Education and improved software, can it improve voice recognition errors in radiology reports?. (September 2015)

• Record Type: Journal Article
• Authors: Najran, Pavan S.; Watson, Nick; Bhalla, Vishal
• Is Part Of: Clinical radiology. Volume 70(2015)Supplement 1
• Page Start: S8
• Publication Date: 2015-09
• Subjects: Medical radiology -- Periodicals; Radiotherapy -- Periodicals; Radiotherapy -- Periodicals; Radiology -- Periodicals; Societies, Medical -- Periodicals; Medical radiology; Radiotherapy; Electronic journals; Periodicals; 616.0757
• DOI: 10.1016/j.crad.2015.06.034
• Languages: English
• ISSNs: 0009-9260

Abstract

Abstract : Purpose: The purpose of this audit was to identify the number of errors in radiological reports generated using voice recognition (VR) software. Additionally to determine if improvements could be made by introducing an updated version of current VR software in combination with re-education of reporters in its use. Methods and materials: All reports generated over four non-consecutive days were analysed for errors which were categorised based on modality and type of error. Subsequently an updated version of the voice recognition software was introduced in combination with education following the initial audit results. Once the new software was introduced a re-audit was performed after three months. Reports were analysed over three non-consecutive days and categorised with the same methodology used in the initial audit. Results: In the initial audit 687 reports were retrospectively analysed with 127 containing errors (18.4%). In the subsequent re-audit 408 reports were analysed with 13 containing errors (3.2%). The re-audit demonstrated statistically significant reduction in errors (p < 0.05). Conclusion: The initial audit demonstrated a significant error rate in VR dictated reports but the combination of updated VR software and radiologist training on its use led to a statistically significant improvement in report transcription accuracy throughout the department.