Longitudinal effects ambient AI scribe use on documentation burden and financial productivity: A quasi-experimental study
Journal:
medRxiv
Published Date:
Jan 14, 2026
Abstract
BackgroundArtificial intelligence (AI) scribes have the potential to reduce documentation burden. Previous studies have mostly relied on aggregated, vendor-provided (e.g., Epics Signal) outcome measures, potentially obfuscating the true effect of AI scribes on documentation burden.
MethodWe conducted a longitudinal, quasi-experimental study with an interrupted time series design to investigate the effect of an AI scribe on outcomes related to note writing time, time outside of scheduled hours (TOSH), pajama time (530PM to 7AM), note length, note closure within 24hours, and work Relative Value Units (wRVU). For each primary care clinician (attending physician or advance practice practitioner) provisioned AI scribe access between Jan 5, 2024-Oct 31, 2025, date of access was considered as day zero; outcome measures were calculated per patient encounter for 150 days prior (pre-period) and after (post-period) day zero.
Results220 primary care clinicians (Mean age=43.7, 70.9% females; 56.4% physicians) who worked in 36 clinics, conducting 314,845 patient encounters were included. All outcomes changed from day zero through day 150. At 150 days after AI scribe initiation, compared to the pre-period, there was a 15% decrease in note writing time (Incidence Rate Ratio, IRR=0.85, 95%CI [0.82, 0.87]), 18% decrease in pajama time (0.82, 95%CI [0.73, 0.91]), 13% decrease in TOSH (0.87, 95%CI [0.77, 0.99]) and 2% increase in total wRVU (1.02, 95%CI [1.01, 1.03]).
Although at day zero, there was a 5% increase in note length (1.05, 95%CI [1.00, 1.10]) and a 31% increase in note closures within 24h (1.31, 95%CI [1.13, 1.53]), there was no evidence for changes from the pre-period in either of these outcomes by day 150.
ConclusionsLongitudinal analysis based on encounter-level measurement of AI scribe showed considerable gains in note writing time, TOSH, and pajama time, and perpetual adaptations in other considered outcomes.
