Separate but Equal Isn't

Every now and then (actually, more frequently than that), two topics from different places somehow collide in my brain to produce synthesis.  This post is a result of that:

I've been paying a lot of attention lately to care planning.  It's been a significant part of the work that I've been involved in through standards development over the last decade, and it comes to special focus for me right now as I work on implementation.

In another venue, the topic of data provenance came up, and the concern that providers have about patient sourced data.  Many of the challenges that providers have is that the patient concerns don't necessarily use the "right terms" for clinical conditions, or that patients "don't understand the science" behind their problems, or that the data is somehow less exact.  My Father's use of the term "mini-stroke" so annoyed one of his providers (a neurologist who reported to him: "there is no such thing"), that it likely resulted in his failing to get appropriate care for what was probably transient ischemia, resulting in an actual stroke, which eventually lead to his ultimate demise through a convoluted course of care.

This concern about veracity or provenance of patient data leads to designs which separate patient health concerns and information from provider generated data.  Yet those same concerns initiate the evaluation process starting first with the patient's subjective experience, gathering of objective evidence through skilled examination, knowledgeable assessment of those details, leading to cooperative and effective planning.

The care planning work that I've been involved in over the past decade originated in early work on patient centered medical homes driven by physician groups, incorporated work from several different nursing communities in HITSP, HL7 and IHE, and eventually resulted in a patient plan of care design, which was subsequently evolved into work implemented in both the HL7 CCDA and FHIR specifications.

The patient's plan of care should NOT originate from a separate but equal collection of data, but rather, from an integrated, patient included approach, that does not treat the patient subjective experience as being any less valuable to the process.  Both FHIR and CCDA recognize that in their designs.  After all, if the patient didn't have any complaints, physicians wouldn't have any customers.

It's past time we integrate patients into the care process with their physicians, and keeping their data "separate" isn't the right way to go.  If my provider want's to be my medical home, he needs to remember, it's my home, not his, and we, as implementers, need to help with that.

   -- Keith

Late Night Rants

"I am a bit frustrated." ... begins the e-mail that comes to me.  It continues "I am struggling to find in the RIM and CDA where the patient gets to express their voice. "

And indeed, I understand the frustration.  There are at least three issues here:

1. Can the information be modeled in the RIM?
2. Do we have common models (C-METs) from which to draw?
3. Can you represent these models in CDA?

Can the information be modeled in the RIM?

The answer to the first question is YES.  The RIM models acts with participants playing roles played by entities and scoped by other entities. This is rather arcane stuff, as models usually are.  Making things a lot simpler:  I, you, or someone else; in a role as patient, knowledgeable 3rd party, parent/guardian or patient advocate, want to author information about, perform, authorize or otherwise take responsibility for healthcare activities.  The subject of those statements or activities may be a healthcare professional, organization, or diagnostic or treatment activity.

So how would this be represented in the RIM?

For the person performing the activity, this represented in the RIM as a Person class, and would show up as a green box in the model.  The role could be patient, agent, caregiver, guardian, family member, or a few other possibilities.  In the case of the patient role, the "scoper" of the role is the healthcare organization where the patient is getting care.  In the case of all other roles, the "scoper" of the role is the person playing the patient role.

Possible participations include Author, Performer, Responsible Party, Verifier, Legal Authenticator and many others.

Other participants, could be healthcare providers, which would be represented as "living subjects" since they are subject of the activity.  Note that patient as person, and provider as living subject is a reversal of the usual role assignment.  The might appear in the act as the record target (in the patients "record" about their providers), the subject, or as an author of health information or performer of care activities.

So, all of this can be represented in the RIM.

Do we have common models from which to draw?

Not really.  Yes, there are some common models, but they are so generic as to be nearly useless in helping a non-HL7 modeler make use of HL7 content.

Can you represent these models in CDA?

Yes and no.  You can represent probably 60-70 percent of the essence of these models in CDA.  One of the challenges is that the use case for CDA as originally envisioned some 15 years ago did not include extensive capturing of patient authored or expressed content.  So, many model components would need to be inferred.  And since we don't have any truly authoritative common models from which to draw, it is hard even to map into CDA.  I could do it, but I would expected someone who is new to HL7 to understand it the way I would, and arguably, I wouldn't expect other HL7 experts to draw the same diagrams I would.

It actually sounds like a good activity for HL7 to engage in, developing patient-centric models for clinical statements being produced by patients or their representatives possibly about healthcare providers or other activities.  First I think we would need a Conceptual Model to describe what we wanted to represent, then I think we could develop some CMETs which could be used to inform other HL7 modeling efforts.  The most interesting piece for me is trying to imagine where this would go.

The Patient Care work group sounds right, but is almost the diametrically opposite place from where I'd put it.  Why?  Because Patient Care mostly contains healthcare providers.  EHR also seems wrong in some way.  Clinical Quality Improvement?  Not really.  This also is less about domain expertise, and more about structure and semantic design, so maybe I should be looking in that division?  Clinical Decision Support?  Nope.  Structured Documents? Maybe.  Clinical Statement?  Yep. That actually seems like the right place

So, here is my thinking:

Propose a project to the Clinical Statement workgroup to develop a conceptual model of patient-centric RIM models of clinical statements to address clinical statements made by patients or their representatives about their healthcare, or providers or provider organizations, and to generate CMETS which can support those representations, which can be used to support development of additional HL7 models.