Medical records, run by clinicians
The Medical Records Trust is a new non-profit that will hold medical records on an open standard, under the control of the clinicians who are responsible for them.
We believe that granting for-profit entities de facto control over medical records poses unacceptable operational and ethical risks. We see direct evidence that this practice drives systemic waste, compromises patient safety and significantly exacerbates health inequalities.
The problem
GPs are the data controllers for their patients' records. In law, you're in charge. In practice, the records live inside a supplier's system, and the supplier decides what the data can do, what it costs to move and what gets built on top of it. We think that's the wrong way round.
The result is that, despite substantial investment in electronic health records, healthcare data remains fragmented across organisations and proprietary systems. Sharing information, coordinating care and developing new clinical services are all much harder than they need to be for the sake of shareholder value.
The established suppliers have little reason to change this. Their commercial position rests on holding the record: the harder it is to move, the more secure the contract. That isn't a comment on anyone's intentions. It's where the incentives point.
Why connecting the existing systems won't fix it
The usual answer is federation: leave every record where it is, and network the suppliers' systems together so that they behave like one national record. A great deal of time, money and goodwill has gone into that idea. We know this can't work, and the reason is engineering, not effort.
A network of separate systems, each holding its own version of a patient's record, can't guarantee that everyone sees the same, current record whenever they need it. That is a well-established limit in computer science, not a gap that better software will close. Federation also asks the most of the oldest systems, run by the organisations with the least reason to make it succeed.
What's missing is a single system of record: one master copy that everything else works from. Until someone takes responsibility for that, interoperability stays out of reach.
Why federation fails: the technical detail →
What the Trust does
The Trust provides an open-standard data layer for medical records. It holds the record in FHIR, the open standard the NHS already uses, and it looks after who can get at it.
It doesn't build clinical systems. Third-party organisations do that, on top of the Trust's data layer, for CQC-registered providers. Because the record sits with the Trust and not inside any one product, a practice can change its systems, or use several, without moving its records.
The Trust isn't another proprietary platform to replace the ones we have. It's an open, standards-based foundation that existing suppliers, incumbents included, and new entrants can all build on. That reduces the need for bespoke integrations, and it lets technology providers compete on the quality and value of their products, not on their control of the record.
Who's behind this, and why
The Trust is being developed with the support of CogStack, which builds clinical informatics and AI technology for healthcare. You're entitled to ask what CogStack gets out of it, so here is the answer.
CogStack has a commercial interest. Its tools are only as useful as the records they can reach. Today, reaching them means a separate integration with each supplier's system, on that supplier's terms. An open record layer would let CogStack, and anyone else, build complete clinical products on a common foundation. We believe that means wider adoption of CogStack's technology, and commercial growth.
That opportunity isn't exclusive to CogStack. Every technology provider gets access on the same fair and transparent terms, set by the trustees. CogStack would be one supplier on the platform among many, bound by the same rules as the rest.
The interests of CogStack and the Trust align for a simple reason: CogStack only benefits if the Trust is trusted, and it will only be trusted if it is independent. A Trust that CogStack controlled would be a new silo with a new owner, which is the problem we set out to fix. So control sits with the trustees, not with CogStack.
Who's in charge
The Trust will be fully controlled by its trustees. The trustees are clinicians, and the majority are GPs. That's deliberate: GPs have the most experience of the current situation, so they need to play a key part in building what replaces it.
The trustees set the rules: who can access records and why, which apps are allowed on the platform, and what the Trust charges.
How it works
Everything rests on one idea: there is a single door to the records, and the Trust keeps it.
- Apps run inside the Trust. Third-party apps are audited and deployed within the Trust's own environment. The apps come to the records; the records don't go out to the apps.
- One way to sign in. The Trust provides the sign-in, and it's the only way an app can find out who the practitioner is. Apps can't bring their own logins.
- One way to the records. Apps can't reach the records directly. Every read and every write goes through a single checkpoint, the permissions choke point, and carries the identity of the practitioner making it.
- Checked, every time. The choke point tests each request against the trustees' rules: who this person is, where they work, and whether they're allowed to see or change this record. If the answer is no, nothing comes back.
- Logged, every time. Every request, allowed or refused, is recorded against a named practitioner.
No identity, no access. Permissions and the audit log both depend on knowing who's asking, and the Trust's sign-in is the only way to establish that. An app can't quietly work round the rules, because there's no other way in.
Under the bonnet: records are stored and exchanged as FHIR. Signing in gives the app a short-lived, signed token that names the practitioner. The choke point sits in front of the FHIR interface and is the only thing with a route to the records.
How it pays for itself
The Trust funds itself with small fees on healthcare services traded on the platform. CQC-registered providers will be able to post prices for specific SNOMED-coded procedures: phlebotomy or radiology, for example. When one is ordered through the platform, the Trust takes a small fee on the transaction.
It's a non-profit, so what it earns goes back into running and improving the platform. It won't sell patient data.
Getting there
Nobody should move a practice's records onto something unproven, so we plan to get there in two steps.
- First, alongside your existing system. The Trust holds a copy of the record, kept up to date from the practice's current clinical system, which stays the system of record. If anything goes wrong with the Trust's copy, it can be rebuilt from the source. IM1, the NHS's standard interface to GP clinical systems, is what makes this step possible, and it's a large part of why this hasn't been done before.
- Later, as the system of record. Once the platform has proved itself, the Trust's copy becomes the master record. This is the harder step, with much higher stakes, and it will only happen when the trustees are satisfied it is safe.
The detail of each step is still being worked out, and we'd welcome your help with it.
The long game
The Trust's long-term mission is to advance the art of medical record keeping, for the benefit of medical practice and of patients. In time, we hope it becomes a centre of excellence for everything to do with how medical records are created and used.
