Cardiovascular safety was tested in the original trial. It passed. Nothing in the data during development suggested it was an issue. But trials can’t detect everything.
It wasn’t until it got to market did a safety signal pop up. Then retrospective analyses of large data sets proved it.
No one is under the illusion it’s perfect or ungameable. A drug slipping by every few years is bad and often tragic, but IMO nowhere close to indicative of a systematic problem. It is a system that is worthy of a high degree of trust.
I'm unfamiliar with Vioxx and whether its approval really was a result of mistakes.
Shouldn't we expect some small percentage of failures in these processes given that they are driven by statistics and confidence intervals? Is that even a failure of the process, or is it a known limitation given how much resources and time we are willing to allocate to the discovery process?
Yes we should expect some small number of failures and so far I agree, I don’t see evidence of a problem that needs fixing.
As patio11 says, the correct amount of fraud in a financial system is not zero, and the correct amount of false positives in drug approvals is not zero.
> Shouldn't we expect some small percentage of failures
Yes, and this is really solved at a more local level. Doctors aren't prescribing new drugs like candy. They, too, are skeptical of their success and will reserve those prescriptions for the most desperate cases. Over years, we (and the doctors) learn how effective these drugs are and what potential side effects they have.
