I just had to read an article that dropped into my inbox this morning. It was titled, “5 Tips for Providers to Advance EHR Interoperability“. Though written by a vendor/research firm (Chilmark Research) I felt I had to read the thing. After all, interoperability is a hot topic – and I feel also a misunderstood topic. I was to reaffirm my perspectives after reading the article. We should also of course agree what we mean by “integration” and “interoperability”. I take a simply path as follows:
- integration is what we do when we connect two systems of different architectural design (meaning they may have their own and different data model)
- interoperability is what takes place when two systems are integrated in such a way that the disparate data model differences, at a semantic level, are reconciled and in some fashion neutralized. Thus all things that are interoperable must have been integrated; but not everything integrated can be interoperable.
I think we should look at how system work together at a data level, a business process level, as well as a technology level. I focus on interoperability at the data and a business process level – the technology level is a given. And this is where I don’t fine value in the article.
Overall it seems that interoperability of EHR is not progressing as fast as any of us would like. Offered up as means to promote the free and easy sharing and exchanging of information were 5 “tips”, as follows:
- Incorporate FHIR and Application Programming Interfaces in a Value-based Reimbursement Strategy
- Push EHR vendors to improve software development kits
- Embrace user-centered design
- Avoid further siloing
- Don’t solely rely on independent software vendors or your EHR vendors for apps
I have issues with most of these suggestion – in fact I have issues with all. I feel that the first tip barely touches one of the main challenges; and the second main challenge is not even hinted at. As usual, the tips to achieving interoperability are posed as technology challenges. This is the least of our problem. And I think this is because so few people actually understand how to make interoperability work.
Disclosure: I am an analyst that focuses on information theory and strategy and I do not focus on any one industry. I am certainly not a healthcare specialist. Now, that out the way, let’s talk about what the real problem is with interoperability.
It so happened that a number of years ago I was very close to a small, industry specific initiative (at the time) referred to as Collaborative Planning Forecasting and Replenishment (CPFR). This initiate was sponsored by retailers and their consumer goods suppliers. The goal was to share and synchronize semantically consistent product and other commercial data between enterprises: to make business information interoperable. As part of this the vendors had to do all the usual technical things: expose the interfaces, avoid silos, and open up access to configure applications, and so on. But the real challenge – the two things that slowed adoption, always came back to:
- No sustainable financial model for either publisher or subscriber, or third party enabler, to sustain the effort, and
- No agreed structure explaining the differences in data models and what was needed to assure semantic consistency in the data being shared
No amount of API programming, or SDK’s, or user-centered design helped – these are all good things to do but they will not make interoperability appear out of thin air. If organizations cannot agree on the fundamental building blocks for the information models to be shared, the rest can’t overturn this gap. If those same organizations have no financial model to enforce and encourage such behavior, no amount of interoperability will persist. Just ‘forcing’ vendors or even regulating end-user behavior, is a poor second place driver too. So while the point of the article is fair, the actual tips offered seem to explain why interoperability is so hard to do: we keep applying the same old tips to the same old problem and we forever wonder why the same old result takes place.
As a footnote – the CPFR initiative itself was not overly successful. Technical enough third party vendors did cooperate such that competing products could physically integrate (connected two disparate systems) easily enough such that information was interoperable (could be swapped out and replaced with equivalent semantics). However, everything always and in every case came down to motivation and the lack of a positive value proposition.