Vioxx was withdrawn from the market in September 2004 after Merck's own trial data definitively showed it doubled the risk of heart attack and stroke. By then, an estimated 88,000 to 139,000 Americans had experienced heart attacks attributed to the drug. Internal documents revealed that Merck scientists had evidence of cardiovascular risk as early as 1997, that the company systematically downplayed these risks in publications and FDA submissions, and that it maintained a list of critical researchers to be 'neutralized.' The legal settlements exceeded $4.85 billion.
Rofecoxib, marketed as Vioxx, belonged to a new class of anti-inflammatory drugs called selective COX-2 inhibitors. Traditional nonsteroidal anti-inflammatory drugs (NSAIDs) like ibuprofen and naproxen block both cyclooxygenase-1 (COX-1) and cyclooxygenase-2 (COX-2) enzymes. COX-2 produces prostaglandins that cause inflammation and pain. COX-1 produces prostaglandins that protect the stomach lining. By selectively inhibiting only COX-2, Merck's scientists theorized they could reduce pain and inflammation without causing the gastrointestinal bleeding that plagued traditional NSAIDs.
The theory was elegant. The execution was catastrophic. What Merck's scientists understood—and what internal documents show they discussed as early as 1997—was that COX-2 inhibition also suppressed prostacyclin, a molecule produced by blood vessel walls that dilates vessels and inhibits platelet aggregation. Traditional NSAIDs also inhibit thromboxane, a molecule that promotes clotting. Vioxx disrupted this balance: it suppressed the anti-clotting prostacyclin without affecting the pro-clotting thromboxane. The result was a drug that could tip patients toward thrombosis—blood clots that cause heart attacks and strokes.
The FDA approved Vioxx on May 20, 1999, based on clinical trials demonstrating efficacy for osteoarthritis, acute pain, and menstrual symptoms. The approval was granted under priority review—a designation for drugs offering significant therapeutic advance. The trials submitted for approval showed that Vioxx worked as intended: it reduced pain with fewer gastrointestinal side effects than traditional NSAIDs. What those trials did not adequately capture was cardiovascular risk, because they were too short in duration and not specifically designed to measure heart attacks and strokes.
The Vioxx Gastrointestinal Outcomes Research (VIGOR) trial was designed to prove Vioxx's core marketing claim: superior gastrointestinal safety. Merck enrolled 8,076 patients with rheumatoid arthritis and randomized them to receive either Vioxx 50 mg daily or naproxen 500 mg twice daily. The trial ran for a median of nine months. The results, published in the New England Journal of Medicine in November 2000, confirmed the gastrointestinal hypothesis: Vioxx caused half as many serious gastrointestinal events as naproxen.
But buried in Table 4 of the published paper was a statistic that would eventually destroy the drug: patients taking Vioxx had a myocardial infarction rate of 0.4 percent, compared to 0.1 percent for naproxen—a relative risk of 5.0. Twenty patients on Vioxx had heart attacks. Four patients on naproxen had heart attacks. The difference was statistically significant.
Merck's explanation, prominently featured in the VIGOR publication, was that naproxen was cardioprotective—that it worked like aspirin to prevent heart attacks. Therefore, the apparent increase in Vioxx patients reflected the absence of naproxen's protection, not Vioxx's danger. This interpretation was immediately controversial. External cardiologists, including Dr. Eric Topol at the Cleveland Clinic, argued that the data suggested thrombotic risk from Vioxx itself, consistent with the known mechanism of COX-2 inhibition. Topol published a perspective in JAMA in August 2001 titled "Failing the Public Health—Rofecoxib, Merck, and the FDA," calling for restricted use and additional studies.
Internal Merck emails told a different story. In March 2000, months before VIGOR's publication, chief scientist Edward Scolnick wrote to colleagues: "The CV events are clearly there." Another email from 1999 discussed the mechanism explicitly: Vioxx inhibited prostacyclin without affecting thromboxane, creating prothrombotic conditions. These emails, disclosed years later in litigation, demonstrated that Merck's senior scientists understood the cardiovascular risk but chose to publicly emphasize alternative explanations.
