Tuesday, November 6, 2012

The AMIA Healthcare Information Exchange Debate


Today I'm in Chicago at the American Medical Informatics Association annual meeting, joining my colleagues Mark Frisse, Bill Yasnoff and Latanya Sweeney to debate the question:

"Resolved - Health information exchange organizations should shift their principal focus to consumer-mediated exchange in order to facilitate the rapid development of effective, scalable, and sustainable health information infrastructure."

Mark and I were assigned "oppose".   Bill and Latanya were assigned "support".   It was made clear that our positions were assigned and did not necessary reflect our personal opinions.  (Note to Christine Bechtel and Leslie Kelly Hall - you know how I feel about the question of patient and family engagement.)

Here's what I said:

I really like the idea of patient mediated exchange and eventually we will widely support both provider and patient mediated exchange (as Meaningful Use Stage 2 will require).  However, in the short term, there are implementation issues that will delay widespread use of patient mediated exchange.

a.  There are 500,000 providers in the US and 300 million patients.   Doing identity management on 500,000 licensed/credentialed professionals easier than issuing credentials to 300 million patients.
b.  Clinicians fear that loss of data integrity will result in increased liability.   How will we ensure the non-repudiatibility of data exchanged between providers if patients collect and edit it first i.e. might Tylenol #3 for pain be changed to Oxycontin for pain?  At present we lack the metadata and digital signatures that will guarantee provenance and integrity of patient mediated data
c.  Many EHRs include features that support provider to provider workflow, but few accept incoming patient generated or stewarded data

These are short term issues that will be address in the next few  years, but the resolution calls for "rapid development".  

Why will provider to provider exchange be more rapid to implement?

1.  Provider mediated exchange is simple

HIEs can push data from organizational entity to organizational entity without having to uniquely identify the patient on a community-wide level.   Although there are many Mary Smith's in the community, there are very few in an individual provider's practice. When a message arrives to a provider concerning Mary Smith, the provider can easily attach it to the correct record.    In Massachusetts there are 20,000 providers and many are associated with a few large organizations running about a dozen different EHRs.  In our HIE we can do connect everyone with a few hundred organization level network connections.   Compare this complexity with the issue of messaging to 7 million unique patients.    

2.  Public and Private provider-based exchanges are already implemented.  Per a recent survey completed at the School of Public Health, over 100 HIEs are actively exchanging real data in the US.  Massachusetts has been exchanging data since 1997 and its HIEs have always been sustainable.

3.  Pushing data between providers does not require complex consent frameworks, it simply replaces the fax machine used in today's processes.   Thus the policies around using an HIE for pushing data are already in place.

4.  Existing EHRs and PHRs support provider directed exchange, since many federal and state demonstration projects have focused on provider-based architectures.

5.  Although we ultimately need both provider and consumer mediated exchanges, I predict 80% of patients will defer to their provider.  My parents, like many older Americans believe  their providers should collect and organize the data, serving as a kind of patient-centered electronic medical home.   Patients can view the collected data via the PHR offered by their primary care clinician.

I wonderful set of point/counterpoint discussion on this topic filled 90 minutes.

The end result - the audience seemed evenly split on the resolution.   We were both right - provider and patient mediated exchanges are needed.

A great discussion.

3 comments:

Adam Trahan said...

Wow, that is a pretty useful resource. Hopefully it will continue to grow.

I'd love to see HITSP/C83 in there.

Unknown said...

The "start simple" use case I imagine is that the patient is at the doctor or at the ED, and can say to the doc: hey, here's my info. It replaces the mis-remembering that goes on when patients arrive unnanounced or without notes from referring doc. In this scenario dont need to tackle identity and consent (yet) just need to give patients easy ways to aggregate own info (e.g, VDT), and ways to demonstrate that clinical info has not been changed (non repudiation).

Anonymous said...

Would you mind sharing the points presented by the other side of the debate? I am interested in knowing the scenarios where patient mediated exchanges are "needed" ... i.e. provider mediated exchanges don't work.