Trends in Transcription

VCU Health System makes a strong commitment to completely and efficiently incorporate continuous speech recognition into physician workflow.

By Rich Pollack

A few years ago,at the suggestion of our then CMIO and with the support of several influential physicians and the acting CIO, VCU Health System (Richmond, Va.) investigated Dragon voice-to-text technology from Nuance Communications. After several demos with excellent physician participation, the IT Steering Committee approved going forward with the purchase of 50 licenses (at a cost of about $55,000) for the Dragon voice-to-text application.

Our overriding thought going into this implementation was to provide an alternative data input device. Specifically, we wanted to help ambulatory physicians who didn't type well to input their notes into our clinical information system (Cerner EMR). A number of attending physicians had privately purchased the Dragon product and were already using it. We acknowledged that moving ahead with the voice-to-text application would be far less expensive than expanding our use of traditional transcription.

Along the way, we learned it was going to be resource-consuming to provide the necessary hardware upgrades and one-on-one training. We did not assign a project manager, which diminished the effectiveness of the initial rollout.

In addition, some physicians expected lots of value with little or no effort. In actuality, the physician has to spend considerable time to train and correct the profile and incorporate it into his/her individual workflow.

Those challenges are still with us today. The pilot group of inpatient hospitalists did not fare well with automated dictation due to high ambient noise levels in the inpatient setting. Besides, residents and many of the hospitalists found they could type faster than they could talk.

Nonetheless, we found that general surgeons seemed to adopt the tool, whereas attending physicians in neurosurgery, vascular and L&D did not. With an introduction to the product, written instructions and a handout, 10 surgeons and mid-level providers have adopted the tool.

Putting the software to use

Dragon has helped improve patient care by giving clinicians an additional way to record electronic data in the course of patient care activities. VCU has about 25 clinicians piloting the software, which is viewed as an additional option to traditional documentation methods such as typing, transcription services or handwritten notes for. While speech-to-text data input is not for every clinician, some find it a significant aid in documenting patient care.

Once the software has been mastered, clinicians cite the following advantages: high degree of accuracy with medical language, efficiency in documentation, and improved, timely communication with other clinicians.

For example, VCU's vascular lab has designed numerous templates that include blanks for patient-specific data elements. Data can be added to the templates quickly; edits to standard phrases can be entered in much less time than it takes to dictate a report.

Surgeons report similar results with their immediate postoperative notes. Additionally, structured data can be entered by verbally selecting from a defined list in addition to the more common block text. The structured data elements subsequently can be used for data queries.

Since Dragon can be used in any "window," blank notes have been set up in the electronic medical record (EMR) and for reports/notes to be entered directly into the data repository. As soon as the report has been signed, the legible electronic document is available for review by all clinicians with appropriate access, thus eliminating the turnaround and editing time inherent with a traditional transcription service.

The neurosurgeons use Dragon for outpatient visit notes. If the patient returns to the ED or contacts the on-call physician, the note is available online as compared to the handwritten note that, in the past, would have been in the physician's office or off-site in the medical records room. Since clinicians have immediate access to reports or notes, therapy decisions can be made early in the care process.

Building on reasonable expectations

The benefits of speech-to-text technology have included the following:

1) an expansion of neurosurgery office notes available in the EMR;

2) decreased transcription costs for that service; and

3) improved turnaround time for vascular lab results.

Your organization needs to establish reasonable user expectations. The clinician has to invest time for training and improving the accuracy of the user profile. The solution requires a strong commitment to continuous use to completely and efficiently incorporate this process into the workflow. It can work well for clinicians who are not proficient typists — or those not used to dictating who prefer to visualize their spoken words and make immediate changes.

At the same time, some of the clinicians who have used transcription services have switched, as they prefer to "touch" the document once and avoid editing delays.

Mr. Pollack is vice president and CIO at VCU Health System in Richmond, Va.

A Drawing Card for Talent

By Michael L. Westcott MD

Sometimes, there is no better proof of a system's efficiency than the strong demand for it. In our case, at Alegent Health, the implementation of speech recognition, specifically Dolbey Fusion Speech, powered by Philips SpeechMagic, was actively fueled by physicians. In today's health care environment, advanced technologies such as speech recognition are indispensable in providing world-class patient care and in attracting world-class medical staff. Alegent Health will invest more than $150 million in state-of-the-art medical technology by 2012.

Speech recognition has become as vital for the reputation of a facility as imaging services. World-class health care and new guidelines and regulations require detailed report information. Easy access to the dictation and speech recognition system promotes a complete dictated record, resulting in comprehensive reports. Speech recognition also enables faster report turnaround times and improved quality.