In March 2005, five months after Vioxx's withdrawal, the New England Journal of Medicine published an unusual notice: an "Expression of Concern" regarding the VIGOR trial publication. The journal revealed that three additional myocardial infarctions had occurred in Vioxx patients before the final data cutoff but were not included in the manuscript submitted to NEJM. Had these been included, the cardiovascular difference would have been even more pronounced.
Merck's explanation was technical: the three events occurred after a pre-specified cutoff date for gastrointestinal outcomes. But critics pointed out that this was a safety analysis, not an efficacy analysis. The cutoff for adverse events should have been the end of observation, not an arbitrary date tied to a different endpoint. The NEJM editors wrote that had they known of the additional events, it "might have affected" their editorial decision. The incident raised fundamental questions about pharmaceutical control over clinical trial publications and the adequacy of peer review when authors have financial conflicts.
"Had we known... it might have affected our evaluation of the benefits and risks."
New England Journal of Medicine Editors — Expression of Concern, March 2005Dr. Alise Reicin, the Merck clinical researcher who was lead author on the VIGOR paper, defended the cutoff decision in subsequent testimony. She argued that the data was disclosed to the FDA and that the published paper appropriately analyzed cardiovascular events. Academic co-author Claire Bombardier stated she had not been aware of the additional events at the time of submission and had relied on Merck's data summaries. The controversy illustrated the problem of academic authorship on industry-sponsored trials: prestigious names lent credibility to publications, but the company controlled the data.
Among the most damaging documents introduced in Vioxx litigation was an internal Merck training presentation titled "Dodge Ball Vioxx." The presentation, used to train sales representatives, listed physicians who had raised concerns about Vioxx's cardiovascular safety. Next to their names were notations: "neutralize," "discredit," "manage," or similar terms. The document outlined strategies for countering criticism, including emphasizing published data favorable to Vioxx, questioning the credentials of critics, and highlighting industry relationships of critics to suggest bias.
Dr. Eric Topol was reportedly on this list. So were other prominent cardiologists who had publicly questioned Vioxx's safety profile. The "Dodge Ball" document became symbolic of Merck's approach: rather than engaging with scientific criticism through additional research or label changes, the company developed marketing strategies to minimize the impact of critical voices.
This was not illegal. Pharmaceutical companies have the right to defend their products and respond to criticism. But the systematic nature of the effort—the creation of lists, the training of sales forces to counter specific individuals, the emphasis on discrediting rather than investigating—demonstrated a corporate strategy prioritizing product defense over patient safety.
The institutional conflict within the FDA mirrored the conflict in the scientific literature. The Office of New Drugs, which had approved Vioxx in 1999, viewed cardiovascular signals as uncertain and manageable through labeling. The Office of Drug Safety, where epidemiologist David Graham worked, viewed the same data as evidence of serious risk requiring regulatory action.
After VIGOR's publication in 2000, the FDA required Merck to add precautionary language to Vioxx's label in April 2002. The new label stated that in the VIGOR trial, cardiovascular thrombotic events occurred more frequently in Vioxx patients than in naproxen patients. But there was no black box warning, no restriction on prescribing, no requirement for additional cardiovascular outcome studies. Vioxx remained on the market with broad indications.
David Graham, working independently in the Office of Drug Safety, conducted an analysis using the Kaiser Permanente database, which included 1.4 million patients and nearly 30,000 cardiovascular events. His study, completed in 2004, found clear dose-dependent cardiovascular risk. Patients taking high-dose Vioxx (more than 25 mg daily) had 3.7 times the risk of sudden cardiac death compared to patients not taking Vioxx or other NSAIDs. Patients taking lower doses had increased but smaller risk.
Graham attempted to present his findings at FDA meetings. He was rebuffed. Office of New Drugs officials questioned his methodology and conclusions. According to Graham's subsequent Senate testimony, he was pressured to soften his findings and told not to present his results at a scientific conference. The institutional hostility illustrated a structural problem: the same office that approves a drug is often reluctant to acknowledge that approval was a mistake.
