Screening for medication errors using an outlier detection system.

Medication errors continue to occur despite implementation of computerized provider order entry and clinical decision support systems. This study sought to assess whether medication error alerts might have a greater impact on mitigating such errors if they were generated based on outlier detection screening. Researchers analyzed data from the electronic health records of 747,985 outpatients to identify outliers that might indicate a medication error. They then chose 300 charts from the 15,693 resulting alerts. The charts were reviewed using a coding system to evaluate the utility of the alerts generated. About 75% of the chart-reviewed alerts created by the screening system identified possible medication errors. The authors suggest that using this type of outlier detection screening to generate alerts might improve existing clinical decision support systems' ability to mitigate medication errors. A prior WebM&M commentary discussed an incident involving an electronic prescribing error.