; Relman.1990; Parish et a1.:1991; Koshland .1992). We found that most CIAR-funded researchers disclosed CIAR as.a.source of funding in their. publications and during their testimony on tobacco- related issues. However, as others hav.e noted, there are varying degrees of disclosure (Rodwin 1989). CIAR-funded investigators tended simply to state that their research had been funded'by the CIAR. However, most readers have never beard of CIAR and do not know that it is financed by the tobacco industry. Even in CIAR's RFAs, the fact that CIAR is sup- ported by the tobacco industry is. mentioned only in the appendix, on page 14 of a twenty-one-page pamphlet (Center for Indoor Air Research 1994). An excellent example of full disclosure is provided by the Health Effects Institute. Publications that acknowledge the institute as a source of funding typically state that it is "an organization jointly funded by the U.S. Environmental Protection Agency and automotive manufacturers." An appropriate way for researchers to disclose funding from CIAR would be to state "This project was funded by the Center for Indoor Air Research (CIAR), an organization created and financed by tobacco oom- panies." In addition, special-reviewed projects should further state 'Tt was selected by tobacco industry representatives, rather than through a process of peer review:' When the true nature of the funding process is disclosed in this way, the full potential for bias is made apparent. industry monies is that it is'often one of the.few, and sometimes the only, conduct biased research as a resuit. The argument in favor of accepting likely to oppose the indnstry in other mattera, and may be more likely to The argument against taking-tobacco industry money is that researchers will provide the industry with undeserved respectability, will be less schools and urged them not to accept support from the tobacco industry. Medical Association, recently wrote to tho deans of many U.S. medical tion, James S. Todd (1994), executive vicp-president of the American Finally, our findings support calls for investigators to refuse all tobacco industry financing. Many authors have argued that researchers should not accept funding of any sort from the tobscco industry ( Ihldng Money from the Devil 1985; Wolinsky 1985; Blocp 1994; White, n.d.). In addi- Impiications for the Ethics of Accepting Tobacco Industry Funding Con CIA peer inde tryry as ai ibilb omn toba spec lobb sion by c m ~ ~ cn J

tcco Smoke: Measuring X: National Academy L6ict of Interest and Sci- Y 30(4):250-251. ad Poisoning. American :ngland Jourva! of Med- s. New England Journal in Journal Supplements rent Journal. Journal of Felson, K. L Mnaker, pported Trials of Non- hri tis. A n: hi ves of Inte r- nitations of Disclosure. Cigarette Smoking and ion. lsm in Science. Journal 1. 683 Federal Supple- :w Process. Journal of e. Journal ofthe Amer- iom Fiction. LosAnge- tics on the Outcome of Wedicine 45:624-629. ol lourna1291(6511): Scientifrc Integrity at sible Cigarette lndus_ tt and dean of North- Turner, S., L. CyF, •and A. J. G}''oss. 1992. The Meastuemenr of Environmental . Tobacco Smoke'in 385 Office Fli*onments. Environmentlnternational 18(1):19-28. U.S. Department of Health and Human Services. 1986. The Health Consequences of._ Involuntary Smoki;tig: A ReporA of the Surgeon General. Rockville. MD: U.S. Pub- Iic Health 5ervice.l•', I U.S. Department of Occupationa~ Safety and Health Administration. 1994. Indoor Air Qualtty Aedrir~g Schedule: ~Panicipants-in the Publie Hearing on OSHA S Pro- posed Standard for Indoor Ai9r Quality (59 FR 15968). Washington, DC: U.S. Department of Labor. ~ U.S. Environmental Protection A4ency. 1992. Respiratory Health Effects of Passive Smoking: Lung-Cancerand Other Disorders. Washington, DC: U.S. Environmen- tal Protection Agency. i U.S. House. 