Impact of speech recognition on radiologist productivity

A survey was conducted of radiology practices with productivity data from at least 3 of the following 4 workflows: film with manual transcription, filmless with manual transcription, film with speech recognition, and filmless with speech recognition. Two surveys were submitted to candidate sites. The first was used to ascertain suitable available data for follow-up. The second survey requested data for report turn around times, full-time equivalent (FTE) staffing levels, and report volume. Data were collected and stored in a Microsoft Access database and statistical analysis performed in Excel. Whereas several metrics were used, the normalized figure of reports-per-day/FTE was found to have improved an average of 1.9 (for filmless with speech recognition) and 2.3 (for film with speech recognition) over the film with manual transcription case. Filmless with manual transcription was only 1.4 times the value of the all manual case. At the 10% confidence level, both filmless with manual transcription and film with speech recognition workflows were found to have statistically significant enhanced productivity. Insufficient data exist to show if the fully automated workflow (filmless with speech recognition) offers benefits over 2 previous semiautomated worlflows.