VA tries voice-friendly EHRs

By Peter Buxbaum
11:09 AM

Using voice-recognition software with electronic medical records could save a Veterans Affairs Department facility in Los Angeles hundreds of thousands of dollars per year in transcription costs, according to one physician advocate. And yet, only a handful of employees at the VA's West Los Angeles Healthcare Center use the technology.

Meanwhile, a VA network in Tennessee has a 750-user license for voice-recognition software and has managed to fill most of those slots.

Those examples demonstrate the range of attitudes toward such software, which works with VA's EMR, the Computerized Patient Record System (CPRS).

"Some networks have made investments for broad deployment," said Dr. Clayton Curtis, a primary care physician and informaticist at the VA Boston Healthcare System. "Elsewhere, it is left to individual facilities."

At VA's West Los Angeles facility, Dr. Steve Zeitzew, an orthopedic surgeon, has been advocating for voice-recognition software for eight years. It took him two years to get administrators to agree to a pilot program, and that happened only after a cardiologist complained to the medical executive committee that his emergency notes were not being transcribed in a timely manner.

Voice recognition "met with a fair amount of resistance," Zeitzew said. "The main obstacle is budgeting. Even though I showed the administration they could save a lot of money if the facility had a large number of VR users, they still said they didn't have money in the budget for software."

Six years later, Zeitzew's pilot program is still a pilot, with six regular users.

In contrast, VA's Tennessee Valley Healthcare System, based in Nashville, holds a 750-user license for Nuance Communications' Dragon Medical voice-recognition software. The network receives a discounted rate for the licenses because it is a test bed for the software.

Bernice Burchfield started the program in 1998 when she was chief information officer at a VA health facility in Memphis, Tenn.

"The facility director instructed me to find some type of automated solution for entering notes" into the EMR, she said. "We brought in five [voice-recognition] vendors, and Dragon emerged as the least of the five evils after being tested by 24 clinical workers."

The Dragon software has improved a great deal in the past 10 years, Burchfield added. The company recently released Version 10 of the medical voice-recognition software.

Burchfield has since moved on to become CIO of the Tennessee network, and she managed to get the software licensed across the network " at six facilities in Tennessee, Kentucky and West Virginia. The greatest number of Dragon users the network has seen at any one time is 400.

"We have worked with [Nuance] to develop templates and macros that help users of Dragon with CPRS," Burchfield said.

She agreed that cost is the primary factor hindering more widespread adoption of the software. "Historically, cost has been the issue," she said. "Individual licenses are not cheap. We do have a community of users across the country. They exchange tips and share templates."

For all its benefits, no one says voice-recognition software is a panacea. "The advantage is you get a reasonably high speed of dictation and accuracy in transcription," Curtis said. "It's great when you're familiar with the patient's chart. You just close your eyes and dictate what you want to say in the notes."

On the other hand, the technology does not allow clinicians to flip through a chart electronically while dictating. "You'd need two computer screens to do that," Curtis said.

"The real key with [voice-recognition] technology is that you can't impose it on people," Zeitzew said. "You need to develop user interest."

You also need to provide the right kind of hardware, such as high-quality sound cards and microphones, and technical support to set up the software correctly, he added.

"Even though the newest version is miles better than years ago, it is not a simple out-of-the-box experience," Zeitzew said.