« Geek Test | Main | Howard Dean earned it »

Pfizer v. Bristol-Meyers-Squibb

From an article in the New York Times reporting a study which shows Pfizer's cholesterol-lowering drug Lipitor might be superior to Bristol-Meyers-Squibb's version Pravachol.

Pfizer sponsored the study, but Dr. Nissen, who prides himself on his independence from financial conflicts of interest, insisted that he control the study and its data analysis, and had the right by contract to publish the results whether positive or negative for Pfizer.

Whew; that's a relief. What's sad is that we can't assume the kind of independence Dr. Nissen claims to have. The pharmaceutical industry has corrupted the process of biomedical research. This analysis and the study it describes gives a flavor of how this corruption works:

Typically, when a firm is studying a therapeutic in human clinical trials, it enters into a contract with an academic institution to perform the studies, and it is these contracts that the report criticizes. The survey found that in only 10% of contracts did site researchers have a say in how data was collected and monitored, and in only 5% did they influence how the data was analyzed and interpreted. In addition, fewer than 1% of contracts guaranteed that results would be published and that an independent committee would control publication.

These survey results describe why researchers like Dr. Nissen feel the need to proclaim their innocence when announcing significant findings. The corporate sponsors of clinical trials, the ones who stand to gain from positive results, are often the ones who do the analysis and interpretation of data. If the results aren't favorable to their commercial position, they aren't published at all. From the New England Journal of Medicine:

As CROs and academic medical centers compete head to head for the opportunity to enroll patients in clinical trials, corporate sponsors have been able to dictate the terms of participation in the trial terms that are not always in the best interests of academic investigators, the study participants, or the advancement of science generally. Investigators may have little or no input into trial design, no access to the raw data, and limited participation in data interpretation. These terms are draconian for self-respecting scientists, but many have accepted them because they know that if they do not, the sponsor will find someone else who will. And, unfortunately, even when an investigator has had substantial input into trial design and data interpretation, the results of the finished trial may be buried rather than published if they are unfavorable to the sponsor's product. Such issues are not theoretical. There have been a number of recent public examples of such problems, and we suspect that many more go unreported.

While some medical journals have taken steps to ensure that they won't become the patsies of the corporate research sponsors, the medical schools and teaching hospitals haven't been as aggressive. Until they do, we should receive studies like the Lipitor/Pravachol comparison with a healthy dose of suspicion.


Bias, both from the testers and those who fund them, has been destroying the potential for this country to have a serious debate on the issue of medical marijuana for decades. Any time I'm ever presented with any study that shows marijuana to be a danger to human health, over and above the simple act of smoking it, I'm instantly skeptical, and can usually find some flaw in either the test itself or some bias in those conducting or funding it.