Nursing Documentation in the AI Era: A Comparative Systematic Review and Meta-Analysis of Efficiency, Mistakes, Stress, and Quality of Care
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background: Nursing documentation underpins patient safety and care continuity but consumes up to 40% of nurses’ working time¹. Traditional charting methods—paper notes or electronic typing—are prone to omissions, errors, and time burden², contributing to stress and reducing bedside presence³. Artificial intelligence (AI)–assisted systems, including voice-to-text, natural language processing (NLP), and predictive charting, are designed to enhance efficiency, reduce errors, and ease stress⁴⁻⁶. Yet, evidence on their comparative effectiveness versus traditional documentation remains fragmented. Objectives: To systematically review and meta-analyze the impact of AI-assisted documentation compared with traditional charting on efficiency, accuracy, mistakes, stress differential, and quality of care. Methods: Following PRISMA 2020 guidelines for quantitative synthesis⁷ and ENTREQ for qualitative evidence reporting⁸, we searched MEDLINE, Embase, CINAHL, PsycINFO, Scopus, IEEE Xplore, and Web of Science (2010–2025). Eligible studies included randomized controlled trials, quasi-experimental, observational, and mixed-methods designs. Quantitative outcomes were pooled using random-effects meta-analysis; qualitative data (e.g., stress perceptions, usability) were synthesized thematically. Risk of bias was assessed with RoB 2 and ROBINS-I; qualitative studies with CASP. Certainty of evidence was graded using GRADE (quantitative) and GRADE-CERQual (qualitative). Results: From 4,986 records, 32 studies (n ≈ 6,200 nurses) were included. AI-assisted documentation reduced documentation time by − 32 minutes per shift (95% CI − 40 to − 24)⁹. Accuracy and completeness improved (RR 1.21; 95% CI 1.10–1.34)¹⁰. Errors decreased for omissions but increased for transcription/autocorrect mistakes¹¹. Stress differentials favored AI (SMD − 0.38; 95% CI − 0.55 to − 0.21)¹², though qualitative findings revealed concerns about deskilling and trust. Quality of care improved via more patients seen per shift and increased bedside time, though patient acceptance of AI-mediated records varied. Conclusions: AI-assisted documentation enhances efficiency, accuracy, and stress reduction, with potential to improve quality of care. However, risks of new error types and nurse concerns necessitate safeguards. A SMART roadmap recommends integrating AI literacy into curricula by 2027, mandatory verification safeguards by 2028, and stress audits in all AI deployments by 2030.