Voice Recognition in the ED

By John DeGaspari for Healthcare Informatics

Why one hospital complemented its EHR implementation with voice recognition

How does voice recognition technology affect patient care quality and efficiency? One proponent for its use is Bud Lawrence, M.D., emergency medicine physician director of risk management at the Henry Mayo Newhall Memorial Hospital, a 250-bed hospital and level-2 trauma center in Valencia, Calif. About three years ago the hospital made speech recognition part of its implementation of its electronic health record.

The voice recognition technology (supplied by Nuance Communications, Inc., Burlington, Mass.) was initially deployed in the hospital’s ED, which sees close to 50,000 visits annually. “We have found that speech recognition has helped us be much more efficient with our patient flow,” Lawrence says.

Entering Patient Information in Real Time

The voice recognition capability is integrated with the hospital’s EHR (which is supplied by MEDITECH, Westwood, Mass., and is implemented in both the ED and the inpatient hospital.) Voice recognition is both faster and more accurate than the transcription service it replaced, Lawrence says. It allows physicians to create a patient chart in real time while seeing a patient, he says. Documentation can begin when the physician visits the patient, rather than completing the documentation after the visit. The physician can sit at a workstation and begin creating a document up to the point when he is with the patient, he says.

In the ED, physicians are able to pull macros into their documents to create personalized documents for patients quickly and easily, he adds. “Voice recognition has helped us create the document in multiple parts faster than we would have created it with standard dictation [using a transcription service], in the sense that the document is done, it’s signed, and we never have to go to medical records two or three weeks later to sign a huge stack of paper charts,” he says. “We don’t have to stay hours after our shifts to finish our documentation—the documents that are created are much more accurate, because they are created in real time.”

In addition to allowing faster and easier patient documentation, voice recognition allows better communication with other caregivers, according to Lawrence. For example, he says, voice recognition has allowed physicians navigate through and fill in the specifics about ordering. “With CPOE, we can fill in the specifics about a study—say you have to order a CT scan, you can fill in very easily the reasoning why you ordered the CT scan and can convey that to the radiologist using voice recognition instead of having to type it into the required field. All of these things have helped us be much more efficient in our daily lives, and in emergency medicine, efficiency is key.”

Lawrence says that he is able to start his document when he sees the patient. If he decides that the patient will need to be admitted, he will contact the admitting physician, who can log on and view the patient’s document remotely. “The primary care physician has access to our notes, all the way through draft mode. He can take a peek and see what our thought processes were and where we’re headed,” he says. (In the state of California, emergency medicine specialists generally do not admit patients, a role performed by primary care physicians.) The ability of the primary care physician to read the patient documentation in real time, just prior to giving his admission orders, in addition to lab reports and insight into the medical decision process, is a huge advantage, he says.

Voice Recognition and the Narrative Process

Lawrence, who took the lead in developing the EHR for the hospital’s ED, says he would not use an EHR without voice recognition. Non-narrative document creation that uses clicks in a template does not accurately convey the physicians’ thought processes and is not easy to read, he says. “It’s very difficult to interpret and get a feel for exactly what is going on,” he says. “I felt that a narrative type document was the only way to convey our thoughts,” he says.

“All of the advantages point to having a voice recognition system to create an immediate document that is visible to all parties that are taking care of the patient, that is timely, accurate and efficient,” Lawrence says. “The only way I can see doing that is with voice recognition, because it comes right out of your mouth onto the screen. You edit as your go, and once you edit it and sign it, it is done.”

The CPOE includes 70 to 80 order sets as well as 700 to 800 macros that exist with the voice recognition portion to use with patient document creation. He says he had expressed the need for a narrative document early in the implementation process.

The hospital went live with voice recognition in its ED three years ago last September, according to Lawrence. Although he expected the change in technology to be a slow process, he was surprised at its uptake by ED physicians. He says the ED was “efficiency negative” for the first 30 days, efficiency neutral at 60 days, and efficiency positive after two months.

“There needs to be a fair amount of physician training on how to use the product appropriately,” Lawrence says. Part of the training is teaching physicians to edit properly,” he says. He notes that the voice recognition package will not misspell words, but may put in an incorrect word. “You have to be able to pick that up. It’s just a matter of appropriate training of where to look and how to find those,” he says.

Lawrence says he reviews his documents twice—once as he is dictating and then a second edit review just before finalizing the document. He completes the process while he is on his shift. He notes that, before voice recognition, most of the physicians in the ED were staying two or three hours after their shifts ended to complete their documentation; using voice recognition, they are now able to leave when their shifts end. “That was one of the biggest satisfiers for them,” he says.

Voice recognition was first implemented in the ED, which is comprised of 15 doctors, which served as a controlled setting. In August the hospital started a pilot with selected physicians before rolling it out to all in-house physicians to use, he says.