Why the NHS keeps choosing hospitals

What Bernstein and Bourdieu can tell us about the drift to the right

Yesterday I saw Fraser Battye’s post on LinkedIn. He was referring to a new BMJ paper by Pettigrew and colleagues and asking why we have allowed hospital doctors to expand at the expense of GPs. It fits with a longer strand of the Strategy Unit’s work — that needs in the community are going to expand, that the increase in healthcare spending is not driven by the ageing population but by the “residual”: the choices made by services. Some of that is certainly supply-led. The expansion of hospital doctors is costing the system more and not necessarily proportionately improving outcomes. To a certain extent it may even be making things worse, because it takes resources away from other things that improve outcomes more.

This is a very big question. It is of the same kind as: why do we have inequality? Why do we keep sending people to prison rather than sorting out education and chances for people in poor areas? The question sounds as if it is about one group of professionals versus another, but it is about something more than that.

I’ve been a GP for 26 years. I have worked in England, Wales and Northern Ireland. I have witnessed the patterns of investment and I have puzzled over them with everyone else. But 15 years ago I also started an EdD programme in a school of social sciences. When I encountered Bernstein’s work on how education systems reproduce hierarchy, it was the first time I had a framework for something I’d felt but never been able to name: why specialism is valued over generalism, not as an individual prejudice but as a recurrent pattern. I think what I learned there might help answer Fraser’s question.

The numbers

In 2004, for the first time, the number of hospital consultants in the NHS overtook the number of GPs. In October 2011, Iona Heath — a GP from Kentish Town — stood up in the Royal College of Physicians to deliver the Harveian Oration, the College’s most prestigious annual lecture, and warned that the balance was tipping dangerously. She was, as far as the record shows, the first practising GP to give the oration in its 355-year history.

As Fraser notes, Simon Stevens flagged this in 2016, writing in the GP Forward View that if anyone ten years earlier had proposed cutting the share of primary care funding and growing specialists three times faster than GPs, they’d have been laughed out of court. But that’s exactly what had happened.

Pettigrew’s data shows what happened next. By December 2024 there were more than twice as many consultants as GPs — 58,382 FTE to 28,197. The consultant workforce has grown steadily. The GP workforce has been flat. But “full-time equivalent” means something different in each case, and the comparison overstates the gap in clinical work. A GP’s FTE is built from clinical sessions — and Parisi and colleagues showed in the BJGP in 2024 that these sessions have expanded to an average of 6.2 hours. A GP recorded as “part-time” at six sessions is doing 37 hours of clinical work a week. A “full-time” consultant’s contract includes protected time for research, teaching, and professional development — about 30 hours of direct clinical care. The system hasn’t just grown one side faster than the other. We also count the work in a way that makes the disparity look larger because it makes GP work less visible.

Fraser thinks we don’t understand what is pulling resources towards hospitals, and that until we do, no policy will reverse the trend. His own explanation is that medical innovation concentrates in research-intensive, hospital-based specialties — and that this drives the differential in power and resources. He asked: what’s your explanation?

I think he is right about the reality. But there is a deeper question: why does innovation concentrate there? The answer is structural, and it doesn’t come from health policy.

How the system reproduces itself

Two sociological frameworks help explain what is happening, and why policy keeps failing to change it.

The first is Basil Bernstein’s concept of “distributive rules” — the deep structures that determine what counts as legitimate knowledge, who produces it, and how it is transmitted. Bernstein was writing about education, not medicine, but Atkinson and Delamont — who first applied his framework to medical education at Cardiff — showed how medical training is organised around precisely these rules. I studied under Atkinson and Delamont. I didn’t finish the doctorate, but I have no regrets about spending the time learning in depth in a different discipline.[^1]

The second is Bourdieu’s theory of capital and field. Hospital medicine operates as a Bourdieusian field — with its own forms of capital (publications, grants, specialist credentials, institutional appointments), its own rules, and its own mechanisms of reproduction. Olsson and colleagues in Sweden used Bourdieu’s framework directly to show how symbolic capital — the prestige attached to different specialties — shapes young doctors’ career choices. Surgery sits at the top; primary care and geriatrics at the bottom. The hierarchy is already fully formed before qualification and reproduces itself through the choices it shapes. General practice sits at the bottom not because of what it does but because of what it cannot accumulate: the symbolic, cultural, and economic capital that the hospital-university axis generates.

