Can People with Conflicts of Interest Make Clear Arguments Defending Their Conflicts?

Calls to limit conflicts of interest affecting advisory panels for the US Food and Drug Administration (FDA) have inspired some curious apologias for the current practices that result in many conflicted board members.

Yesterday, Dr Thomas P Stossel, identified as a Professor at Harvard Medical School and director of the Translational Medicine Division at Brigham& Women's Hospital wrote about "Drugs and Demagogues" in the Boston Globe. His arguments, in my humble opinion, were weak and beset with logical fallacies.

The Arguments and My Responses

Invalid accusations of conflicts of interest may deprive the government of good advice:

Thucydides noted how some people invoke financial conflicts of interest to discredit worthy opponents: 'If a man gives the best possible advice but under the slightest suspicion of being influenced by his own private profit, we are so embittered by the idea that we do not allow the state to receive the certain benefit of his good advice.'

Although Stossel's use of a historical anecdote is appealing, this argument appears to be based on a false dichotomy: some accusations of conflicts of interest may be invalid, so all such accusations must be invalid. Clearly, the importance of conflicts of interest depend on their nature, and the context.

The scientific evidence shows that conflicts of interest do not adversely affect FDA advisory panel decisions:

The answer is not that financial conflicts of interest provably influence FDA panel recommendations. A study of advisory panel decisions published in The Journal of the American Medical Association found that advisers' financial relationships had no effect on their approval recommendations.

There are multiple problems with this argument. First, Stossel seemed to confuse lack of evidence with evidence of a lack. Stossel asserted that the data from one study did not provide evidence for the influence of conflicts of interest on FDA advisory panel recommendations. Lack of evidence for such influence in one study, however, does not constitute evidence that conflicts of interest have no influence on the panels' recommendations.

Second, Stossel ignored the major weakness of the JAMA study, one that would have made it unlikely that the study would have found a relationship between conflicts of interest and FDA panel recommendations. (Stossel did not identify the study, which was apparently: Lurie P, Almeida CM, Stine N et al. Financial conflict of interest disclosure and voting patterns at Food and Drug Administration advisory committee meetings. JAMA 2006; 295: 1921-1928.) The study was a cross-sectional study that assessed the relationship between the prevalence of members with conflicts on advisory panels and the panels' votes. Since the panels were the unit of analysis, and the study only considered panels that voted on specific drug products, the sample size was very small (76). The study used three different types of analyses, all rather complex. Two types of analyses failed to show relationships between conflicts of interest and voting outcomes. The article noted "both analyses were hampered by small sample sizes." This study was too small to have much statistical power to find a relationship between conflicts of interest and panel recommendations.

Finally, Stossel ignored evidence in the study that went against the thesis of his commentary. Even though the JAMA study was small, one of the three types of analyses did find an effect of conflicts of interest, that "exclusion of advisory committee members or voting consultants with conflicts would have produced absolute vote margins less favorable to the index drug in the majority of meetings...." Stossel completely ignored this finding.

Excluding people with conflicts from FDA advisory panels would exclude the best and the brightest,

Furthermore, medical experts who consult for private companies have the most research agency grant funds. They are the most knowledgeable, inventive, and productive scholars -- which is why corporations want to work with them.

But is this a post hoc fallacy? Do pharmaceutical, biotechnology, and device companies select the best and most accomplished doctors and researchers to consult for them? Or do people who are supported by these companies have more academic success, and hence come to be known as the best and the brightest?

At best, that such companies hire the best and brightest as consultants is a testable hypothesis. Yet even if this is a hypothesis that Dr Stossel favors, it is not a proven fact.

The alternative hypothesis is also tenable. Maybe pharmaceutical, biotechnology, and device companies hire as consultants and fund as researchers the people the companies feel are most likely to help market their products, one way or the other. Since, as we recently posted, the strongest current determinant of success in medical school now appears to be acquisition of external funding, it could be that commercial funding causes faculty members to be successful, rather than successful faculty members are the ones who get commercial funding.

People who do not have conflicts may make terrible recommendations,

Meanwhile, some groups selected for their independence can make terrible recommendations. After having to retract its assertion that certain baby car seats were unsafe, Consumer Reports magazine recently changed its policy of consulting only with experts without commercial ties. The second-best experts do not produce the best results.

This is simply reasoning by anecdote. Furthermore, it seems to again present a false dichotomy: either an expert is perfect, or terrible. Thus, because Consumer Reports made a single error, it must be terrible.

Can Conflicted People Make Clear Arguments in Defense of Their Conflicts?

So Stossel's defense of conflicted FDA panel members seems to be based on a mixture of logical fallacies and poor understanding of existing evidence.

To return to the main issue, conflicts of interest. In the British Journal of Medicine, Joe Collier made the argument that having conflicts of interest leads to conflicted, and hence unclear thinking: "people who have conflicts of interest often find giving clear advice (or opinions) particularly difficult." [See: Collier J. The price of independence. Br Med J 2006; 332: 1447-9. and our relevant post here.] That is an argument against having people with conflicts of interest on FDA panels.

It may also help explain the lack of clarity in Dr Stossel's arguments.

Toward the end of the article, Stossel noted,

I have not participated in FDA panels, but I do have some corporate relationships that could exclude me under the proposed rules.

A little internet-based research revealed Stossel's conflicts of interest.

In a commentary in the New England Journal in which Stossel attacked the rigorous conflicts of interest rules that were once again imposed on the US National Institutes of Health (NIH), Stossel disclosed "having received consulting fees from ZymeQuest, owning stock options in ZymeQuest and Biogen, and having pending and issued patents, owned by Brigham and Women's Hospital, some of which are licensed to ZymeQuest." [Stossel TP. Regulating academic-industrial research relationships - solving problems or stifling progress? N Engl J Med 2005; 353: 1060-1065.]

But in another paean to a laissez-faire approach to conflicts of interest in Perspectives in Biology and Medicine, Stossel also disclosed that he "is a member of the Board of Directors of Zymequest Inc, and the Scientific Leadership Advisory Board of Merck & Co. The author is a founding scientist of Critical Biologics Corporation and a consultant to Boston Scientific, Inc., and Gerson-Lehrman, Inc." [Stossel TP. Regulation of financial conflicts of interest in medical practice and medical research. Perspect Biol Med 2007; 50: 54-71. ] Note that members of boards of directors of for-profit corporations are legally expected to exercise the duty of loyalty, and exhibit "unyielding loyalty" to the corporation's stockholders. [See Monks RAG, Minow N. Corporate Governance, 3rd ed. Malden, MA: Blackwell, 2003.] Such unyielding loyalty may require directors to act in ways that challenge the mission of academic medicine. Furthermore, a 2004 ZymeQuest press release revealed that Stossel has been on its board at least since that year, suggesting that Stossel's disclosures in his 2005 NEJM article were not complete.

So two of the most visible published arguments against more stringent conflict of interest rules for FDA advisory panels were made by people who have relevant conflicts of interest, and seem not always to have disclosed them completely. (See our post on Epstein's article in the Wall Street Journal here and on his conflicts of interest here.) Furthermore, both these commentaries were badly argued, exhibiting misconceptions about relevant data and depending on logical fallacies. These are only anecdotes, but it does seem that the confusion caused by conflicts of interest make it difficult for those with conflicts to make clear arguments that such conflicts are harmless.

I am waiting to see if any individuals without any relevant conflicts will stand up to argue the harmlessness of major conflicts of interest of FDA advisory panel members.