Clinical Barriers to Hands-Free, Eyes-Free Voice Input for Nursing Records: Field Usability Study.

Background: Nursing records are essential for maintaining patient care quality but impose a substantial workload on nurses, thus contributing to burnout and diverting attention from direct care. Voice input technology enables hands-free and eyes-free documentation, allowing simultaneous patient care and record entry. Despite its potential, its adoption in clinical nursing practice remains limited owing to concerns about patient privacy, technical instability, and the complexity of entering structured data into electronic medical record interfaces. Furthermore, most previous studies have been conducted in simulation settings or have focused on post hoc dictation. Thus, the feasibility of true simultaneous documentation in real-world clinical environments remains largely unexplored. Objective: This pilot study was designed to explore the feasibility of using hands-free and eyes-free voice input for concurrent nursing documentation in the highly structured clinical environment of a catheterization laboratory. Methods: This study was conducted at Kyoto University Hospital using a mixed methods exploratory design. Eight cases of percutaneous transhepatic cholangiodrainage and transcatheter arterial chemoembolization were observed between December 2022 and January 2023. Five nurses participated in this study and documented intraoperative events using both traditional handwritten records and a prototype voice dialogue system comprising a smartphone (Google Pixel 6a) and wireless earphones (Pixel Buds Pro). Researchers observed the nurses' behavior in adjacent control rooms to minimize interference. Log data from voice input and the corresponding handwritten notes were compared to determine the proportion of events successfully captured. In addition, semistructured interviews and usability surveys were conducted to obtain qualitative feedback on usability, practicality, and perceived barriers. Results: Voice input successfully recorded 40% to 100% of events during the preoperative and intraoperative phases, but only 0% to 12% during postoperative documentation. Observations and interviews revealed that the postoperative phase involved higher cognitive and communication demands, thereby making simultaneous voice documentation difficult. A significant barrier identified was the "social awkwardness" of interacting with the system; nurses reported feeling embarrassed when speaking loudly to activate the device and often stepped away from patients to record data, negating the benefit of concurrent entry. Concerns about disturbing patients or interrupting medical communication have also hindered use. Conclusions: This pilot study identified two major barriers in applying concurrent voice input in clinical settings: (1) the instability and unreliability of using voice input as the sole recording method and (2) the conflict between voice interaction and the social dynamics of care. To overcome these, future implementations should consider visible devices to signal recording status, support for whisper recognition, and protocols for code words to handle sensitive information. Gradual implementation may reduce nurses' cognitive and psychological burdens. Further large-scale longitudinal studies are warranted to validate these strategies in routine nursing documentation.