Published June 10, 2007
Rep. Henry Waxman, D-Calif., launched a congressional investigation this week into the popular diabetes drug, Avandia. It’s too bad his inquiry appears headed in the wrong direction.
A May 16 study in the New England Journal of Medicine reported that Avandia users had a 43 percent higher rate of heart attack and a 64 percent higher rate of death from cardiovascular causes.
Given the 13 million prescriptions written for Avandia’s 1 million U.S. users last year, the report sent a shockwave through the diabetes community and the financial markets. In the week following the report’s release, the number of Avandia prescriptions dropped by 16 percent and manufacturer GlaxoSmithKline lost more than 10 percent of its market valuation — about $16 billion.
Not surprisingly, Waxman, the king of congressional show trials, inserted himself into the controversy with his June 6 hearing focusing on his favorite targets — a Republican-run regulatory agency and big business.
His hearing’s ostensible purpose was to explore whether the Food and Drug Administration and GlaxoSmithKline mishandled scientific data concerning the drug’s safety.
Waxman’s inquest, however, presumes that the New England Journal of Medicine study is meaningful, which it most probably isn’t.
First, no one knows what actually caused any of the heart attacks in the Avandia study. No one knows how much Avandia any patient actually took. This entire scare is a statistical fantasy, where the statistics are concocted from few facts.
Although the reported 43 percent and 64 percent increases in risk sound large, they likely are too small to have been reliably detected by the researchers.
“In studies of disease patterns in human populations, risks of less than 100 percent are considered small and usually difficult to interpret. Such increases may be due to chance, statistical bias or effects of confounding factors that are sometimes not evident,” opined the National Cancer Institute in 1994.
The margins of error for the two results are rather large — more than twice size of the reported risks — further indicating unreliable statistical results.
In addition to being “small and difficult to interpret,” the reported 64 percent increase in risk of death from cardiovascular causes isn’t even statistically significant — if possible, that result is less than meaningless.
The study authors didn’t observe any Avandia users directly. Instead, they statistically combined the results of 42 previous studies involving about 28,000 patients — a methodology known as “meta-analysis” — to derive their results. But the mere use of this technique raises many red flags.
The idea behind meta-analysis is simple enough — combine the results of many small and statistically weak studies into a single large and statistically strong study. While this sounds logical enough, it overlooks two realities.
First, the individual studies likely differ in significant ways — like apples and oranges — that make a valid combination a dubious proposition. There was, for example, no standard method among the studies for identifying or validating whether patients actually suffered heart attacks.
It’s debatable whether the many potentially significant differences in study methodologies and data can be eliminated or smoothed out by cavalierly throwing them into a statistical blender.
Next, meta-analysis tends to be used when researchers refuse to believe what the smaller studies are telling them. The Avandia studies are a prime example of this.
In about two-thirds of the studies, which typically numbered hundreds of patients, the difference in the number of heart attacks between the Avandia group and the control group was either one or zero.
Neither of the two largest among the 42 studies, accounting for more than one-third of the 28,000 total patients, reported statistically significant results linking Avandia with health effects.
The bottom line is that none of the 42 studies raises red flags about Avandia — and the meta-analysis does not change this situation.
More interesting than the Avandia meta-analysis, and something Waxman ought to consider looking into, are the origin and funding of the attack on Avandia.
Pharmaceutical companies can be ruthlessly competitive and, as discussed in my book “Junk Science Judo: Self-Defense Against Health Scares and Scams,” have a history of conducting scare campaigns against their rivals’ products.
In this case, study author Steven Nissen has acknowledged that his work is supported by many GlaxoSmithKline competitors including Pfizer, AstraZeneca, Sanofi-Aventis, Eli Lilly and the Japanese drug company called Takeda, which makes the diabetes drug Actos that competes directly with Avandia.
Nissen’s study has been terrific for Takeda. While the number of Avandia prescriptions plunged following the release of the study, the number of new Actos prescriptions jumped 56 percent.
Additionally, Nissen’s study was accompanied in the New England Journal of Medicine by an editorial authored by Bruce Psaty (University of Washington) and Curt Furberg (Wake Forest University) who opined that in view of the reported risks, the rationale for prescribing Avandia was “unclear.” Drs. Psaty and Furberg are no strangers to the pharmaceutical health scare game.
In the mid-1990s, they launched a scare against the class of hypertension drugs known as calcium channel blockers, or CCBs, another fierce drug company battleground. Doctors were deluged by calls from frightened patients, two patients reportedly died after they stopped taking their medication, and the value of Pfizer’s stock fell by a billion dollars in a few hours.
But the Psaty-Furberg CCB scare later turned out to be so bogus that Psaty subsequently was forced to apologize to angry colleagues at a meeting of the American Society of Hypertension.
Whether or not there’s any rival drug maker skullduggery afoot with respect to Avandia, Waxman ought to skip his usual political inquisition and view this scientifically unsound attack on Avandia — and on the peace of mind of its diabetic users — with the utmost skepticism.
Steven Milloy publishes JunkScience.com and CSRWatch.com. He is a junk science expert, and advocate of free enterprise and an adjunct scholar at the Competitive Enterprise Institute.