Evaluating the Impact of Artificial Intelligence Scribes on Clinical Workflow and Documentation Quality: A Randomized Controlled Trial.
Abdu Etagiuri, Daniel Chaudhry, Jano van der Merwe et al.
✦ AI-curated · Sources linked
In 30 seconds
This randomized controlled trial compared AI-generated clinical documentation with conventional dictation or templates in an adult reconstruction practice. The study involved 304 patients at a single-surgeon practice, although the reported group counts were 140 conventional and 139 AI scribe patients. AI scribes significantly reduced physician and administrative processing times (P < 0.0001). Spelling and grammatical error rates did not differ significantly, while AI-generated letters were rated more complete, professional, and clear. The AI group had 19 medical errors versus nine in the conventional group; patients generally found the scribe helpful and were not negatively affected.
Key findings
- AI scribes significantly reduced physician and administrative processing times (P < 0.0001).
- AI-generated letters were rated significantly more complete, professional, and clear; spelling and grammatical errors did not differ significantly.
- There were 19 medical errors in the AI scribe group compared with nine in the conventional group.
- Patients were not negatively affected, and the vast majority found AI scribing helpful.
Why it matters
Documentation tools that reduce clinician and administrative processing time may ease workflow burden while preserving aspects of documentation quality. However, the higher number of medical errors in the AI group is an important counterpoint to the favorable ratings and efficiency results. The findings highlight the need to consider accuracy and patient experience alongside time savings when evaluating AI documentation.
What to keep in mind
The trial was conducted in a single-surgeon practice. Medical errors were reviewed by the surgeon, while grammar, spelling, and wording were assessed by unblinded researchers.