The Food and Drug Administration (FDA) estimates that Vioxx may have contributed to 27,785 heart attacks and sudden cardiac deaths between 1999 and 2003. The estimate is based on the number of prescriptions issued for Vioxx between 1999 and 2003.

David Graham, the associate director for science in FDA's office of drug safety, made the estimate based on 92.8 million U.S. prescriptions for Vioxx between 1999 and 2003. It's part of a study Graham conducted in cooperation with Kaiser Permanente.

Merck pulled Vioxx, a popular pain reliever widely used by arthritis patients, off the market in September, saying it was "putting patient safety first" but the Wall Street Journal reported earlier that company officials had fought for years to protect the highly profitable drug and to keep news of the health risks quiet.

Vioxx was a big moneymaker for Merck, generating about $2.5 billion in yearly sales.

In his study, Graham's team examined records for 1.39 million members of Kaiser Permanente, including 26,748 who took Vioxx and 40,405 who were on Pfizer's Celebrex, another COX-2 inhibitor.

The study found that high doses of Vioxx, or rofecoxib, tripled risks of heart attacks and sudden cardiac death. Graham planned to present the findings at an epidemiology conference Aug. 25, but his supervisors said the results were "too preliminary" and recommended that the study be submitted first to a medical journal so it could undergo peer review or be presented at the conference with an alternative FDA opinion.

When the study was presented Aug. 25, the abstract said, "[T]his and other studies cast serious doubt on the safety of rofecoxib ... and its use by physicians and patients" at doses exceeding 25 milligrams. "When Graham submitted a revised, final version to FDA on Sept. 30, FDA's announcement of the study's release did not mention specific data on cardiovascular risks.

About 20 million Americans had taken Vioxx by the time Merck withdrew it.

The company responded to the latest report by saying there is "no reliable way to estimate the actual events."

"Because heart attacks and strokes occur in the general population, one cannot say that if someone had an event while taking Vioxx, that Vioxx caused it," a Merck spokesman said.