Clinician burnout is reshaping emergency medicine in ways that compromise both provider wellbeing and patient care quality. Documentation has long been identified as the single most time-consuming non-clinical task emergency physicians face, so any intervention that measurably reclaims that time carries real significance — especially at scale across high-volume departments.
This 13-month retrospective cohort study conducted at a tertiary academic emergency department analyzed 10,344 patient encounters managed by 100 attending physicians across four distinct care settings. Using electronic health record audit log data — a more objective measurement method than self-report — researchers applied mixed-effects linear models with physician-level random intercepts to isolate the effect of ambient AI scribe use from confounding patient and encounter characteristics. The primary finding: encounters in which the ambient AI scribe was used were associated with a 72.6-second reduction in on-shift documentation time compared to encounters where it was not used, even after adjusting for physician and patient-level variables. Scribe use was entirely at physician discretion, creating natural within-physician variation that strengthened causal inference without a formal randomized design.
This study addresses a meaningful gap. Earlier ambient AI scribe evaluations were constrained by short observation windows — often weeks rather than months — and low or inconsistent adoption rates that limited statistical power. A 13-month window captures seasonal patient acuity variation and reflects more stable, real-world utilization patterns. That said, the retrospective design and physician self-selection of when to use the tool introduce selection bias: physicians may be more likely to activate the scribe during less cognitively demanding encounters, potentially inflating the apparent time savings. The 72.6-second reduction per encounter, while modest individually, aggregates meaningfully across a high-volume ED — potentially reclaiming hours of documentation burden weekly per physician. Whether these time savings translate into reduced burnout scores, improved patient throughput, or better note quality remains an open and critical question for subsequent prospective research.