Intended for healthcare professionals


Sentinel systems are needed for long term adverse drug reactions

BMJ 2005; 330 doi: (Published 27 January 2005) Cite this as: BMJ 2005;330:255
  1. Torbjörn Callréus (torbjorncallreus{at}, research fellow in pharmacoepidemiology
  1. Institute of Public Health, IPH Research Unit of Clinical Pharmacology, University of Southern Denmark, Winslowparken 19, 3rd floor, DK-5000 Odense C, Denmark

EDITOR—In the wake of the withdrawal of rofecoxib, the limitation of the current system of pharmacovigilance is once again under discussion. With the system's focus on spontaneous reports, unexpected and rare reactions occurring shortly after exposure are usually detected with reasonable sensitivity. However, adverse effects are not readily identified if they occur after long time exposure or manifest themselves as an increase of a common disease. The cardiovascular events caused by rofecoxib belong to the latter group.

Although attractive in theory, the proposal of Dieppe et al to require independent, large scale, randomised trials before definitive drug licensing, is associated with some problems.1 As Oakley points out,2 the sample sizes needed to investigate rare adverse reactions in traditional randomised controlled trials would increase costs for clinical development and ultimately raise drug prices. Furthermore, over time, treatment switches and discontinuations often transform large and simple randomised trials into observational cohort studies. Instead, Oakley wants to promote the use of cost effective case-control studies. In selected cases, when a clear hypothesis exists about the nature of the reaction and some idea of the suspected drugs, case-control studies can provide important information. However, as a strategy for continuous post-marketing surveillance of unexpected effects, they are not so suitable.

In recent years large databases originating from routine healthcare procedures have become widely available. Although collected for other purposes, the information in these data sources could have an important role as a cost effective sentinel system for long term adverse drug reactions. However, the abundance of information in healthcare databases requires techniques for research to move beyond traditional epidemiological study designs, inherited from an era when data collection was expensive and yielded only a few facts for each study subject.3


  • Competing interests As a former employee of AstraZeneca, TC has dividends that currently are in the process of being divested.


  1. 1.
  2. 2.
  3. 3.
View Abstract

Log in

Log in through your institution


* For online subscription