The nature of the scientific evidence leading to drug withdrawals for pharmacovigilance reasons in France

Abstract

Introduction: Because of design, objectives and number of included subjects, clinical studies are insufficient to assess the safety of new drugs. Sometimes, serious adverse drug reactions (ADRs) led to withdrawal of the drug from the market after their approval. The objective of our study was to determine the scientific evidences leading to drug withdrawal for pharmacovigilance reasons in France.

Methods: Data coming from French Health Products Safety Agency, literature and Toulouse Pharmacovigilance Center allowed to identify all drugs withdrawn from the French market for pharmacovigilance reasons from 1998 to 2004. We classified data according to their study design (Randomized Clinical Trial [RCT], case serie or case report, case-control study, cohort study, observational study, animal study), the organ/system affected and the type of ADR.

Results: A total of 21 drugs were withdrawn for safety reasons between 1998 and 2004 in France. The most frequent ADRs were hepatic (n = 7), cardiovascular (n = 4) or neurological (n = 3) ones. Eleven withdrawals were due to type-B ('unexpected') reactions (52%). For 19 out of 21 drugs, scientific evidence leading to drug withdrawal came from spontaneous case reports (or case series). Among these, case reports were the sole evidence in 12 cases. Withdrawals were based on evidence from case reports in combination with case-control or cohort study in four cases, in combination with observational study in two cases or in combination with animal study in two other cases. In only one case, a RCT supported the decision.

Conclusions: This study underlines the importance of spontaneous case reports in detecting signals and supporting withdrawal of drug for pharmacovigilance reasons in France. Health authorities suffer from lack of comparative data resource. In this perspective, a pharmaco-epidemiological population-based database could represent a helpful tool to both generate and test safety hypotheses.