← back to the index
RX-001 COX-2 inhibitor 2004

Vioxx (rofecoxib) — Merck’s Heart-Attack Painkiller Pulled in 2004, $4.85 Billion Paid

cardiovasculariatrogenic-mortalitythromboticcardiologypain-managementrheumatology2000s
Patients exposed
~80M prescriptions / ~20M U.S. users
Documented harm
~88,000–139,000 excess heart attacks/strokes; ~30–40% fatal
On market
1999→2004 (5.4 years)
Status
Withdrawn

Summary

When Merck & Co. pulled Vioxx from pharmacy shelves worldwide on 30 September 2004, chief executive Raymond Gilmartin framed it as a responsible company acting on new data; the documented record shows a five-year gap between that posture and what the company's own clinicians had measured. Vioxx (rofecoxib), a COX-2 selective anti-inflammatory approved by the U.S. Food and Drug Administration on 20 May 1999 and marketed as a gentler painkiller that spared the stomach, was instead associated with one of the largest drug-safety catastrophes in regulatory history. FDA drug-safety reviewer David Graham would testify to the Senate Finance Committee on 18 November 2004 that the drug had caused on the order of 88,000 to 139,000 excess heart attacks and strokes in the United States, of which perhaps 30 to 40 percent were fatal — a toll he compared to between two and four jumbo-jet crashes a week sustained over five years.

The gap between promise and reality was not discovered late; it was visible in Merck's own trials. The VIGOR study, published in The New England Journal of Medicine on 23 November 2000 under lead author Claire Bombardier, showed that the 8,076 patients taking rofecoxib had roughly four to five times the rate of myocardial infarction of patients taking the older drug naproxen — a relative risk the published paper reported as 0.2 in naproxen's favor. Merck attributed the difference to a supposed protective effect of naproxen rather than a cardiac hazard of Vioxx — a hypothesis that was never proven and that the company continued to advance while sales climbed past $2.5 billion a year and a sales force of more than 3,000 representatives carried the message to prescribers.

The withdrawal came only when the harm became impossible to explain away. The APPROVe trial — designed to test whether Vioxx prevented recurrent colon polyps, not to study the heart — was halted early on 23 September 2004 after showing that rofecoxib roughly doubled the risk of confirmed cardiovascular events after eighteen months of use, against placebo. The verdict here is therefore plain at the outset: an approved, heavily advertised, trusted medicine reached tens of millions of patients while the signal that would eventually condemn it sat in the company's data, the FDA's review files, and the medical literature, fully legible to anyone with access and the will to read it.

What followed was the largest pharmaceutical mass-tort reckoning of its era. In November 2007 Merck agreed to pay $4.85 billion to resolve roughly 27,000 lawsuits covering some 47,000 plaintiff groups, NEJM issued a formal "Expression of Concern" alleging the VIGOR authors had withheld cardiac data, and the case became the standard byword for how surrogate-endpoint approval, aggressive marketing, and suppressed safety signals can converge into mass harm before any regulator pulls the cord.

