Accuracy of pharmacy benefit manager medication formularies in an electronic health record system and the Epocrates mobile application.

Background: Physicians commonly use formulary medication coverage information generated by electronic heath records (EHRs) and the Epocrates mobile drug database application when making medication selection decisions. Nonformulary selections may lead to higher out-of-pocket patient costs and nonadherence with prescribed regimens. Nonformulary selections also contribute to higher overall health plan spending. However, the accuracy of these systems compared with actual insurance coverage is not known. 

Objective: To assess the accuracy of formulary status icons generated by an EHR system and the Epocrates mobile application for patients with Alabama Medicaid and Blue Cross Blue Shield of Alabama (BCBS), the primary insurance providers in the state of Alabama.

Methods: Patients of all ages who had a chart review performed at the outpatient family medicine or pediatric clinics at the University of Alabama at Birmingham Huntsville Medical Regional Campus from May to October 2013 were included in this retrospective analysis. Patients who were not insured by either Alabama Medicaid or BCBS were excluded. Patients who did not have new medications added at the time of the visit were also excluded. For each medication prescribed, the formulary status provided in the EHR system and Epocrates was compared with the actual Medicaid and BCBS formularies published online, and the accuracy of the 2 databases was determined. 

Results: A total of 1,529 medication records were analyzed. The EHR and Epocrates provided accurate formulary information for 93.1% and 89.4% of medications, respectively. Formulary information generated by the EHR was 96.3% accurate for Medicaid patients and 80.1% accurate for BCBS patients. Epocrates was 88.2% accurate for Medicaid patients and 94.4% accurate for BCBS patients. A total of 936 medication records from the pediatric clinic were analyzed, and the majority of these patients (88.4%) had Medicaid insurance. In this population, the EHR was more accurate (96.9%) than Epocrates (86.6%). Of the small number of pediatric medication records (n = 109) associated with patients who had BCBS, Epocrates was more accurate (92.7%) than the EHR (83.5%). In family medicine, 593 medication records were analyzed. Again, for Medicaid patients, the EHR was more accurate (95.3%) than Epocrates (91.5%). For those with BCBS, Epocrates was more accurate (95.3%) than the EHR (78.2%). When over-the-counter (OTC) medications (n = 232) were analyzed separately from prescription medications, it was found that overall the EHR was 90.5% accurate, and Epocrates was 41.4% accurate. It is important to note that when only prescription medications were analyzed, the accuracy rates were high (93.6% for the EHR and 98.0% for Epocrates).

Conclusions: Formulary information generated by the EHR system and the Epocrates mobile application is a useful tool for physicians when prescribing medications, but neither source is completely accurate. Prescribers should be particularly cautious when making OTC formulary decisions using the Epocrates mobile application, since the formulary information provided for these medications was found to be the least accurate.