Prescription errors before and after introduction of electronic medication alert system in a pediatric emergency department

Acad Emerg Med. 2015 Jun;22(6):714-9. doi: 10.1111/acem.12678. Epub 2015 May 21.


Objectives: Prescription errors occur frequently in pediatric emergency departments (PEDs).The effect of computerized physician order entry (CPOE) with electronic medication alert system (EMAS) on these is unknown. The objective was to compare prescription errors rates before and after introduction of CPOE with EMAS in a PED. The hypothesis was that CPOE with EMAS would significantly reduce the rate and severity of prescription errors in the PED.

Methods: A prospective comparison of a sample of outpatient, medication prescriptions 5 months before and after CPOE with EMAS implementation (7,268 before and 7,292 after) was performed. Error types and rates, alert types and significance, and physician response were noted. Medication errors were deemed significant if there was a potential to cause life-threatening injury, failure of therapy, or an adverse drug effect.

Results: There was a significant reduction in the errors per 100 prescriptions (10.4 before vs. 7.3 after; absolute risk reduction = 3.1, 95% confidence interval [CI] = 2.2 to 4.0). Drug dosing error rates decreased from 8 to 5.4 per 100 (absolute risk reduction = 2.6, 95% CI = 1.8 to 3.4). Alerts were generated for 29.6% of prescriptions, with 45% involving drug dose range checking. The sensitivity of CPOE with EMAS in identifying errors in prescriptions was 45.1% (95% CI = 40.8% to 49.6%), and the specificity was 57% (95% CI = 55.6% to 58.5%). Prescribers modified 20% of the dosing alerts, resulting in the error not reaching the patient. Conversely, 11% of true dosing alerts for medication errors were overridden by the prescribers: 88 (11.3%) resulted in medication errors, and 684 (88.6%) were false-positive alerts.

Conclusions: A CPOE with EMAS was associated with a decrease in overall prescription errors in our PED. Further system refinements are required to reduce the high false-positive alert rates.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adolescent
  • Child
  • Child, Preschool
  • Decision Support Systems, Clinical / statistics & numerical data
  • Drug Dosage Calculations
  • Drug-Related Side Effects and Adverse Reactions
  • Emergency Service, Hospital / statistics & numerical data*
  • Female
  • Hospitals, Pediatric / statistics & numerical data*
  • Humans
  • Infant
  • Infant, Newborn
  • Male
  • Medical Order Entry Systems / statistics & numerical data*
  • Medication Errors / prevention & control*
  • Medication Errors / statistics & numerical data*
  • Physicians
  • Prospective Studies