User-Directed vs Interruptive Real-Time Benefit Tools for Medication Changes in Primary Care.

Importance: Real-time benefit tools (RTBTs) in the electronic health record (EHR) provide medication cost estimates and lower-cost alternatives. RTBT use leads to improved adherence and lower costs among patients with chronic conditions; however, clinician use is low. Objective: To assess RTBT display rate and medication order change frequencies associated with the redesign of an RTBT from user-directed (clinicians must actively open it) to interruptive (automatically displays when a cost threshold is met). Design, Setting, and Participants: This retrospective cohort study used EHR data collected in primary care at Duke Health, a large academic medical center, between January 1, 2022, and December 31, 2023. Duke Health redesigned their RTBT in December 2022. Participants included adults with multimorbidity who were prescribed at least 1 medication and had an RTBT that could display. Exposure: Binary indicator denoting the RTBT period: user-directed (2022) vs interruptive (2023). Main Outcomes and Measures: Outcomes were the percentage of encounters in which the RTBT was displayed and the percentage of encounters in which the clinician made a medication order change. Analyses were conducted using generalized linear models adjusting for patient and clinician characteristics. Results: The analytic cohort included 39 788 patients with similar characteristics across 2022 and 2023. The mean (SD) age was 67.3 (13.2) years in 2022 and 67.9 (13.2) years in 2023. The mean (SD) number of chronic conditions was 3.6 (1.7) in 2022 and 3.5 (1.6) in 2023. Most patients were female (60.6% in 2022 and 60.9% in 2023) and non-Hispanic White (61.3% in 2022 and 61.4% in 2023). Patients were involved in 156 405 encounters with 758 primary care practitioners. RTBT display frequency increased from 1.3% of encounters in 2022 to 13.5% in 2023. In adjusted analysis across all encounters, the interruptive design was associated with significant increases in RTBT display frequency (risk ratio [RR], 9.62 [95% CI, 8.16-11.35]) and in medication changes (RR, 8.37 [95% CI, 7.08-9.91]). In the subgroup of encounters in which the RTBT was displayed, the frequency of medication order changes was similar across both periods (RR, 0.93 [95% CI, 0.84-1.04]). Conclusions and Relevance: In this retrospective cohort study of RTBT use in primary care, the change in RTBT design from a user-directed configuration to an interruptive configuration was associated with an increase in RTBT display rates. The lack of a corresponding decline in medication order changes suggests that clinicians continued to pay attention and responded similarly to the RTBT after the design change.