For three years, two hospital trusts on either side of the Humber estuary have run as a single enterprise sharing one executive team, one board and, increasingly, one set of problems. That arrangement is now ending. Following an external review commissioned by NHS England and carried out by the consultancy Carnall Farrar, the board of NHS Humber Health Partnership has voted to dissolve the group structure. Hull University Teaching Hospitals and Northern Lincolnshire and Goole will each revert to standing alone, with their own chairs, chief executives and management teams restored in due course.

The decision closes a case study that NHS England would rather not have generated. Group models, in which neighbouring trusts pool executive leadership while remaining legally distinct organisations, were promoted nationally as a way of spreading scarce management talent and embedding shared standards across struggling providers. Humber became instead a demonstration of the opposite risk: that combining two troubled trusts under one thin layer of leadership can dilute accountability faster than it builds capability. HUTH finished bottom of NHS England's acute trust league table. NLAG sat only marginally higher. Both organisations were served Section 29A notices by the Care Quality Commission demanding urgent improvement, and the partnership was placed in the lowest tier of the National Oversight Framework. Consultants at NLAG went so far as to write to the group chair stating they had no confidence in the chief executive's handling of the structure, arguing that decisions affecting services on the south bank of the Humber were being made without adequate local engagement. In August, NLAG's council of governors passed a motion unanimously calling for separation.

None of this happened because group working is inherently unsound. It happened because scale was applied to a set of problems that were more structural than administrative: financial strain, workforce shortages and estate constraints that predate the merger and will outlast the demerger. Recombining leadership teams did not resolve those pressures; it added a layer of governance that made it harder to see who was responsible for fixing them. That is the uncomfortable lesson for NHS England as it continues to encourage group models elsewhere in the country as a lower-cost alternative to full trust mergers.

For NHS leaders watching from other regions, the practical implication is sharper than a cautionary tale about one partnership. Group structures are attractive to national policymakers because they promise efficiency without the cost and disruption of formal merger. Humber suggests that promise depends entirely on execution: on whether shared leadership genuinely strengthens grip on operational performance, or simply spreads a finite amount of executive attention across a wider footprint of failing services. Where the underlying trusts are already stretched, doubling their span of control is unlikely to be the answer.

For patients and staff in Hull, Scunthorpe and Grimsby, separation will not itself shorten waiting lists or ease the pressure on emergency departments. The clinical and financial constraints that produced the CQC notices remain in place, and two smaller executive teams will now have to tackle them with less combined capacity than before, even if with clearer lines of accountability. Whether that trade-off produces better outcomes will depend on how quickly new leadership can be appointed and how much support NHS England is willing to provide beyond the Intensive Recovery Programme that got the partnership this far.

The wider test is for the department overseeing NHS reform. Group models sit at the heart of current thinking about how a financially constrained health service should organise itself without resorting to costly full mergers. Humber's unwinding does not discredit that thinking outright, but it does puncture the assumption that shared leadership is a low-risk default. Structure was never going to substitute for capacity, and the Humber board has just spent three years, and two trusts' worth of performance data, confirming it.

‍