Wednesday, June 10, 2015

Standards Alone are not the Answer for Interoperability

Today I have the honor of presenting a guest blog by David McCallie MD, SVP Medical Informatics, Cerner.   He summarizes the collective feeling of the industry about the trajectory of interoperability.

"I have been honored to have served on the HIT Standards Committee from its beginning in 2009. As I reach my term limits, I have reflected on what we have all learned over the past six years of helping to define the standards for the certified EHR technology that lies behind the Meaningful Use program.

But before diving into those reflections, we should acknowledge what we have accomplished.  It's fashionable to complain that we haven't 'solved interoperability.'  This is certainly true, but we have come a long way, and have achieved significant and lasting advances.

In particular, we have mostly settled the vocabulary questions for encoding the record.  We have widely deployed a good ePrescribing standard. We have established a standard for secure email that will eventually replace the fax machine, and we have widely (but not yet universally) deployed a good standard for document-centric query exchange.  We are close to broad support for a stable standard for encoding complex clinical data into summary documents.  And perhaps most interestingly, we have opened the door to the promising world of API-based interoperability.

Nonetheless, the refrain we hear from Capitol Hill is that we have failed to achieve the seamless interoperability that many had expected. This has lead to numerous legislative attempts to 'fix' the problem by re-thinking government approaches to the standard setting processes authorized by HITECH.

We should be careful not to overreact in light of any disappointments and perceived failures around interoperability.  There are many things we must improve, but we should not inadvertently take steps backwards.

I think the biggest mistake Congress appears to be legislating is to assume that standards alone are what creates interoperability.

Standards are necessary, but are not sufficient, for interoperability to occur.

Standards organizations (SDOs) can create standards, but they do not create the additional entities that are necessary for standards to deliver useful interoperability.

I believe that the sufficient conditions for interoperability include the following:

*A business process must exist for which standardization is needed. As Arien Malec put it recently, 'SDOs don't create standards de novo. They standardize working practices.'

*A proven standard then needs to be developed, via an iterative process that involves repeated real-world testing and validation.

*A group of healthcare entities must choose to deploy and use the standard, in service of some business purpose. The business purpose may include satisfying regulatory requirements, or meeting market pressures, or both.

*A 'network architecture' must be defined that provides for the identity, trust, and security frameworks necessary for data sharing in the complex world of healthcare.

*A 'business architecture' must exist that manages the contractual and legal arrangements necessary for healthcare data sharing to occur.

*A governance mechanism with sufficient authority over the participants must ensure that the network and business frameworks are followed.

*And almost no healthcare standard can be deployed in isolation, so all of the ancillary infrastructure (such as directory services, certificate authorities, and certification tests) must be organized and deployed in support of the standard.

The Jason Task Force (which I was privileged to co-chair with Micky Tripathi) summed up these requirements into what they called a 'Data Sharing Arrangement (DSA).' DSAs do not emerge from SDOs. DSAs cannot be simply created by legislative fiat.  They require the active engagement and collaboration of the many entities who are willing to do the hard work to create the necessary infrastructure for an interoperability standard to find real-world, widespread use.

We've learned that forcing providers to simply deploy standards and then to expect interoperability to happen is ineffective policy. The most rapid progress has happened when DSAs have emerged because provider organizations, Health IT developers and SDOs agree to work together to achieve a clear mission, focused on meaningful outcomes. Congress should accordingly define the 'what' and let the U.S. health care stakeholders define and achieve the 'how.'"


Grahame Grieve said...

Thanks Dave and John. I just wanted to add to this that legacy data and legacy practices are a big part of the problem of getting interoperability. I'm sure congress and other cheerleaders for 'interoperability' don't want us to abandon legacy data - that is, everyone's existing healthcare status, nor to simply dump existing work practices and try some new synthesis with unknown safety characteristics. And unless we do that, working interoperability will remain a slow process, no matter how much push there is.

David said...

This was a very thoughtful, realistic, and balanced guest blog. "Necessary but not sufficient" was an appropriate way to say it. Stated another way, there is no single silver bullet to solve interoperability, but rather an interplay of several factors, technical and nontechnical. Iterative improvement based on learning experiences are key. The standards, regulations, policies, and infrastructure already in place indicate significant progress, but all have much room for improvement, and several efforts are underway to address the gaps. Thanks for this post.

Will Ross said...

David & John - Thanks for bringing this message back to the discussion. It is not news to those of us who develop, deploy, and operate technology that standards are "necessary but not sufficient." When interacting with our nontechnical colleagues it is always helpful to look for teachable moments when this nuanced message can be reinforced. It is obvious to us but our colleagues need an occasional reminder that to be adopted a standard must pass the crucial test, which is clear utility for the user. Recent regulatory and legislative activities make this a perfect time to repeat this message. Btw, one of my favorite examples of this message is "A Few Years Of Magical Thinking" by Clay Shirky and Carol Diamond, published in Health Affairs in 2008 ( Here's a great quote from that article: "It seems tautological, but standards aren’t really standard unless they are widely adopted, and this step can’t be easily mandated." Keep up the great work!