Human factors problem analysis of a voice-recognition computer-based medical record

A voice-recognition computer-based medical record system was purchased by a large urban emergency department (ED) to enable quicker dictation, clearer documentation, improved quality assurance, and decreased liability. However, the system was abandoned after being used briefly. We retrospectively analyzed this medical record system to determine reasons for its failure. The human factors problem analyses used were function/task analysis, job design and training analysis, and evaluation of the organizational context. Methods included on-site observations, interviews, and literature/documentation review. The system did not meet the needs and expectations of the ED physicians: the system required too much time to dictate a case; proficiency training was lengthy, yet minimal time was allotted or available: frustration with the vocabulary and preformatted template for entering patient data; a lack of computer skills and time needed to modify the templates and vocabulary; and occasional technical problems with voice recognition, leading to slower dictation. Flow chart analysis of the medical documentation process revealed the computer system was not integrated into existing patterns of work, and the task of documentation was interspersed among other more important physician tasks. Suboptimally, some physicians completed all their documentation at the end of their shift. Reports from other user sites and claims by the vendor about training and usability were usually more positive than the findings from this study. Recommendations include: reduced training time, a more malleable vocabulary and template, increased ease of modification, improved technical performance, and better integration into the overall flow of the ED documentation process