Timeline

1999-05-20
FDA approves Vioxx
Rofecoxib is approved for osteoarthritis, acute pain, and menstrual pain on the strength of efficacy and reduced gastrointestinal bleeding versus older NSAIDs — a stomach-sparing surrogate benefit, not a cardiovascular safety mandate. Merck launches it in May 1999 alongside Pfizer's rival Celebrex, opening the COX-2 market.
2000-03
VIGOR trial completes
The 8,076-patient Vioxx Gastrointestinal Outcomes Research study finds roughly half as many serious GI events on rofecoxib but a markedly higher rate of myocardial infarction in the rofecoxib arm than in the naproxen arm.
2000-11-23
VIGOR published in NEJM
The paper (Bombardier et al., N Engl J Med 2000;343:1520-8) reports ~4–5x more heart attacks on rofecoxib but frames the gap via the unproven 'naproxen is cardioprotective' hypothesis. Three later MIs in the Vioxx arm are not in the published table; editors later said the manuscript's analysis was locked while events were still accruing.
2001-02-08
FDA Arthritis Advisory Committee reviews VIGOR
The panel examines the cardiovascular imbalance; the discussion produces a label change rather than any restriction on marketing.
2001-08-22
JAMA flags cardiovascular risk
Mukherjee, Nissen, and Topol publish a pooled analysis in JAMA warning that COX-2 inhibitors, including rofecoxib, may raise cardiovascular risk and calling explicitly for a dedicated cardiovascular trial — a public signal more than three years before withdrawal.
2001-09-17
FDA warning letter to Merck
The agency cites Merck for promotional conduct that 'minimizes the potentially serious cardiovascular findings' of VIGOR and misrepresents the drug's safety profile to physicians.
2002-04
Label updated, not withdrawn
The FDA adds VIGOR cardiovascular information to the Vioxx label's precautions section, but the drug stays on the market and direct-to-consumer advertising continues; annual sales approach $2.5 billion and roughly 100 million prescriptions have been written worldwide.
2004-09-23
APPROVe stopped early
The colon-polyp prevention trial's external data safety monitoring board halts dosing after rofecoxib roughly doubles confirmed cardiovascular events versus placebo after 18 months of continuous use.
2004-09-30
Worldwide withdrawal
Merck voluntarily withdraws Vioxx globally. An estimated 80 million people had been prescribed it; ~20 million Americans had taken it. Merck shares fall roughly 27 percent that day, erasing about $26–28 billion in market value.
2004-11-18
Graham testifies to the Senate
FDA reviewer David Graham tells the Senate Finance Committee that Vioxx caused an estimated 88,000–139,000 excess heart attacks and strokes and calls the FDA as structured 'incapable of protecting America' from the next Vioxx.
2005-02-18
FDA advisory panel votes Vioxx could return
A joint advisory committee narrowly concludes the drug's benefits could outweigh risks under restrictions; Merck does not reintroduce it.
2005-03-17
APPROVe published
Bresalier et al. (N Engl J Med 2005;352:1092-1102) formally documents the doubled cardiovascular risk in the 2,586 randomized patients of the colorectal-adenoma trial.
2005-12-08
NEJM Expression of Concern
NEJM editors publicly state the VIGOR manuscript omitted three myocardial infarctions known to authors before publication, alleging the printed data did not reflect the true cardiac hazard.
2007-11-09
$4.85 billion settlement
Merck agrees to a $4.85 billion fund resolving roughly 27,000 cases covering some 47,000 plaintiff groups — at the time the largest drug-injury settlement in U.S. history — without admitting liability.

The Stomach-Sparing Promise: How a Surrogate Benefit Built a Blockbuster

Vioxx was engineered to solve a real problem. Traditional non-steroidal anti-inflammatories such as ibuprofen and naproxen relieve pain by blocking both the COX-1 and COX-2 enzymes, and the COX-1 blockade erodes the protective lining of the stomach, causing tens of thousands of bleeding ulcers and thousands of related deaths a year in the United States alone. Rofecoxib selectively inhibited COX-2, the enzyme tied to inflammation and pain, while largely sparing COX-1. On that mechanism Merck built a clean marketing story: the same pain relief, fewer ulcers. The FDA's May 1999 approval rested largely on this gastrointestinal surrogate benefit — a measurable proxy for harm avoided — rather than on any demonstration that the drug was safe for the heart over years of chronic use. That distinction would prove decisive. The same COX-2 selectivity that spared the stomach also tilted the balance of prostaglandins that govern clotting: by suppressing prostacyclin, which keeps platelets from aggregating, while leaving thromboxane untouched, the drug plausibly made blood more prone to forming clots. The biology that made Vioxx gentle on the stomach was the same biology that threatened the cardiovascular system, and the approval pathway never required Merck to resolve that trade-off before mass exposure. Backed by a direct-to-consumer advertising budget that reached roughly $160 million in 2000 — more than was spent that year promoting Pepsi or Budweiser — and fronted in television spots by figures including Olympic skater Dorothy Hamill, Vioxx reached roughly 80 million prescriptions worldwide and became a multibillion-dollar franchise within a few years of its launch.