The Adenomatous Polyp Prevention on Vioxx (APPROVe) trial was not designed as a cardiovascular safety study. It was testing whether Vioxx could prevent recurrence of colorectal polyps—a potential new indication that would expand the drug's market. But APPROVe had a critical advantage over previous studies: it was placebo-controlled and longer in duration, with patients taking Vioxx or placebo for up to three years.
On September 23, 2004, Merck's independent Data Safety Monitoring Board conducted a routine interim analysis of APPROVe. The results were unambiguous: after 18 months of treatment, patients taking Vioxx had twice the risk of serious cardiovascular events compared to placebo. The relative risk was 1.92 (95% confidence interval 1.19 to 3.11). There was no naproxen in APPROVe to explain away the difference. This was Vioxx versus placebo, and Vioxx lost.
Merck executives were briefed on the results on September 27, 2004, a Monday. By Thursday, September 30, CEO Raymond Gilmartin announced that Merck was voluntarily withdrawing Vioxx from the global market. The press conference emphasized that the company was acting responsibly based on new scientific information. The stock market reacted immediately: Merck's shares dropped 27 percent in a single day, erasing $27 billion in market capitalization.
On November 18, 2004, less than two months after Vioxx's withdrawal, David Graham testified before the Senate Finance Committee chaired by Charles Grassley. His testimony was devastating—not just to Merck, but to his own employer, the FDA.
Graham presented his Kaiser Permanente analysis, estimating that Vioxx had caused between 88,000 and 139,000 excess heart attacks in the United States. Given that 30 to 40 percent of heart attacks are fatal, this translated to between 26,000 and 56,000 deaths. "By way of comparison," Graham testified, "this is equivalent to 500 to 900 aircraft dropping from the sky." He called Vioxx "the single greatest drug safety catastrophe in the history of this country."
Graham then turned his criticism to the FDA itself. He described a "culture of denial" where the Office of New Drugs, which approved drugs, resisted post-market safety concerns. He detailed how he had been pressured to suppress his findings, told not to present them publicly, and subjected to intimidation. "I was pressured to change my conclusions and recommendations," Graham testified, "and basically threatened that if I did not go along, there would be consequences."
"Vioxx is the single greatest drug safety catastrophe in the history of this country... I would argue that the FDA, as currently configured, is incapable of protecting America."
David Graham, FDA Scientist — Senate Finance Committee Testimony, November 18, 2004Graham's conclusion was radical: "I would argue that the FDA, as currently configured, is incapable of protecting America against another Vioxx." He recommended structural reforms including an independent drug safety board with authority separate from the drug approval process, mandatory post-market studies for drugs approved under accelerated procedures, and protection for FDA scientists who raise safety concerns.
Senator Grassley praised Graham as a hero and criticized FDA leadership for attempting to prevent his testimony. The hearing produced headlines for weeks and prompted multiple investigations. The FDA subsequently implemented some reforms, including creation of the Drug Safety Oversight Board and enhanced post-market surveillance systems, though critics argued the changes were insufficient.
Merck faced approximately 27,000 lawsuits from patients who had suffered heart attacks or strokes while taking Vioxx. The company adopted an unusual strategy: it would try every case individually rather than negotiate a global settlement. Merck calculated that it could prevail in enough trials to make individual plaintiffs give up or settle for low amounts.
The first case to reach trial was Ernst v. Merck, tried in state court in Texas in August 2005. Plaintiff Carol Ernst claimed that Vioxx had killed her husband Robert, a 59-year-old marathon runner who died of a cardiac arrhythmia in 2001 after eight months on Vioxx. Attorney Mark Lanier represented Ernst and focused his case on Merck's internal documents—the Scolnick emails, the "Dodge Ball" list, the VIGOR data omissions. The jury awarded Ernst $253 million: $24 million in compensatory damages and $229 million in punitive damages.
