Voting behavior during FDA Medical Device Advisory Committee panel meetings.

Objectives During premarket review, the US Food and Drug Administration may ask its Medical Device Advisory Committee (MDAC) Panels to assess the safety and effectiveness of medical devices being considered for approval. The objective of this study is to assess the relationship, if any, between individual votes and Panel recommendations and: (1) the composition of Panels, specifically the expertise and demographic features of individual members; or (2) Panel members’ propensity to speak during Panel deliberations. Methods This was a retrospective cohort study of routinely collected data from voting members of MDAC panels convened between January 2011 to June 2016 to consider premarket approval. Data sources were verbatim transcripts available publicly from the FDA. Number of words spoken, directionality of votes on device approval, profession, and demographics were collected. Results 658,954 words spoken by 536 members during 49 meetings of 11 Panels were analyzed. Based on multivariate analysis, biostatisticians spoke more (+373 words; P = 0.0002), and women (-187 words; P = 0.0184) and other non-physician voting members less (-213 words; P = 0.0306), than physicians. Speaking more was associated with abstaining ( P = 0.0179), and with voting against the majority ( P = 0.0153). Non-physician, non-biostatistician members ( P = 0.0109), and those having attended more meetings as a voting member ( P = 0.0249) were more likely to vote against approval. In bivariable analysis, unanimous Panels had a greater proportion of biostatisticians (mean 0.1580; 95% CI 0.1237–0.1923) than non-unanimous Panels (0.1107; 95% CI 0.0912–0.1301; p = 0.0201). Conclusions Panelists likely to vote against the majority include non-physician, non-biostatisticians; experienced Panelists; and more talkative members. The increased presence of biostatisticians on Panels leads to greater voting consensus. Having a diversity of opinions on Panels, including in sufficient numbers those members likely to dissent from majority views, may help ensure that a diversity of opinions are aired before decision-making.

Abstract licence: CC BY

10 more references, whose wording the paper’s licence keeps private.

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.