Speech Recognition Proves its Worth

Vol. 14 • Issue 3 • Page 18

There is a common perception that the widespread adoption of electronic medical records (EMRs) has been slowed – at least in part – by physicians’ reluctance to spend time typing patient information into a computer.

That perception lingers for good reason: Physicians don’t want to be clerks. They want to practice medicine.

Some doctors, already under pressure to see as many patients as possible on a daily basis, are indeed hesitant to embrace EMRs, out of fear that they will be less efficient and lose valuable “face time” with patients.

At the University of Pennsylvania Health System (UPHS), speech recognition technology has helped put those fears to rest.

Since implementing Nuance’s Dragon speech-recognition software in 2007, the technology has helped simplify the use of EMRs for the 1,800 physicians in our Philadelphia-based health care system, saved the organization hundreds of thousands of dollars in transcription costs and other services, and enabled clinicians at UPHS to actually deliver more personalized care.

Changing perceptions

Eliminating the need for transcription and the associated costs and delays was one of our goals when we implemented speech-recognition software. The technology also seemed to offer a way to eliminate physicians’ objections about becoming data-entry clerks with the emergence of electronic medical records (EMRs).

Rather, we integrated the Dragon technology into our EpicCare EMR; we installed the software on exam room PCs and in clinicians’ offices. That setup has helped simplify physician use of EMRs throughout our health system.

While clinicians receive instruction on using the speech-recognition technology, the bulk of the training is actually conducted in order to familiarize the software with individual doctors’ speech patterns, dialect and accent.

Once the software is “trained,” physicians’ profiles are created. Physician profiles, portable throughout the system, effectively follow each doctor around. Whichever computer the physician uses, the profile loads and allows the physician to speak.

Faster and more efficient

Beyond ease of use, the key advantage of speech-recognition technology for UPHS physicians is that it enables them to see more patients each day.

For physicians who adopt speech recognition, the technology also enables them to have a more collaborative discussion with patients about their care.

The provider verbally summarizes his/her notes, which the patient then gets to hear and see on the screen during the visit. This helps foster a partnership with the physician and patient.

When a patient visit concludes, the information recorded by the physician is instantly available to every other clinician in the system. Now the next clinician who wants to see that patient – an hour later, a day later, a week later – has more rapid access to vital data.

Instant access to patient records is especially important at UPHS, where many patients are referred from other health care providers. We have to send out notes to patients and referring doctors. Using this technology allows us to get referring physician letters out quickly; we don’t have to wait 24 or 48 hours to get a transcription service to send out a letter.

Speech recognition has yielded benefits beyond the workplace as well, with some physicians noting that the time saved by using the software has helped improve their work/life balance. At the end of the day, doctors don’t need to go back through their notes. They don’t have to read charts at home.

Something to think about

For all the aforementioned benefits, the IT executive must weigh a number of factors before deploying speech-recognition technology.

We see about a 50 percent drop in productivity when physicians first use the EMR software, but they generally get back to where they were – or become even more productive – when they get used to the EMR software.

Temporary drops in productivity aside, there are other issues to consider.

Speech-recognition software is a memory-intensive application, requiring two gigabytes of memory. Many health care systems are running half a gigabyte to a gigabyte on desktop computers. As such, the software doesn’t really work on laptops, and IT may need to assess the impact on desktop computers throughout the enterprise. You may find that you need to upgrade PCs currently in use in order to utilize speech recognition.

The leadership team must also take the organization’s patient population into consideration before purchasing speech-recognition technology.

In a children’s hospital, for example, pediatric patients or their siblings may be crying; it tends to be a noisy workplace. In that environment, speech recognition may be problematic. Crowded areas of the hospital such as the ER may present similar problems.

You also need to consider physicians with thick accents or those who speak very quickly. The software may have a hard time deciphering their speech.

Worth the effort

Physician buy-in is the most important piece of the puzzle for IT executives looking to deploy speech recognition. To help get it, seek out proponents of the technology within the organization. Generally, one to three champions can get the word out throughout the organization, depending on its size.

In the last two years, we’ve identified those clinical champions at UPHS. With about 70 percent of physicians in the ambulatory setting using the EMR technology, doctors in other areas of the system are eager to get on board. Right now we have physicians asking, “When am I going to get it?”

Ultimately, we plan to achieve full adoption throughout the organization, and have already realized substantial savings – about $300,000 in transcription costs alone. But that amount pales in comparison to the technology’s value in helping us get closer to achieving meaningful use of electronic health records (EHRs).

You can’t put a price on the adoption of EHRs; we’re all facing meaningful-use requirements and the prospect of ARRA incentive payments. In the end, incentive payments should far exceed the price of speech-recognition software.

Mr. Wells is chief technology officer, information services, at the University of Pennsylvania Health System.