Technological advancement may facilitate a healthcare organization’s drive to efficiency. However, wide adoption is usually governed by the provider’s tolerance for change.
Advanced technologies that eventually succeed in establishing significant usage in a market typically follow well-charted adoption curves. Such models suggest that innovations take root with early adopters who like to try the new, move through various stages of wider adoption as successes are shown and more conservative users adopt, and eventually achieve maturity with a broad, mainstream user base. The central issue for a technology is usually the rate at which it passes through this curve.
Technological advancement may facilitate a healthcare organization’s drive to efficiency. However, wide adoption is usually governed by the provider’s tolerance for change.
Advanced technologies that eventually succeed in establishing significant usage in a market typically follow well-charted adoption curves. Such models suggest that innovations take root with early adopters who like to try the new, move through various stages of wider adoption as successes are shown and more conservative users adopt, and eventually achieve maturity with a broad, mainstream user base. The central issue for a technology is usually the rate at which it passes through this curve.
Perhaps nowhere is this adoption rate issue more evident than in healthcare. At the heart of the drive toward electronic health records (EHR) is the debate over how rapidly physicians and other caregivers will adopt a host of new technologies designed to promote consistency, data exchange and better quality. While growing, the usage of EHR software is one example of slow uptake that demonstrates how difficult the process is in healthcare.
Emergence
One of the potentially transforming healthcare technologies passing through the adoption phases has been speech recognition. Observers of speech over at least the past ten years have witnessed it develop through several stages, attended by substantial skepticism along the way. In the early years there was the question “does it work at all?” As it improved, particularly with the advent of more powerful desktop computers, the question became “Are there any successful healthcare users?” Early adopters succeeded, leading to prospective buyers seeking installations with more than a handful of doctors. As these sites emerged, the more current questions revolve around breadth and depth of usage. Buyers ask, “Can you show me sites with hundreds of speech users, with the technology completing a significant volume of patient reports?”
The Dean Health System in Madison, Wis., provides one strong answer to the last question. With some 350 providers and 45 transcriptionists actively using speech recognition, Dean represents one of the largest speech installations in the country. Its speech utilization can offer lessons to others on how to navigate successfully in this area.
Dean Health Systems is a multispecialty integrated healthcare delivery system. Its network of nearly 60 completely or jointly owned clinics provides primary, secondary and tertiary care throughout southern Wisconsin, delivered by approximately 500 physicians. Dean also is linked with SSM Healthcare of Wisconsin, which owns or is affiliated with a number of inpatient facilities including St. Mary’s Hospital in Madison and St. Clare Hospitals in Baraboo. Dean offers other health services and insurance to its customers.
Health Information Management (HIM) and IT administrators at Dean had been tracking speech recognition with great interest for many years. During that time, the organization did not see enough evidence to convince it to become an early adopter. Nevertheless, one principal problem facing Dean drove continued investigation of the technology.
Dean is a complex organization with hundreds of doctors practicing in many facilities across numerous specialties. It is also expanding. This profile produces substantial transcription volume with the corresponding burgeoning labor costs, which has been estimated at over $10 billion annually industrywide. The total cost of patient documentation is a pressure point for many healthcare institutions, and was of particular concern for Dean given its ownership status. The for-profit health system is 95 percent owned by physicians, so the bottom line directly affects the staff.
Minimizing Change
Dean used a combination of in-house and outsourced transcription services. Needs and costs for both were increasing, which, along with the continued challenges of hiring transcriptionists compounded the problem. These pressures drove the continual return to speech recognition as a potential long-term way to break the cycle. As Laura Cantrall, Health Information Supervisor for Transcription and Abstraction, stated, “We knew a solution had to be found.”
At the same time, Dean administrators had a strong desire to minimize any changes technology would bring to the physicians who would be using it. The balancing act of seeking major process efficiencies while avoiding great change for doctors is central to today’s healthcare technology adoption debate and one that no doubt is familiar to many health information management professionals.
The solution to the dilemma for Dean had its source in the institution’s use of Dictaphone’s Enterprise Express, the dictation/transcription workflow management system. This software captures and distributes dictated notes originating from the telephone. The company’s EXSpeech system was introduced to work directly with the core dictation system, and, very importantly for Dean, to provide speech recognition with telephony-based input.
This integration proved a winner for Dean in deciding to move to speech. As Cantrall explained, “With 95 percent of our doctors dictating into a phone, the majority of the physicians didn’t have to make any adjustments to the dictation styles they are so accustomed to.”