The verdict stunned Merck. Though the judgment was later overturned on appeal (Texas law required proving Vioxx was more than 50 percent responsible, and arrhythmia was not a proven Vioxx effect), the trial established a template. Plaintiffs' attorneys now had the documents, the narrative, and proof that juries would hold Merck accountable.
Merck won some cases and lost others. As trials accumulated through 2005 and 2006, the company's litigation costs mounted. In November 2007, Merck agreed to a $4.85 billion settlement covering approximately 27,000 claims. Under the settlement structure, individual payouts ranged from $25,000 to over $1 million depending on severity of injury and strength of medical evidence. Merck did not admit liability. The company stated the settlement was to avoid the uncertainty and expense of protracted litigation.
The internal documents introduced in litigation painted a consistent picture: Merck knew about cardiovascular risk early, discussed it internally, received external warnings from credible cardiologists, and chose to continue aggressive marketing while emphasizing alternative explanations for concerning data.
Edward Scolnick's 1999 email discussed the specific mechanism: COX-2 inhibition would reduce prostacyclin (which inhibits clotting) without reducing thromboxane (which promotes clotting), creating a prothrombotic state. His March 2000 email stating "the CV events are clearly there" came before VIGOR was published, indicating senior scientific leadership recognized the signal in real time.
Marketing documents showed Merck spent over $100 million annually on direct-to-consumer advertising, featuring Olympic figure skater Dorothy Hamill in commercials emphasizing active lifestyle. This advertising created demand among patients who then asked doctors for Vioxx by name. Many of these patients were not in the high-gastrointestinal-risk category where Vioxx offered meaningful benefit over cheaper alternatives. They were exposed to cardiovascular risk for marginal or no GI benefit.
The "Dodge Ball" document systematized the response to criticism. Rather than treating scientific concerns as signals requiring investigation, Merck treated them as marketing problems requiring neutralization. This approach is legal but ethically troubling. It prioritizes brand defense over scientific inquiry.
No Merck executive was criminally charged. No one went to jail. Raymond Gilmartin retired in 2006 with a $37.8 million compensation package. Edward Scolnick had retired in 2002. Alise Reicin remained with the company. The Justice Department considered criminal charges but ultimately declined to prosecute. The standard for corporate criminal liability is high, requiring proof that executives knowingly violated specific laws with criminal intent. Aggressive marketing of a dangerous drug, even with internal knowledge of risk, does not necessarily meet that standard.
The civil settlement, though massive, was manageable for Merck. The company's annual revenue exceeded $20 billion. The $4.85 billion settlement was paid over multiple years and was partially offset by tax deductions and insurance. Merck's stock recovered. The company continued operations without fundamental structural change.
This outcome illustrates a pattern in pharmaceutical accountability: civil settlements without criminal prosecution, payments that constitute a fraction of revenue, no individual executive consequences. The incentive structure remains: if a blockbuster drug generates $10 billion in profit before problems emerge, a $5 billion settlement is a cost of business, not a deterrent.
Vioxx prompted the FDA Amendments Act of 2007, which gave the agency new authorities including power to require post-market safety studies, authority to mandate label changes without manufacturer agreement, and resources for the Sentinel Initiative—an electronic surveillance system monitoring adverse events across multiple databases.
These reforms addressed some structural problems David Graham identified. The Drug Safety Oversight Board created a mechanism for elevating safety concerns above the Office of New Drugs. Post-market study authority closed a gap where manufacturers could ignore FDA requests for additional safety data.
But fundamental tensions remain. The same agency approves drugs and monitors their safety. Career advancement within FDA still depends more on approvals than on safety actions. Industry funding through user fees creates financial dependence. The revolving door between FDA and pharmaceutical industry employment creates conflicts. These structural issues persist.
