Merck Paid $5.8 Billion for Vioxx and Admitted Almost Nothing. That Was the Plan.

On September 30, 2004, Merck pulled a drug that had sold $2.5 billion the year before, voluntarily, worldwide, in a single morning.¹ The trigger was a trial Merck itself was running — APPROVe — which was supposed to show Vioxx could prevent colon polyps and instead showed that after eighteen months on the drug, patients had a higher relative risk of cardiovascular events.¹ A company does not casually walk away from a blockbuster. The decision looks, on its face, like conscience. It was something more interesting: the opening move of a multi-year project to pay for a catastrophe without ever conceding it caused one.

The official story is that Vioxx was a tragic surprise — a hidden risk that emerged from a 2004 trial and that Merck acted on the moment it knew. The closer story is that the signal had been visible for years, that Merck had a public explanation ready for it, and that the recall was less a confession than the start of a containment exercise that worked almost exactly as designed.

The warning that arrived four years early

The cardiovascular signal did not appear in 2004. It appeared in VIGOR, a study published in late 2000 that was meant to show Vioxx was gentler on the stomach than older painkillers. It did show that. It also showed that patients on rofecoxib had four times as many heart attacks as patients on naproxen.⁶ Merck's response was not to investigate the heart-attack number; it was to explain it away. The company's standard line was that naproxen must be cardioprotective — that the comparison drug was unusually good for the heart, rather than Vioxx being bad for it.⁶ It is a tidy hypothesis. It is also the explanation that happens to keep your drug on the shelf. The signal was real, acknowledged, and reframed — all at once.

The reframing reached the literature itself. When VIGOR was published, three myocardial infarctions — all in the rofecoxib group — were not in the data the journal received.⁷ Five years later the New England Journal of Medicine issued a formal Expression of Concern, saying the omission raised questions about the integrity of the cardiovascular data, and noting that its editors had first learned of the additional heart attacks from FDA-published figures back in 2001.⁷ Three data points, left out of the most-read cardiovascular table in the study, in the direction that mattered. That is the texture of this failure: not one dramatic act of fraud, but a long series of small, defensible-sounding decisions that all leaned the same way.

The settlement was the strategy

Here is the thesis. The withdrawal was not Merck's masterstroke — the settlement architecture was. In November 2007, Merck agreed to put a fixed $4.85 billion into two funds: $4.0 billion for heart-attack claims, $850 million for ischemic-stroke claims.³ Read the structure, not the number. It was deliberately not a class action. Claimants had to be evaluated individually against medical criteria. And Merck admitted no fault.³ Each of those three features does specific work. A fixed fund caps the company's total exposure — Merck recorded it as a single $4.85 billion pretax charge and moved on.⁴ Individual claim evaluation filters out the weak cases that a class action would have swept in. And no admission of liability means no precedent, no concession that can be cited against you in the next lawsuit, the next regulator's letter, the next drug.

The pattern repeats in the criminal resolution. In 2011 Merck agreed to pay $950 million to the Department of Justice — but look at what it actually pled guilty to.⁵ Not concealing heart-attack data. A single misdemeanor count of misbranding: promoting Vioxx for rheumatoid arthritis before the FDA had approved that use.⁵ The cardiovascular-safety statements — the part that actually killed people — were handled in the civil half of the deal, the $628 million piece, where you write a check without a guilty plea attached.⁵ The criminal admission was kept as small and as off-target as the law would allow.

Isn't paying $5.8 billion the opposite of getting away with it?

The honest objection is the obvious one: Merck paid out enormous sums — well over $5.8 billion across product-liability, criminal, and shareholder channels — so calling the response a clever escape feels perverse. That money is real, and it hurt. Fair. But the argument here is not that Merck dodged the bill. It's that Merck controlled the shape of the bill, and in litigation the shape is everything. A capped fund is a known number a board can absorb; open-ended jury verdicts are not. A no-fault settlement is a closed door; an admission is a key handed to every future plaintiff. The company chose to pay a large, finite, defined cost in order to never owe an undefined one. That is not getting away with it. It is buying certainty — and for a serial drug-maker that will face the next safety fight, certainty and a clean legal record are worth more than the cash they cost.

There is a second objection worth steelmanning: maybe the naproxen-cardioprotection theory was a sincere scientific belief, not a cover story, and maybe the omitted heart attacks really did fall after a pre-specified data cutoff. Both are plausible defenses Merck genuinely made. But that is precisely the structural problem. When the same company designs the trial, interprets the signal, and writes the public explanation, every sincere judgment call has a way of landing on the side of the product. You don't need malice for self-regulation to fail. You only need a series of close calls and a strong incentive about which way they should break.

Strip away the recall theater and the size of the checks, and what Vioxx really exposes is the quiet architecture of pharmaceutical self-policing: a system where the entity that profits from a drug is also the first to judge its danger, and where the most damning evidence can be acknowledged internally and reframed externally without anyone crossing a line they'd recognize as fraud. Merck did not win the Vioxx fight by being innocent. It won it the way a good litigant always does — by deciding, well in advance, exactly what it would ever be made to say out loud. The drug came off the market in a day. The admission Merck refused to make is still off the market today.