The ability to obtain good speech recognition from telephone quality voice input—as opposed to use of a special noise-canceling microphone—is a technological advancement that has changed the adoption equation over the past few years. Dean’s situation is not dissimilar to that of most hospitals and larger clinics in which the bulk of dictation occurs through the phone. The device is ubiquitous, quick and comfortable to use—a powerful combination for doctors.
Accommodating this style means that the benefits of speech recognition can be realized with little change of habit. The physician dictates reports in normal fashion, voice-to-text conversion takes place in the background at server level, and a transcriptionist edits the report to completion. Some doctors are even unaware that speech recognition is in the mix. The telephony support should not be underestimated in evaluating this technology today.
Adding an EMR to the Mix
Dean is deploying an electronic medical record from Epic Systems. However, as at many institutions that also moved to an EMR, a substantial volume of patient documentation continues to be dictated. Physicians cite not only convenience and speed for dictation’s persistence in such an environment, but also the desire to retain the narrative form of documentation—one that promotes the detail, context, and comprehensiveness they often find lacking in structured documentation tools.
Interfacing the Enterprise Express text system with the Epic EMR so that transcribed and speech-recognized reports would be part of the repository and therefore available for subsequent review meant convenience for doctors. It also meant that the speech recognition system was helping to optimize the EMR for Dean and promoting its overall usage.
With the decision made, getting started was a measured affair. Cantrall explains that the initial group of speech recognition users to go-live included 40 doctors and 10 transcriptionists. There was no special bias to the selection of these individuals. Cantrall simply sought a reasonable mix and, most importantly for the transcriptionists, a willingness to participate. Like so many best practices guidelines in advancing technology, gaining committed early users remains one of the top strategies.
Once successful with the initial group, Dean proceeded through a steady march upward, adding users to the speech system until arriving at its current level of 350 providers and 45 transcriptionists. How does an organization attain these substantial levels of doctors? The answer to this question shows how the technology has again developed in ways that foster adoption.
One key is the ability to support doctors from a wide range of specialties. When asked to name the medical disciplines of her physicians using speech, Laura Cantrall jokes, “Stop me when you don’t want to hear any more,” and goes on to list more than 20 ranging from internal medicine to orthopedics to psychiatry. Unlike years ago, today’s enterprise-level speech recognition software has medical language models, or vocabularies, that span the spectrum of disciplines, making enrolling doctors easier.
Those models also have been built by processing a range of typical inpatient and outpatient report types. This breadth is another key to attaining volume usage in speech. Cantrall cites most of the usual work types, including the narrative-heavy classics such as H&Ps, Consultations and Progress Notes.
Measuring Success
Achieving steady usage across 350 physicians suggests that speech recognition is realizing its promise to become enterprisewide. It also indicates that the software is becoming a daily production tool, like the workhorse dictation systems in use today. But the number of users on the system is only one measure to gauge progress toward becoming a production tool. The other is the volume of transcription work being processed.
On this count, Dean is again leading the charge. Cantrall says that about 22,000 reports each month are being completed with the speech recognition system. This figure represents roughly one-third of the organization’s volume and Cantrall figures it will continue to grow. Other HIM departments around the country are registering half or more of their volume accomplished with speech. Clearly another milestone is being passed on the road to broad adoption.
What about the technology’s impact on the transcriptionist? This is a typical concern of HIM administrators. If the telephony-based background editing mode shields doctors from major practice changes, the transcriptionist’s world changes significantly. Having worked through the ramp-up, Cantrall points to the fact that 45 transcriptionists now employ speech editing. Most importantly, “Transcriptionists who now edit would not go back to typing.” Celia Fine, Dean’s Quality Analyst and lead for speech, concurs and says the transcriptionists are happier and more comfortable in their work. According to Cantrall, productivity of the transcriptionists has shown an average gain of 40 percent. Some of her employees have been measured with output gains of nearly 150 percent. This increase in throughput has contributed to the department’s ability to handle Dean’s growing workload. “Overall, we are able to do more with less staff.”
It also has generated a savings in outside transcription costs. Cantrall says she is tracking well over $300,000 under her outsource budget. She also feels that additional cost savings have resulted from the considerably reduced effort expended by Dean’s human resources group in the struggle to hire transcriptionists.
Dean plans to bring more doctors and transcriptionists into the speech recognition workflow. But for now, this organization’s experience substantiates the idea that this technology is fast moving from the niche, early-adopter stage to the long-predicted full enterprise maturity in healthcare.
For more information on Dictaphone Enterprise Express and EXSpeech,
www.rsleads.com/702ht-210
