How UK healthcare works
Patients in the United Kingdom cannot normally see an NHS specialist without first being referred by a general practitioner. This is called gatekeeping, and it has been the design of the NHS since 1948. Here is why it was built that way, what the clinical and economic cases for it look like, and where the design has limits.
Dr Seth Rankin
MBChB MRCGP. Founder of LoveMyLife. Former NHS Commissioner and Managing Partner of Wandsworth Medical Centre.
23 April 2026
11 min read
One of the things that surprises people new to the United Kingdom most about its health system is that they cannot simply ring a hospital specialist and book an appointment. Planned access to secondary care in the NHS, for almost every specialty, runs through a general practitioner (GP) referral. The GP is the gate, and the gate is, by design, almost the only way in.
This arrangement is called gatekeeping. It is a feature of UK healthcare, of the Dutch, Scandinavian, and several southern European systems, and of some parts of Commonwealth primary care. It is not a feature of German, French, Swiss, or most US-based health systems, where direct access to specialists is more common. This article sets out what gatekeeping actually means in the NHS, why the UK chose this design, what the clinical and economic cases for it are, how other countries handle the same question, and where the honest limits sit. Sources are at the end.
The core rule is straightforward. Any non-emergency referral to a National Health Service (NHS) secondary-care specialist in England, Scotland, Wales, or Northern Ireland is initiated by a GP. A patient who wants to see an NHS rheumatologist, endocrinologist, orthopaedic surgeon, psychiatrist, or almost any other specialist route books an appointment with their GP first, the GP makes the clinical assessment, and the referral follows (or does not) based on that assessment.
There are a few exceptions.
Emergency care. Accident and Emergency (A&E), 999, and NHS 111 routes do not require a GP referral.
Some sexual-health, termination-of-pregnancy, and mental-health crisis services can be accessed directly.
National screening programmes (cervical, breast, bowel, abdominal aortic aneurysm, diabetic eye) come to eligible patients by direct invitation.
Optometry and NHS dentistry are accessed directly rather than through a GP.
Some self-referral mental-health services, notably NHS Talking Therapies (formerly IAPT), can be accessed without a GP referral.
Outside these, the NHS is a gatekeeper system. A patient with a new knee problem does not call a knee surgeon; they call their GP. A patient with unexplained weight loss and fatigue does not book a haematologist; they book with their GP. The gate is the front door.
The gatekeeper design has been part of the NHS from the start. When the NHS was founded in 1948, general practitioners were preserved as independent contractors providing a registered-list primary-care service, rather than nationalised as direct NHS employees. A patient registering with a GP was registering with a specific named practice, in their local area, responsible for their primary-care record.
That design had a consequence. Access to the new state-run hospital sector, which had been nationalised and consolidated into regional hospital boards, was coordinated through the family doctor. The GP held the patient's file, made the clinical judgement about what specialist attention was warranted, and wrote the referral.
No one sat down in 1948 to design a gatekeeper system from first principles. It emerged from the combination of three things: a universal list-based primary-care system, a nationalised hospital sector answerable to a fixed budget, and a shared clinical conviction that general practice was a meaningful clinical discipline in its own right. Those three conditions have been stable for seventy-seven years, and so has the gatekeeper design that follows from them.
The clinical rationale for having a generalist see the patient first is well established and is not unique to the UK.
Undifferentiated presentation. A patient's first clinical presentation is rarely labelled with the specialty it belongs to. Tiredness may be thyroid, haematological, depressive, infective, autoimmune, cardiac, or simply the outcome of a bad week. A generalist trained to hold that wide differential in one consultation can often direct investigation more efficiently than a patient choosing a single specialty route for themselves.
Continuity of record and relationship. A GP who has seen a patient repeatedly over years has information a specialist receiving a single consultation cannot have: the patient's baseline, their patterns, their family history, their social context, and the medications that have already been tried. This context shortens the specialist's work when a referral is made.
Co-morbidity. Older patients and those with long-term conditions often have multiple things going on at once. A generalist can hold the interplay across specialties, which is harder for a single-specialty clinic to do.
Appropriate use of specialist time. Specialist clinic appointments are scarce. A generalist filter that ensures the specialist sees the patients who most clearly need the specialist consultation makes fuller use of that scarce capacity.
Reassurance and safety-netting. A significant proportion of primary-care consultations are resolved without a referral, through explanation, simple treatment, or active monitoring over time. A direct-access system that skips this step can produce more investigations than are clinically warranted.
