Five commitments, and the grammar that sorts them

The Neighbourhood Health Framework, published by the Department of Health and Social Care and NHS England on 17 March 2026, defines its subject in the first line of the introduction: "Neighbourhood health puts the person at the centre of how we deliver their health and care by organising services so they can work together to serve a defined population" (Neighbourhood Health Framework, 2026). Everything else in the document is an attempt to make that sentence operational.

The shortest route into the framework is the letter that accompanied it. On 18 March 2026 the two bodies sent a joint letter to integrated care boards (ICBs) and providers, signed by the national medical director and the director general for system development, which reduced the first year to five requests. Working through health and wellbeing boards, they asked ICBs and local government to: agree neighbourhood footprints around natural communities for the future development of integrated neighbourhood teams; agree plans to establish those teams focused on high priority cohorts, including how devolving care budgets could work locally; confirm intentions to use pooled funding under the Better Care Fund in line with fund guidance; confirm organisational ownership of planned deliverables; and confirm plans for the data-sharing arrangements needed for patient identification and evaluation (Next steps on neighbourhood health and new delivery models, 2026).

Read the verbs. Two of the five ask for agreement to a plan. Three ask only for confirmation. That distinction is not decoration. To "agree neighbourhood footprints" and to "agree plans to establish INTs" requires a decision with a boundary and a cohort attached. To "confirm intentions" and to "confirm organisational ownership" requires a statement of position, and the letter does not say who must sign it or what happens if the position later changes. The thesis of this week follows from that grammar. A system that reads the five asks closely can tell which of them a plan has actually delivered and which it has only restated.

Three reform agendas and a floor, not a ceiling

The framework organises delivery under three reform agendas. The first is to improve services for people who need routine healthcare, so that neighbourhood health benefits everyone, which turns on general practice access, a red tape challenge at the primary and secondary care interface, and a larger role for community pharmacy. The second is to improve proactive care for people, delivered mainly through integrated neighbourhood teams and through a new approach to planned care. The third is to deliver better alternatives to hospital care: urgent community response, virtual wards, intermediate care, and, for some systems, neighbourhood mental health centres (Neighbourhood Health Framework, 2026).

Under those agendas the framework sets what it calls a floor. It asks ICBs "to implement a series of minimum interventions in every community over the next 3 years", and it is explicit about their status: "These are not the ceiling of neighbourhood health, but the foundation upon which local priorities will be built." The national goals carry numbers. For people with mid to severe frailty, in a care home or housebound, the aim is to reduce non-elective admissions and bed days of one day or over by 10% by March 2029. The case for concentrating on this cohort is that people over 75 living with frailty, people at end of life and care home residents are 3 to 5% of the population yet account for over 25% of non-elective admissions and 50% of bed days. Community health service activity should reach 78% within 18 weeks in 2026 to 2027 and 80% by 2028 to 2029, and 95% of people with complex needs should hold an agreed care plan by 2027.

The permissive parts sit alongside the floor, and they are just as deliberate. The NHS "will not define nationally what should constitute an INT". DHSC and NHS England "will take an enabling, non-prescriptive approach, allowing local systems to determine optimum models". On size, the working assumptions are a single neighbourhood provider for around 50,000 people and a multi-neighbourhood provider for 250,000 or more, but the framework states that it "will not mandate nationally the size of neighbourhood health geographies". Two red lines remain: hospital standard contracts and general medical services contracts stay the primary vehicles of delivery for the two biggest groups of NHS providers, and neighbourhood health is to be delivered "through commissioning reform" rather than through organisational upheaval.

What changes in 2027/28, when the plan arrives

The five asks are groundwork for a single document. From 2027/28, ICBs and local government, working through health and wellbeing boards, must develop a local neighbourhood health plan. The letter lists what it must contain, and the framework repeats it in almost the same words: a broad overview of how the national NHS objectives will begin to be delivered through the three reform agendas; how neighbourhood health will support wider local goals on outcomes and inequalities; how local objectives are informed by the joint strategic needs assessment; the final geographies partners will work within; which organisations are responsible for different elements of delivery; the arrangements, including governance, that will deliver it; and how other local services or initiatives, such as Best Start Family Hubs, housing, mental health hubs, Pride in Place and employment support, align with the strategy (Next steps on neighbourhood health and new delivery models, 2026).

Two shifts matter. The first is from initial footprints to final geographies, which converts a planning assumption into a contracting boundary. The second is the move from confirming ownership to confirming "which organisations are responsible for different elements of delivery". The 10 Year Health Plan is blunter about where the power sits. A neighbourhood health plan will be drawn up by local government, the NHS and their partners at single or upper tier authority level under the leadership of the health and wellbeing board, incorporating public health, social care and the Better Care Fund; the ICB then brings the local plans together into a population health improvement plan for its footprint, and integrated care partnerships are abolished (10 Year Health Plan, 2025). Once agreed, the local plan is incorporated into the ICB's refreshed five-year strategic commissioning plan, which becomes the formal NHS commissioning strategy for neighbourhood health (Neighbourhood Health Framework, 2026).

