Dr. Rosie Benneyworth, Interim Chief Executive Officer of the Health Services Safety Investigations Body (HSSIB)

Picture an outpatient booking office on an ordinary Tuesday. An administrator scrolls a worklist of referrals awaiting appointments, some weeks old and some far older. One of them has been sitting unactioned for six months. On day 180 it leaves the list without a warning, an alert or a record. The GP who made the referral assumes the patient is waiting. The hospital sees nothing to book. The patient waits for a letter that nobody is going to send.

That scenario is drawn from findings published by the national patient safety investigator in two interim reports on the electronic referral service. The service removes a referral from a provider's worklist after 180 days without action, a function designed to stop very large lists degrading system performance. The investigators found referrals that vanished this way, and others that disappeared for reasons nobody could establish. They also found that the hazard had never been formally documented in the national clinical safety paperwork for the platform. The risk is hard to size, but it has probably grown as outpatient waits have lengthened.

The detail that matters most is where responsibility has landed. Management of the 180-day risk was delegated to provider organisations, and the investigation found that many did not know the function existed. Providers could not show compliance with the clinical safety standard that governs local deployment, and several lacked a hazard log. Where controls existed, they were administrative, informal and dependent on staff time. Repeat incidents occurred where administrative capacity had been cut or roles had become unclear. A design decision taken nationally became a housekeeping task locally, with no guarantee that anyone owned it.

This is a story about state capacity as much as software. The platform was not built for the referral volumes the NHS has handled since the pandemic. In any specialty where waits approach six months, a function that once removed stale entries now removes live patients. Planned upgrades have been delayed, and the investigators describe the function as necessary, with the risk impossible to eliminate for now. That is a candid admission that the safest answer is unavailable and that vigilance must substitute for engineering.

Vigilance is exactly what current policy is squeezing. Integrated care boards face steep reductions in running costs, national bodies are being restructured, and trusts are under pressure to convert administrative posts into clinical ones. The staff who check worklists, chase rejected referrals and reconcile complaints against booking data rarely feature in productivity narratives. Their work is invisible until it fails.

The wider referral model compounds the problem. Advice and guidance, and the single point of access arrangements built around it, are central to the effort to divert demand from hospitals. The investigators found local pathways that obliged GPs to seek advice before referring, and referrals declined despite continuing clinical concern. They do not claim that list management drove these decisions. The incentive is plain all the same. A patient who is advised and never referred does not join the waiting list, and a referral that expires does not stay on it. Both outcomes flatter the headline figure.

The practical implications are specific. Provider boards should confirm that a named clinical safety officer exists for the referral service, that the 180-day hazard sits in a local hazard log, and that someone audits for referrals that have dropped off. Patient safety, digital and complaints teams need a shared view of lost referrals, because no single team currently sees them. Ministers and national bodies should accept that a national platform carries national risk, which cannot be fully devolved to organisations that did not design it. Suppliers and health-tech firms working on interoperability will find that the case for modern referral infrastructure is now evidenced in patient harm.

The government's plans for a more digital, community-facing NHS assume that records can be trusted to carry patients between organisations. A waiting list is only as reliable as the system that records who belongs on it, and until that system is made safe, every promise to shorten the list rests on figures that may be missing some of the people waiting.

‍