None of these points should be read as a criticism of specialist medicine. They are observations about how patient presentations and specialist clinics work in combination. A generalist consultation followed by a specialist referral where needed is an efficient sequence for a system that is trying to serve a whole population on a fixed budget.
The economic case for gatekeeping is that it is a demand-management mechanism.
Specialist care is typically several times more expensive per episode than primary care. Investigations initiated in specialist clinics tend to be broader, more numerous, and more expensive than those initiated in primary care. Referrals that bypass a primary-care filter, in the systems that allow it, tend to produce more tests and more follow-up than referrals that pass through a filter.
For a health system trying to deliver universal coverage within a fixed budget, the arithmetic of this matters. If every UK resident could book a specialist directly, the number of specialist consultations per year would rise substantially, the number of scans and tests initiated would rise, the total cost would rise, and the capacity available for the patients with the clearest clinical need would fall. The gatekeeper design is, among other things, a way of keeping the system inside its budget.
It is worth being honest about the trade-off. Gatekeeping protects the population-level budget by creating friction at the individual level. The friction is useful in most cases, because most cases do not need specialist care. It is costly in a minority of cases where the gatekeeper assessment underestimates what the patient needs. More on that below.
Gatekeeping is common in publicly funded universal systems. OECD analysis shows that most Commonwealth primary-care systems and most publicly funded European systems use some form of GP gatekeeping.
Explicit gatekeeping (must be referred by a GP): United Kingdom, Netherlands, Denmark, Norway, Finland, Spain, Portugal, New Zealand, Australia (for most publicly funded specialist care).
Incentivised gatekeeping (direct access possible but financially penalised): France from 2005 onwards, where patients who bypass their *médecin traitant* receive a lower reimbursement rate.
Direct access to specialists (no gatekeeping): Germany, Switzerland (for most cantons), Austria, Japan, South Korea, the United States (subject to insurance plan design).
Mixed or evolving: Ireland, Israel, Italy.
The direct-access systems are not necessarily more generous or better resourced. They are organised around a different underlying logic. Insurance-funded direct-access systems tend to pass more of the cost directly to the insurer (and so ultimately to the patient's premium), which is a different way of controlling demand. Publicly funded gatekeeper systems control demand at the point of access instead.
No one model of organising specialist access is the universally correct choice. Each represents a specific trade-off between patient autonomy, cost control, and clinical efficiency.
Research comparing gatekeeper with direct-access systems shows reasonably consistent findings.
Gatekeeper systems tend to have lower total specialist utilisation, lower imaging and testing rates, and lower total healthcare costs per capita.
Gatekeeper systems tend to have longer average waits for elective specialist appointments, driven by the filter effect and the fixed specialist capacity in a publicly funded system.
Clinical outcomes at the population level are broadly comparable across gatekeeper and direct-access systems for most conditions, although the evidence for specific conditions is mixed.
Patient satisfaction differs by what the patient is measuring. Satisfaction with access tends to be higher in direct-access systems; satisfaction with co-ordinated care and continuity tends to be higher in gatekeeper systems. These measure different things.
The large international comparative analyses (for example by the OECD, the Commonwealth Fund, and the King's Fund) are broadly consistent with this pattern. What they do not show is that one structural design produces clearly better clinical outcomes than the other for the whole population.
Gatekeeping is not frictionless, and it does not always get the assessment right. Several well-documented limits of the UK model are worth naming.
Missed or delayed diagnosis. A non-specialist assessment that fails to identify a serious condition early can produce worse outcomes than a direct-to-specialist assessment would. This risk is highest for rare cancers, rare neurological conditions, and some autoimmune presentations that masquerade as common benign conditions.
Patient frustration with "having to see the GP first". For patients who feel confident about their diagnosis, the gatekeeper step can feel redundant and time-consuming, especially for conditions the patient has lived with for years.
GP workload pressure on assessment quality. A ten-minute NHS GP consultation leaves limited time for a thorough differential in every complex case. The quality of the gatekeeper assessment depends on the time and resource available to the assessor.
Access inequality. Patients with higher health literacy, stronger relationships with their GP, or more time to follow up on referral delays tend to navigate the gatekeeper system more easily than patients who have none of those. The design is formally equal; the practical experience is not always.
The access problem in NHS general practice since 2019, documented in the GP Patient Survey, means that the gatekeeper step is sometimes not difficult because of clinical reasoning but because reaching the gatekeeper takes longer than patients expect.
None of these limits undermines the underlying clinical and economic logic of gatekeeping. They describe the difference between how the system works at its best and how it sometimes works at its worst. Every structural design has this gap.
A patient facing the NHS gatekeeper system in England has defined legal rights under the NHS Constitution for England.
