August 07, 2006

Disclosure Doesn't Eliminate Conflict of Interest

Writes Jim Fossett:
Shirley Wang has an excellent piece in last Friday's WSJ [subscription] on the effectiveness of disclosure as a means of dealing with conflicts of interest among biomedical researchers who have financial ties to the drug or medical device companies whose products they evaluate. Many journals, including bioethical ones, now require disclosure of employment, consulting arrangements, royalties, stock ownership and the like for would-be authors as part of the publications review process. Awareness of such economic interests among researchers, together with the peer review process, presumably puts readers on their guard and minimizes the effects of potential bias in reporting research results. Many journals impose higher standards for writers of editorials and other review pieces, limiting the value of such financial stakes for authors of such pieces.

Are such disclosures effective in limiting bias in the reporting of research results? Much of the evidence Wang’s article cites suggests they are not, but for complex reasons that are hard to identify and repair. Most researchers are absolutely adamant that such financial ties, whether disclosed or not, have no effect on their scientific objectivity, just as most physicians insist that getting free lunches, pens, and other small gifts from drug company representatives have little effect on their prescribing practices. What evidence there is frequently suggests otherwise. A review of over 160 studies where such ties were disclosed published last October in the American Journal of Psychiatry (R Perlis, et al “Industry Sponsorship and Financial Conflict of Interest in the Reporting of Clinical Trials in Psychiatry” American Journal of Psychiatry 162:1957-1960, October 2005 ) found that researchers with personal financial ties to drug companies were roughly five times as likely to report favorable results than those without such ties. There’s less formal evidence on the effects of lunch on prescribing, but as reported last week in the NYT, an increasing number of large medical centers are banning free lunches over concern about conflicts of interest.

Wang cites some fascinating research by Max Bazerman of the Harvard Business School, who has studied similar conflicts of interest among financial auditors. Bazerman’s work, which can be found by links from his personal website here, finds relatively few “bad apples” or auditors who actively collude with company management to deceive investors. Rather, he argues that most unethical behavior is less conscious and stems from informal attachments between evaluators and evaluated. Most people are prone to interpret situations in ways that favor their own interests and give greater weight to evidence which favors those interests even when they are not conscious of doing so. Researchers who get significant income from companies have a stake in avoiding negative outcomes (less than positive drug trial results, significant side effects) that may lose them that income, but they may not be aware that they have such a stake and may continue to believe that their results are objective and unbiased. Disclosure of financial ties, Brazerman argues, may not reduce biased advice; it may even increase it, as a means of counteracting expected skepticism.

What to do? Bazerman’s arguments suggest that simple disclosure of financial ties are of limited value and don’t insure audit independence. In similar fashion, disclosing financial ties to drug companies may not have much effect on improving the independence and objectivity of research reports. Having to publish corrections of data that wasn’t disclosed may be embarrassing, but the disclosure of such data is still very much a hit or miss affair. Requiring complete disclosure might help some, as might a few large verdict civil cases. The pervasiveness of financial ties between researchers and industry,however, makes silver bullet solutions hard to come by.

- Jim Fossett

View blog reactions

| More