Antithrombotic drug-drug interaction alerts in MedChart

• Published in: Proceedings for Annual Meeting of The Japanese Pharmacological Society Vol. WCP2018; p. PO3-12-21
• Main Authors: Chin, Paul K. L., Crawford, Amanda M., Chuah, Qianyi, Doogue, Matt P.
• Format: Journal Article
• Language: English
• Published: Japanese Pharmacological Society 2018
• Subjects: antithrombotics; clinical decision support; drug-drug interactions
• ISSN: 2435-4953; 2435-4953
• DOI: 10.1254/jpssuppl.WCP2018.0_PO3-12-21

Introduction. Concomitant use of two antithrombotic medications (antiplatelets or anticoagulants) is associated with increased risk of bleeding. Public hospitals in Christchurch, NZ use MedChart;, a computerised physician order entry system that has been locally configured with prescribing alerts that trigger against antithrombotic drug-drug interactions (DDIs). Aims. To evaluate the rate of antithrombotic DDI alerts and prescriber responses to these alerts. Methods. MedChart alert data from 1 August to 31 December 2016 were extracted and rates of antithrombotic DDI alerts were determined. A subset of the alerts were reviewed in detail. Analysis of prescription changes attributed to the alert and issues contributing to alert fatigue were also performed. Results. During the study period 1,011 antithrombotic DDI alerts were recorded (mean 7 per day) corresponding to an alert rate of 48/10,000 total prescriptions. Oral anticoagulant-oral antiplatelet alerts comprised of 62% (624/1,011) of these DDI alerts. Of 280 alerts assessed, 81% (228/280) were 'clinically appropriate'. Prescribers changed antithrombotic prescriptions within 30 minutes of triggering DDI alerts on 28% (79/280) of occasions. The combination of enoxaparin and dabigatran was associated with 34 alerts, of which 88% (30/34) were 'clinically appropriate' and 74% (25/34) were associated with a change in antithrombotic prescription within 30 minutes of the alert firing. Discussion. Targeted clinical decision support can reduce high risk prescribing. However, even carefully constructed alerts targeting the highest risk prescribing has specificity considerably below 100%. It is important to assess temporally proximal changes to prescriptions following an alert, and not just focus on the decision at the point of the alert. These data will inform further development of clinical decision support.