The Signal in the Data: VIGOR, the Naproxen Defense, and Five Lost Years

The warning did not arrive after withdrawal; it arrived in 2000, inside Merck's own outcomes trial. VIGOR, designed to prove the gastrointestinal advantage in patients with rheumatoid arthritis, instead showed that those on rofecoxib suffered myocardial infarctions at roughly four to five times the rate of those on naproxen — a difference that surfaced in absolute terms as a small but consistent excess of cardiac events. Merck's published explanation was not that Vioxx harmed the heart but that naproxen protected it — a 'cardioprotective naproxen' hypothesis for which no adequate randomized evidence existed and which independent cardiologists, including Eric Topol and Steven Nissen of the Cleveland Clinic in JAMA in August 2001, openly disputed, noting that naproxen's supposed aspirin-like effect had never been demonstrated in any outcome trial. Worse, NEJM editors later alleged that three additional heart attacks in the rofecoxib arm, known to at least two authors before publication, were absent from the printed paper, so that the published relative risk understated the hazard. The FDA's own response was incremental: a 17 September 2001 warning letter rebuking Merck for minimizing the findings, then an April 2002 label change that buried the cardiovascular data in the precautions section — but no withdrawal, no boxed warning strong enough to halt promotion, and continued direct-to-consumer advertising. For roughly five years the drug stayed on the market while the dispute played out as a scientific disagreement rather than a safety emergency, and tens of millions of additional prescriptions were written across that interval.

The Reckoning: APPROVe, the Senate, and the $4.85 Billion Verdict

Vioxx was undone by a trial that was not even looking for the harm. APPROVe asked whether rofecoxib could prevent recurrent colorectal adenomas; its external data safety monitoring board stopped the 2,586-patient study on 23 September 2004 because the drug roughly doubled confirmed cardiovascular events against placebo after eighteen months of use — placebo, not naproxen, eliminating the cardioprotection alibi at a stroke. Merck withdrew the drug worldwide a week later, on 30 September, wiping roughly $26 billion to $28 billion off its own market value in a single trading day. The reckoning then moved from the clinic to the Capitol and the courtroom. On 18 November 2004, FDA safety reviewer David Graham testified to the Senate Finance Committee that Vioxx had caused an estimated 88,000 to 139,000 excess heart attacks and strokes in the United States, with perhaps 30 to 40 percent fatal, and warned that the agency as structured was 'virtually incapable of protecting America' from a repeat — testimony he gave after describing internal pressure on his own Kaiser Permanente study estimating the harm. In December 2005 NEJM published its Expression of Concern over the VIGOR data. Litigation, fed by internal documents pried loose in discovery — including emails in which Merck staff discussed avoiding studies that might expose cardiac risk and a marketing campaign reportedly nicknamed 'Dodge Ball' to deflect physician questions — culminated in a 9 November 2007 agreement under which Merck paid $4.85 billion to resolve roughly 27,000 lawsuits, the largest drug-injury settlement of its time, while admitting no liability. The wonder-drug narrative was retired; the body count became the headline.