From six components to a floor: the direction of travel

The direction of travel is visible only against the previous year. The 2025/26 neighbourhood health guidelines, published in January 2025, described themselves as "deliberately short and permissive about how neighbourhood health should be implemented", and set out six core components: population health management; modern general practice; standardising community health services; neighbourhood multidisciplinary teams; integrated intermediate care with a Home First approach; and urgent neighbourhood services (Neighbourhood health guidelines 2025/26, 2025). Those components were to be standardised, connected and scaled, with an initial focus on a cohort estimated at around 7% of the population and around 46% of hospital costs, narrowed in practice to perhaps 2% to 4%.

What changed in 2026 is not the component list. It is that the list acquired an owner and a timetable. The framework added a national floor of minimum interventions over three years, a set of goals with dated trajectories, and two stages: ten minimum basic requirements for 2026/27, and a locally owned plan from 2027/28 within a period that runs to March 2029. The guidelines asked systems to consider six components. The framework asks ICBs to deliver a defined minimum and to show their working.

Pooled money and shared data: where the paper test bites

Two of the five asks are process commitments that carry most of the delivery risk, and both have an evidence base.

Pooled funding was first mandated in England through the Better Care Fund, and it has been evaluated. Harlock and colleagues interviewed 40 staff responsible for the fund across 16 sites and concluded that pooling "can lever closer collaboration between sectors and services", but that "major cultural, operational and territorial barriers" remained, along with "the practical and political challenges of managing pooled budgets and the complexity of working across geographical boundaries" (Harlock et al., 2020). Stokes and colleagues tested whether pooling changed hospital use across a cohort of 14.4 million patients and found "no effects of budget pooling on secondary care use for the whole population", with bed days for multimorbid patients rising by 4.9% per patient per year in the short term, while noting that pooling "does appear to stimulate additional integration activity" (Stokes et al., 2019). A system that confirms an intention to pool money has therefore signalled a change in governance. It has not promised a fall in admissions.

Data sharing is the ask on which the others depend. Ling and colleagues evaluated the English Integrated Care Pilots and found that "different IT systems in partner organisations caused difficulties in data-sharing and communicating, especially across health and social care teams", that one pilot's "necessary agreement to share data between health and social care was never forthcoming", and that money "was often tied up within that organisation's spending plans" (Ling et al., 2012). Lalani and colleagues, studying quality management in integrated care systems, found that barriers "included slow progress in establishing data sharing agreements and the disparate availability of integrated care records" (Lalani et al., 2025). The framework asks for data-sharing arrangements for patient identification and evaluation rather than for a single shared record, and the evidence suggests the arrangement, not the intention, is the binding constraint.

Freedom to innovate, or cover to do very little?

That is the question the five asks force. The framework's own answer is that it grants freedom: it sets a minimum, refuses to prescribe a model, and leaves footprints, team composition and provider form to local decision. The evidence does not support reading that freedom as neutral.

Alderwick and colleagues interviewed 32 senior leaders across three integrated care systems serving deprived populations and found that leaders "had contrasting, sometimes conflicting, perceptions of the boundaries" of their system's role, that "clarity from national policy-makers was frequently perceived as limited or lacking", and that they "worried that objectives on reducing health inequalities were being crowded out by other short-term policy priorities". The authors noted that "lack of clarity was not always seen as a drawback by local leaders, given they often wanted flexibility to address local needs", while warning of "interventions being poorly targeted" and of systems that "may even inadvertently widen inequalities" (Alderwick et al., 2024). A permissive framework is compatible with the strongest and the weakest local response, and that study contains both.

The integration literature sharpens the risk. Baxter and colleagues reviewed 167 documents and rated the evidence stronger for patient satisfaction, perceived quality of care and access, but "inconsistent or limited" for system-wide impacts on primary care, secondary care and cost, concluding that integration "rarely leads to unequivocally positive effects" and that new models "may be best targeted to particular patient groups, such as those with complex needs" (Baxter et al., 2018). Damery and colleagues, in an umbrella review of 50 systematic reviews, found that 29 reported significant improvement in at least one outcome, but that interventions "rarely demonstrated unequivocally positive effects", with the clearest gains for single conditions such as heart failure and where care was delivered at home (Damery et al., 2016).

So the framework gives real freedom of model and very little cover on results. What separates innovation from inaction is whether a plan passes three tests that the five asks themselves supply. Does a named organisation own each deliverable, as the fourth ask requires? Does a boundary on the map carry a cohort and a caseload, as the first and second require? And does the data-sharing arrangement exist before the outcomes are claimed, as the fifth requires? The wider evidence suggests that the failure mode is not a system that refuses to act. It is a system that confirms every intention, agrees every plan, and never names the person who will hold the budget or the measure by which the plan will be judged. A reader can test their own system against the five asks in an afternoon, and the answers will say more about the next three years than the framework's ambition does.