Pharmacoepidemiology, the study of the effects of drugs in large numbers of people, is a relatively new discipline that applies the methods of epidemiology to clinical pharmacology. Premarketing clinical trials remain the only way to scientifically determine whether a drug is causally effective, yet these trials do not provide information that allows for estimates about rare, late, or off-label toxicities associated with the drug in question. Increasing concern about common and debilitating adverse effects of medications has highlighted the need for enhanced postmarketing surveillance to follow up on clinical trials. There is a further concern regarding the broader issue of the risk-benefit ratio of certain medications. Computerized databases are a good source of information on which to lay the foundations of postmarketing research. A system of cross-linked computerized medical records may better enable researchers and physicians to realistically monitor postmarketing safety and incorporate monitoring benefits. The same research could also elucidate the net public health effect of regulatory decisions. The Veterans Affairs database in the United States and PHARMO in the Netherlands may represent good models on which to base future postmarketing surveillance studies.

In this expert roundtable supplement, Brian L. Strom, MD, provides an overview of the appropriate use of large computer databases in pharmacoepidemiology studies. Philip Wang, MD, DrPH, discusses the advantages of pharmacoepidemiologic studies over clinical trials in correctly detecting drug safety issues. Francesca Cunningham, PharmD, reviews a large computer database that is currently in use within the VA healthcare system, and Ron M.C. Herings, PharmD, PhD, discusses the development of the PHARMO system in the Netherlands.