BMJ Study: COI Associated with Substantial Bias on Rosiglitazone Dangers

In HOOKED, I mentioned a classic study by Stelfox et al. (NEJM 338:101-6, 1998) showing that financial conflicts of interest were strongly associated with the position authors took on the potential risks of calcium channel blocking drugs. In those olden days, Stelfox et al. found that only 3% of the articles they surveyed explicitly disclosed authors' COI.

To see what progress we have made since then, Dr. Amy T. Wang and colleagues in internal medicine at the Mayo Clinic decided to do a review of comments made after the publication of two major studies on rosiglitazone, one claiming that there was increased risk of heart attack, the other minimizing such risk. Their results have been published on line in BMJ (subscription required). (Hat tip to Dr. Roy Poses and the Health Care Renewal blog for mentioning this paper.)

Using careful statistical methods the authors first retrieved a set of papers (including editorials and letters to the editor as well as research studies) that cited the two initial studies. They did extensive searches to identify whether the authors had any COI with the makers of rosiglitazone, the makers of its major competititor drug, pioglitazone, or of any other oral diabetes drug. They then blindly classified the contents of each paper as positive/negative as to the heart attack risks of rosiglitazone and/or as recommending/not recommending its continued use.

Bottom line: even in this supposedly more enlightened age they found explicit COI statements in only 53% of the 202 papers they surveyed. The found authors with COI more than 3 times as likely to be positive about rosiglitazone or to recommend its use.

The results also argue against a rationalization sometimes heard by defenders of taking industry largesse--that if you take bribes from enough different companies, they all cancel out in the end. COI was associated with a greater likelihood of minimizing heart risk and recommending rosiglitazone whether the company from which the author had received payment manufactured rosiglitazone, pioglitazone, or another diabetes drug. Wang et al. did note, however, that of the 29 articles that they found in which strongly recommended using pioglitazone rather than rosiglitazone, 25 were written by authors being paid by pioglitazone manufacturers.

In short, the evidence continues to mount that COI does bias physicians' judgment, even though an association cannot prove cause-and-effect. And it seems that once you start getting financial goodies from drug companies, it becomes less important exactly which companies paid you; you begin to see drugs generally in a favorable light and to minimize risks of adverse reactions.

Wang AT, McCoy CP, Murad MH, Montori VM. Association between industry affiliation and position on cardiovascular risk with rosiglitazone: cross sectional systematic review. BMJ 2010; 340:c1344, doi: 10.1136/bmj.c1344