Study Published in JAMA Finds Conflicts Rarely Result in Recusal and Are Related to Voting Behaviors

WASHINGTON, D.C. – Conflicts of interest at drug advisory committee meetings for the U.S. Food and Drug Administration (FDA) are common and often of considerable monetary value, but rarely result in recusal, finds a study conducted by Public Citizen and published in The Journal of the American Medical Association (JAMA). The study details financial conflicts of interest between drug advisory committee members and the companies producing the drugs they evaluated; it also examines conflicts with competing companies. The study exposes a weak but statistically significant relationship between certain conflicts and votes in favor of the drug under consideration.

The study appears in the April 26 issue of JAMA, and was authored by Dr. Peter Lurie, Dr. Cristina Almeida, Nicholas Stine and Dr. Sidney Wolfe, all of Public Citizen’s Health Research Group, and Alexander Stine of the Department of Earth and Planetary Science at the University of California, Berkeley.

The Center for Drug Evaluation and Research (CDER) at the FDA approves 25 to 30 new chemical entities each year, often relying in its decision-making on the advice of outside advisory committee members and FDA-invited voting consultants. In September 2001, Public Citizen threatened a lawsuit against the FDA to force it to disclose detailed information about the financial interests of committee members, as required by law. In response, the FDA instituted new guidelines in January 2002, requiring more detailed financial conflict of interest disclosure publicly at the beginning of advisory committee meetings held to consider specific drugs.

The study examined those disclosed conflicts, using all posted agendas and transcripts from FDA drug advisory committee meetings from 2001 to 2004, approximately one year before and three years after the new guidelines.

During this period, 28 percent of advisory committee members and voting consultants had a conflict, and at least one member or consultant had a conflict in 73 percent of the meetings. Despite this, only one percent of members were recused from attending the meeting. The data were limited to self-reported conflicts disclosed at the beginnings of advisory meetings, which could underreport actual conflict rates.

For advisory committee members and voting consultants present, the study found that 19 percent of consulting arrangements involved more than $10,000, 30 percent of investments were worth more than $25,000 and 23 percent of contracts or grants exceeded $100,000.

“Conflicts of these magnitudes should result in automatic recusal from advisory committee meetings,” said Peter Lurie, MD, MPH, deputy director of Public Citizen’s Health Research Group. “With as many highly qualified professionals as we have in this country, there should be little difficulty identifying members with more limited or, ideally, no conflicts of interest.”

The study also found that speakers in the public session were frequently supported and/or flown to the meeting by the manufacturer of the drug under review. In fact, public speakers were more likely to disclose a conflict than advisory committee members or voting consultants.

The FDA guidelines appear to have improved conflict of interest disclosure but should be expanded to require naming the competitor company, explaining the reasons for a member’s recusal and to cover meetings that do not address specific products.

The study employed six analytic methods to determine whether a relationship between conflicts and voting behavior exists. These considered the potential impact of conflicts both on individual voting behaviors as well as on the overall outcome of meetings. A weak but, in several cases, statistically significant relationship between certain types of conflicts and voting behaviors was detected in a variety of different analyses.

For every voting member with a conflict of any type (index drug or competitor drug), there was a 10 percent greater likelihood that the meeting would favor the drug being considered. The exclusion of advisory committee members and voting consultants with conflicts would have produced margins less favorable to the drug being considered in a majority of meetings, but would not have altered the overall outcome of any of the meetings in this study. 

In the study, the authors compared conflict standards for advisory committees with those applied to juries, which make decisions with more limited societal impacts than advisory committees. “If a juror is found to have so much as read a newspaper, he or she can be dismissed in certain trials,” Lurie said. “We should have zero tolerance for accepting levels of conflict that would have even a small impact upon voting outcome, but in our study we found such impacts. What we need is not simply disclosure of conflicts, but also greater efforts to identify members without conflicts.”