Background: Clinical decision support systems (CDSS) can help identify drug-related problems (DRPs). However, the alert specificity remains variable. Defining more relevant alerts for detecting DRPs would improve CDSS.

Aim: Develop electronic queries that assist pharmacists in conducting medication reviews and an assessment of the performance of this model to detect DRPs.

Method: Electronic queries were set up in CDSS using "triggers" from electronic health records: drug prescriptions, laboratory values, medical problems, vital signs, demographics. They were based on a previous study where 315 patients admitted in internal medicine benefited from a multidisciplinary medication review (gold-standard) to highlight potential DRPs. Electronic queries were retrospectively tested to assess performance in detecting DRPs revealed with gold-standard. For each electronic query, sensitivity, specificity, positive and negative predictive value were computed.

Results: Of 909 DRPs, 700 (77.8%) were used to create 366 electronic queries. Electronic queries correctly detected 77.1% of DRPs, median sensitivity and specificity reached 100.0% (IQRs, 100.0%-100.0%) and 99.7% (IQRs, 97.0%-100.0%); median positive predictive value and negative predictive value reached 50.0% (IQRs, 12.5%-100.0%) and 100.0% (IQRs, 100.0%-100.0%). Performances varied according to "triggers" (p < 0.001, best performance in terms of predictive positive value when exclusively involving drug prescriptions).

Conclusion: Electronic queries based on electronic heath records had high sensitivity and negative predictive value and acceptable specificity and positive predictive value and may contribute to facilitate medication review. Implementing some of these electronic queries (the most effective and clinically relevant) in current practice will allow a better assessment of their impact on the efficiency of the clinical pharmacist.