Merck's Deleted Data
Robert Langreth and Matthew Herper, 12.08.05, 8:10 PM ET
A top editor of The New England Journal of Medicine says that he was stunned to find out that data linking Vioxx to cardiovascular risk was deleted from a major study his journal published five years ago--and that it appears that Merck researchers may have deleted that data.
"I was somewhere between surprised and stunned," Dr. Gregory Curfman, executive editor of The Journal, says. "They allowed us to publish an article that was just incomplete and inaccurate in some respects and was misleading and may have contributed to the detriment to the public health."
Merck recalled Vioxx after its own study linked long-term use of the drug to an increased risk of heart attacks and strokes. Now, the question is if that alarm should have been sounded much earlier.
Just days after Merck recalled Vioxx from the market, editors at The Journal discovered....early versions of the manuscript [that] contained a blank table entitled "CV events"--which is standard jargon for cardiovascular events. Time stamps in the software indicated that the table was deleted two days before the manuscript was submitted to The New England Journal on May 18, 2000. "When you hover the cursor over the editing changes, the identity of the editor pops up, and it just says 'Merck,'" Curfman says.
The editors weren't sure what to make of the finding, so they kept quiet. It wasn't clear that the information that had been in the table would have changed the conclusions of the study...."We talked internally. Should some action be taken? We did not feel we had sufficient evidence to act on it," Curfman says.
an internal Merck document dated July 5, 2000--after the VIGOR manuscript was submitted to the journal, but well before the study went to press....it indicated that two Merck authors on the VIGOR study knew of three additional heart attacks among Vioxx patients in the study, which had not been disclosed to The New England Journal of Medicine. The heart attacks occurred in the final five weeks of the trial and in patients at low risk for heart problems.
Curfman says the editors had assumed that the VIGOR manuscript only included limited data on heart attacks because that was all that was available at the time. "It turns out that they had quite a bit more already worked up," he says. He raced back to the office and spent the next few weeks analyzing all the VIGOR drafts, raw data and correspondence.
Curfman says he called lead author Claire Bombardier of the University of Toronto, on Monday, indicating that the statement [noting the potential trouble] would be published. She told him that she would begin working on a correction.....Bombardier said that the VIGOR paper appropriately disclosed the data and that the authors were working on finalizing a response to the editorial.
Hints that Vioxx might cause heart attacks had existed for years, as basic research pointed to a mechanism by which the drug might cause more clots. When VIGOR was published in The New England Journal of Medicine, there were already questions among scientists about the potential of Vioxx and related drugs
In a statement, Merck disputed The New England Journal analysis. "The VIGOR publication, which was peer-reviewed, fairly and accurately described the results of the study as of the prespecified cutoff for analysis. The additional events referred to in the editorial were events that were reported after the prespecified cutoff date and, therefore, these were not included in the primary analysis reported in the article.
"Nevertheless, the additional events were disclosed to the FDA in 2000, presented publicly to the FDA's Advisory Committee in February 2001 and included in numerous press releases subsequently issued by Merck. We also note that these additional events did not materially change any of the conclusions in the article."
Curfman responded, "We're not buying into that."
Here's a bit of hearsay: when these stories came out last year, a close friend and former Pfizer statistician told PFP that her department had been warning management for quite awhile that their studies on Celebrex and heart trouble were using the wrong endpoints and potentially distorting the conclusions.
Categories: Legal, Vioxx, Cox-2