Medication prescribing errors: a pre- and post-computerized physician order entry retrospective study.

Background The computerization of prescriptions with a computerized physician order entry contributes to securing the error-free drug supply, but is not risk-free. Objective: To determine the impact of a computerized physician order entry system on prescribing errors immediately after its implementation and 1 year later. Setting The Cardiology and Diabetology Departments at Toulouse University Hospital, France. Method The prescriptions were analysed by pharmacists over three 30-day periods for 3 consecutive years (N: computerization period, N - 1, N + 1). For each identified error, the prescriber was informed by a pharmaceutical intervention. The pharmaceutical interventions were counted and arranged according to the classification by the French Society of Clinical Pharmacy. Their average numbers and clinical impacts were compared for each period using t-tests and Kruskal-Wallis tests. Main outcome measure The average numbers of pharmaceutical interventions. Results In total, 12.1 pharmaceutical interventions per 100 patient days were done during the N - 1 period, 14.1 during N and 9.6 during N + 1. Among those, 3.6 (N) and 2.1 (N + 1) were related to the computerization itself, and 10.5 (N) and 7.5 (N + 1) were not. The average number of computerization-related pharmaceutical interventions significantly decreased from N to N + 1 (p = 0.04). The average number of classic interventions decreased from N - 1 to N + 1 (p = 0.02). The clinical impacts of the computerization related errors were similar to those of other errors. Conclusion The implementation of the computerized physician order entry induced the appearance of specific computerized-related errors, but the number of classic errors decreased. The entry-system related errors were not more severe than other errors, and the number decreased after 1 year.