Dec 2005 — Merck & Co. felt the first fallout from a top medical journal’s allegation that key data were excised from a 2000 study of its painkiller Vioxx, as plaintiffs in a federal liability trial over the drug seized on the disclosure to call for a mistrial.
The request showed how the disclosure from the New England Journal of Medicine, which was made late Thursday afternoon, could damage Merck in the thousands of pending lawsuits over Vioxx. The omitted data concerned three Vioxx patients who had suffered heart attacks. The New England Journal said the omission made the drug appear safer than it should have.
The motion in federal court in Houston, where a jury was spending a second day deliberating in the third Vioxx trial, came as more details emerged about how editors at the medical journal learned of the excised data from the Merck-funded study, which was known as Vigor. The medical journal said Thursday that it learned of the problem only last month, when its executive editor was deposed by plaintiffs’ attorneys.
The disclosure has stirred questions in the medical world over Merck’s role in the study, the role of two Merck employees who were co-authors on the study and the New England Journal’s failure to identify the omitted data sooner, despite knowing about the three heart attacks that were omitted.
“I think for the New England Journal to be blaming Merck at this stage is diverting attention from their own responsibility in this,” said John Abramson, a clinical instructor at Harvard Medical School who has given lectures about what he calls the “misrepresentation” of the Vigor data. When the medical journal “recognized that they published something misleading, they had a responsibility to let the physician community know. Why are they letting the doctors know now?”
Yesterday, the medical journal’s executive editor, Gregory Curfman, strongly defended his staff, describing its actions as “prudent,” and saying: “In hindsight, we did the right thing.”
In pitting one of the world’s most influential medical journals against an icon of the pharmaceutical industry, the episode exposes a fault line in the relationship between industry and journal editors over how to ensure and protect the integrity of scientific information. It also underscores a current controversy in clinical trials: how to handle potentially important data that emerge outside the time-line or other protocols of a clinical trial.
Phil Fontanarosa, executive editor of the Journal of the American Medical Association, said he believed the authors had a responsibility to include information on the additional heart attacks in the study, regardless of what the prespecified analysis called for.
“This is another in the long list of recent examples that have generated real concerns about trust and confidence in industry-sponsored studies,” he said.
Some details about the study remained unclear, including just who had deleted the data from a manuscript shortly before it was submitted to the New England Journal. The medical journal said it had determined that the omission was done on a Merck computer. Merck documents indicate that company employees were aware that the heart attacks had taken place and that they weren’t included in their article.
Merck executives declined to comment. In a statement released Thursday, the company said the article “fairly and accurately” represents the results of the study, and that the omitted heart attacks took place after a predetermined date on which the study ended. It also said it would respond to the New England Journal.
Claire Bombardier, lead author of the Vigor study and a researcher at Mount Sinai Hospital and the University Health Network in Toronto, said she wouldn’t comment on any specific statements from the New England Journal of Medicine until the authors complete an evaluation of the scientific facts and data contained in the editorial. “I am first and foremost a scientist and I will allow the data to speak for itself when it is complete,” she said in an email.
In 4 p.m. composite trading on the New York Stock Exchange, shares of Merck were off 55 cents, or 1.8%, at $29.13. Merck pulled Vioxx in September 2004 amid studies that tied the drug to heart attacks in patients who had taken it for longer than 18 months.
Possible Impact on Litigation
Yesterday’s motions showed how the episode could affect the wave of litigation facing Merck. U.S. District Judge Eldon E. Fallon, who is overseeing the Houston trial, told plaintiffs’ attorneys he would take their motion for a mistrial under advisement but appeared likely to allow the jury to reach a verdict before ruling.
Judge Fallon also told the lawyers in a closed meeting in his chambers that if the jury finds Merck liable in the death of Richard “Dicky” Irvin, who took the painkiller for less than a month before suffering a fatal heart attack, he would consider allowing the plaintiffs to present new evidence about the missing heart-attack data. That would occur during a separate phase of the trial in which the jury will decide on punitive damages. Deliberations continue on Saturday.
Merck launched the Vigor study in January 1999. It involved more than 8,000 patients with rheumatoid arthritis, and was aimed at showing that Vioxx caused fewer ulcers and episodes of stomach bleeding than the common painkiller naproxen. Merck funded the study and two of its employees were among its 12 authors.
Publication of the paper in November 2000 in the New England Journal showed that Vioxx indeed was associated with fewer such problems and helped set the stage for Merck’s huge marketing of the drug.
The paper included a short section on cardiovascular risk, showing that 0.4% of Vioxx patients, compared with 0.1% of those on naproxen, had suffered heart attacks during the average nine months of followup. That reflected 17 heart attacks among Vioxx patients, compared with four on naproxen.
The medical journal now says that Merck knew about three other heart attacks among Vioxx patients and had included them in the manuscript, but deleted data on at least some of them just two days before it submitted the study for publication on May 18, 2000.
Merck issued a press release on the studies findings, noting noted that “fewer heart attacks were seen in patients taking naproxen (0.1%) compared to the group taking Vioxx (0.4%) in this study.” But Merck documents subpoenaed in the Vioxx litigation make clear that the company had known about the additional heart attacks for months.
