Naproxen Hypothesis: How a Plausible Counter-Explanation Delayed Vioxx Withdrawal for Four Years
Merck's {{naproxen hypothesis}} — the claim that {{naproxen}} was unusually cardioprotective rather than {{Vioxx}} being harmful — kept a flawed explanation alive for four years until the placebo-controlled {{APPROVe trial}} ruled it out.
When the VIGOR trial (2000) showed 5 times more heart attacks in the Vioxx arm versus the naproxen arm, Merck faced two possible explanations: Vioxx caused heart attacks, or naproxen prevented them. The company publicly defended the second reading, calling it the naproxen hypothesis, and maintained it for roughly four years. The hypothesis was plausible enough on paper to delay scientific consensus. Naproxen is a non-selective NSAID that inhibits both COX-1 and COX-2, and COX-1 inhibition reduces thromboxane-mediated platelet aggregation — the same mechanism that makes low-dose aspirin cardioprotective. The argument that naproxen produced a similar effect could not be cleanly disproved without a controlled experiment that removed naproxen from the picture entirely. That experiment was the APPROVe trial, a placebo-controlled study of Vioxx for colon polyp prevention. With no naproxen arm, the naproxen hypothesis had nothing to hide behind. APPROVe showed roughly 2 times the heart attack rate in the Vioxx arm at 18 months, and the trial was halted by its independent safety monitoring board. Merck withdrew Vioxx within days. The broader meta-lesson is that a false hypothesis which cannot be ruled out without a specific controlled experiment can survive for years against accumulating observational evidence. Independent researchers like Eric Topol of the Cleveland Clinic published challenges in JAMA as early as 2001, and David Graham's Kaiser Permanente study of over one million real-world patients flagged a 3.7-times heart attack risk at high doses in 2004 — but the hypothesis held in mainstream regulatory discourse until APPROVe forced the issue.