Pharmacovigilance is all about the safety of drugs in their conditions of normal routine use. It does involve collection and analysis of information about drugs as they are used in a community. No longer is the major focus that of the randomised controlled clinical trial where a well-defined subset of the population is exposed under carefully controlled circumstances to a medicine of interest and followed for a defined duration thereafter. We now enter the area of observational studies with all the problems in interpretation that such observational studies entail. It is important to realise that the interpretation of observational data is much more complex than the interpretation of randomised controlled clinical trials. Such studies are, by their very nature, full of incomplete information and subject not only to controllable biases and distortions, but some are strictly unable to be interpreted because of insurmountable distortions. The various types of studies involved have been covered in earlier chapters in this book. They fall broadly into three categories:
1. The anecdotal study in which reports of suspected problems are solicited and analysed to see if they can give hints about possible drug-related problems, exemplified by the spontaneous reporting schemes.
2. More detailed observational studies, but still without appropriate comparator groups who are not exposed to the medicine of interest, e.g. ad hoc follow-up studies.
3. Controlled studies, including follow-up and case-control studies.
Some 15 years ago, giving the keynote address on pharmacoepidemiology and public health policy at the International Society of Pharmacoe-pidemiology meeting in Minneapolis, I made several points about this subject which I think are worthwhile repeating here. These are
1. It is the duty of pharmacoepidemiologists to ensure that spontaneous reports of suspected adverse reactions are used wisely in the full knowledge of their substantial limitations.
2. It is our duty as pharmacoepidemiologists to ensure that other sources of information are available which can be interpreted in a reasonably rapid time frame. Good data in six years is no substitute for usable data in six months or less.
3. Pharmacoepidemiology will not prosper if it develops as an intellectual subject which plots the history of why drugs fall from favour. It must be a live and contemporary subject, providing answers to current problems of drug use and drug safety in real time.
These aphorisms are as relevant today as they were when first spoken. They apply across the board to all types of studies. Thankfully we have made progress in the intervening years, albeit not as much as we would have liked.
Was this article helpful?