A few months later, Merck took a different tack. In February 2001, as Merck prepared for a high-profile Food and Drug Administration advisory committee hearing to contemplate changes to the Vioxx label, it put out a press release noting that the Vigor study showed that 0.5% of patients using Vioxx had heart attacks, compared with 0.1% in the naproxen users.
On May 22, 2001, Merck put out a press release, which was later criticized by the FDA in a warning letter, titled: “Merck Confirms Favorable Cardiovascular Safety Profile of Vioxx.” That press release also said the Vioxx risk was 0.5%. Physicians who asked Merck for more information in the summer of 2001 were sent the New England Journal of Medicine reprint as well as a four-page letter that summarized the article. On the last page, the letter read: “In the subsequent final analysis, which included all events reported after the February cut-off date, the rate of cardiovascular events was 0.5% among patients taking Vioxx.”
Including the three events would lead to a modest increase in a reported risk of a heart attack. Multiplied by millions of users, that difference becomes larger. The additional 0.1 percentage point increase (to 0.5% from 0.4%) would represent an additional 1,000 heart attacks for every million pa
tients who took the drug, though from an individual perspective the difference may seem small.
“It’s not that this changes the interpretation,” said Robert Califf, head of Duke University’s Duke Clinical Research Institute, Durham, N.C. “It’s that it raises the question of integrity. They knew about these events and they didn’t put it in the paper.”
Three Heart Attacks
The discovery of the deleted heart attacks was prompted by information contained in internal Merck documents and provided to Dr. Curfman during the deposition last month, he said. The documents indicated to Dr. Curfman that Merck researchers on the study knew of the three heart attacks in July 2000, more than four months before the journal published the Vigor paper—plenty of time to have included them in the report.
Merck said Thursday the heart attacks weren’t included because they occurred after a prespecified cut-off point for counting adverse events in the study. But Dr. Curfman said he didn’t know until he was presented with the Merck memos during his deposition that there was an earlier cut-off date for cardiovascular events. “That information was not available to the editors,” when the study was being considered for publication, he said.
Cut-off dates are typically used in clinical trials to define the period during which benefits and risks of a treatment are measured.
The medical journal editors have not stated explicitly that the three heart attacks took place before the cut-off date. Whenever they occurred, they argue, the heart attacks were important enough to set the formalities of the study aside.
About five months after the New England Journal published the study on Nov. 23, 2000, Dr. Curfman said in an interview, Merck did make an updated report to the FDA about the Vioxx study that included mention of the three additional heart attacks. Dr. Curfman said, however, that at that point he presumed these were “late” events that occurred after the study’s end and that the Journal does not routinely investigate late events.
Having not heard otherwise from the studies authors, “it was our assumption that the three additional heart attacks were simply late events, too late to be entered into the article,” Dr. Curfman said. “It’s fairly common for there to be late events in a clinical trial….It’s not something we would routinely investigate.”
Soon after Vioxx was pulled from the market on Sept. 30, 2004, Dr. Curfman and other editors went back to the Journal’s office and retrieved related submission files from storage. The file included a storage disk that no one had seen at the time of publication since the Journal was then just transitioning to electronic editing and still did almost all editing from paper copies.
“We wanted to refresh our memories about what was in the file in case the media called,” Dr. Curfman says. “And it briefly crossed our mind that maybe at some time we might receive a subpoena.”
On the disk, Stephen Morrissey, the medical journal’s managing editor, found a presubmission version of the manuscript that no one had reviewed before. The file was such that editors could see editing changes that had been made and when. They determined that certain changes had been made on May 16, 2000, two days before the manuscript was submitted for publication. They found that the number of cardiovascular events in the body of the manuscript had been changed to percentages, which is how the story was published.
“I wished I had had more reservations,” about that, Dr. Curfman said. He said the disk file also contained a chart labeled “cardiovascular events” but that it was blank.
Dr. Curfman said he and Dr. Morrissey called in computer technicians to see if more information could be retrieved, but it couldn’t. They also spent several days discussing the matter with other editors. In the end, they decided they did not have enough information to launch a deeper investigation.
“We really didn’t have enough to go on,” Dr. Curfman said, “and to launch an investigation at that time, on the basis of what we had, I think would have been counterproductive, potentially very time consuming and would have probably led to a blind alley.”
He said they decided to keep an eye on the matter but took no further action and did not contact the studies authors about their concerns. Still, he said, the matter was unsettling enough that they hid the disk in Dr. Morrissey’s office.
Asked why the medical journal didn’t issue a correction or an update soon after both Merck and the FDA began publishing corrected numbers including the three additional heart attacks, Dr. Curfman said he continued to presume they were late events and didn’t see the need to take such action. “The (Vigor) trial was primarily a gastrointestinal trial and we were working under the assumption that cardiovascular data were preliminary and had not been completely worked up and assessed,” he said. “That was our working assumption.”
Asked whether he thinks the New England Journal should have investigated further after learning of the changed and deleted data editors found on the storage disk last year, Dr. Curfman described its response as “prudent.”
He also said that the medical journal does not plan to overhaul its manuscript review process. “We think our review procedures are very rigorous and very sound,” he said. “For many years, the scholarly community has functioned well because of the principle of trust.”