Conflict of interest disclosures common at FDA advisory meetings, rarely result in recusals

A cross-sectional study of the agendas and transcripts of FDA drug advisory committee meetings from 2001 to 2004 found that conflict of interest disclosures occurred at 73% of the meetings examined, but that the recusal of advisory committee members from decision-making as a result of those conflicts occurred only 1% of the time.

Researchers divided meetings into "product meetings," which discussed specific products, and "nonproduct meetings," which discussed more broad, scientific issues. They then examined oral statements of conflicts from the executive secretary at the start of all meetings, which included whether members had been restricted or recused. Each conflict was identified as a separate entry, and a single attendee could have multiple conflicts of interest.

Researchers divided 983 declared conflicts of interest with identifiable details among advisory committee members (256), voting consultants (208), and public speakers (519) by using meeting transcripts. Investments and consulting were the sources of the majority of the conflicts for both advisory committee members (67%) and for voting consultants (70%), while contracts, grants, and CRADAs were the source of 37% of public speaker conflicts. In conflicts in which monetary values were identified, 19% of consulting contracts among advisory committee members and voting advisors were >$10,000, 23% of contracts, grants, and CRADAs were >$100,000, 30% of investments were >$25,000, and 44% of lectures and honoraria were >$10,000.

A total of 1,860 unique individuals at the meetings accounted for 3,718 person-meetings. Of those meetings, only 22 person-meetings involving advisory committee members resulted in recusals (1%), with 3 advisory committee members being recused from 2 meetings and 1 member being recused from 3 meetings. Restrictions for advisory committee members and voting consultants occurred at a combined rate of 1%, with all but 1 occurring in product meetings.

In 2 initial analyses, no relationship was found between conflict rates (encompassing conflicts between individuals and the index product's manufacturer, individuals and a competing pharmaceutical company, and either the sponsor or a competitor) and voting outcome. In a third analysis, researchers found that excluding committee members or voting advisors with conflicts would have produced less favorable results for the index product in a majority of meetings (P value not provided). However, a fourth analysis found that the less favorable results would not have affected the overall vote result toward the index drug.

Additional analyses of individual voting behavior found statistically significant relationships between the likelihood of voting for an index drug and conflicts. In a Mantel-Haenszel analysis, competitor conflicts (RR=1.2; 95% CI, 1.12–1.28) and any conflict (RR=1.1; 95% CI, 1.03–1.17) were significant.

"An RR of 1.1 (that calculated for any conflict) means that for each 1 person with any conflict, there was a 10% greater likelihood that the meeting would favor the index drug," the authors stated. "Such a level of bias would never be tolerated in a jury."

Study limitations included the size of the samples used in the voting analyses, and the possibility that the actual number of conflicts of interest could be underreported because individuals may not disclose all of their potential conflicts.

SOURCE Lurie P, Almeida CM, Stine N, Stine AR, and Wolfe SM. Financial conflict of interest disclosure and voting patterns at Food and Drug Administration drug advisory committee meetings. JAMA. 2006;295:1921–1928.