Group Says FDA, Advisory Panels Show Bias Toward Drug Approvals

By Shankar Vedantam
Washington Post Staff Writer
Tuesday, August 29, 2006

The panels of experts assembled by the Food and Drug Administration to advise it on whether to approve new drugs and medical devices are often biased in favor of recommending approval, according to a consumer group's analysis released yesterday.

The advisory panels often vote unanimously to recommend approval even after members express doubts about the product or say the evidence submitted is insufficient, said the report by the National Research Center for Women & Families, a policy research and advocacy group. The FDA has recently come under fire over drug safety issues, including its decisions to approve the popular arthritis pain medications Vioxx and Celebrex that were later associated with an increased risk of heart attacks and strokes.

The group's report also said the FDA is more likely to accept its panels' advice when they recommend approving a product than when they vote against approval.

"When the votes are for approval, they go along with them 96 percent of the time," said Diana Zuckerman, president of the center. "When the votes are against approval, they are much more likely to disagree and approve the product anyway. . . . They are 10 times more likely to overturn a recommendation against approval than for approval."

Agency officials said the critical report is wrong both in its assumptions and conclusions.

"First and foremost, it is an advisory committee process, and in that sense panels do not make decisions, they offer advice," said Randall Lutter, associate commissioner for policy and planning at the FDA. "To say they serve as a rubber stamp is not a way of contributing to informed debate."

Lutter said that based on promising clinical trial data, the FDA sometimes knows it is going to approve a product when it convenes an advisory committee -- so it is hardly surprising that committees would find in favor of such products. Much of the value of such meetings, he added, comes in offering nuance and guidance to the agency on labeling and use of the product.

The report tracked the votes of 11 FDA advisory committees, chosen at random, from 1998 through 2005. That represents about a third of the panels convened and 89 products.

Although some committees came up with a mix of approvals and disapprovals, others, including the panels evaluating arthritis drugs and vision-related devices, almost always voted for approval and almost always did so unanimously, said Zuckerman.

Zuckerman suggested that conflicts of interest may have biased some scientists.

But Lutter rejected this criticism. The FDA's own analysis of conflicts of interest among advisory committee members, he said, found that such conflicts did not predict how panel members would vote.

© 2006 The Washington Post Company