OBJECTIVES: Drug-related problems (DRPs) are events or circumstances involving drug therapy that actually or potentially interfere with desired health outcomes. This study tested the applicability of clinical decision support software in identifying and managing DRPs among cardiovascular surgery inpatients. METHODS: Two clinical pharmacologists attended ward rounds on a low-dependency cardiovascular surgery ward every 2 weeks over a 7-month period. Three hundred and three patients were assessed. On average, patients received 17 scheduled and 'as required' medicines. DRPs were identified 'manually' via assessment of electronic prescription charts and patient records and 'electronically' using clinical decision support software (Pharmavista). The numbers of alerts for optimizing medication safety generated by the two methods were compared. RESULTS: Manual checking identified 346 DRPs leading to 346 alerts in 201 patients (overall 1.1 alerts/patient). Relevant interactions accounted for 44% of DRPs detected by clinical pharmacologists. Clinical decision support software, which could only report interactions, however, generated 1,370 alerts (average 4.5 alerts/patient). Only 147 (11%) drug-drug interaction alerts were identical to those identified by manual checking; the remaining 89% were considered not clinically relevant. CONCLUSIONS: Compared to identification of DRPs by clinical pharmacologists, the clinical decision support software performed poorly due to over-alerting and inability to assess for problems not caused by drug-drug interactions.
OBJECTIVES: Drug-related problems (DRPs) are events or circumstances involving drug therapy that actually or potentially interfere with desired health outcomes. This study tested the applicability of clinical decision support software in identifying and managing DRPs among cardiovascular surgery inpatients. METHODS: Two clinical pharmacologists attended ward rounds on a low-dependency cardiovascular surgery ward every 2 weeks over a 7-month period. Three hundred and three patients were assessed. On average, patients received 17 scheduled and 'as required' medicines. DRPs were identified 'manually' via assessment of electronic prescription charts and patient records and 'electronically' using clinical decision support software (Pharmavista). The numbers of alerts for optimizing medication safety generated by the two methods were compared. RESULTS: Manual checking identified 346 DRPs leading to 346 alerts in 201 patients (overall 1.1 alerts/patient). Relevant interactions accounted for 44% of DRPs detected by clinical pharmacologists. Clinical decision support software, which could only report interactions, however, generated 1,370 alerts (average 4.5 alerts/patient). Only 147 (11%) drug-drug interaction alerts were identical to those identified by manual checking; the remaining 89% were considered not clinically relevant. CONCLUSIONS: Compared to identification of DRPs by clinical pharmacologists, the clinical decision support software performed poorly due to over-alerting and inability to assess for problems not caused by drug-drug interactions.