Notwithstanding the many advantages of speech recognition, some challenges remained for Alegent during implementation.

We chose an incremental implementation to acclimatize physicians to a speech recognition-based and paperless workflow. We also gave our physicians the chance to grow with the system, instead of saying, 'Here's your new system — now work with it.'

The implementation took time but acceptance is high and our physicians are pushing for the technology for their offices, too. Solutions that increase physician productivity and decrease turnaround time have a high acceptance rate. Recruiting a successful medical team requires the implementation of technologies that streamline physician workflow, and one of those is certainly speech recognition.

Dr Westcott is chief medical information officer at Alegent Health — the largest non-profit, faith-based health care system in Nebraska and southwest Iowa.

Trends in Transcription

By Peter Durlach

We have entered the next generation of transcription, and it has already had a dramatic impact on health care organizations. With today's technology, physicians can get documents back faster, with fewer errors, and can treat their patients more efficiently, with better quality care and safety, than they could in the past. As a typical example, Aultman Hospital in Canton, Ohio, implemented speech recognition technology in 2006 and was able to decrease documentation turnaround time from 24 hours to under 4 hours. This type of rapid access to completed reports has ushered in significant quality and safety improvements as providers are able to start treating patients based on finalized electronic reports much more quickly, often while the patient is still at the facility.

While quality and patient safety are always top priorities for health care organizations, rapid report turnaround also yields significant financial benefits. Because billing has to wait until reports are completed, signed and submitted, a significant turnaround improvement affecting a large volume of reports can lead to a measurable reduction in DNFB (Discharge Not Final Billed) charges and a corresponding improvement in cash flow. In the Aultman example, the improvement from 24 to 4 hours could mean the difference between billing two days after the test to billing the same day. Multiply those two days across thousands of reports and the financial impact quickly becomes significant.

Speech recognition has also made an impact in terms of cost savings. By transitioning from MTs performing full dictation to having them edit and perform quality assurance (QA) on documents created using speech recognition, MT productivity can increase 100 percent or more, allowing organizations to generate increased volume at high quality with fewer resources. This is accelerated by the "self-edit" effect, where the total number of reports going to transcription is decreased by physicians who increasingly find it expedient to simply edit and sign their own reports. Memorial Hermann, in Houston, for example, achieved annualized transcription cost savings of just under $1.5 million, including cost reductions at individual facilities ranging from $15,000 per month to between $500 and $1,000. One site saw monthly costs decline from $30,000 to less than $500.

In addition to improvements in documentation and reduced costs, speech recognition has changed the role and productivity of the MT. With this change has come a new set of skills for employees to learn and we have witnessed improved quality of work life for MTs. No longer are MTs overwhelmed with the pressure of typing reports word-for-word. Instead, as document editors and QA specialists, they can focus on reviewing and editing a much larger number of documents, and adding value at a much higher level. Advanced Healthcare, a multi-specialty clinic group in southeastern Wisconsin, has seen its top MTs perform at 210 percent the normal rate prior to speech recognition technology.

Beyond speech recognition

Speech recognition technology has pioneered the way for new uses of speech recognition in health care and for the emergence of new technologies that complement speech. One of the biggest changes has been the growing movement toward physicians self-editing their reports, especially within diagnostic areas such as radiology and pathology, and in combination with new electronic medical record (EMR) applications. This development has been driven by continuing improvements in the technology and increased pressure to fully and rapidly document findings within PC-based clinical information systems. Self-editing has also redefined reporting workflow. Physicians can use templates or macros to quickly create reports that accurately describe the patient's story in a time-efficient manner.

One of the next trends we see for the future of language processing technology is in the area of natural language processing (NLP). Today, this technology can be effectively used for non-real-time data analysis. For example, radiology customers are using NLP to help measure utilization and clinical appropriateness of imaging studies in ways previously not possible. In the future, this technology will allow computers to scan through documents, using algorithms to "understand" the content. The technology will be able to pore through transcribed reports (speech-recognized or not) to find key data points such as medications, diagnoses and vital signs, and then turn them into measurable data elements that can be manipulated, searched and used in reports. As part of this process, NLP will be used to populate EMR applications.

So the future looks bright for the transcription industry with sustained gains in productivity, cost-efficiency and the creation of new capabilities. We expect new technologies and other innovations to continue to help drive down costs while simultaneously improving care.

Mr. Durlach is senior vice president of health care marketing and product strategy for Nuance Communications.