Time Savings Through an AI Speech Assistant for Nursing Documentation: Pre-Post Time-Motion Study in German Long-Term Care

Abstract

Background: Nurses in long-term care spend up to one-third of their working time on documentation, contributing to administrative burden and limited time for direct care. Artificial intelligence (AI) speech assistants have shown potential to accelerate documentation, but longitudinal evidence from real-world long-term care settings remains scarce. Objective: This study aimed to evaluate whether implementing a domain-specific, mobile AI speech assistant is associated with reduced documentation time in German long-term care under routine conditions. Secondary objectives included examining usability, perceived documentation effort, interruptions, and workplace satisfaction. Methods: A pre-post time-motion study with full-shift observation was conducted. Continuous, event-based observations were performed before (t0) and after (t1) implementation of the mobile speech assistant voize. The primary outcome was total documentation time per morning shift based on direct observations. In addition to observations, questionnaires were administered to assess perceived documentation effort, interruptions, satisfaction with the documentation system, and workplace satisfaction. The primary end point was analyzed using a linear mixed-effects model. Secondary, self-reported outcomes were analyzed exploratorily via paired pre-post differences with pooling across multiple imputations and Holm-Bonferroni correction. Results: A total of 52 registered nurses from 14 long-term care facilities participated (mean age 42.37, SD 12.37 years; 42/52, 80.8% female). Across 770 observed hours, the observed total documentation time per morning shift decreased significantly by an adjusted mean of 15 (SE 3.36) minutes, t46.29=-4.46, P<.001, with a 95% CI of -21.75 to -8.23, corresponding to an approximately 28% reduction relative to the baseline mean. Holm-Bonferroni-corrected exploratory analyses indicated significant declines in self-reported documentation time and interruptions, and satisfaction with the documentation system improved, while workplace satisfaction showed no significant change. Usability was rated as acceptable. Conclusions: This study provides real-world evidence from a single-group pre-post design that an AI-based speech assistant is associated with reduced documentation workload in long-term care. In this sample, the integration of a mobile, domain-specific speech system into daily workflows coincided with substantially decreased documentation time and improved perceived efficiency. Beyond these observed time savings, such technology has the potential to alleviate workload, free time for resident care, and enhance working conditions. These findings are also relevant for policy discussions on addressing the nursing workforce shortage, showing that well-integrated, speech-enabled documentation systems can support more sustainable long-term care environments.

Ähnliche Arbeiten