The person who most decisively configured that axis was Abraham Flexner. Before his 1910 report, medical education in the United States was plural — university programmes, apprenticeships, proprietary schools. Much of it was community-based. Flexner, commissioned by the Carnegie Foundation, visited 155 medical schools and declared most inadequate. His remedy: medical education must be grounded in laboratory science and housed in universities with teaching hospitals. The Rockefeller Foundation backed the model with over $500 million. The model it funded was not just scientifically prestigious — it was commercially productive, generating patentable knowledge, marketable technologies — genomics, devices, expensive new drugs — and an industry with a material interest in its expansion. Primary care generates little of this, and what it does generate converts poorly into funding or prestige. More than half of America’s medical schools closed. What survived was a particular configuration: the hospital as the site of legitimate clinical knowledge, the university department as its organisational unit, the laboratory as its method. Knowledge produced outside that axis — in communities, in homes, in the ongoing relationship between doctor and patient — was structurally excluded from the system that trained, credentialed, funded, and rewarded doctors.

George Newman’s 1918 Memorandum on Medical Education brought Flexner’s logic to England. Even then, Clifford Allbutt, the Regius Professor of Physic at Cambridge, warned of “a contrary process, that of gradually eviscerating the general practitioner.” He could see it happening, but it happened anyway.

This wasn’t a peculiarly British problem. Everywhere Flexner’s model was adopted, primary care was structurally devalued. Barbara Starfield’s cross-national evidence showed that health systems with stronger primary care had better outcomes at lower cost — and that the specialist-heavy workforce Flexner’s model produced was driving up costs without commensurate gains. Her remedy: “We need another Flexner Report.” The Carnegie Foundation obliged in 2010. But the resulting report recommended integrating formal knowledge with clinical experience — without questioning whether the hospital-university axis should remain the organising principle. Even when someone tried to redo Flexner, the structure survived.

General practice — what Bernstein would call a “region” that draws on multiple disciplines without owning any of them — did try to build its own institutional base:

• 1944 — The Goodenough Report recommends that students should be “properly trained to practise as family doctors” — but also that they must complete a pre-registration year in hospital before entering general practice. The need is recognised; the route to it runs through the hospital.

• 1950 — The Collings report finds English general practice “bad and still deteriorating.”

• 1952 — A College of General Practitioners is founded in response.

• 1972 — Royal charter granted.

• 1970s–80s — Academic departments established. Mandatory vocational training introduced.

• 1986 — Bristol becomes the last UK medical school to appoint a professor of general practice.

• 2002 onwards — Standalone departments begin to be absorbed into broader units of “primary care” or “population health” — any title, as one commentator noted, as long as it is not general practice.

Each gain was real, but none of them changed who controlled the curriculum, the funding, or the prestige. And now even those gains are being reversed.

Where the capital accumulates

Bourdieu described four forms of capital that operate within any field.

The forms convert into each other. Research generates publications (cultural capital), which generate prestige (symbolic capital), which generates funding (economic capital), which generates posts and departments (social capital), which generate more research. The cycle is self-reinforcing.[^2] The career path from medical student through specialty training is embedded in this loop — institutionally supported at every step by hospitals, royal colleges, deaneries, and research councils. General practice has the same loop in theory. But with university departments halved, research funding marginal, and premises decaying, the cycle has weakened to the point where it can no longer reproduce itself with confidence. The effects run deep enough to operate within general practice itself. When a GP researcher described the structural barriers — social science journals with impact factors below 2 against the BMJ’s 13, conference scoring biased toward clinical and quantitative work, the RCGP ignoring its own experts in commissioning research — the response was not structural reform but a promise of better programming at next year’s conference. Meanwhile, the original debate had been framed as a choice between Scientia and the “cosy cardigans of Caritas“ — as though general practice’s problem were insufficient scientific rigour rather than insufficient institutional power. The RCGP’s own motto is Cum Scientia Caritas — science with compassion. The problem was never that general practice didn’t value research. It was that the structures to support it — the departments, the funding, the protected time — were never built at the same scale. When the debate is framed as a deficit in GPs rather than a deficit in the system that supports them, the doxa — Bourdieu’s term for an assumption so deep it is not experienced as a position but simply as how things are — is doing its job.

The policy gap

The Darzi review put it plainly: since at least 2006, successive governments have promised to shift care from hospital to community, and in practice the reverse has happened. As Fraser notes in the Strategy Unit’s recent Neighbourhood Insights webinar, the lineage goes back further — to the 1962 Hospital Plan and a Royal Commission in 1979 that found “considerable practical difficulties” in shifting resources and warned that without additional resources, “progress would be slow.” Every major strategy since has said the same thing. None has reversed the trend. The Strategy Unit’s own analysis shows that non-demographic growth in hospital activity — driven by innovation in diagnostics, treatments, and technology — massively outstrips community services. Innovation flows where the capital is. Left undirected, it gives you what we’ve currently got.

