Pharmacovigilance and the Null Hypothesis

Abstract

Over the past few years, the media has been replete with criticism of the failure of pharmacovigilance (PV) systems to prevent harm to people. In both the US and Europe this has resulted in calls for more effective PV. Indeed, on both sides of the Atlantic there is a flurry of funded activity to find better data sources, to use better analysis tools and to be more proactive in risk management. It was arguably rofecoxib (Vioxx ) that instigated the activity because of the increased incidence of myocardial infarction. Because heart attack is serious and common, and the drug was very widely used, the public health impact was great and the public outcry loud; the regulators had been much too slow in taking action. Much of the above is summarized beautifully in a review by Greener. Just after the Vioxx issue became public, Edwards wrote ‘‘The Vioxx situation was not a failure of regulation itself, neither was it an issue of data collection, nor of the quality of studies performed. It was and is a complex decision-making/communication challenge in which some improvements are possible. Making wise drug safety decisions is not easy y’’. There was an early signal about that problem, and about 6 months after the drug was launched there were warnings in the product literature. The increase in risk over background was small, even though the absolute numbers affected and the public health impact were large, and a fairly large study was needed to evaluate this. The key question was, and is, who makes the decision to do such a study, when, and who should fund and perform the study.