Right to choose. An NHS patient is legally entitled to choose the hospital or clinic they are referred to, from any NHS-funded provider in England. The referral itself is initiated by the GP; the destination is the patient's choice. Full detail at NHS patient choice.
NHS Right to Choose for specific conditions. For a defined list of conditions, notably adult attention-deficit/hyperactivity disorder (ADHD) assessment, patients can request referral to any NHS-funded provider in England, including certain independent providers contracted to the NHS.
Right to a second opinion. If a patient is unhappy with their care or diagnosis, they can ask their GP for a second opinion referral. There is no statutory right to one, but in practice it is generally accommodated.
Right to access a GP and be registered. Patients cannot be refused NHS GP registration for reasons of immigration status, proof of address, or identity documentation, under NHS England guidance.
These rights are the operational counterweight to the gatekeeper structure. The system tells the patient that the route is through the GP; it also tells the patient that, once the referral is made, the choice of where to be seen is the patient's.
For most patients, most of the time, the gatekeeper design works well. The GP is the person who knows you, has your record, can make a quick assessment of what kind of help you need, and can send you in the right direction. For an undifferentiated symptom, this is usually the right first call.
Where the design is less helpful is when the patient is sure they know what they need and the gatekeeper step is a delay. In those cases, three options open up.
Insist on the route you think is clinically right, and be specific about the reason. A GP responding to a concrete request ("I would like a referral to a neurology clinic because I have a family history of X and the following symptoms have developed") is different from a GP responding to a vague one.
Use your [NHS Right to Choose](https://www.nhs.uk/nhs-services/hospitals/going-into-hospital/your-right-to-choose-where-you-have-your-nhs-treatment/), where it applies, to direct the referral to a provider whose waiting times or specialty mix fits your need.
Consider a private specialist route for a defined, time-sensitive question. Private specialists can be self-referred in most cases. Care done privately can be handed back to the NHS through a GP-mediated shared-care arrangement if ongoing NHS prescribing or follow-up is needed.
None of these options involves criticising the gatekeeper design. They work with it.
The NHS uses GPs as gatekeepers because that is what its founding structure (nationalised hospitals, list-based independent-contractor general practice, fixed national budget) naturally produces. The clinical case is that a generalist first assessment fits how most patient presentations actually arrive: as undifferentiated symptoms rather than specialty-specific labels. The economic case is that a gatekeeper filter lets a publicly funded system deliver specialist care to the patients who most clearly need it within a controlled budget.
The design is not free of trade-offs. Most universal systems in Europe and the Commonwealth use something similar. Direct-access systems (Germany, the US, Japan, Switzerland) make different trade-offs, not better or worse ones.
For a patient, the practical answer is to understand the gate, use it well, and know the rights that sit alongside it. The referral comes from the GP; the choice of where to go next is yours.
gov.uk, NHS Constitution for England. Patient rights and NHS principles.
NHS, Your right to choose where you have your NHS treatment. The Right to Choose explained.
NHS England, Registering with a GP. Rights of access and registration.
NHS, NHS Talking Therapies. Self-referral mental-health service.
OECD, Health at a Glance 2025. International comparison of health systems.
OECD, Health systems performance. Cross-country analysis of access and outcomes.
Commonwealth Fund, Mirror, Mirror 2024. Comparative analysis of high-income health systems.
King's Fund, kingsfund.org.uk. Independent health policy analysis of UK general practice and access.
Nuffield Trust, nuffieldtrust.org.uk. Independent research on NHS primary care.
GP Patient Survey, gp-patient.co.uk. Patient-reported NHS primary-care experience.
NHS England, NHS history. Origin of the 1948 settlement.
Royal College of General Practitioners, rcgp.org.uk. Professional body for UK general practitioners.
Clinically reviewed
Dr Seth Rankin · MBChB MRCGP - Founder and Medical Director, LoveMyLife
Dr Seth Rankin qualified in medicine at Auckland School of Medicine in New Zealand in 1990 and worked as a junior doctor across New Zealand, Australia, and the UK before qualifying as a Member of the Royal College of General Practitioners (MRCGP) through the London Deanery in 2004. He was Managing Partner of Wandsworth Medical Centre from 2006 to 2016 and served as a Board Member of Wandsworth Clinical Commissioning Group for nine years. He is the founder of London Travel Clinic, London Doctors Clinic, London Medical Laboratory, and LoveMyLife.
If you would like a consultation with an MRCGP-registered GP, the next step is to book.
Begin your consultation at this link. Online or in person at Westfield London.