It's true that 1992 was a watershed year for the FDA and pharmaceutical manufacturers. That was the year Congress passed the Prescription Drug User Fee Act, or PDUFA, under which the agency began collecting application fees from drugmakers seeking new drug approvals.
In turn, the FDA was to use the revenue to hire more scientists to review applications, cutting what had become an embarrassingly long average time between application filings and final decisions.
In a new, more or less comprehensive analysis published this week in Health Affairs, a group of academic researchers along with Public Citizen's Sidney Wolfe, MD -- a longtime FDA gadfly -- argue that the agency succeeded in that mission but sacrificed due consideration of drug safety along the way.
Whether their argument holds water is another matter, though.
Led by Cassie Frank, MD, of Cambridge Health Alliance and an instructor at Harvard Medical School, the group pulled data on all 748 new chemical entities approved from 1975 to 2009. They found that, after approval, 114 received boxed warnings (some had more than one) and 32 were withdrawn from the market because of safety issues.
These events were more common for drugs approved after 1992, the group found, with an odds ratio of 1.35 (P<0.05; 95% CI not reported). Average review times fell from 34 months during 1979 to 1986 to just 16 months for the period of 1997 to 2002.
So, if you accept that adding a post-approval boxed warning or yanking it from the market indicates that the FDA screwed up the initial review, then, yes, PDUFA might have modestly increased the likelihood of screw-ups.
But even so, there are caveats to consider:
- Half the withdrawals were for drugs approved before PDUFA rules were actually implemented, which was in 1993.
- Because the vast majority of events in the study were warnings, they effectively drove the stated results.
- The effectiveness of medications, in terms of quality-adjusted life-years saved or other metrics, was not accounted for.
- Changes in postmarketing surveillance and FDA policies over time, other than PDUFA, were not considered.
- When NSAIDs were removed from the analysis, the pre- versus post-PDUFA difference shrank markedly.
- At least one of the drugs cited as withdrawn for safety reasons was actually not, according to the FDA.
- Of the 748 NCEs approved during the study period, 80% have had no post-approval warnings or withdrawals.
The researchers made much of their statistical estimate of 34 events (new boxed warnings or withdrawals) per 100 drugs per 25 years of follow-up. But is it fair to judge the FDA on its 25-year predictive acuity? Is it remotely feasible to demand clinical trials powered for ultra-rare adverse events?
If, instead, we take 5 years of follow-up as more appropriate for judging whether the FDA screwed up a safety review, then its screw-up rate was only about 8%. And there was no difference pre- versus post-PDUFA. Only at 10 years do the rates show clear divergence.
This is not to say that the FDA has always done great work in the PDUFA era. As Wolfe told MedPage Today by telephone, the time line of new warnings and withdrawals may be more indicative of a problem than the summary statistics.
He noted that these events spiked dramatically for drugs reviewed in the mid-1990s, when the first implementation of PDUFA was in full swing and the agency was eager to show that review times had been cut. Eight of the 32 withdrawn drugs were approved in 1997.
Wolfe -- always a favorite with journalists because of his accessibility and outspokenness -- also contended that many of those drugs weren't clinically significant.
Several were weight-loss drugs and me-too fluoroquinolone antibiotics, which "were not remotely a breakthrough" that might have justified expedited reviews, he said. (However, the 1997 class also included troglitazone, the first thiazolidinedione drug for diabetes.)
And one of the authors' conclusions in the Health Affairs paper is hard to dispute: that new drugs should carry a disclaimer that their safety hasn't been thoroughly established and older drugs are likely to have fewer unknown risks. On the other hand, we didn't need this study to persuade us.
HypeWatch is a blog by MedPage Today staff that tracks media coverage of research findings. This installment was written by John Gever.