Tobacco money, the BMJ, and guilt by associationBMJ 2003; 327 doi: https://doi.org/10.1136/bmjusa.03070001 (Published 19 November 2003) Cite this as: BMJ 2003;327:E236
- Steven H Woolf, MD, MPH
From BMJ USA 2003;July:344
Bans against smoking in public are ubiquitous. Cigarette smoke is irritating, and passive inhalation increases tobacco-related morbidity and mortality. Or perhaps not, say Enstrom and Kabat (BMJ USA p 369). Based on an analysis of data from the American Cancer Society, they conclude that the association “may be considerably weaker than generally believed.” A firestorm erupted when the BMJ published this article in May. The American Cancer Society released a stinging press statement. The BMJ received 130 Rapid Responses and was advised to issue a retraction. Two main concerns predominated. First, the authors acknowledged being funded by an organization tied to tobacco interests. At least one author had been a tobacco industry consultant.
Second, the study had an obvious design flaw. The control group meant to represent unexposed persons consisted of spouses of nonsmokers, who in that era encountered cigarette smoke almost everywhere they went. The American Cancer Society had warned Enstrom about this problem beforehand (see BMJ USA p 352). In 1996 JAMA rejected what many assume is an earlier version of this study. Some believe the BMJ would have done the same if its peer reviewer had been an expert on the topic.
The decision to reprint this study in BMJ USA was difficult. The editors oppose Big Tobacco and poor research but are also committed to open scientific discourse. The study suffers from an imperfect control group and overstated conclusions, but hundreds of such studies are published regularly. It would be publication bias to reject a study of similar quality because it reached the “wrong” conclusion.
Some might discard any study tainted by tobacco money, but in 1996 the BMJ rejected such a policy, saying that it would intrude on the “open marketplace of ideas.” “[A]ll studies undertaken must be available in some form: if some studies are systematically suppressed then we will reach false and biased conclusions when reviewing a body of research” (BMJ 1996;312:133–134).
The decisive factor should be study quality, not its financing or even its authors. As one reader put it, “The cogency of an argument should not depend on who makes it” (BMJ USA p 374). Enstrom was faulted for having tried, as early as 1975, to obtain tobacco money for his research. This is suspect, but top researchers do the same with drug companies. A perfectly good study of an unpopular hypothesis may find funding only through industry. The funding does not impugn the methods.
Or does it? The integrity of research depends on trust—that the methods described were followed and the data are complete and undistorted. It would be foolish to trust tobacco companies, which have knowingly promoted a product that kills its users. But can we trust researchers who accept tobacco money, or the journals that publish their work? How far does guilt by association extend?
Articles cited in Editor's Choice are listed below, beginning with their BMJ USA page number
BMJ USA p 369 Environmental tobacco smoke and tobacco related mortality in a prospective study of Californians, 1960–1998 (Enstrom and Kabat) http://bmj.com/cgi/content/full/326/7398/1057
BMJ USA p 352 More misleading science from the tobacco industry (Thun) http://bmj.com/cgi/doi/10.1136/bmjusa.03070002
BMJ USA Rapid responses to article by Enstrom and Kabat, http://bmj.com/cgi/eletters/326/7398/1057