To evaluate error reduction and behavior modifications with the use of a Web-based controlled substance prescription writer that included weight-based dosing logic and alerts.Methods:
This is retrospective study of a Web-based controlled substance prescription writer that was deployed in November 2005 for an urban tertiary care pediatric hospital. Participants included all prescribers who wrote discharge or outpatient narcotic prescriptions for patients between birth and 18 years from December 2005 to October 2006. Measured outcomes were number of prescriptions, number of attempts exceeding maximum dose, information on alerts that were generated and on how/if these alerts resulted in prescription alterations, and number of prescriptions abandoned. Data were analyzed using frequencies and χ2 analyses.Results:
There were 266 prescribers. The most common user groups were pediatric (112; 42%), orthopedic (38; 14%), and general surgical (33; 12%) house staff. There were 4995 prescription attempts, resulting in 3239 actual prescriptions of which 75% were for oxycodone (297 prescriptions with alerts/2942 prescriptions without alerts). The 713 prescriptions attempts with alerts generated 1007 alerts. Seventeen percent of the alerts were hard alerts, resulting in zero prescriptions. A prescription attempt with an alert was abandoned 58% of the time compared with 31% of the time if no alert were generated (P < 0.001).Conclusions:
This Web-based controlled substance prescription writer prevented analgesic medication errors by alerting users that their doses exceeded hard limits for weight-based dosing. The use of alerts changed prescriber behavior and decreased the potential for medication errors as demonstrated.