Implementation of Speech Recognition in a Community-based Radiology Practice: Effect on Report Turnaround Times

Abstract

Large academic practices have reported important benefits with the implementation of speech recognition software (SRS). However, the applicability of these results has been questioned in the community hospital setting because of major differences in workflow. The aim of this study was to evaluate the impact of SRS on radiology report turnaround times (TATs) at a community-based hospital practice with no radiology training program. The secondary goal was to evaluate the impact of SRS on radiologist productivity. SRS was implemented at a 150-bed community hospital between May 2011 and July 2011. Radiology report TATs and normalized radiologist productivity were determined during 5 months before and after SRS implementation. Median and 80th and 95th percentile report TATs were compared between the preimplementation and postimplementation periods. The trend in productivity was also assessed. Median and 80th and 95th percentile report TATs decreased multiple-fold between the preimplementation and postimplementation periods (median, from 24 to 1 hour; 80th percentile, from 60 to 10 hours; 95th percentile, from 165 to 33 hours; P < .0001). No significant trend in report TATs was appreciated beyond the initial implementation of the software, a sustained effect on TATs. Normalized radiologist productivity was stable throughout the study period. The implementation of SRS was associated with 24-fold improvement in the median radiology report TAT in a community hospital setting with no radiology trainees. Improvements were obtained without affecting normalized radiologist productivity.