1978> Subcommittee on Tobacco. Effect of Smoking on Nonsmokers: Hearing beforee the Subcommittee on Tobacco of the Committee on Agriculture. 95th Congress. ( .1986. Subcommittee on Health and the Environment. Designation of Smok- ing Areas In Federal Building): Hearings before the Subcommittee on Health and the Environment of the Committee on Energy and Commerce. 99th Congress. 1987. Suboommittee on Aviation. To Ban Smoking on Airline Airrratt: Hear- ing before the Subcommittee dn Aviation of the Committee on Public Works and Transporiation..100th Congre4s. . 1989. Subcommittee on Aviation. To Ban Smoking on Airline AlrrraJt: Hear- ing before the Subcommittee an Aviation of the Committee on Public Works and Transportation. 101st Congres,s. . 1993a. Subcommittee on Health and the Environment. Environmental Tobacco Smoke: Hearing befor~,r the Subcommittee on Health and the Environmrnt :--_ .°.-°-.............°.......': ~ 99th Congress. of the Committee on Energy at~d. Commeree.103d Congress. 1993b. Subcommittee or4 Public Buildings and Grounds. Tb Prohibit Smok- ing in Federal Buildings: Hea{ings before the Subcommittee on Public Buildings and Grounds of the Committee on.Public Works and Transportation. 103d Con- gress. ! 1993c. Subcommittee on Specialty Crops and Natural Resources. Review of the U.S. Environmental Proteltion Agency's Tobacco and Smoke Study: Hearing before the Subcommittee on Specialty Crops and Natural Resources qf the Com- mittee on Agriculture. 103d Cqngess. 1994a. Subcommittee on Health and the Environment Environmental Tobacco Smoke Investigation. ~03d Congress. 1994b. Subcommittee n Health and the Environment. Environmental Tobacco Smokc (Part 2): He rings before the Subcommittee on Health and the Environment of the Conrtnitte on Energy and Commerce. 103d Congress. U.S. Seoate.1985. Subcommitteeion Civil Setvice, Post Office, and General Services. Non-Smokers Rights Act of 1985: Hearings before the Subcommittee on Civil Ser- vice, Post Office, and General ~ervices of the Committee on Gove?nmental Affairs.

M AR's peer-reviewed ry is using them to .utant, because most possible that peer- blicity and improve t suggests that the c data it can cite in ;. For example, data presented at OSHA •d, and data from a ngressional hearing rcial aircraft. In all co industry position Zg research through influence policy on by the lack of peer -le, the Oak Ridge • review appears to :)r bias. Participants firm with ties to the :ntists employed by uudy design, policy of the study, which smoking should not t on the HBI study ved projects appear i the subcommittee he approval of its policy makers will HBI with a partic- a Deflecting Attenti Our study leaves uni funding peer-reviea peer-reviewed proja as an indoor air poll reviewed projects~ .., is supported by a i subcommittee inc~ within HBI in late 1S concept is to broade challenge" (U.S. Ho belief that ETS is "a The tobacco indu by funding projects health hazard, and harmful. For exam reviewed project tlu after his study foun rotic plaques than d The tobacco ind ically increased by based organization Good Publicity ~ il ~I I The tobacco industr CIAR to provide it i ® tobacco industry c improve its public Tobacco Institute e 0 Cigarette Industry, Industry support~ 1988 and has re papers.. . . Pursui industry support Research to awar quality in enclos~ lved the question of why the tobacco industry is projects through CIAR. One possibility is that re being used to deflect attention away from ETS because more