Rethinking Opioid Prescribing at Surgical Discharge: User-Centered Approach to Clinical Decision Support in the Electronic Health Record

Background: Electronic health record (EHR) clinical decision support (CDS) can individualize postoperative opioid prescriptions using inpatient consumption, but clinician acceptance may not translate into guideline-concordant prescribing. We evaluated barriers, adoption, and prescribing fidelity following user-centered CDS redesign. Study Design: We conducted a single-institution, mixed-methods, two-phase pilot at an academic health system. Phase 1 used patient-level randomization between procedure-based and oral morphine equivalent (OME)-based recommendations. Following a 7-month pilot, 20 clinicians participated in semi-structured interviews analyzed using the Sociotechnical Domains Framework. Feedback informed CDS redesign and a second pilot (February–July 2026) across three general surgery teams. Phase 2 outcomes included adoption (CDS order-set use) and fidelity (prescriptions within ±7.5 OME of expected daily requirements). Results: Among 215 Phase 2 encounters, 143 (66.5%) received discharge opioid prescriptions. CDS adoption occurred in 48 of 143 encounters (33.6%), but only 11 of 143 (7.7%) achieved end-to-end fidelity. Among CDS users, 11 of 48 prescriptions (22.9%) matched expected requirements; 25 of 37 nonmatching prescriptions (68%) exceeded them. Among 72 opioid-free discharges, 61 (84.7%) met appropriateness criteria. Phase 1 engagement occurred in 3 of 32 intervention encounters (9.4%). Interviews identified alert fatigue, prescribing habits, rigid taper instructions, fragmented clinical information, and service-level variability as barriers. Phase-specific measures were not directly comparable. Conclusions: Despite clinician acceptance and user-informed redesign, CDS adoption was limited and consumption-concordant prescribing remained uncommon. Measuring adoption alone substantially obscured implementation fidelity. Effective postoperative opioid stewardship may require sustained clinician engagement, workflow integration, and prescribing feedback beyond EHR interface optimization.