Reports allege deception over Vioxx

WASHINGTON – Two teams of researchers with access to thousands of documents gathered in lawsuits over the painkiller Vioxx allege that Merck & Co. waged a campaign of deception to promote its drug, moving slowly to warn of possible hazards while at the same time dressing up in-house research as the work of independent academic researchers.

The reports in today’s Journal of the American Medical Association in effect charge one of the world’s biggest pharmaceutical makers with various forms of scientific fraud.

One study alleges Merck gave an incomplete accounting to the Food and Drug Administration of deaths in a clinical trial of Vioxx in people with mild dementia. Federal regulators eventually got the data, which added to the growing evidence that that Vioxx increased the risk of heart attack or stroke.

Simultaneously, Merck was using what the JAMA authors term “guest authorship and ghostwriting” to make it appear that research done by its own employees or contractors was the work of scientists in medical schools and universities. That presumably gave the findings more credibility when they were published in medical journals, boosting Vioxx’s profile in the crowded painkiller market.

Vioxx, whose generic name is rofecoxib, went on the market in 1999. It became a “blockbuster” with $2.3 billion in sales in 2003 but was voluntarily withdrawn in September 2004 after several studies showed that it increased the risk of heart attacks and strokes.

Since then, Merck has been named in 26,500 lawsuits by people saying they were harmed by the drug. Last fall, the company created a $4.85 billion fund to settle the claims while not admitting that Vioxx caused heart attacks, strokes or deaths.

The two JAMA papers – based on access to company documents made public through the lawsuits – claim to provide a look at practices widespread in the pharmaceutical industry.

This view was endorsed in an editorial signed by the journal’s editor, Catherine DeAngelis, who wrote, “But make no mistake – the manipulation of study results, authors, editors, and reviewers is not the sole purview of one company.”

Although the two studies question the integrity of dozens of physicians and scientists, the JAMA authors did not seek responses from them. Several of those people Tuesday called the conclusions incorrect, incomplete or unfair. A spokesman for Merck’s legal team dismissed the JAMA authors as “people in the pay of trial lawyers.”

In the study questioning Merck’s reporting of deaths, Bruce Psaty, a physician and epidemiologist at the University of Washington, examined documents from two clinical trials testing the theory that Vioxx could slow the progress of dementia in elderly people.

In July 2001, Merck filed a report with the FDA saying that out of the roughly 1,000 people in each “arm” of the study, 29 taking Vioxx had died, compared with 17 taking a placebo. However, an internal company document three months earlier contained a more ominous and lopsided report – 34 deaths in the Vioxx group compared with 12 deaths on placebo.

The difference in the two sets of numbers apparently arose from two different ways of counting the deaths.

In the in-house document, the Merck researchers included both people who died while taking either Vioxx or a placebo and people who died after stopping, as quite a few did because they had side effects, changed their minds, or moved away. This cautious method of accounting automatically captures any delayed effects of a drug (sometimes measured in months), although it unavoidably includes deaths that are very unlikely to be linked to the drug, such as a fatal car crash occurring a year after a person stops taking a medicine.

In its FDA filing, however, Merck reported only deaths that occurred while “on treatment” and up to two weeks later (for one study) or three months later (for the other). This gave a mortality risk that was less dramatic than the 3-to-1 ratio calculated with the method Merck used in-house.

Merck reported all the mortality data to FDA two years later, and the information appeared in journal articles in 2004 and 2005.