The Evidence Against Vioxx
Blogging UCSD economist James Hamilton waded into epidemiology
yesterday but appears to have been dragged under by the libertarian Reason magazine
. Hamilton devotes most of his post to arguing that "studies suggest but can't prove" that Vioxx causes heart attacks. Hamilton is concerned about a recent jury verdict that found manufacturer Merck liable for the heart attack death of a Vioxx user.
Before going into the specifics, it's worth noting that the editors of the top medical journals don't share Hamilton's doubts about Vioxx's harmfulness. Lancet is the most vociferous:
"The licensing of Vioxx and its continued use in the face of unambiguous evidence of harm have been public-health catastrophes." [Lancet editorial, requires paid registration]
I've put excerpts from the JAMA and NEJM editorials at the end of the post.
Why are these editors so sure that Vioxx causes heart attacks? Because, unlike Hamilton, they don't rely on only one study
[the link is to the free, preliminary version. Published version here
.]. Since Vioxx was approved in 1999, studies have been piling up demonstrating the harmful effects of Vioxx. The best of these studies
(because it was a randomized trial) found that Vioxx nearly doubled the risk of heart attack compared to a placebo.
Hamilton's most misleading statement is that Vioxx causes a "1 in 4,000 risk of dying from a heart attack." The casual reader might fail to notice that Hamilton says "dying from a heart attack," not having a heart attack. In fact, the FDA study Hamilton discusses says that 1 in 75 of those using Vioxx at high doses will suffer a heart attack due to using the drug (and 1 in 397 of those using low-dose Vioxx). These are "excess" heart attacks, in addition to those the patients would normally suffer. Before Vioxx was pulled off the market, it was pretty commonly used at high doses -- 18% of prescriptions, according to the FDA study.
Why the difference? Why does Vioxx seem to cause heart attacks but not deaths from heart attacks? Well, there probably is no difference. It seems much more likely that the issue is simply that many people survive heart attacks, so studies examining deaths rather than attacks will simply have a smaller sample, and so be unable to prove much of anything. In addition, the Vioxx studies have relatively short follow-up periods, mostly two years or less. Longer studies would be needed to observe the fatal second (third, fourth, ...) heart attacks.
Ultimately, Hamilton is upset that a jury is responsible for "evaluating the scientific merits of statistical evidence." While jury trials may not be the perfect system, in this case the 12 members of the Vioxx Jury managed to come to the right conclusion.
------------------------------------------Editorials in Top Medical Journals
The recent withdrawal of rofecoxib from the market because of adverse cardiovascular events identified in the unpublished Adenomatous Polyp Prevention on Vioxx (APPROVe) study has raised major concerns about the undue control of industry over postmarketing safety data. Topol pointed out that although he and his colleagues published a clear warning about the cardiovascular toxicity of rofecoxib in 2001, the FDA never insisted on a trial to determine the extent of the problem and the manufacturer countered with a "relentless series of publications . . . complemented by numerous papers in peer-reviewed medical literature by Merck employees and their consultants." [Jama Editorial, requires free registration]
The Adenomatous Polyp Prevention on Vioxx (APPROVe) trial, a study of patients with a history of colorectal adenomas, was stopped early because rofecoxib doubled the risk of major cardiovascular events (relative risk, 1.92; 95 percent confidence interval, 1.19 to 3.11). These findings confirmed the increased risk of myocardial infarction previously seen in the Vioxx Gastrointestinal Outcomes Research (VIGOR) trial [NEJM Editorial, requires paid registration].