The NHS interoperability problem gets discussed a lot, usually with the same conclusion: proprietary systems, expensive integrations, fragmented records. Less discussed is what a structurally durable answer might look like — one that doesn’t require replacing everything, but does require agreeing on something.
I’ve been writing about this over at ProtoFlex, where I’ve been doing some early exploratory work with OpenEHR.
What makes OpenEHR different
OpenEHR’s core idea is that instead of every system owning a proprietary data model and exposing bespoke APIs, you agree on a shared clinical data model — using clinician-authored archetypes — that systems write to and read from. Data lives in a canonical form. Systems can still be different, still be specialist, still compete on what they’re actually good at. They just speak the same language underneath.
What I find genuinely compelling about that model is the distinction it draws between integration and interoperability. Integration is bilateral — it requires someone to build and maintain a connection between two specific systems, and the cost compounds as the landscape grows. Interoperability, if the standard has enough depth and adoption, means you write once and you’re talking to everyone else who’s also writing to the same model.
Best-of-breed without the silos
The tension I see in NHS digital strategy is that EPRs, which do a great deal well — coherent patient records, clinical safety infrastructure, wide coverage across care settings — often end up as the default answer for workflows they weren’t really designed for. There are areas where specialist tools, built for specific clinical domains, can genuinely deliver something better than the generic equivalent inside a large platform.
The problem has always been that deploying specialist tools risks creating more silos: another system, another integration, another export. OpenEHR is an interesting structural answer to that specific problem. If the data flows into a shared canonical model regardless of which tool captures it, you get the specialist capability without the fragmentation.
I’m not suggesting it’s simple - archetype governance requires genuine clinical engagement, trusts need CDM infrastructure, and the benefits compound meaningfully only when adoption reaches critical mass - but the direction feels right, and it aligns well with where NHS interoperability policy seems to be heading.
I’ve been exploring this at ProtoFlex — see the detail over there, including an early proof of concept and where the harder problems lie.
If you’re working on this from any angle — trust informatics, system supplier, ICS — I’d be interested to compare notes.