Medical Scribe and Ambient Artificial Intelligence Impact on Emergency Physician Documentation Burden and Clinical...

Emergency physicians experience substantial documentation burden, contributing to physician burnout. Human scribes reduce documentation workload but are expensive and pose staffing challenges. Ambient artificial intelligence (AI) scribes offer a potential alternative by automating note generation from clinician-patient conversations using AI. We compared ambient AI and human scribes against encounters with no scribe on emergency physicians' documentation time and clinical productivity.

This retrospective cross-sectional observational study evaluated emergency department encounters from January 2025 to September 2025 at 4 hospitals within a large integrated health care system. Encounters were categorized as ambient AI scribe, human scribe, or no scribe. Patient demographics, documentation time, work relative value units (wRVUs), and shift data were extracted. Attending documentation time was modeled using median quantile regression with standard errors clustered at the physician level and controlling for encounter-level variables. Clinical productivity measured as total wRVUs per shift hour was modeled using generalized estimating equations adjusting for physician and shift-level variables.

Among 198,178 emergency department encounters, 8,489 (4.3%) used ambient AI scribes, 15,947 (8.0%) used human scribes, and 173,742 (87.7%) had no scribe. Median patient age was 49 years (interquartile range 30 to 68 years), and 53.1% were female. Compared with encounters with no scribe, ambient AI scribes were associated with a 1.6-minute reduction in adjusted median attending documentation time per note (95% confidence interval 0.3 to 2.9), whereas human scribes were associated with a 3.3-minute reduction (95% confidence interval 2.3 to 4.3). Total wRVUs per shift hour did not differ among groups.

Ambient AI and human scribes were associated with reduced physician documentation time. Clinical productivity did not differ between study groups.