than two-thirds of CIAR's peer- substances unrelated to tobacco. This hypothesis ent that was obtained as part of the conpessional Into HBI. The document, which was circulated or 1992, stated that "the key objective of the HBI e debate on indoor air quality to deflect the ETS 1994a: 5). In particular, HBI was to promote the nor contributor" to indoor air problems. may also be trying to divert attention from ETS lieves will support its position that ETS is not a denying funds to projects that find that ETS is one PI, who had received financing for a peer- Ch1R, stated that his funding was not renewed at roosters exposed to ETS had larger artheroscle- nexposed roosters (Stolberg 1994). 1 's power to deflect attention from ETS is dramat- fact that, to our knowledge, no other nationally ding research on ETS. ay also be funding peer-reviewed projects through good publicity and to enhance its credibility. The n cites its funding of independent research to ge. For example, a publication produced by the ed Three Decades of Initiatives by a Responsible es: " ndependent research exceeds $162 million through ed in publication of more than 3,000 scientific its effort to advance scientific knowledge, tobacco to establishment of the Center for Indoor Air ads to independent investigators in the field of air paces. (Tobacco Institute, circa 1988) ® 0 ,,: 4

rticles to determine iects would be more s to be pro-industry. lity and conclusions ized, articles result- iociated with poor- ;ulting from special- riewed journals. In ad 20 percent were other types of pub- :cial-reviewed proj- : the sole source of !cial-reviewed proj- wed projects to sup- )port evidence from Ls'sociated with pro- jers 1988; Lexcbin 3rovide further sup- f may be funding o produce data that of research funded ed research has led acco i-n,r_tustryposi= s that support its ;ial-reviewed proj- >rojects to have tes- R-funded research by CIAR-funded earchers who have iTS: the list of per- d Health Adminis- lity standard (U.S. iinistration 1994); ; to the U.S. Envi- Barnes and Bero  industry-Funded Research 527 Tabie 3 Quality and Outcomes of Publications Resulting from C1AR- Funded Research ercentage of Articl • Resulting from Peer-Review Projects (n = 48) Percentage of Articl, • Resulting from Special-Reviewed .. Projects (a = 21) / Statistical Test e of Article Ty p Symposium 17 33 Chi square = 4.37 Original 71 48 df = 3 Review 2 . 0 p=0.22 Other 10 19 Souces(s) of Funding Acknowledged • Tobacco Industry 15 62 Chi square=24.12 Multiple 79 19 df = 3 Other 2 0 p<0.0001 None Stated 4 06 19 Conclusion Pro-industry 2 CMIN 29 Chi square = 11.89 And-industry 6 ~ ` 9 df=2 Neutral 92 G~ 14 62 p = 0.003 10 ~ ronmental Protection Agency regarding its draft risk assessment of ETS, which had been obtained for a previy,t,c, stt;dy `rR,,,`'ro anc[G}an{z }993): and the tables of contents of congressional hearings related to ETS. We identified congressional hearings by searching the Congressional Infor- mation Services Index from 1970 to 1994 using the key words "smoking" and "tobacco:' We included only those.hearings that focused on ETS or proposals to restrict smoking in public places. We identified nine hear- ings, including three on proposals to restrict or ban smoking in federal buildings (U.S. Senate 1985; U.S. House 1986,1993b), two on proposals to restrict smoking in public places nationwide (U.S. House 1993a, 1994b), two regarding proposals to ban smoking on airline flights (U.S. House 1987, 1989), one on the health effects of exposure to ETS (U.S. House 1978), and one involving a critical review of the U.S. Environ- mental Protection Agency's risk assessment of ETS (U.S. House 1993c).

.cco Smoke. Journal of se to a Risk Assessment 03-113. +vithin Academia in the ~f Investigative Derma- eatch. Tobacco Control I Interpretation of Epi- -lrnica! Epidemiology ida: RequestforApplf- 'ions. Linthicum, MD: plications. Linthicum, r Indoor Air Research. wided by V. Christine h, 13 October. led research, provided door Air Research, 13 led research, provided Center for Indoor Air ions. Linthicum, MD: :um, MD: Center for -ig the Three Stages of a 263(10):1392 - 1345. g Studies Published in panies: Guidelines for :hives of Dermatology Pr Tobacco Researrii- cal Trials. Journal of agarette Smoking on ne. Journal of Family Barnes and Bero , Industry-Funded Research 539 Fontham, E. T. H., P. Correa, P. Reynolds, A. Wu-Williams, P. A. Buffler, R. S. Greenberg, V. W. bben, T. Altertnan, P. Boyd, D. F. Austin, and J. Liff.1994. Envi- ronmental Tobacco Smoke and Lung Cancer in Nonsmoking Women:. A Multi- center Study. Journal of theAmerican Medical Associatton 271(22):1752-1759. Freedman, A. M., and L P. Cohen. 1993. Smoke and Mirrors: How Cigarettc Mak- ers Keep Health Question "Open" Year after Year. Wall Stnet Journal, 1 I Febru- ary,p.A1. Fry, J. 1995. HBI Piesident Denies Charges, Decries Report. IndoorAir Review 4:1. Glantz, S. A., and W. W. Parmley. 1991. Passive Smoking and Heart Disease: Epi- derniology, Physiology, and Biochemistry. Circulation 83:1-]2. Hedge, A. 1991. Psychosocial and Environmental Influences on "Sick" Building Syn- drome. Indoor Air Pollution: A Complete Update on fAQ Pollutants, Sources, E,ffects, Liability,!Research, and, Control. Tulsa. OK University of Tulsa. 1994. Letter to the Occupational and Safety Health Administration Declar- iog Notice of Intention to Appear at Hearings on Proposed Indoor Air Quality Standard. Cornell University, Ithaca, NY, 4 August. Hedge, A., W. Erickson, and GAtubin. 1992. Effects of Personal and Occupational Factors on Sick Building Syndrome Reports in Air-Conditioned Offices. In Work and Well-Being: Assessmentsand Interventions for Occupational Mental Health, ed. J. Qu•ick, L. Ivltuphy, and J. Hurrell. Washington, DC: American Psychological Association: 286-298. Hiliman, A. L., J. M. Eisenberg, M. V. Pauly, B. S. Bloom, H. Glick, B. Kinosian, and J. S. Schwartz. 1491. Avoiding'Bias in the Conduct and Reporting of Cost-effec- tiveness Research Sponsored by Pharmaceutical Companies. New England Jour- nal of Medicine 324(19):1362-1365. Hirayama, T. 1981. Non-smoking'Wives of Heavy Smokers Have a Higher Risk of Lung Cancer. A Study from Japan. British Medical Journal 282(6259):183-185. International Committee of Med'[cal Journal Editors. 1988. Uniform Requirements for Manuscripts Submitted to Biomedical Journals. Annals of Internal Medicine 108:258-265. Jenldns, R. A. 1995: Transcript of Oral Testimony Presented befotr the Occupational Safety and Health Administration Regarding the Proposed Standan3for IndoorAir Quality. Washington, DC: Bayley Reporting, Inc. Koshland, D. E. J. I992. Conflict of Interest Policy. Science 257:595. Levin, M. 1993. Whes behind the Building Doctor. The Narion 9(16):168-171. Lexchin, J. 1993. Iritetactions between Physicians and the Pharmaceutical Industry: What Does thb Literature Say? Canadian Medical Association Journal 149(10):1401-407. Lilienfeld, D. E. ]99L The Silence: The Asbestos Industry and Early Occupational Can- cer Research-A Case Study. Ametfccmt Journal of Public Healtle 81(6):791-800. Lundbetg. G. D., dnd. A. Flanagin. 1989. New Requirements for Authors: Signed Statements of Apthorship Responsibility and Financial Disclosure. Journal of the American MedicalAssociation 262(14):2003-2004. Malmfors, T., D. Thorburn, and A. Westlin. 1989. Air Quality in Passenger Cabins of DC-9 and Mn-80 Aircraft. Environmental Technology Letters 10:613-628.

® N ® I a I Conclusions Barnes and Befo w Industry•Funded Research 537 available sources of funding. Our findings suggest that the tobacco indus- try's credibility is enhanced through its association.with independent investigators. The •scientists who sit dn CIAR's science advisory board, and those who accept funding throug}i.its peer-reviewed projects, all pro- vide the tobacco industry with the op~ortunity to claim that it is funding high-quality, objective research. Hov~,ever, the true purpose of CIAR is made clear by the industry's use of itsspecial-reviewed projects to lobby against regulation of smoking in publi places. All researchers associated with CIAR, even those who do not study tobacco-related issues, are con- tributing to the tobacco industry's agenda. CIAR is funding two types of projects°for two different purposes. CIAR's peer-reviewed projects are avKarded ~ased;on the recommendations of independent scientists and most are unrelateii to ETS. The tobacco indus- try may be funding peer-reviewed p' ects t`o deflect attention from ETS as an indoor air pollutant, to gain goo~ publicity, and to enhance its cred- ibility. CIAR's special-reviewed projicts are awarded based on the rec- ommendations of tobacco industry e!cecutives and tend to support the tobacco industry position. The tobacFo industry appears to be funding special-reviewed projects to develog scieriti$e data that it can use to lobby against regulation of smoking i~t public places. CIAR's stated mis- sion of funding high-quality, objecti~Je research has been compromised by conflict of interest. ~ I ; References i . Barnes, D.. and L. Bero. Unpublished daia. Methodological Quality of Original Research Articles on Environmental Tobacco 'Smoke and the Effects of Peer Review. Barnes, D. E., P. Hanauer, J. Slade, L. A: Be , and`:S. A Glaotz 1995. Environmen- tal Tobacco Smoke: The Brown and WiIliamson Documents. Journal of the Arncr- ican Medical Association 274(3):248-25 . Bero, L., D. E. Barnes, P. Hanauer, J. Slade, nd S. A. Glantz. 1995. Lawyer Control of the Tobacco Industry's External Res h Program: The Brown and Williamson Documents.JournaloftheAmerican.6le tcalArsociation274(3):241-247. ' Bero, L. A., A. C~albraith, and D. Rennie. 1 92. 71e Publication of Sponsored Sym- posiums in Medical Journals. New,Bng I Journal ofMedicine 327:1135-1140. 1 , ;outhgate 1987; 3; Lundberg and land 1992). We R as a source of ny on tobacco- varying degrees s tended simply However, most it is financed by it CIAR is sup- e appendix, on )r Air Research by the Health tute as a source y funded by the nanufacturers:' 1g from CIAR • for Indoor Air y tobacco com- urther state "It than through a ding process is ?parent. use all tobacco archers should Taking Money n.d.). In addi- the American y U.S. medical )acco industry. iat researchers y, will be less more likely to rr of accepting times the only,

inded researchers who itements. When avail- As of oral testimony to' research, whether the ry was disclosed, and ustry position on ETS es). fs on special-reviewed compared with 3 per- rojects (Fisher's exact ct had testified on five I only once. led Prs supported the to health. The P1 who toking can be accom- D PIs argued that evi- ;ease; one maintained ,onclude that ETS is xplain the statistical tat ETS is not a major ; and one argued that at the exposure is too ed ptojects have been al-reviewed projects proposed indoor air banning smoking in Scussed at hearings : Sights. In all three :ts were used to sup- 3ger A. Jenkins, who ved project entitled tal Tobacco Smoke" fational Laboratory . kins disclosed in his ; however, he stated or against OSHA's iducting a study and gathering some information I thinl ins 1995: 9690). Jenkins testified orally that the exposure to ETS in the workplac place and non-workplace exposu had participated in the study: Be firm based in Winston-Salem, N participants; R. J. Reynolds had determine the levels of exposure t Ridge National Laboratory was study design and for analyzing a that Bellomy often conducts mark The study design Jenkins descri of bias. Specifically, it is highly u in the outcome of a study to be al and to conduct the laboratory an could have been hired. During his eral steps had been taken to ensur late the data in any way. Howev would have approved a study desi Jenkins also noted that the de slightly skewed. For example, th women, to have high soeioeconoo els, to be white; and to work in w that many or all of these factors normal workplace exposure to ET According to Jenkins, the preli that most persons are exposed to sion supports the tobacco indus should not regulate smoking in th The second CIAR special-revie was conducted by Alan Hedge, of ronmental Analysis at Cornell Univ Hedge received $431,177 from- 1 which studied the effects of differ conditions on indoor air quality in purpdse of the study was to m,easure andjfo compare the levels of work- e. H~ 9tated that three organizations omy Research, a marketing research h Carolina, had recruited the study ondu~ted the laboratory analyses to ETS in the various groups; and Oak sponkible for approving the overall inte~reting the data. Jenkins noted. ting r,dsearch for R. J. Reynolds. contains several potential sources usua~ ~or firms with strong interests owed;to select the study participants Iyses' :particularly when other firms testiWny, Jenkins stressed that sev- that Il:. J. Reynolds did not manipu- r, it ;4 unlikely that peer reviewers such a high potential for bias. ographics of the study group were parti~ipants were more likely to be stattl , to have high educationallev- ite-co occupations. It is possible ould `e associated with lower-than- ~. i ary; ults from the study suggested ETS on the job. This conclu- posi , ri that•the federal government workoace. prbject presented before OSHA Department of Design and Envi- ccording to CIAR's tax forms, 9 to 1 993 ~to conduct this project,s nt sn1Qktng policies and ventilation Mce btsildings (Center for Indoor Air

ble for our analy- 2) noted that his i very minor role, " In addition, he .ation systems in le" (Hedge 1994: try position that ed at a congres- !ial aircraft (U.S. !mfors, of Malm- iircraft. Larry C. ited preliminary Itant to CIAR on ed to submit his 1mb, "the results very low in pas- 3ter published in e to ETS on air- -ither passengers i CIAR special- fa; Barnes et al. I that 25 percent i may have been is conducted by indoor air qual- . The study was 2) and reported similar to levels ood ventilation, ate amounts of te Environment I study in which i with data sub- inalysis showed that measures of respirable suspended partictilates, which are used to estimate ETS levels, were substantially altered more than 25 percent of the time. Several other discrepancies were identified that would have tended to decrease the levels of ETS measured in rooms used for smok- ing (U.S. House 1994a). HBI employees confirmed that their data col- lection forms were routinely changed to minimize measurements of ETS. The subcommittee requested an independent ' analysis of HBI's data, which concluded that "the data [are] so marred by unsubstantiated data entries, discrepancies, and misclassifications that it raises serious ques- tions of scientific fraud" (U.S. House 1994a: 6): Robertson has denied the congressional findings (U.S. House 1994a; Fry- 1995). According to information submitted to the subcommittee by the Tobacco Institute, representatives of HBI have testified at least129.times before federal, state, or local government agencies on issues related to ETS. In our analysis, Robertson submitted a total of five statements, and he typically argued that smoking could be accommodated indoors through adequate ventilation. • . Our findings on the use of CIAR-funded research provide further evi- dence to support our hypothesis that the tobacco industry is using CIAR's special-reviewed projects to produce data it can use in legislative and legal settings. In addition, our findings suggest that the industry may be using CIAR to develop relationships with researchers who might be will- ing to testify on behalf of the tobacco industry. Discussion • Together our findings suggest that CIAR is funding two types of projects for two different purposes. Peer-reviewed projects are funded based on the recommendations of independent scientists. They are awarded pri- marily to academic investigators and most of them are not related to ETS. Special-reviewedprojects, on the other hand, are awarded based on the recommendations of tobacco industry executives. They are more likely than peer-reviewed projects to be related to ETS, to be awarded to private sector. consultants, and to support the tobacco industry position that ETS has not been proved harmful to health. In addition, data from several of CIAR's special-reviewedrojects have been presented at hear- ings related to ETS to support the tobacco indostry agenda. These findings sugge'st that CIAR's peer-reviewed and special-reviewed projects may be serving different pur~poses. The tobacco industry appears to be funding special-reviewed projects to develop scientific data that it