[Use of computerized data in pharmacoepidemiology]

Positive and negative effects of long-term therapy are sometimes difficult to highlight, for instance in gerontology or in respiratory medicine, and the use of computerized data may be a good alternative. Data collection is made of three steps. First, a study population is identified from disease criteria. All treatments used during a given study period are then extracted from computerized files; these data may be linked, when needed, to detailed patient surveys. Finally, all outcomes of the study period are retrieved, possibly using validated proxies. This method also allows for assessment of the quality of prescribing and the burden of disease. Asthma studies are used as examples.