Three hours of bureaucracy for every hour with patients
The most-cited study on physician time allocation comes from Sinsky and colleagues (2016): in an observational study across four US specialties, physicians spent only about 27 % of their consultation time directly with patients — the rest on electronic documentation, administrative tasks and desk work. Per hour of patient contact, roughly two hours of EHR and desk work accrued (Sinsky et al., Annals of Internal Medicine 2016).
The situation in Germany is no better. Surveys by the Kassenärztliche Bundesvereinigung (KBV) show that practices identify bureaucratic burden as their greatest daily challenge. The Marburger Bund and the German Medical Association confirm this picture in their annual member surveys: documentation has been a central driver of burnout and early practice abandonment for years.
What ambient AI scribes measurably change
The first robust evidence on AI-assisted documentation tools ("Ambient AI Scribes") was published in 2024. The largest pilot study was conducted by Permanente Medical Group with over 3,000 physicians and published in NEJM Catalyst Innovations in Care Delivery: after introducing an ambient AI scribe, physicians reported a significant reduction in their perceived documentation burden and up to 1 hour more productive time per day.
Follow-up analyses also showed higher patient satisfaction, because physicians looked at the screen less frequently during consultations — eye contact noticeably increased. In Germany too, the German Medical Association addresses documentation burden in its annual report as a central challenge of medical work.
Limitations of existing evidence
Most available studies come from the US healthcare system. Transferability to German practice is not trivial: different coding systems (ICD-10-GM instead of ICD-10-CM), different billing logic (EBM/GOÄ instead of CPT), different language, different data protection requirements (GDPR/BDSG). What works in California does not necessarily work 1:1 in a general practice in Mecklenburg.
Independent published comparisons between different AI providers are also lacking for many solutions. Accuracy of ICD-10 suggestions, hallucination rates and effects on billing accuracy are topics where we need robust German data.
What Sanadoc aims to contribute
Together with German general practices and a university medical institution, we are preparing two studies that aim to transfer the international evidence to German practice:
- Accuracy of ICD-10-GM coding — comparison of Sanadoc suggestions with manual coding across several thousand anonymised consultations (study protocol in preparation).
- Reduction of documentation time — multicentre observational study in general practices in Berlin and Brandenburg (pilot phase starting 2026).
As soon as results are available, we will publish them openly — as a preprint and in a peer-reviewed journal. Until then, we deliberately refrain from making our own effect-size claims.
Cited sources
- Sinsky CA, Colligan L, Li L, et al. Allocation of physician time in ambulatory practice: a time and motion study in 4 specialties. Ann Intern Med. 2016;165(11):753–760. doi.org/10.7326/M16-0961
- Tierney AA, Gayre G, Hoberman B, et al. Ambient artificial intelligence scribes to alleviate the burden of clinical documentation. NEJM Catal Innov Care Deliv. 2024. Search study
- Kassenärztliche Bundesvereinigung — care reports and surveys on bureaucratic burden. kbv.de
- Bundesärztekammer — annual reports and statements on documentation burden in medical care. bundesaerztekammer.de
As of May 2026. This page is continuously updated as new evidence or own results become available.