In September 2004, Merck voluntarily withdrew Vioxx (rofecoxib) from the world market after an interim safety analysis showed that Vioxx was associated with an increased risk of cardiovascular events. 

After the APPROVe trial was released in November 2004, Merck's CEO testified before a U.S. Senate committee that "until data from [APPROVe] . . . the combined data from randomized controlled clinical trials showed no difference in confirmed cardiovascular event rates between Vioxx and placebo."  Three meta-analyses sponsored and conducted by Merck had reported no increased cardiovascular risk from Vioxx, leading physicians to believe the drug was safe.

Could the FDA, Merck, and the medical community have known before September 2004 that Vioxx increased the risk of cardiovascular events and death? In yesterday's Archives of Internal Medicine, researchers analyzed all clinical trials comparing Vioxx to placebo up to the time of withdrawal, including unpublished data made available through the Vioxx litigation. They found a statistical trend (p=0.07) toward increased risk of a cardiovascular event or death in clinical trials completed as of December 2000, which should have raised safety concerns several years before the drug was taken off the market.

Examining clinical trials completed as of June 2001, there was a 35% increased risk, which had become statistically significant (p=0.05). As of April 2002, there was a 39% increased risk (p=0.02), and using data available as of September 2004, there was a 43% increased risk (p<.001).

This study shows that information about the risks of Vioxx should have been known several years before it was withdrawn from the market, allowing physicians and the public to make informed choices about risks and benefits. The investigators, all of whom have served as consultants to plaintiffs' attorneys in the Vioxx litigation, write:

Such analyses should be ongoing for all drugs for which trials are being conducted, with attention to the rapid addition of new data about potential harm to any cumulative pooled analyses.

The authors also describe how their analysis reached different conclusions from the meta-analyses conducted by Merck. Specifically, Merck used an inappropriate endpoint that accounted only for cardiovascular, bleeding-related, and deaths from unknown causes, and nonfatal heart attacks and strokes (including strokes caused by bleeding in the brain). This endpoint is appropriate for evaluating the overall safety of blood thinners, which prevent blood clots but can also cause dangerous bleeding. It was not appropriate for evaluating Vioxx's risks, which are related to blood-clotting effects.

In addition, the study published today includes deaths from any cause. Including all deaths is important in evaluating cardiovascular safety, since many cardiovascular deaths are not identified as such.

This study shows how drug safety should be monitored once a drug goes on the market, and the manipulation of the endpoints for the Merck-sponsored meta-analyses shows the importance of such analyses being conducted by independent investigators. The increased availability of clinical trial data on clinicaltrials.gov should help make such studies possible.

-- Marilyn Mann