Postmarketing Surveillance Is Good And Normal

A sharp corrective to a Scientific American framing that 'a third of FDA-approved drugs were later recalled or warned about' means the FDA is too lax. Scott shows only 3/222 (1.3%) were actually withdrawn — an impressively LOW false-positive rate that should instead make you fear the false negatives. Rare harms can't be caught pre-approval: nefazodone causes one liver failure per 300,000 patient-years (you'd need a 600-year or 600x-sized trial), and you can't pre-specify every subgroup (the Viberzi/no-gallbladder, Tegretol/Han-Chinese, linezolid/sauerkraut cases) — post-marketing surveillance catching these IS the system working, not failing. Many boxed warnings are trivial or obvious (warfarin 'can cause bleeding'; benzo+opioid — 'well, duh'). The real question (is the FDA too strict or too lax?) can't be answered from the event rate alone — only by weighing people helped by faster approval against people harmed by slips, and the only actual cost-benefit work (Isakov, Lo & Mostazerhodjat) finds the FDA too conservative. Vivid, statistically literate, and a clean reframe of base-rate reasoning about regulation.

Quality 70 · Strong. Strong (70). A clear, memorable, statistically literate reframe — 'post-marketing events are the system working' and 'a 1.3% withdrawal rate means we're probably too strict' are durable, oft-cited points, with vivid examples. Held to mid-Strong by its narrow, responsive scope (rebutting one study) and brevity.

Claude’s paradigm shift 50 · Moderate. Moderate (50). The FDA-too-conservative and false-positive/false-negative framings are Scott's recurring theme drawing on health econ (Peltzman/Isakov-Lo); the fresh, clarifying angle is 'safety communications = the system working,' a reframe of a then-current debate.

Real-world impact 3 · Moderate. Moderate (3). A clear contribution to the consequential FDA-reform / drug-approval-economics discourse, but a single blog argument, not itself moving policy.