Contributing Factors

01
Surrogate-endpoint approval without a cardiovascular safety mandate
Vioxx was cleared in 1999 on a gastrointestinal proxy benefit — fewer ulcers — rather than on long-term cardiovascular outcomes. The COX-2 mechanism that protected the stomach plausibly raised clotting risk by suppressing prostacyclin while sparing thromboxane, yet no adequately powered cardiac-safety trial was required before tens of millions were exposed. Approving on what is easy to measure while deferring what is hard to measure left the most dangerous question unanswered until after mass harm had already accrued.
02
Signal suppression and the manufactured alternative hypothesis
When VIGOR (2000) showed a 4–5x excess of heart attacks, Merck reframed the result through the unproven 'naproxen is cardioprotective' hypothesis and, per NEJM editors, omitted three known infarctions from the published paper. A sponsor that controls the data, the analysis, and the framing can convert a safety signal into a debatable footnote — and a debate is something a company can survive commercially in a way that a warning is not. The hypothesis cost Merck nothing to assert and bought it years of continued sales.
03
Marketing velocity outrunning the safety review
One of the most aggressive direct-to-consumer advertising campaigns in pharmaceutical history — roughly $160 million in 2000 alone, carried by a sales force of more than 3,000 — pushed Vioxx to roughly 80 million prescriptions before the cardiac question was settled. The FDA's September 2001 warning letter for minimizing VIGOR's findings shows promotion and prescribing accelerating faster than the regulatory response; every quarter of delay translated directly into additional exposed patients.
04
Regulatory incrementalism and a captured post-market system
The FDA's reaction to mounting evidence was a label change in 2002, not a withdrawal — a posture David Graham attributed to an agency where the same office that approves a drug also judges its post-market safety, creating institutional reluctance to reverse its own decisions. Weak post-market authority and structural conflict turned a five-alarm signal into a slow series of label edits while the franchise kept selling.
05
Litigation as the true disclosure engine
The most damning facts — internal memos on cardiovascular risk, the missing VIGOR infarctions, the 'Dodge Ball' marketing playbook, the limits of the naproxen defense — surfaced not through regulation or peer review but through subpoenas in mass-tort discovery. The case demonstrates a recurring pattern: when sponsors control the evidence, the courtroom, not the agency, becomes the mechanism that finally forces the data into daylight.

Aftermath

The material toll was vast and uneven. Merck's $4.85 billion fund, agreed in November 2007, resolved roughly 27,000 lawsuits covering some 47,000 plaintiff groups without an admission of liability, paying typical heart-attack and stroke claimants on the order of $100,000 to $200,000 before legal fees — a figure many critics judged small against fatal and disabling injuries numbering in the tens of thousands, and one the company funded against reserves it could absorb. The durable ripple reshaped drug regulation: the FDA Amendments Act of 2007 substantially expanded the agency's post-market authority, empowering it to require post-approval safety studies and Risk Evaluation and Mitigation Strategies (REMS) and to compel label changes — powers whose absence the Vioxx episode had exposed in plain view. The entire COX-2 class was forced into boxed cardiovascular warnings, Pfizer pulled the related valdecoxib (Bextra) in 2005, and the surviving member, celecoxib, carried restrictions and lingering suspicion for years until the large PRECISION trial in 2016 partly rehabilitated it. The NEJM Expression of Concern became a landmark in the debate over sponsor control of clinical-trial data, accelerating demands for mandatory trial registration on ClinicalTrials.gov and independent analysis of raw results. What remains is a name that functions as shorthand. 'Vioxx' is now invoked whenever a profitable, heavily promoted, regulator-blessed product is found to have carried a measured, suppressible harm for years before revocation — the canonical case of how an approved and trusted medicine can become a mass-casualty event while the warning sits, fully legible, inside the maker's own files.

Lessons

  1. Treat a surrogate-endpoint approval as a hypothesis, not a verdict: when a product is cleared on a proxy benefit, demand the hard outcome trial before scaling exposure — never after.
  2. When the maker of a thing also controls and frames its safety data, weight independent analysis above the sponsor's explanation; an unproven alternative hypothesis that happens to protect revenue deserves the heaviest scrutiny, not the benefit of the doubt.
  3. Read a label change as the floor of a response, not the ceiling: if a signal is strong enough to alter the warning, ask why it is not strong enough to halt the promotion, and count the patients exposed during every quarter of delay.
  4. Build the disclosure mechanism before you need it — mandatory data registration, independent monitoring, and real post-market authority — because if the only force that surfaces the truth is litigation, the harm is already done and merely awaiting a verdict.
  5. Audit the institution for structural conflict: when the body that approves a product also judges its safety in the field, expect reluctance to reverse course, and design independent review in from the start.

References