Now, in "Merck admits a data error on VIOXX", New York Times, May 31, 2006, I observe that medical informatics expertise might have been of some preventive value on an issue that's likely to be costly in the courtroom:
In an admission that could undermine one of its core defenses in Vioxx-related lawsuits, Merck said yesterday that it had erred when it reported in early 2005 that a crucial statistical test showed that Vioxx caused heart problems only after 18 months of continuous use.
That statistical analysis test does not support Merck's 18-month theory about Vioxx, the company acknowledged yesterday.
But Dr. Peter S. Kim, Merck's chief scientist, said the company stood by the overall findings it reported in 2005 — including the conclusion that the drug's heart risks were not apparent if patients took it less than 18 months.
But outside scientists said yesterday that Merck's admission, when considered along with other clinical trials of the drug and studies tracking real-world Vioxx use, supports critics' longstanding claims that Vioxx caused heart problems quickly.
"There never was any evidence for the 18-month story," said Dr. Alastair J. J. Wood, a drug safety expert at Vanderbilt University.
... When it reported the Approve results in The New England Journal of Medicine early last year, Merck said that it had performed a statistical test to examine whether Vioxx's risk changed over time. That test found with almost total certainty that the drug had significantly higher risk than placebo only after the 18-month benchmark — but no extra risk before that time.
Yesterday, Merck said it had made a mistake in reporting that result last year.
In reality, the test that the company said it had used to check the results shows that there is a 7 percent chance that Vioxx has an equally high risk of causing heart attacks both before and after the 18-month benchmark is reached.
That 7 percent figure may seem like a relatively small chance of error, but scientists say it is high enough to mean that Merck has not proved its theory.
Clearly, these studies could benefit from a broader inclusion of skills and insights. As Gartner observed in its "2006 Industry Predictions" report for pharma, "only a small percentage of biopharmas routinely utilize personnel with medical informatics backgrounds to search for adverse events in approved drugs."
What will it take to change this?