The linkage of administrative health care utilization databases offers advantages for postmarketing surveillance. The data have already been collected , and, although such data systems are not ideal for research purposes and need careful evaluation to ensure adequate data quality , they usually do not suffer from problems arising from a significant nonresponse. Saskatchewan has been recognized as having some of the most comprehensive, linkable health care utilization data in the world. The datafiles are considered to be complete and reasonably valid.

The feasibility of using the Saskatchewan health care utilization datafiles as a data source for an acute adverse event signalling scheme has been evaluated using information about new patients dispensed either piroxicam or sulindac . It was concluded that the use of administrative health care utilization datafiles in a systematic scheme to signal potential ADRs is a feasible and practical prospect.

However, it was considered important to evaluate the methodbeyond a single class of . Therefore, information on patients dispensed either alprazolam or lorazepam was obtained in the same project, and the use of the datafiles to signal potential ADRs following the prescription of these drugs was assessed. We selected benzodiazepines because they are a widely used group of drugs whose adverse effects have been examined in numerous studies.