Characterization of prescribing errors in an internal medicine clinic

A pilot study was conducted to characterize the epidemiology of prescribing errors, comparing those that occurred pre- to postimplementation of an electronic prescribing system; this article describes the results of the study during the preimplementation phase, when a handwritten prescription process was still in place.

A retrospective review of 1411 prescriptions that were handwritten during a five-month time frame was used to identify and characterize medication errors and potential medication errors. The review was conducted in an internal medicine clinic in a large health system that was preparing to implement an electronic prescribing system. The first phase was the implementation of a basic system-one that facilitated the writing of a more complete and legible prescription. The second phase consisted of adding more sophisticated clinical decision support (CDS) capabilities. Three data sources were reviewed: the handwritten prescription, the electronic health record and the prescription as it had been entered into the pharmacy computer system. Almost 28% of the prescriptions evaluated contained one or more errors or potential errors. Over 90% of the errors were potential errors. Only 0.2% of the errors caused patient harm. Non-clinical errors (illegibility, missing information, wrong dose) may be affected by a basic electronic prescribing system, and clinical errors (drug-disease interaction, contraindication of a drug) may be affected only when more sophisticated levels of CDS programming are added.

Potential prescribing errors occurred frequently but few reached the patient or caused harm. The most severe errors were those that may be reduced by the implementation of an electronic prescribing system with CDS capabilities.