Q&A: Saving the Whole Patient Story
 |
Working through the HL7 standards development organization, volunteers will consolidate standards exchange of eight types of clinical documents, along with the HL7 Continuity of Care Document (CCD) standard, into one foundation for HIE.
CMIO recently spoke with Bob Dolin, MD, chair of Health Level 7 International, president and chief medical officer of Lantana Consulting Group, about the Health Story Project and clinical document exchange.
Can you talk about the Health Story Project, and the effort to save the full patient story?
As part of meaningful use, there’s a need to capture discrete data elements. There are two ways you can go after discrete data elements. One is, you can say to the clinician, ‘I know you already have your own workflow, but can you also go ahead and capture these 20 data elements?’ That can be disruptive. We believe a more efficient strategy is to embrace and build upon existing clinician workflow.
As a practicing internist, I see patients in the ER who have been treated elsewhere. First and foremost, at the point of care, I need to read their last discharge summary, their last history and physical, their last cardiac cath report [or] their operative report. Frontline clinicians today need to have clinical documents; they need to have the rich narrative. It’s embedded in our workflow. The Health Story strategy says, how can we continue to support the frontline clinicians, how can we build upon this foundation that we need, but do it in a way that supports meaningful use?
What we have found is that if we establish this bedrock of clinical documents using the Health Story/CDA/HL7 strategy, we can continue to embrace the existing clinician workflow by simply adding in today a couple of discrete data elements into these otherwise narrative documents. The elements that we add in today are based on the priorities in meaningful use Stage 1. Next year, there will be new priorities as part of meaningful use Stage 2.
I’m still going to be sending my clinical documents back and forth because from the get-go, I need to support the needs of frontline clinicians who are seeing patients and I need to preserve and create minimal disruption to existing workflow, so I’ll put in a couple more data elements.
It also means implementers [such as] the transcription industry [can] jump in and get involved in meaningful use today; we can continue doing what we’re doing, and next year we’ll put in a couple of additional data elements. So it’s not only from a technical perspective how you preserve the full narrative, it’s from a strategic perspective—if your definition of meaningful is you want to get discrete data elements, my belief is the best way to get there is to build on our established workflows and embrace what we’ve got today with clinical documents and with the narrative.
In an EMR, is this an effort to recognize natural language and abstract, or natural data?
I’ve debated that with the Health Story group quite a bit. … We want to get toward this notion of meaningful use, which is what a lot of people think of as discrete data elements. I believe a way to get there is to build upon a foundation of narrative clinical documents where you put in discrete data elements over time.
What we’re finding is that in these narrative documents there are many techniques, some we know, some we’re just beginning to learn about, for pulling discrete data elements from these notes. NLP [natural language processing] is an emerging technology: Everyone saw Watson on Jeopardy, so now NLP is the new hot thing.
NLP is clearly a promising technology for extracting discrete elements out of narrative documents. Will it be a panacea, will it be the only way? I think it will be one of the ways. In some cases, [maybe] you’re creating a document that’s going to a particular registry [that has] a need for 25 fields. And you can’t simply rely on a dictated note where you expect an NLP engine to pull out that data, because I may not have stated that data to begin with. We will find many innovative ways to get at discrete data from these documents, and NLP will play a very important role.
Does this effort include other kinds of data, such as diagnostic images?
When we talk about sending narrative docs back and forth, we could talk about sending PDFs, or Word documents back and forth. What I’m talking about is sending documents back and forth that are structured based on the HL7 CDA standard; it’s an XML format for representing clinical documents. So I send you a history and physical document, but it’s an XML CDA format or I send you an operative report, it’s an XML CDA format.
CDA provides a formalism whereby I can send you a narrative document today, but I can start to layer in discrete documents over time, which you can’t do in these other formats. CDA is a multimedia document representation formalism. So in a CDA document, just like with an HTML document, we can support drawings that the clinicians makes in the office, can support multimedia.
We also have a CDA document for diagnostic imaging that we developed in collaboration with DICOM. So the ability to support multimedia was one of the early-in requirements that’s been built into the CDA document standard.