Merck's Vioxx Recall Is Taught as a Model of Responsibility. The Paper Trail Says Otherwise.

On September 30, 2004, Merck pulled Vioxx off every shelf in the world in a single day. More than 80 million people had taken the painkiller, and it was throwing off over $2.5 billion a year.¹ The company killed it on its own, citing its own study, before any regulator forced its hand. In business-school case studies and corporate-ethics seminars, this is the part you remember: a company that found a problem and did the right thing. It is taught next to Tylenol as a textbook crisis response.

The story everyone tells is that Merck discovered the cardiac risk, and the moment it knew, it acted. Almost every word of that is doing quiet work to hide a date. Merck did not act the moment it knew. It acted years after the first trial safety signal — and the gap between knowing and acting is where the entire story actually lives.

The trial safety signal emerged by late 1999. The recall arrived in 2004.

Run the timeline forward instead of backward and a very different company appears. In September 2001 — three years before the celebrated withdrawal — the FDA sent Merck a formal warning letter saying its promotion discounted a finding from the VIGOR study: patients on Vioxx had a four- to five-fold increase in heart attacks compared with naproxen.⁹¹⁰ Merck did not pull the drug. It added a cardiovascular warning to the label in April 2002, nearly two years after the VIGOR data had first been made public.¹⁰ Then it kept selling. The withdrawal didn't come because Merck finally saw the danger. It came when a fresh trial made the danger impossible to keep selling around.

This is the thesis, and it is uncomfortable because the recall really did happen and really did cost Merck a fortune: the Vioxx withdrawal was not a model crisis response — it was the moment a five-year strategy of selling through a known risk finally became untenable. The clean ending everyone cites is real. What the case studies amputate is the long middle, where the choice was made again and again to keep the revenue and manage the science.

How you sell through a risk you already know about

Selling through a known risk isn't passive. It takes machinery, and Merck built it. The company spent more than $100 million a year on direct-to-consumer advertising for Vioxx — the ask-your-doctor blitz that turned a painkiller into a household name.⁵ To arm the sales force, it produced a 'Cardiovascular Card,' a marketing device the FDA never approved, which leaned on a flattering pooled analysis of pre-approval studies and left out the VIGOR cardiovascular data entirely. Reps were instructed to steer away from discussing heart risk with physicians.⁵ The demand engine ran at full throttle while the safety conversation was being engineered to a whisper. That is the mechanism: not a company caught off guard, but a company whose marketing outran its own science on purpose.

The same instinct reached into the scientific record itself. In December 2005, the New England Journal of Medicine — the most prestigious venue in medicine, and the journal that had published VIGOR — took the extraordinary step of issuing an 'Expression of Concern,' stating that 'inaccuracies and deletions' in the manuscript called the integrity of the data into question.³ An internal Merck memo dated July 5, 2000, surfaced later, showed that at least two of the study's authors knew of three additional heart-attack cases at least two weeks before submitting their manuscript revision — and those cases did not appear.³ The journal's editors did not soften it on review. They reaffirmed that the published article 'did not contain relevant safety data available to the authors more than four months before publication' and therefore 'did not accurately reflect the potential for serious cardiovascular toxicity with rofecoxib.'⁴

Wasn't writing a $4.85 billion check the responsible ending?

The fair objection is that Merck did pay, and paid enormously. In November 2007 it agreed to put a fixed $4.85 billion into a fund for U.S. heart-attack and stroke claims — a settlement that wasn't even a class action, with each claim evaluated individually.⁶ That is not the behavior of a company running from accountability. True. But read the same filing closely and the shape of the 'responsibility' sharpens: Merck did not admit fault.⁶ And the legal chapter did not close in 2007. In 2011 the company paid $950 million to the Department of Justice over improper Vioxx marketing; in 2016 it settled an investor class-action for $830 million⁷ — bringing the calculated total of those three documented settlements to over $6.6 billion. A check that large, paid without admission, across nearly a decade of further litigation, is not the punctuation mark on a clean recall. It is the cost of having sold through the risk for as long as the market allowed.

And the human stakes are why the framing matters. The harm is hard to count precisely — the figures are models, not confirmed tallies. FDA safety officer Dr. David Graham testified to the Senate in 2004 that delayed action may have caused as many as 55,000 premature deaths, and senior FDA officials immediately cautioned that his model 'must be interpreted with caution.'⁸ The exact number is contested. The pattern that produced it is not.

Merck is remembered for the day it did the hard, expensive thing — and it did. But a crisis response is not the dramatic exit; it's everything you chose in the long quiet before it. By the time the withdrawal came, more than 80 million people had filled a prescription, the FDA had warned in writing, the leading journal in medicine had questioned the data, and the demand machine had spent years outrunning the safety science. The recall didn't prove Merck would do the right thing. It proved how long a company will keep selling before the right thing becomes the only thing left.