Pharmacoepidemiology and gastroenterology: a close couple.

UNLABELLED: The discipline of pharmacoepidemiology has always been strongly linked with the problem of gastrointestinal injury caused by non-steroidal anti-inflammatory drugs, and has led to a close liaison between gastroenterology and pharmacoepidemiology. AIM: This paper discusses three important areas of pharmacoepidemiological interest relevant to gastroenterologists: confounding by indication and channeling bias, drug exposure patterns, and postmarketing surveillance studies. (i) Drug prescribing is associated with a patient's prognosis and disease status, called indication for treatment. These patient disease characteristics can drive drug channeling in risk populations and create confounded drug-effect associations. (ii) Drug exposure is the result of the 'natural' experiment of pharmacotherapy over time. Reliable information on the time-sequence of events in relation to drug exposure is required to evaluate both beneficial and adverse effects of drug therapy. (iii) Postmarketing surveillance studies provide intensive learning about drug effects when large spectra of patients with various medical backgrounds, prognoses, co-morbidity, and the like, are exposed to the drug. Information on drug use and patient outcomes in 'real life' populations is necessary to bridge normal practice experiments with randomized clinical trials.