Monitoring Physicians' Prescription Patterns on Electronic Health Record: The Prescription Pattern around Clinical Event (PACE) Algorithm

Electronic health records (EHRs) have gained attention as a valuable data source for medical research, as its adoption rate continues to rise. However, no method for the monitoring of physicians' prescription patterns has been established. Since EHR maintain all prescription data as well as clinical events that occur during the care of patients, we hypothesized that a physician's prescription pattern can be monitored from EHR. In this study, we developed a novel algorithm named PACE, Prescription pattern Around Clinical Event. This algorithm analyzes distribution of the prescription of specific drugs around the time of a clinical event. In the proof of concept study, prescription changes with regard to hyperkalemia were well represented by the algorithm, and the observed patterns well correlated with the physician's knowledge on hyperkalemia (Cohen's kappa, 0.457-0.653). We expect that this algorithm can be used to monitor the guideline adherence of physicians.