SanadocWhat studies tell us about documentation workload and AI assistants, and where we still need robust answers.
Sinsky and colleagues studied how time is spent across four outpatient specialties in the US. Their observational study highlights the role of electronic documentation and desk work.
Original publication Sinsky et al. · Annals of Internal MedicineTierney and colleagues describe the use of AI scribes in clinical care. Their work explores the potential to reduce documentation workload and support doctor–patient interaction.
View publication Tierney et al. · NEJM CatalystEach study examined a different AI assistant, not Sanadoc. Listed here are the measured results and the limitations the authors state themselves.
Sanmark and colleagues analysed 236,153 notes from 1,295 clinicians at a Swedish care provider. After introduction, estimated documentation time was 4.72 instead of 6.69 minutes per note, around 29 percent less.
In context: Observational study without a control group. The analysis covers a selected group of sustained users, and the time figure is self-reported by 177 respondents. 138,196 notes without recorded editing time were excluded. Several authors work for the vendor under study.
Original publication Sanmark et al. · JMIR Med Inform 2026;14:e90052Duggan and colleagues followed 46 clinicians across 17 specialties at a US health system. Time in notes per appointment fell from 10.3 to 8.2 minutes, same-day appointment closure rose from 66.2 to 72.4 percent, and after-hours documentation fell from 50.6 to 35.4 minutes per workday.
In context: A single health system, a short observation period, voluntary participation and no control group. Clinician feedback on accuracy and editing effort was mixed.
Original publication Duggan et al. · JAMA Netw Open 2025;8(2):e2460637Olson and colleagues surveyed 263 ambulatory clinicians across six US health systems before and 30 days after introduction. The share reporting burnout symptoms fell from 51.9 to 38.8 percent. Participants also reported less cognitive documentation load and more attention for patients.
In context: No control group, only 30 days, mostly self-reported data from a voluntary group. One co-author works for the vendor under study. The work shows an association, not proof that AI assistants prevent burnout.
Original publication Olson et al. · JAMA Netw Open 2025;8(10):e2534976Tan and colleagues had trained observers time 169 consultations directly instead of relying on system timestamps. Documentation time fell from 5.3 to 4.5 minutes per consultation and the share of eye contact rose from 69.6 to 77.1 percent. Consultation length was unchanged.
In context: Nine clinicians at a single hospital in Singapore, not randomised, mostly English-language consultations. Because of the direct observation the study is still more informative than surveys alone.
Original publication Tan et al. · JMIR Med Inform 2026;14:e85580Brunner and colleagues examined the documentation quality of AI assistants in simulated patient encounters across several specialties. The work asks the counter-question to the time studies: not how fast documentation is produced, but how good it is.
In context: Simulated rather than real care and a small sample. The individual results are in the original publication; no abstract is available through PubMed. Work like this is exactly why every note must be reviewed by a clinician.
Original publication Brunner et al. · JAMA Netw Open 2026;9(1):e2552870US studies cannot be directly generalised to German practices, billing systems and workflows.
Results from other AI assistants are not evidence of Sanadoc's effectiveness.
Sanadoc is preparing studies on documentation time and ICD-10-GM coding in everyday German practice.
Interested in evaluating Sanadoc scientifically? We'd love to talk.
Get to know Sanadoc in 20 minutes.
Personal, with no commitment, and tailored to your practice.