Healthy Skepticism Library item: 17295

Warning: This library includes all items relevant to health product marketing that we are aware of regardless of quality. Often we do not agree with all or part of the contents.

Publication type: news

Melville NA
Avandia analysis: Authors with financial conflicts draw more positive conclusions
Medscape Medical News 2010 Feb 23
http://www.theheart.org/article/1049423.do

Full text:

An analysis of authors who published reports on the antidiabetic drug rosiglitazone (Avandia, GlaxoSmithKline) shows that those with ties to industry were more likely to conclude that the drug did not increase MI risk, as compared with authors with no industry ties, according to a study presented here at Preventive Medicine 2010, the American College of Preventive Medicine (ACPM) annual meeting [1].

The new analysis-presented as a poster by Dr Mohammed Hassan Murad (Mayo Clinic, Rochester, MN)-comes in the midst of a media maelstrom over a Senate report, released on the weekend, concluding that there are serious health risks associated with rosiglitazone that were not appropriately acknowledged by the FDA or the manufacturer.

In their poster, Murad and colleagues performed a review of 202 articles that addressed the possible association between the risk of MI and use of rosiglitazone in patients with diabetes. Articles selected for review had addressed the findings of two studies-a meta-analysis of small trials and a subsequent larger trial-that largely contributed to the rosiglitazone controversy. The authors’ aim was to characterize the prevalence and effects of conflicting relationships and to examine the extent to which the position taken in the paper could be explained by the authors’ conflicting relationships. Two reviewers who were blinded to the authors’ financial relationships evaluated each article and classified it as being favorable (supporting the argument that the drug does not increase the risk of MI), neutral, or unfavorable (questioning rosiglitazone’s safety).

Among the 202 papers that were evaluated, 107 (53%) included a conflict-of-interest statement and 90 (45%) had a conflicting financial relationship. Among authors who concluded rosiglitazone does not increase risk of MI, 91% had financial relationships with antihyperglycemic agent manufacturers and 86% had relationships with rosiglitazone’s manufacturer.

Among authors of articles representing unfavorable reviews, only 25% had financial relationships with antihyperglycemic agent manufacturers and 18% had relationships with rosiglitazone’s manufacturer that were disclosed in the papers (odds ratio 32 [95% CI 8.7-117.2] and 41.5 [95% CI 10.9-157.4], respectively).

Murad noted that in some cases, when they did an online search for some of the authors, they found some who listed financial conflicts in other publications that were not disclosed in their rosiglitazone paper. “Disclosure rates of conflicting financial relationships were misleadingly low despite their clear and strong linkage with authors’ expressed views,” the authors said. “These findings underscore the need for further progress in reform for the scientific record to be trusted. The quality of care patients receive is clearly affected by these findings.”

The study demonstrates the problematic pairing of researchers with financial interests in a company and studies involving that company’s product, said Dr David L Katz (Yale University Prevention Research Center, New Haven, CT), who was not involved in the meta-analysis.

“If the researcher has a direct financial interest in the item being studied, the researcher-who, unlike the funder, is responsible for interpreting the data-will share the funder’s bias,” Katz said in an interview. “The trend in the literature indicates that most people do not adequately resist this inclination.”

When the study is backed by high-quality methodology, however, the facts borne out from the methods should be solid enough to speak for themselves and override any author biases, said Katz. “The stronger the methodology, the more potential bias on the part of the researchers it can bear, because one measure of strong methodology is its capacity to overcome and minimize any influence of bias.”

Katz noted, however, that the driving motivation behind most research is to test some kind of theory or conclusion and that, in itself, could constitute a bias.

“There is always some bias. I would not waste my time doing a study, no matter who funded it, where I had no ‘preference’ regarding the outcome,” he said. “Another name for that preference is bias.”