Over the past year, leadership of almost every mental health trust have urged significant changes to be made around data access. Most mental health trusts run their clinical records on SystmOne, the electronic patient record built by the Leeds-based company TPP. Executives have discussed issues covering workforce, demand, beds and money. These conversations all ended in the same place. Each trust, in its own words, asked for help getting a better data feed out of the system.

It is worth pausing on how strange that is. The information in question was recorded by NHS staff, about NHS patients, in a system the NHS pays for. Yet the organisations responsible for some of the most vulnerable people in the country describe getting hold of it as a problem they need rescuing from. Tissue deserves attention well beyond one supplier, because it goes to the heart of whether the health service can actually run the systems it already owns.

Consider a ward manager on a Monday morning. She needs to know which of her patients are clinically ready to go home, why each of them is still there, and who is responsible for moving things on. National reporting already expects precisely this. Under the Mental Health Services Data Set, trusts record the date a patient becomes clinically ready for discharge and the reason they cannot leave: no care coordinator allocated, a funding panel yet to decide, a dispute over which commissioner pays for what comes next. When the reason changes, a new record begins. The data is designed to tell a story over time. A dashboard showing only occupied beds tells the manager she has a problem. It cannot tell her what the problem is, and each of those reasons demands a different phone call.

Mental health makes this harder than almost anywhere else in the NHS. So much of what matters is narrative: the risk formulation, the family circumstances, the conversation in which a discharge plan was agreed or fell apart. And the cost of not knowing is unusually high. Every bed occupied by someone who no longer needs it is a bed unavailable to someone in crisis, who may be waiting in an emergency department or sent many miles from home for care.

In fairness, TPP does not claim to keep data locked away. Quite the opposite. Its own description of SystmOne Mental Health promises an in-built analytics suite, the extracts needed for national submissions and a complete, open data extract for trust data warehousing. Its principal route, known as Strategic Reporting, is a bulk extract intended to feed an organisation's own warehouse, and some trusts use it to real effect.

The gap lies in the detail. Technical documentation maintained by researchers in Cambridge University's psychiatry department describes the Strategic Reporting extract as a nightly download that, by default, leaves out free text and documents. Setup notes published by one analytics supplier for GP practices describe finding a computer running SystmOne in a special gateway mode, logged in as a user with particular permissions, then checking a folder the following day to see whether files have arrived. None of this is unusual by the standards of NHS technology, which is rather the point. An overnight file serves last month's board report well enough. It is of limited use to the manager deciding about today's admissions, and the free text it omits is where mental health care is most fully described.

A file is also the start of the work, not the end of it. Someone must understand hundreds of tables, link them correctly, interpret local coding and notice when an update quietly fails to arrive. Large trusts can build teams to do this. Smaller ones may have a handful of analysts already stretched by national returns. Local practice matters too: two trusts on the same system can record the same event in different ways, and no supplier can repair data that was entered inconsistently. Any honest account has to accept that part of the difficulty sits with the NHS itself.

That is an argument for clarity, not fatalism. When the same plea arrives from nearly every trust using one system, it stops being a story about local implementation. Either the offer is harder to use than it appears, or the support behind it is too thin, or the NHS has never specified clearly enough what it needs. Probably all three. Each of those has a different owner, and at present nobody has been asked to take responsibility for the whole.

The law has now changed in a way that makes that question harder to dodge. The Data (Use and Access) Act 2025 makes information standards mandatory for the IT suppliers that serve health and care in England, creates an accreditation scheme to show compliance and gives the Health Secretary power to publicly censure or fine suppliers who fall short. Parliament has handed the NHS a lever it did not previously have. The test is whether it will be used for something as unglamorous as a data feed.

It should be. Every electronic record supplier ought to set out, in terms a trust can test, what data it can provide, how quickly, with what documentation, at what cost and how failures are detected and put right. Trusts, in return, should commit to consistent recording and to the analytical capacity needed to use what they receive. And the centre should define what good looks like for mental health specifically, rather than leaving every trust to negotiate its own version of the same answer.

There is a further reason for urgency. Ministers speak often, and rightly, about the promise of artificial intelligence in the NHS. A model that predicts which patients are at risk of a delayed discharge is only as good as the information fed into it. A health service that cannot reliably get yesterday's data out of its own records is not ready to ask machines to forecast tomorrow.

The trusts that made this plea are not asking for anything exotic. They want to see clearly what is happening to their own patients, on their own wards, as recorded by their own staff. That they have had to ask at all should trouble anyone who believes the NHS can modernise. A patient record that cannot tell a hospital what is happening on its own wards is doing only half its job.