Study: FDA Should Simplify Drug Approval Programs

A new study published Tuesday in JAMA suggests that the evolution of drugs and biologics’ regulatory approval programs at the US Food and Drug Administration has resulted in a system that approves products based on weaker data, without reducing the overall development times.

Researchers at Brigham and Women's Hospital, a teaching affiliate of Harvard Medical School, analyzed FDA's regulation of pharmaceuticals between 1983 and 2018. FDA has received additional authority from Congress contingent on the agency further facilitating both product development and the marketing of new medical products since the early 1980s.

The agency has several new special approval programs, including Fast-Track, Accelerated Approval, as well as those for priority review vouchers and Breakthrough Therapy designations.

The number of expedited product development and regulatory approval programs “expanded greatly since 1983,” resulting in more approvals with limited evidence and greater uncertainty around clinical benefit. The introduction of new programs added billions to healthcare costs, too. 

The use of the Accelerated Approval, Fast-Track, and priority review pathways for new drugs increased, with 81 percent (48/59) of new drugs having used at least one of these as of 2018. 

New drug approvals, including biologics, went from 34 products between 1990 and 1999, 25 between 2000 and 2009, and 41 between 2010 and 2018, the researchers found. Yet the time from the effective date authorizing human studies to the effective date of approval lengthened from an average from 7.8 years from 1986 through 1996 to 9.1 years from 2008 through 2017.

FDA has approved about 80 percent of new drug applications and biologics license applications since 1988, the researchers found. The use of surrogate measures, meanwhile, that were “not yet designated as well established” but were still used for justification of approval under the agency’s Accelerated Approval pathway rose from 9 percent to 13 percent from 2011 to 2018. 

FDA review times went from more than three years in 1983 to under one year in 2017. Median review time was 10.1 months for standard applications and 7.6 months for priority applications by 2018.

The findings raised concerns with approvals granted without definitive evidence and incentives regulatory innovations have created where the clinical benefit is not the focus. The researchers, however, recommend a continued evolution of FDA's regulations, as opposed to a revolution.

“While retaining policies that encourage efficient review, Congress and other government officials should also consider the implications of less rigorous clinical outcome requirements and whether the current complex array of regulatory programs should be simplified,” they argue.

Congress and FDA have been grappling with striking the right balance between pre- and post-approval study data collection for years. Nothing prevents FDA from requiring a PAS. 

“The overall picture is not of a struggling FDA, but rather of a regulatory process that has evolved over time into a thicket of special programs, flexible review criteria, and generous incentives,” former FDA principal deputy commissioner Joshua Sharfstein, writes in an accompanying editorial. “As a result, it is challenging to understand the totality of these reforms on drug approval in the United States. He recommends four starting points to “enhance the benefits and value and reduce the inefficiency and risks of the US system of drug approval.”

Sharfstein's staring points relate to rationalizing the existing programs for expedited review, strengthening oversight where a medication offers significant clinical benefits and poses serious risks and recalibrating programs that provide special marketing protections. The fourth calls on Congress to "use patent and pricing incentives to accelerate the generation of definitive evidence under accelerated approval," citing research on limiting pricing or market exclusivity for study sponsors until the studies that assess clinically relevant endpoints are completed.