Open discourse about scientific disagreements is common among scientists. However, the CTR's references to the “political” aspects that Dawber might be less concerned about post-retirement, and Hockett's discussion of the “delicate diplomatic situation” (particularly when understood within the subsequently-exposed larger context of the CTR's work), suggest that these disputes were not merely scientific in nature, but were about controlling the analyses and interpretation of results potentially unfavorable to industry. This case illustrates why researchers who accept tobacco industry funding risk losing control of data, analysis, and publication. CTR clearly sought to fund research aimed at exonerating cigarette smoking as a cause of disease, providing funding for the primary purpose of gaining full control of the data for re-analyses aimed at supporting the “constitutional hypothesis.” Once the data were obtained, Dawber was no longer needed, particularly after he apparently resisted the post hoc subgroup analyses and interpretations that the industry sought (Fig 2).
As a scientist anxious to continue his work and constrained by lack of funding, Dawber's position was not unlike that of many scientists today. Epidemiological studies, with their long time frames of decades or more, are particularly vulnerable to changes in funding. A lack of stable funding can jeopardize earlier investments in research by hindering the publication of the full results of an epidemiological study. Industry funding, often encouraged by academic-business “partnership” arrangements increasingly embraced by many universities, can allow important scientific work to begin or continue that otherwise might be delayed or never done. However, this study illustrates how such arrangements risk pressures to compromise scientific independence for the benefit of a sponsor. Funding source can introduce biases related to data control and publication in epidemiological studies as well as experimental studies.[63
Many questions remain about the best models for corporate-academic research relationships. These models must protect not only against bias in the design, conduct and reporting of research, but also guard against the unethical use of researchers and study data to support industry objectives. [64
] Consortia of funding from industry have been suggested [65
] as one model that can reduce the influence of a single sponsor, in line with Dawber's comment that “with multiple sources of support it is not possible to allow one tail to wag the dog.” However, it has been difficult to get corporate sponsors to participate in such consortia. [64
Dawber's role in this case is complex. As many scientists have likely done, he appears initially to have dangled before CTR the “carrot” of a potentially industry-favorable finding in order to secure funding. In two papers published during the period during which CTR funding was received, he referred to “susceptible” and “predisposed” persons in discussing effects of smoking on heart disease [66
]--terms he did not use in his later work. These terms are consistent with the industry's favored “constitutional hypothesis.” However, as pressure increased to conduct analyses he felt were inappropriate, Dawber proved unwilling to compromise on analyses he must have felt could undermine his integrity as a scientist—and lost funding and control of the data as a result.
The tobacco industry has sought researchers' data through requests, lawsuits, and policy changes to require its availability. [68
] A unique aspect of epidemiological research is that a corporate sponsor does not need to control the design and conduct of the study in order to influence the findings. A sponsor can obtain the data after they are collected and re-analyze them in a way that is different than what the original investigators intended. The tobacco industry, for example, attempted to obtain and reanalyze data from Elizabeth Fontham's multi-center case control study of the effects of secondhand smoke.[69
This study explores industry activities related to one of the major epidemiological cohort studies of the 20th
century. Seeking access to data merely to reanalyze it to fit a self-serving interpretation is scientifically and ethically inappropriate. Seltzer's subsequent use of Framingham data for analyses completely independent of (and attacking) the original Framingham investigators violated the spirit of authentic scientific collaboration, which requires researchers to be open about disagreements and to make their case for reviewers and readers. In addition, many of Seltzer's attacks took the form of opinion pieces, letters, and editorials [58
] which were used for industry public relations, rather than peer-reviewed research papers. [74
This case study also has implications for contemporary publication practices. It points to the inadequacy of conventional journal policies related to disclosure of funding sources.[80
] Newer policies may require disclosing the “role of the sponsor,” typically asking for explicit disclosure of whether any funder was involved in study design; data collection, analysis, and interpretation; writing; and/or decisions about submissions. However, even this level of disclosure can be insufficient to ensure that editors and readers understand how decisions about data analysis and presentation may strategically mask more important findings or why such decisions were made. Authorship and contributorship statements can also help define the role of the sponsor by asking authors about who controls the data. [81
] Publishing competing interest statements in addition to funding sources, role of sponsor statements and authorship contributions could provide more information about sponsor control of data. For example, some journals now require disclosure of any nonfinancial ties that might influence the work.[81
] Lastly, disclosure may never be sufficient to protect against influence of a research sponsor, so sponsored studies should be viewed with skepticism.[82
More broadly, this study raises the question of how legitimate science—even large studies like Framingham—may be misused to serve corporate interests and thereby harm public health. Part of Framingham's attractiveness for tobacco companies was the study's near-legendary status among researchers, the media and the public. This lent legitimacy to publications referencing its data, which would likely be picked up by mainstream media. CTR's concern about being able to publish its re-analyses independent of NIH oversight suggests that it was trying to avoid addressing questions of bias in analyses and data presentation that NIH might have raised, while benefiting from the cachet of the Framingham name. Ultimately, CTR's efforts to control and use Framingham data via Seltzer's publications likely had the effect of confusing and delaying for several years full appreciation of Framingham's important findings that smoking tobacco causes CHD.