Study looks at the pros and cons of voice recognition
Voice recognition dramatically decreases the turnaround time for radiology reports -- referring physicians are often getting results the same day their patients have the radiologic examinations -- but technical problems with these systems are reducing some radiologists to typing rather than dictating those reports, a new study shows.
"There are many benefits of voice recognition, but unfortunately we have been facing some technical problems that are impacting our productivity, " says Joel Gross, MD, assistant professor of radiology at Harborview Medical Center in Seattle.
The biggest benefit of voice recognition is the reduction in report time turnaround, says Dr. Gross. "In 2001, when we used the traditional dictation system most of the time, it took an average 133 hours after the examination was completed before a report was finalized, says Dr. Gross. A preliminary report was available with 56 hours, notes Dr. Gross. (The traditional dictation method includes the radiologist dictating the report, a transcriptionist transcribing it, then the radiologist reading the transcribed report and making any necessary corrections before it is available to the referring physician as a finalized report.)
"With voice recognition as our main method (approximately 75%) of reporting in 2002, we have reduced that to 66 hours, with preliminary reports available within 14 hours," he says. "This number is slightly skewed because it includes cases in which the patients undergo the exam late in the evening, and the films aren't read until the following day. With most exams, voice recognition is allowing us to offer same day reporting service," Dr. Gross says.
Another advantage of voice recognition is that a full report is available immediately to the radiologist. "At our teaching facility, the attending physician can more easily review cases with residents. There is not the delay caused by waiting for reports to be transcribed," he says. "This allows full evaluation of all the details in the report while reviewing the images or while the study is still fresh in the radiologist's mind. Signing off on traditional dictated reports several days after reviewing images, one is often reduced to looking for obvious errors in the reports," he says.
However, "voice recognition is not a plug and play system," says Dr. Gross. Faster computers are needed so the system doesn't freeze up. A good support team, available 24/7, that can fix problems or make necessary technical changes is essential, and must be budgeted for when making cost projections, he says. "We've had problems with poor voice recognition, difficulty stopping the transcription, spell-checking, filtering reports, losing macros and speech files, and other glitches that have slowed us down," says Dr. Gross.
"In addition we've also had problems with the system interfacing with our RIS system. We've had situations where our voice recognition system has overwritten changes we've made to our reports in our RIS system, which could potentially lead to medico-legal problems," he says. These technical problems have led some radiologists to simply type in the report themselves instead of dictating into the voice recognition system, says Dr. Gross. "These problems lead to a decrease in radiologists productivity and satisfaction," he adds.
"I'm optimistic that voice recognition will become a viable part of the radiology practice," says Dr. Gross. "However, not only must there be improvements in the voice recognition technology itself, but other technical problems that impede efficiency must be addressed," he says.
Dr. Gross will present his study May 5 during the American Roentgen Ray Society Annual Meeting in San Diego.