Context: Many adverse drug errors may be prevented through electronic order entry systems that provide decision support to physicians by screening prescriptions for dosing errors, drug-disease, drug-allergy and drug-drug interactions. The adherence to such decision aids is varied and the reasons for this variance not well understood.
Objective: To assess the feasibility and performance auto-mated drug alerts within an electronic decision support system for physician prescribing.
Methods: Drug alert data were collected from a pilot project with 30 participating general practitioners who were provided with interactive electronic prescription capabilities through a personal digital assistant (PDA).
Results: 66,642 electronic prescriptions resulted in a total of 1,869 drug alerts. The most common alert types were analysed, along with reasons for non-adherence to automated drug alerts.
Conclusions: Non-adherence to alert information appears to be associated with additional knowledge of the clinical situation, beyond that inherent in the decision support tool, for the specific patient context. Further work is required to understand how best to provide this type of support to physicians.