Dr. Jerry Avorn at Harvard has written extensively on Vioxx as a case study in pharmaceutical dysfunction. His analysis emphasizes that Vioxx was not just a drug that turned out to be dangerous—it was a drug that never should have been used as widely as it was. Traditional NSAIDs like ibuprofen and naproxen cost pennies per dose. Vioxx cost dollars. For most patients, the marginal GI benefit did not justify the cost, even before cardiovascular risk was known. For the subset of high-GI-risk patients (elderly, history of bleeding, concurrent steroid use), COX-2 inhibitors offered real benefit.
But Merck marketed Vioxx broadly through direct-to-consumer advertising rather than targeting the narrow population where benefit outweighed risk. This pattern—broad marketing of expensive drugs for conditions where cheaper alternatives exist—is endemic in pharmaceutical industry practice. It is driven by the blockbuster model: to justify development costs and generate profits demanded by investors, manufacturers must achieve mass-market penetration, not niche prescribing.
"Most patients prescribed Vioxx were not in the high-risk category where COX-2 inhibitors offered meaningful benefit. They were exposed to cardiovascular risk for marginal or no GI advantage."
Jerry Avorn — Powerful Medicines, 2004The Vioxx case also demonstrated the inadequacy of pre-approval clinical trials for detecting rare but serious adverse events. Heart attacks occur at a baseline rate in the population. To detect a doubling of risk requires thousands of patients followed for months or years. Pre-approval trials are typically powered to demonstrate efficacy for the primary indication, not to rule out rare adverse events. Post-market surveillance is where serious safety signals often emerge—but by then, millions of patients have been exposed.
Why were no executives criminally charged? The evidence that Merck knew of cardiovascular risk and continued marketing seems clear from internal documents. But criminal prosecution requires proof beyond reasonable doubt that specific individuals knowingly violated specific criminal statutes with intent to defraud or harm.
Wire fraud and mail fraud statutes require proving that defendants made false statements knowing they were false, with intent to obtain money under false pretenses. Merck could argue that cardiovascular risk was uncertain, that alternative explanations (naproxen's protective effect) were scientifically plausible, and that all data was disclosed to FDA. The company continued marketing under FDA approval—which, while not a complete defense, complicates prosecution.
Conspiracy charges would require proving that multiple executives agreed to suppress information. Internal emails show discussions of cardiovascular mechanisms and data, but they don't clearly show agreement to hide information from regulators. Scolnick's "CV events are clearly there" email is damning in civil litigation context, but in criminal context, prosecutors would need to prove what "clearly there" meant—acknowledged risk requiring withdrawal, or uncertain signal requiring monitoring?
The Justice Department's decision not to prosecute likely reflected these evidentiary challenges plus the reality that pharmaceutical companies are typically charged as corporations rather than individuals. Corporate prosecution allows settlement without threatening company viability or jobs. Individual prosecution of executives is rare in pharmaceutical cases and requires exceptionally clear evidence of intentional criminal conduct.
Vioxx fundamentally changed pharmaceutical regulation and public perception of drug safety. The case demonstrated that FDA approval does not guarantee safety, that pharmaceutical companies have financial incentives to minimize risk signals, and that post-market surveillance is critical. It prompted reforms but did not fundamentally alter the incentive structure that produced the problem.
The cardiovascular risk of COX-2 inhibitors turned out to be a class effect. Celebrex, Pfizer's competing COX-2 inhibitor, also shows cardiovascular risk though it remains on the market with black box warnings and restricted indications. The entire COX-2 class has been scaled back from blockbuster drugs to niche products for specific high-risk GI patients.
Twenty years after Vioxx's withdrawal, the estimated death toll remains contested. FDA's estimate of 88,000 to 139,000 heart attacks, with 30 to 40 percent fatality, suggests 26,000 to 56,000 deaths. Other epidemiological analyses produced different numbers. Merck has never accepted these estimates. The company maintains that it acted responsibly based on available data and that cardiovascular risk only became clear with the APPROVe trial in 2004.
The internal documents tell a different story—one of early warnings, mechanistic understanding, external criticism, systematic marketing-driven response, and delayed action. That is the documented record. The interpretation remains contested, but the facts are clear.