Mixed methods implementation evaluation of electronic deprescribing tools informed by behavioral science in primary care in the NUDGE-EHR pragmatic trial

Despite the promise of electronic health record (EHR) tools to support deprescribing of potentially inappropriate medications, evidence on their real‑world implementation remains limited. The NUDGE-EHR hybrid type 1 effectiveness-implementation trial evaluated 14 EHR tools, based on eight behavioral science principles, to support primary care providers (PCP) in deprescribing. The trial demonstrated that some EHR tools were more effective than others. To contextualize these findings and inform future adaptations or implementation strategies, we conducted an implementation evaluation guided by the Consolidated Framework for Implementation Research (CFIR) and the Technology Acceptance Model (TAM). We used an explanatory sequential mixed-methods design, combining structured EHR data from the trial with PCP surveys and qualitative semi-structured interviews with PCPs after trial completion. CFIR and TAM informed our data collection, coding, and interpretation. Quantitative data were analyzed descriptively, and qualitative data through thematic analysis. Alert actions that were considered “deprescribing-supportive” included sharing a patient handout, scheduling a 4-week follow-up reminder or asynchronous message, advising a follow-up appointment, cancelling the medication order, or opening the orderset. We triangulated qualitative and quantitative findings to identify barriers and facilitators to EHR tool use. Overall, 216 PCPs participated in the trial. Among 183 PCPs still practicing at trial completion, 81 completed the survey (44%) and of the 50 invited to participate in an interview, 20 agreed (40%). Overall, we found that engagement with the EHR tools was relatively low but varied by behavioral science principle. Depending on the principle, 36–97% of PCPs performed deprescribing-supportive alert actions along the deprescribing pathway. Key barriers to adopting the EHR tools included the alert timing, limited awareness of tool components, and challenges in care coordination. These barriers aligned with CFIR constructs, particularly Innovation Design, Compatibility, Motivation, and Relational Connections, which helped explain variation in tool use. Findings were supported by the TAM, as perceived usefulness and ease of use shaped the adoption of tools. EHR deprescribing tools based on behavioral science principles substantially varied in implementation, with alert timing, clinician targeting, and actionable information influencing clinician engagement. Addressing workflow, communication, and tapering-related barriers may improve uptake of deprescribing tools and reduce high-risk prescribing in older adults.