David Heiman, M.D., just loves his digital voice recognition system.
"I have not paid a transcriptionist to type a single word," says the Tampa, Fla., gastroenterologist. "It saves me money, time, labor. My note is finished the moment I'm done speaking, proofreading and editing."
Heiman, a solo practioner, says he saves $700 a month, or nearly $10,000 a year, in labor and mailing costs by having speech recognition technology.
"My transcriptionist is free," Heiman says. "I'm spending less time dealing with paper. I don't have to worry about three-week turnarounds. I'm communicating more efficiently.
"I cannot imagine why you wouldn't do this."
But many others can, for all sorts of reasons.
Voice recognition has encountered much resistance because it requires more deliberate speech and the speaker must actually say words like "comma" and "period" to get the proper punctuation. Then the speaker must stop after each sentence to proofread and make corrections.
What would be a three-minute dictation into a conventional recorder can end up taking six minutes of a physician's time.
According to most estimates, more than 90% of physicians still rely on the tape recorder and transcription service for dictation work. Voice recognition technology has improved immensely in recent years, but nobody has yet come up with the so-called "killer app" that would make the microcassette obsolete, according to industry skeptics.
"There's a high level of interest but not a whole lot of installations because most people are in a wait-and-see mode," says Dan Emig, director of technology marketing for Siemens Medical Solutions Health Services Corp. in Malvern, Pa.
Clearly, though, voice recognition technology has improved in recent years.
Microsoft Corp. has built speech recognition features into the Office XP Professional productivity software suite it introduced last May.
"This is not a feature that is very widely advertised," says Ahmad Hashem, M.D., global industry manager for healthcare at Microsoft. "It's not quite ready for mass adoption."
Hashem says that Microsoft's system is about 98% accurate right now. "I don't think that the average physician is going to tolerate that yet."
But technology is advancing fast enough for Hashem to forecast a 90% adoption rate within two to five years. "The key is, at what level will (speech recognition) be good enough?"
According to Sam Brandt, M.D., vice president for clinical informatics at Siemens Medical, a sea change in technology is about to take place.
Nascent innovations that could be the next big thing include "ink"--electronically captured handwriting--and embedded voice files that are treated as distinct data types rather than media that must be converted to text.
"Our prior dictating system failed miserably and our retention of transcription staff was poor," says Anthony Mancuso, M.D., chairman of radiology for the University of Florida's Shands HealthCare system in Gainesville, Fla.
But the current Dictaphone PowerScribe system works just fine in what Mancuso says is the first radiology department in the country to go 100% to voice recognition.
"The downtime is nil on this system," he says. "After about 40 minutes of training, the recognition (accuracy) is about 98%." Most errors are "logical and thus fairly obvious," making them easy to edit, according to Mancuso.
"If you fully dictate the report, it requires about 15 to 20 seconds of editing," he adds. The doctor signs off on the report and stores it in the patient's electronic record, where it is accessible to physicians scattered across the system's three hospitals.
This is especially important in radiology, Mancuso says, so reports can go out at the same time as images, making life easier for referring physicians.
"There's no way to accomplish this short of having a typist next to you. It's an extraordinay patient care system," he says.