Bernstein would recognise the problem. He distinguished between the “official recontextualising field” — policy documents, strategies, ministerial announcements — and the “pedagogic recontextualising field,” the actual institutions that train, employ, fund, and reward the workforce. Politicians and policymakers write strategies that say shift left. But the people who control training, funding, and career structures don’t redistribute power, prestige, or resources — because they can’t see that these need redistributing. The timeline above is the evidence: policy interventions were absorbed without changing the underlying structure, and some of the hardest-won institutional gains — the academic departments — are now being dissolved. A concrete example: GP placements are now part of every medical curriculum. But the structure of medical education has already shaped students’ perceptions before they arrive. Jerjes and Kelada (2024) found they needed a separate 120-hour programme — outside the normal curriculum — just to counter students’ assumption that general practice lacked complexity and opportunity. The content was added. The status was not.

What would it take?

If the diagnosis is structural, the treatment has to be structural too. Sixty years of strategies have said the same thing. At some point, the question stops being “what should we do?” and becomes “why does what we do keep not working?”

“Shift left” — the idea that care should move out of hospitals into primary and community settings — is one of the government’s stated “big shifts” in the forthcoming ten-year health plan. But the NHS Confederation has asked whether it is “mission impossible,” noting that the hospital share of the NHS budget rose from 47% in 2006 to 58% by 2022. The Strategy Unit calls this a “right drift”: funding for acute care grew by 21.4% between 2016 and 2023, while funding for community healthcare shrank by 4.2%. In practice, “shift left” has meant: move the work left, but leave the investment where it is. The labour shifts. The capital doesn’t. Other forces amplify the pattern: political responsiveness is skewed toward visible crises — A&E waits, waiting lists — while primary care’s pressures are diffuse and harder to see. When access to general practice does become politically visible, the response tends toward urgent care centres, walk-in clinics, and pressure to increase appointment volumes — fragmenting the relational continuity that is the thing that actually makes general practice work, and that reduces demand for appointments in the first place.

A genuine shift would mean moving funding, prestige, research infrastructure, and career capital out of hospitals.[^3] Everyone who works within the current structure — and that includes most of the people who design health policy — sees it as normal. Not through malice. The doxa is simply not visible to them. They don’t see that the economic, symbolic, cultural, and social capital the hospital holds is not a natural fact but a product of the structure it operates within. Bernstein’s distributive rules explain why policy doesn’t work — the official field produces strategies, but the institutional field that trains, funds, and rewards the workforce absorbs them without changing. Bourdieu’s capital theory explains why the institution doesn’t change — the capital loop is self-reinforcing, and everyone within it has a stake they can’t see.

Fraser said we don’t understand what is pulling resources towards hospitals. The assumption that hospital medicine is simply what real medicine looks like runs so deep that nobody argues for it, because nobody needs to. The Strategy Unit’s own work — on the right drift, on continuity, on how we count the workforce — often does the work of making the doxa visible, even without using that language. Bourdieu was better at diagnosing reproduction than prescribing change. But he did argue that the first step is making the invisible visible — forcing doxa into the open where it has to defend itself. Once the assumption has to be stated rather than taken for granted, it becomes contestable. That is already happening. Fraser’s question, the Strategy Unit’s analysis, and rooms like the one Caren Walsh created in Belfast — where a Permanent Secretary asked to be challenged and clinicians responded with evidence, honesty, and hope — are all part of making the doxa visible. The conversation has started.

[^1]: A longer account of how Bernstein’s framework applies to the history of undergraduate medical education in the UK, including the historical sources drawn on here, is in my earlier essay: Bernsteinian Perspectives on the History of UG Med Ed in the UK.

[^2]: This pattern of capital convertibility sustaining structural advantage despite policy intervention has recently been identified in other fields — notably in energy transitions, where fossil fuel systems maintain dominance through the same self-reinforcing capital loop despite decades of renewable energy policy (Husu, Energy Research & Social Science, 2022), and in sustainability accounting, where EU regulation intended to shift corporate behaviour is absorbed by existing institutional hierarchies (Lehner, Accounting, Auditing & Accountability Journal, 2026). Both note that Bourdieu’s framework has been underused at the structural level — applied more often to individuals than to the fields that shape them. Healthcare appears to be a field where this analysis remains largely unapplied.

[^3]: Denmark offers an instructive comparison. GPs there have protected time for research and quality improvement, are well remunerated, and hospital consultants view general practice positively — over three-quarters of UK medical students have encountered dismissive attitudes to general practice by fifth year, but the same dynamic does not hold in Denmark (Shall we all move to Denmark?, BJGP, 2023). Denmark has not dismantled the hospital’s capital loop. It has resourced primary care’s loop enough that it can reproduce itself.

---

Views expressed are my own and do not represent my employer.

---