If you wanted to buy the most health for your money, you'd invest in general practice. In the United States, every extra 10 primary care doctors per 100,000 people was linked to 51.5 more days of life expectancy. The same number of extra specialists was linked to 19.2 days (Basu et al., 2019, JAMA Internal Medicine). States with more general practitioners used more effective care and spent less doing it (Baicker and Chandra, 2004, Health Affairs).
Governments have always known this. In 1978 the World Health Organization declared primary health care "the central function and main focus" of a country's health system (World Health Organization). What has changed is how hard they're now pushing to train more general practitioners (GPs). England plans 6,000 GP training places a year by 2031/32 (NHS England, 2023). The University of Toronto is adding 40 family medicine residency positions (Temerty Faculty of Medicine, 2024). New Zealand's next medical school is built around primary care (University of Waikato).
The catch is that you can't train a GP in a lecture theatre. GPs are trained in working practices, by GPs who teach between their own patients. Every one of those plans depends on how much teaching a full appointment book can absorb.
Why general practice is good value
The evidence goes well beyond one country. Across Europe, countries with stronger primary care had fewer potentially avoidable hospital admissions, fewer deaths from conditions including ischaemic heart disease and cerebrovascular disease, and slower growth in health spending. Their total spending was higher, not lower (Kringos et al., 2013, Health Affairs). So the case for general practice isn't that it's cheap. It's that it buys more health for what's spent, and keeps costs from climbing as fast.
Continuity is part of the reason. A systematic review found that 18 of 22 high-quality studies reported lower mortality when patients kept seeing the same doctor (Pereira Gray et al., 2018, BMJ Open). