In May 2003, the British Medical Journal (BMJ) published “Environmental tobacco smoke and tobacco related mortality in a prospective study of Californians, 1960–1998”.¹ In this paper, Enstrom and Kabat used a longitudinal cohort collected by the American Cancer Society beginning in 1959 (the CPS-I dataset, Cancer Prevention Study) to conclude that secondhand smoke exposure does not increase the risk of lung cancer and heart disease. This conclusion conflicts with comprehensive reviews which find that secondhand smoke increases these risks by 20–30%.² The paper was quickly and widely cited in the lay press^3,4 and by tobacco industry supported journalists who criticise government sponsored “junk science”.⁵

Subsequent criticisms of the paper focused on the methodology of the study and the authors’ disclosed financial ties. The main methodological issue was that there was no real “unexposed” group in the CPS-I dataset that Enstrom and Kabat used.^6,7

The article included a lengthy statement of funding sources and competing interests, listing as one of its sponsors the tobacco industry’s Center for Indoor Air Research (CIAR). Peer reviewers tend to be more critical of articles with industry sponsorship than those without such sponsorship.⁸ One rapid response to the Enstrom and Kabat article suggested that it might be useful for editors to “require a statement about the role of the funding source in the design, conduct, analysis and reporting of the data”.⁹ Earlier tobacco industry funded studies of the health effects of secondhand smoke have failed to fully disclose the sponsor’s role in the research.^10,11 In contrast to these earlier situations, the Enstrom and Kabat paper was consistent with the requirements for such statements established by the BMJ.

Financial disclosure policies for authors of journal articles are becoming the norm.^12,13 However, experts on scientific journal disclosure policies have questioned the adequacy of these policies.¹⁴ Studies comparing disclosures of financial ties in research articles with information on the financial ties of authors obtained from independent sources have shown that only a small percentage of research articles (less than 20%) fully disclose the financial ties of authors.^14,15 A more recent study investigated the lead and last authors of research articles who declared that they had no conflicts to disclose in 163 original articles that appeared in four journals (New England Journal of Medicine, Journal of the American Medical Association, Environmental Health Perspectives, Toxicology and Applied Pharmacology). By comparing publicly available databases and the journals’ conflict of interest policies, the study found 13 articles (8%) where relevant conflicts of interest were not disclosed to readers.¹⁵ In most cases, the authors failed to disclose the relevant information to the journal.¹⁵

The inadequacy of journal article disclosures when compared to information on financial ties of authors obtained from other sources has led experts on disclosure to suggest more comprehensive policies and stronger enforcement of existing policies.^14,16 For example, experts have suggested that journals require disclosure of exact amounts and duration of financial ties, and that journals clarify how they define the “relevance” of a financial tie to the research.¹⁵ A number of scholars have argued that corporate sponsorship of research or financial ties of authors should be prohibited.^14,17,18 These proposed bans eliminate the need for disclosure to “manage” the conflict of interest and protect against even the appearance of conflict.

The Enstrom and Kabat paper provides an interesting case study to test this suggestion precisely because they provided an extensive disclosure that met the current BMJ standards for such disclosure. Using methods similar to that of Krimsky, we compared the disclosure in a scientific journal article with information on financial ties of the authors obtained from an independent source.^19,20 We compare the disclosure Enstrom and Kabat provided in their article with internal tobacco industry documents describing their relationship with the tobacco industry in order to obtain details of the history and nature of their association, with the goal of understanding whether a disclosure statement that met BMJ’s requirements gave the reader a clear description of the role of the tobacco industry in the study and the historical relationship of the authors with the industry.

METHODS

We retrieved documents from the Legacy Tobacco Documents Library (www.legacy.library.ucsf.edu) using the terms “Enstrom”, “Kabat”, “CIAR”, “CPSI”, “CPSII”, “Womble Carlyle”, and “Shook, Hardy and Bacon” and names of key individuals, followed by “snowball searches”. We collected 156 documents; 81 discussed the development of the CPS-I data analysis or other research by Enstrom or Kabat.

RESULTS

Table 1 summarises financial ties between the authors and the tobacco industry. The disclosure statement was long, 305 words, compared to the average of 127 words in the 20 other research articles published in BMJ that month. Of these 305 words, 98 related to the tobacco industry (table 2).

Acknowledgement of American Cancer Society and California Tobacco-Related Disease Research Program

The authors disclosed that the American Cancer Society (ACS) developed the CPS-I dataset.²¹ The disclosure notes that some work was funded by the California Tobacco-Related Disease Research Program (TRDRP), then states that continued funding was “denied”. The renewal for the project was not funded because it had inadequate scientific merit as determined by a peer review panel in a year that TRDRP had an 85% budget reduction.²²

Early interactions with the tobacco industry

The disclosure states that the research was supported by the tobacco industry after continuing support from TRDRP was denied and that “[i]n recent years” Enstrom received research funding from the tobacco industry because he was unable to obtain equivalent funds from other sources. Although these statements are accurate and comply with Enstrom’s reporting obligations under BMJ policy, the tobacco industry documents also show that Enstrom had a long history of association with the industry. In 1975, he approached the Council for Tobacco Research (CTR), a tobacco industry research organisation, to fund a study of cancer among US Mormons.^23,24 He argued that the study “should be helpful in assessing the possible role which other factors besides smoking play in the etiology of lung cancer”.²⁴ The proposal was deferred for discussion by the Executive Committee, Scientific Advisory Board of CTR in June 1975.²⁵ Although Enstrom requested a letter of support from the Tobacco Institute, the industry’s trade organisation in the USA,²⁶ and modified his experimental design in April 1976,²⁷ there is no record that the proposal was funded. In 1978, Enstrom submitted another request to CTR to study “Smoking cessation and mortality trends among California physicians”.²⁸ There is no record that this proposal was funded.

In 1979, Enstrom published a review article that provoked concern and criticism from the tobacco industry.^29–31 The article, which was consistent with journal policy at the time, disclosed no sources of funding, used data from the Mormon cohort, US veteran cohort, CPS-I, and National Mortality Survey to conclude that lung cancer rates had been rising among non-smokers between 1914 and 1968 and that this rise was related to factors other than “personal cigarette smoking” including “environmental pollution including environmental tobacco smoke”.³²

Developing a relationship with the tobacco industry

The industry documents reveal little interaction between the tobacco industry and Enstrom from 1979 until 1990, when the tobacco industry began to recruit Enstrom as a potential critic of research on secondhand smoke. In 1990, Thomas Borelli, Manager Scientific Issues, Philip Morris, asked Enstrom, among others, to comment on a PhD dissertation by Luis Varela.^33,34 Borelli was interested in the dissertation because it contained findings suggesting that secondhand smoke exposure was not associated with disease. The industry frequently cited the dissertation as evidence that should be included in risk assessments of secondhand smoke.³⁵ In contrast to the preliminary results in the dissertation, however, the final publication in the New England Journal of Medicine³⁶ concluded that childhood exposure to secondhand smoke was associated with adult lung cancer.

Enstrom declined to review the dissertation, stating: “My epidemiologic research does not deal directly with the issue of environmental tobacco smoke”.³³ However, he used the invitation to inquire about funding from Philip Morris to update his work on lung cancer mortality trends among non-smokers.³³ Borelli responded that “Philip Morris does not usually fund research projects” and referred Enstrom to the industry’s CIAR.³⁷ The CIAR was supported solely by the tobacco industry and funded research projects using two different mechanisms. Some projects were peer reviewed by scientists and were more likely to examine health effects of indoor air contaminants other than secondhand smoke.³⁸ Other projects, which were more likely to focus on secondhand smoke, were reviewed by tobacco industry executives and lawyers.³⁸

Enstrom sent Borelli a number of references that suggested a need for further research on lung cancer trends among non-smokers, noting that “Attention could be given to the issues of personal smoking and environmental tobacco smoke”.³⁹ Myron Johnston, Research and Development, Philip Morris, was asked by Bob Pages, Director, Science and Technology, Philip Morris, to comment on the correspondence between Borelli and Enstrom and concluded that:

[Enstrom] wrote a section of the 1983 Surgeon General’s Report but now seems to have changed his mind on the smoking/mortality relationship. I say seems to have changed his mind because his writing is a little obtuse…⁴⁰

Enstrom’s section of the 1983 Surgeon General’s Report concluded that smoking cessation decreases the risk of heart attack and coronary heart disease. However, the conclusion went on to suggest that other risk factors for heart disease, such as blood pressure and cholesterol levels, needed further study.⁴¹

Enstrom submitted a pre-proposal to CTR for a study entitled “Mortality Trends Among Smokers and Nonsmokers”. In September 1991, two of the three CTR Scientific Advisory Board executive committee reviewers recommended that Enstrom be encouraged to submit a formal application.⁴²

In November 1991, a formal application was submitted to CTR⁴³ encouraged by the executive committee.⁴⁴ The project sought to investigate mortality trends from lung cancer, other smoking related causes, and all causes among smokers and non-smokers between 1966 and 1987 using data from the 1966–68 National Mortality Survey, 1986 National Mortality Survey, 1971–87 NHANES I Follow-up Study, and the American Cancer Society “follow-up of the 118,000 California residents in the 1959 Cancer Prevention Study (CPS-I)”.⁴³ CTR funded the study for $34 500 for the first year of a three year project in 1992 and funding was renewed in 1993⁴⁵ and 1997^46,47 with a $25 000 supplement in 1997.⁴⁸

Part of the data from the CPS-I portion of the analysis was published in Epidemiology, but CTR funding was not acknowledged in the published paper.⁴⁹Epidemiology required authors to disclose their funding sources in a cover letter to the editor and the editor decides if these funding sources should be published (www.epidem.com/pt/re/epidemiology/authorinfo.htm). Enstrom’s cover letter is not available among the tobacco industry documents. Some of the data also are reported in the recent BMJ article by Enstrom and Kabat without acknowledging CTR funding (table 2). The analysis from the NHANES data was published with an acknowledgement of CTR funding.⁵⁰

Reanalysis of Fontham article

In April 1996, Enstrom submitted to the Journal of the American Medical Association (JAMA)⁵¹ a reanalysis of a major cohort study linking secondhand smoke and lung cancer published by Fontham and colleagues.^52,53 The paper disclosed that the reanalysis was “supported in part by a special grant form the R.J. Reynolds Tobacco Company and Philip Morris”.⁵⁴ Enstrom’s analysis was conducted for Womble and Carlyle, one of the industry’s law firms.⁶⁴ Before Enstrom submitted the manuscript to JAMA, he circulated it for comment among tobacco industry law firms, executives, and INBIFO (a Philip Morris scientific laboratory in Germany).⁵⁶

JAMA rejected the paper.⁶⁵ Although Enstrom’s response to the peer reviewers’ comments was circulated within the law firm Womble and Carlyle,⁶⁶ it is unclear whether the paper was resubmitted to JAMA. It was not published.

Strengthening the relationship with the tobacco industry

Relationship of low levels of active smoking to mortality

In January 1997, Enstrom submitted a research proposal to the Philip Morris Research Center,⁶⁷ where it was reviewed by the Scientific Research Review Committee (SRRC), a committee whose purpose was to “ensure that all scientific research, related to tobacco or smoking, conducted or funded by Philip Morris, …serves relevant business needs”.⁶⁸ The proposal, “Relationship of low levels of active smoking to mortality”, sought to analyse data from four epidemiological cohorts: 1980–94 US Veterans study, 1971–92 NHANES I, 1976–92 NHANES II, and 1960–94 CPS-I in California. The CPS-I analysis was an expansion of the analysis funded by CTR. In his cover letter to Richard Carchman, Director of Scientific Affairs, Philip Morris, Enstrom stated:

These data are highly relevant to the ETS issue… A level of trust must be developed based on my past research on passive smoking and epidemiology in general in order to work out the best way for me to conduct this research. A substantial research commitment on your part is necessary in order for me to effectively compete against the large mountain of epidemiologic data and opinions that already exist regarding the health effects of ETS and active smoking.⁶⁹

Despite his status as a faculty member, Enstrom noted that a relationship of “mutual trust” with the tobacco industry in the context of its funding of his proposed research would minimise university involvement in the project. The proposal stated: “an unrestricted gift to James E. Enstrom / UCLA with mutual understanding/trust would minimize university restrictions and eliminate overhead costs.”⁶⁷ During January and February 1997, the proposal was reviewed by high level Philip Morris executives, lawyers and scientists who comprised the SRRC.^70,71 At this time, the SRRC consisted of R Carchman, R Cox, C Ellis, A Kassman, J Nelson, GM Nixon, H Reif, W Reininghaus, and R Walk.⁷² Personnel at INBIFO commented that the amount of money requested seemed high when considering the amount of new scientific information that would likely be produced by the research, but stated that Enstrom “seems to have good connections/resources which might be useful in the future for other issues”.⁷³

Philip Morris funded the project in April 1997 for $150 000 to be paid in two instalments.^74–79 (fig 1)

Research on secondhand smoke

During the 1990s, Enstrom also began to seek funding from the CIAR. In May and June 1996, Enstrom submitted a pre-proposal to CIAR for research on secondhand smoke,^80,81 including the re-analysis of the Fontham study (subsequently funded by RJ Reynolds and Philip Morris, as described above), reanalysis of the Varela dissertation data, a new case–control study, and follow up of several cohort studies.⁸¹ One of the proposed cohort studies was:

Conduct analysis of California CPS I cohort (51,000 males and 67,000 females) from 1960–94 and conduct follow-up of CPS I from selected states where individualized computerized death records are available (up to 200,000 and 300,000 females). Analysis will examine spousal smoking and death from lung cancer, coronary heart disease, and all causes.⁸¹

A formal proposal was submitted to Max Eisenberg, director CIAR, on 15 July 1996. In his cover letter, Enstrom noted that the results of his past research were favourable to the industry:

For the past three years I have done consulting and research on passive smoking for Jeffrey L. Furr of Womble Carlyle [law firm] on behalf of RJ Reynolds and Philip Morris. This research has found a number of results that raise serious questions about several published findings on the relationship of passive smoking to lung cancer and other diseases⁶⁴

Enstrom’s proposal was discussed at the meeting on 14 August 1996 of CIAR’s board of directors (consisting of tobacco industry executives and lawyers), which agreed that Eisenberg would meet with Enstrom to discuss the proposal in more detail.^82,83

Two sets of review comments on the 1996 Enstrom proposal were in the industry documents. One set, from INBIFO,⁸⁴ was critical of the proposal for its lack of a hypothesis and vague description of methods. The other review, probably from the CIAR board of directors,⁸⁵ was critical for similar reasons, as well as the fact that Enstrom once had acknowledged that secondhand smoke might be harmful. It noted:

Engstrom [sic] says that “[I]t is not possible to rule out a very weak relationship between passive smoking and mortality, especially for individual causes of death.” How he reached a relationship for individuals when the statistics are not significant cannot be seen.⁸⁵

The reviewer went on to recommend an entirely different study for Enstrom:

In order to achieve something really new, CIAR could head for a pre-project which would check the protocols used for the above mentioned studies and put together the “ideal protocol” with all unthinkable adjustments for confounders and necessary investigations. This would give CIAR a “model protocol” which could then be used in order to create a “gold standard” against which other study protocols could be evaluated.⁸⁵

We did not find any documents showing that this “model protocol” was developed.

At its meeting on 25–26 November 1996, the CIAR board of directors decided that Max Eisenberg (Director, CIAR), Richard Carchman (Vice President, Scientific Affairs, Philip Morris), and Charles Green (CIAR, Chairman of Board and Principal Scientist, Research and Development, RJ Reynolds)⁸⁶ would visit Enstrom at the University of California, Los Angeles. Enstrom’s proposal was not funded at that time.

One result of the UCLA meeting was apparently to encourage Enstrom to collaborate with Geoffrey Kabat.^87,88 Enstrom and Kabat had no prior record of collaboration. A search of PubMed on 25 July 2003 revealed that their only joint publication was the article in BMJ in 2003.¹

Kabat’s ties with the tobacco industry

The documents provided no evidence that Kabat received any direct funding from the tobacco industry. Kabat had an ongoing indirect relationship with the tobacco industry since at least 1981, through Ernst Wynder, whose American Health Foundation had been funded by Philip Morris.⁸⁹ Two of the 21 papers Wynder and Kabat co-authored between 1981 and 1995 were related to passive smoking, concluding that there was no association of secondhand smoke with lung cancer.^90,91 He and Wynder published a review of secondhand smoke and lung cancer in a symposium proceeding⁹² which concluded that further evaluation of secondhand smoke and lung cancer risk was necessary. Although the symposium had been organised by individuals who were affiliated with the industry, tobacco industry funding was not expressly acknowledged in the symposium publication.^{92a,92b,92c,93}

In 1997 Kabat co-authored a paper with several members of the tobacco industry’s secret International ETS Consultants Project⁹⁴ critical of the evidence linking secondhand smoke with lung cancer. The project was run by industry lawyers to recruit, train, and pay scientists who would advocate the industry’s position.^11,95

Receiving funding from CIAR

In September 1997, Carchman and Eisenberg had additional face-to-face meetings with Enstrom to discuss possible CIAR funding.^96–103 On 3 November 1997, Eisenberg recommended that Enstrom’s revised proposal, “Proposed research on passive smoking”, be considered under the Directed Studies Program which is controlled by high level industry executives.¹⁰⁴ The revised proposal only included the analysis of the CPS-I dataset.¹⁰⁵ The proposal stressed Enstrom’s past relationship with the ACS, noting that he “made use of this large and rich data base [CPS-I] with the cooperation of ACS”.¹⁰⁵ Although Enstrom listed former ACS Vice Presidents for Epidemiology Clark Heath and Lawrence Garfinkel as unpaid consultants in the proposal (and in the BMJ disclosure, table 2), the proposal in the industry documents did not include letters of support from them, as is usual with consultants. In addition, ACS had advised against the use of the CPS-I dataset for Enstrom’s analysis.⁶

In a memo commenting on the Enstrom proposal in November 1997, Chris Coggins of Lorillard Tobacco recommended funding the proposal.¹⁰⁶ An (unsigned) internal industry critique of the final proposal raised many of the same concerns about the Enstrom study that were subsequently raised by ACS⁶ and in the BMJ rapid responses (http://bmj.bmjjournals.com/cgi/eletters/326/7398/1057). For example,

The proposal fails to distinguish between “ETS exposure” and “living with a spouse who smokes,” makes light of substantial loss-to-follow-up expected in this cohort [CPS-I], freely assumes adequate adjustment can be made for the non-representativeness of the initial cohort, underplays the import of smoking cessation on the analysis, and is uncritical of positions established by anti-smoking community.¹⁰⁷

Despite these concerns, the critic was “impressed” with the listing of Heath and Garfinkel from ACS as consultants on the proposal, as well as Kabat.¹⁰⁷ The reviewer suggested that their participation “should add credibility to the interpretation of the results”.¹⁰⁷

On 19 November 1997 the CIAR board of directors voted to fund the Enstrom CPS-I proposal with modifications¹⁰⁸ for $525 000 from 1 June 1998 through 31 May 31 2001.¹⁰⁹

Additional financial ties

Enstrom had numerous other financial ties with the industry. He was paid to prepare analyses of scientific documents for tobacco industry law firms and served as a peer reviewer for Philip Morris’ external research programme.¹¹⁰ In 1998, Bill Rickert, chair of the consulting firm Labstat, Inc, was asked by the Canadian government to convene an expert committee to examine and make recommendations concerning cigarette toxicity.¹¹¹ Carchman, of Philip Morris, recommended Enstrom and long time tobacco industry consultant Peter Lee to serve on this committee.¹¹² Enstrom was a presenter at a scientific meeting organised by the tobacco industry, a June 2000 epidemiology conference organised by Philip Morris at which he discussed CPS-I results.^113,114

DISCUSSION

The published disclosure in the Enstrom and Kabat paper,¹ which meets the requirements established by BMJ, does not reveal the full extent of the relationship the authors had with the tobacco industry. Tobacco industry funding of research is associated with favourable outcomes for the industry.^115–117 Reasons for this observed association are complex and include sponsor involvement in the research questions asked, design, conduct, and publication of the study.¹¹⁸ The Enstrom and Kabat paper is another example of how the content of even an extensive funding disclosure that meets the journal’s requirements may not allow readers to understand fully the nature of the relationship between the authors and the research funder.^10,11,119 The history of Enstrom’s contacts and collaborations with the tobacco industry illustrates several reasons why the industry funds scientific research. First, although industry insiders were critical of Enstrom’s methods, they nonetheless funded the research to enhance the industry’s credibility, particularly by touting an association with the American Cancer Society. Second, the industry funded Enstrom because it perceived that his connections might be useful in the future. Enstrom had contributed to a Surgeon General’s report on smoking and some of his early work suggested that secondhand smoke might be associated with cancer. Funding Enstrom allowed the industry to work with a scientist to advance its goals, as it has with other scientists.⁸⁹ Third, funding Enstrom provided access for industry executives to seek to influence his research protocols by suggesting modifications. Lastly, funding Enstrom allowed the industry to connect him with other tobacco industry affiliated researchers, such as Kabat.

The Enstrom and Kabat BMJ disclosure statement, when compared to the full extent of funding for the study as revealed in the internal industry documents, suggests that competing interest disclosures can be an inadequate mechanism for obtaining a complete understanding of the role of the sponsor in research. BMJ’s competing interest disclosure policy specifically asks authors to disclose only competing financial interests related to their article; the disclosure of any other competing interests is left to the judgment of the author. Although the BMJ, like many other journals, asks for disclosure of relevant financial ties within the last five years, a requirement with which BMJ found that Enstrom and Kabat had complied, it might be more informative for readers to know that an author has a long, steady history of funding from a particular sponsor. Thus, journals could ask for disclosures of financial ties over a longer period of time.

Our findings raise the question of what additional steps journals can take to obtain the most meaningful disclosures from authors. An elaborate policing operation is not feasible or necessarily desirable, but a simple search could provide additional information. At least if the paper in question relates to tobacco, BMJ and other journals could conduct a quick search of the tobacco industry documents for the names of authors of papers on tobacco. The documents are freely available on the internet at legacy.library.ucsf.edu and bat.library.ucsf.edu. The industry documents could identify more details of the relationship between researchers and the tobacco industry than required in a simple funding disclosure. Beyond the possibility of identifying undisclosed funding, such searches could provide a description of the tobacco industry’s role in the design, conduct or dissemination of the research. As an alternative to conducting tobacco industry document searchers themselves, journals could seek a peer reviewer with tobacco industry document research experience. Journals should discuss how the tobacco industry documents can be used most effectively to improve and inform the peer review process.

Journals should be alert to disclosures of funding from tobacco industry supported research organisations such as CIAR or the more recently formed Philip Morris External Research Program. Journals could keep on file peer reviewed articles that have investigated these tobacco industry programmes. These articles would describe the involvement of tobacco industry lawyers and executives in selecting projects for funding.^38,120

Readers of the Enstrom and Kabat paper are likely to critique it more rigorously if they are aware of the authors’ long standing relationship with the tobacco industry and lack of support from the ACS. A randomised experiment to test the effects of financial disclosures on readers’ evaluations of scientific journal articles found that readers rated articles with financial tie disclosures as less interesting, important, relevant, valid, and believable than those without financial disclosures.⁸ Requiring the disclosure of financial ties in all academic publications and presentations is frequently used as a way to manage financial conflicts of interest of researchers.^14,16,121 An increasing number of scientific and medical journals are instituting requirements of disclosure of financial ties by authors of articles in the journals. Although a number of journals were initially opposed to financial disclosure policies,¹²² journal editors have begun to acknowledge that scientists might be influenced by financial interests and that disclosure is becoming the norm.^12,13

In addition to reporting financial ties, it is also important to know what, if any, involvement the sponsor had in the design, conduct, and presentation of the research. One rapid response to the Enstrom and Kabat article suggested that it might be useful for editors to “require a statement about the role of the funding source in the design, conduct, analysis and reporting of the data”.⁹ We believe that journals should require such a description as part of the Methods section in the paper, as required by Lancet (http://www.thelancet.com/authorinfo).

The case of the Enstrom and Kabat paper also raises the issue of whether academic researchers should accept funding from the tobacco industry. A number of universities have adopted policies refusing tobacco industry funding for research.¹²³ These institutions have decided to stop the tobacco industry’s long standing efforts to manipulate research while hiding behind the respect and credibility of the academic institutions.