Dictation is a major part of the physician culture and verbalizing notes is a practice they heartily embrace. As John Umekubo, MD, observes: “We have the capability to do our documentation on the computer, but that requires typing and many are not very good at it…but we are very good at dictating.”
Umekubo, a primary care practitioner at St. Mary’s Medical Center in San Francisco, has been working on automating the hospital’s electronic medical records system for three years, leading up to physician documentation this year. Patient histories, physicals, discharge summaries, transfer summaries and operative reports are typically dictated by physicians, with transcription appearing hours or even days later.
“The weak link is daily progress notes, which is usually done on paper,” Umekubo said. “It is not an easy process for documentation entry.”
The ideal scenario for physicians is to continue letting them dictate while speeding up the transcription and placement in the patient’s medical record, which is why Umekubo has been working with Atlanta-based Webmedx to devise such a system for his hospital. The Webmedx clinical documentation system gives physicians a choice for data entry, said Chris Cashwell, vice president of clinical development.
“As an industry in the past, there has been a need to shove doctors in boxes where they must change their behavior and either be ‘all in’ or ‘all out’ with speech programs,” he said. “Each individual encounter needs to be up to the doctor as to whether to use speech recognition or more traditional methods.”
Dave Crawford, Webmedx vice president of business services, adds that a sophisticated new engine “is remarkable in quality and speed” for transcriptions. Using a concept of “speech understanding” as opposed to speech recognition, the system puts a context to words while transferring the data into the medical record in minutes.
St. Mary’s is currently undergoing a pilot program for 13 physicians, which will eventually expand to the entire medical staff of 550 if things continue to go as planned, Umekubo said.
“When Webmedx came up with this new system, my eyes opened up because of the ease and accuracy of it,” he said. “I thought this was something we could use right away for doctors who are better at dictating than typing. Often we have to dictate a transfer summary on the spot – this system makes it simple.”
Coupling technology
To be sure, voice recognition systems are getting smarter, but not without help from middleware firms like Denver-based Health Language. Billing itself as a “terminology services” provider, the company offers a language engine that stores and maintains the terminology standards and makes them accessible through the engine, said Brian Levy, MD, chief medical officer and senior vice president. Its optical character recognition process codifies information on scanned documents.
“We create mappings between the standards – consumer- and provider-friendly synonyms and content,” he said.
“We work with many vendors, we are the black box that sits under other applications.”
The system focuses on ensuring “the right parts of the right terminology,” Levy said, explaining that “as the voice is converted into text, there are errors and as the text is converted into knowledge, our engine converts to the codes.” Context is the hardest part of the equation, he said, “so it’s important to narrow things down. To convert text into code for patient history, we need to match the code against the diagnosis, not the procedure. Context can help the specificity of it.”
Voice communications
Healthcare call center communications are also part of the speech recognition category and technology is advancing in the direction of “high availability, high skill” systems, said Steve Kaish, vice president and product manager for Melville, NY-based CosmoHealth.
Integrated delivery networks are increasingly becoming scattered geographically, with facilities located across regions while needing to maintain one call center.
“They need the flexibility of serving different business units in different locations and share the information across the enterprise,” Kaish said.
CosmoHealth’s pedigree is serving telecommunications companies, which Kaish says translates well into the healthcare market for telehealth applications, billing, prescription orders, appointments and scheduling. The system creates “virtual environments” for each division, “like carving up apartments in a building,” he said. “Within each tenant, there is “permissioning” of the capabilities at their disposal, with multiple levels of permission.”
Supported by both landline and wireless, the system is designed to help healthcare organizations maximize productivity. For instance, if a patient needs a phone consultation and calls one location but no clinicians are available, the call can get routed to another site where someone can help.
“It enables healthcare organizations to effectively distribute the load so they can provide a consistent level of service across all locations,” Kaish said. “That way they don’t have five people sitting idle at one location while another one is backed up 20 deep."
