In 2000 Vioxx promised arthritis sufferers a way to ease pain with less trauma to the stomach than naproxen, an older and cheaper drug in such products as Aleve. Merck, the respected drug giant that created Vioxx, funded the VIGOR study, headed by Claire Bombardier, Director of Rheumatology at the University of Toronto, and this study was published by the prestigious New England Journal of Medicine in November 2000 (Claire Bombardier, et al. “Comparison of Upper Gastrointestinal Toxicity of Rofecoxib and Naproxen in Patients with Rheumatoid Arthritis,” NEJM Volume 343:1520-1528.) The study revealed that Vioxx users had four times the rate of heart attacks as naproxen users. The higher risk came about, the study explained, because naproxen protected the heart, not because Vioxx hurt it. But the authors have been accused of withholding evidence about the drug's heart risks by editors of the New England Journal of Medicine. It was found that three heart attacks out of a total of 20 were omitted from the data just two days before the article was published. Were they included, the study would have revealed an even higher risk of heart attacks among Vioxx users. Merck stopped selling Vioxx last year because another study found that patients taking the drug for more than 18 months doubled their risk of heart attacks and strokes. Since then more than 7,000 people have sued Merck, blaming Vioxx for injuries or deaths. Recently, Dr. Gregory D. Curfman, Executive Editor of the NEJM, said that he should have been more aggressive in challenging the research when it was first submitted for publication. An editorial by Gregory D. Curfman, Stephen Morrissey, and Jeffrey M. Drazen, the editors of NEJM, appears in print in the today's issue of NEJM.
This incident is the latest to raise troubling questions about selective disclosure of data in industry sponsored clinical trials. To be sure, there will always be risks and contraindications with any medication, but health care regulators, providers, and patients can't assess those risks if they aren't given all the information. (See the latest regulatory contibution, note this registry is voluntary.) Important questions that emerge in this case are ones that Jerry Avorn asks us to think of as “pharmacoepistemological”. Epistemological questions are those that involve the nature and the scope of knowledge, truth, and justification of claims. Thus pharmacoepistemological questions have to do with how it is that we-- regulators, patients, and providers-- know that a drug is effective and safe enough to use. How do we --regulators, patients, and providers--get drug information and how can we determine if the information is trustworthy? It is the norm that the most trustworthy sources are scientific articles written in prestigious peer-reviewed journals by experts in the field who work at accredited institutions of higher learning. But in this procedure there is the assumption that people who submit the manuscripts will be telling the truth and not fudging data.
The fraudulent data announcing the creation of human embryonic stem cell lines that were submitted to and published by the world respected journal Science again reminds us that we get to know about empirical data as they are presented to us by trustworthy sources. But all along the chain of transmission of data, from lab investigator to department chair to external funding agency to the editorial board of the journal, we find human beings. And human beings have interests. Bias and interest are not the same thing. Bias can occur because of some interest or other that is contrary to the interest of the truthful and full transmission of data. The problem is not that there are values and interests inherent in the scientific project. This in itself does not make science “subjective." The values of truth telling and full disclosure are essential to good science. Thus, it is the kinds of values and human interests that affect the process that invite scrutiny. We have to ask what values are being advocated and whose interests are being served.