Saving healthcare with the clinical narrative

A recent exposé on the dangers of electronic health records and in particular "copy-and-paste" functionality accentuates the battle between innovative technology and the essential inclusion of the nuanced clinician narrative. As one specialist put it, limiting a patient's medical record to a point-and-click template decreases the ongoing value of that record and weakens its contribution to clinical decision making. In fact, the value of the narrative is even evident from medical records dating back to 1547; consider the medical history of Henry VIII that even today continues to provide new insights into his clinical conditions and cause of death.

While templates help drive standardization and can ease data analysis, without the narrative we risk losing important clinical data that is buried, as shown in an example found on the Medical Transcription Industry Association website. The same note captured using a typical EHR system balloons to five pages, burgeoned with excess and difficult-to-read content, but from the narrative shrinks to two concise, readily digestible pages. The challenge is how to integrate the narrative into the EHR so as to benefit from the numerous EHR advantages of digitization and rapid sharing of information.

Speech remains the preeminent means of knowledge capture in healthcare. According to the Centers for Disease Control and Prevention, approximately 1.2 billion clinical documents are produced in the U.S. each year, of which 60 percent come from dictation and transcription, the MTIA says. A solution that thousands of clinicians are already using to preserve their often preferred dictation method, while still being able to capture the narrative as part of the EHR, is to apply speech recognition technology to the EHR process.

Interestingly enough, there was a conference in Boston late last month that included a panel session: "Getting Clarity--Developing Effective Health IT Policies and Standards." During that session, Harvard Medical School CIO Dr. John Halamka asked the audience the following:

"How many doctors are in the audience?" A sea of hands went up.

"How many are using an EMR?" About two-thirds of the hands remained up.

"How many of you love it?" Just two people still had their hands in the air: Dr. Larry Garber, who is a medical director for informatics at Fallon Clinic, and Dr. Michael Lee, a practicing pediatrician at Atrius Health.

What is striking about these two individuals is that they are both avid users of speech recognition.

The dataset may not be large, but the principle is clear: doctors oftentimes prefer to use their voice to capture clinical information and speech recognition allows them to do this, even as part of the data-centric EHR workflow. To seamlessly bridge this divide will require further technological innovation that blends the narrative with the data essential to the EHR. In some cases physicians will capture data in structured form, but in many cases data do not tell the whole story. With speech recognition, physicians can have their cake and eat it--they can produce the narrative efficiently within the EHR and generate the structured data to feed the data-hungry EHR.

Nick van Terheyden is chief medical information officer for clinical language understanding at Nuance Communications and a board member of MTIA.