Comparative evaluation of three clinical decision support systems: prospective screening for medication errors in 100 medical inpatients
- First Online:
- Cite this article as:
- Fritz, D., Ceschi, A., Curkovic, I. et al. Eur J Clin Pharmacol (2012) 68: 1209. doi:10.1007/s00228-012-1241-6
- 377 Views
Clinical decision support systems (CDSS) are promoted as powerful screening tools to improve pharmacotherapy. The aim of our study was to evaluate the potential contribution of CDSS to patient management in clinical practice.
We prospectively analyzed the pharmacotherapy of 100 medical inpatients through the parallel use of three CDSS, namely, Pharmavista, DrugReax, and TheraOpt. After expert discussion that also considered all patient-specific clinical information, we selected apparently relevant alerts, issued suitable recommendations to physicians, and recorded subsequent prescription changes.
For 100 patients with a median of eight concomitant drugs, Pharmavista, DrugReax, and TheraOpt generated a total of 53, 362, and 328 interaction alerts, respectively. Among those we identified and forwarded 33 clinically relevant alerts to the attending physician, resulting in 19 prescription changes. Four adverse drug events were associated with interactions. The proportion of clinically relevant alerts among all alerts (positive predictive value) was 5.7, 8.0, and 7.6%, and the sensitivity to detect all 33 relevant alerts was 9.1, 87.9, and 75.8% for Pharmavista, DrugReax and TheraOpt, respectively. TheraOpt recommended 31 dose adjustments, of which we considered 11 to be relevant; three of these were followed by dose reductions.
CDSS are valuable screening tools for medication errors, but only a small fraction of their alerts appear relevant in individual patients. In order to avoid overalerting CDSS should use patient-specific information and management-oriented classifications. Comprehensive information should be displayed on-demand, whereas a limited number of computer-triggered alerts that have management implications in the majority of affected patients should be based on locally customized and supported algorithms.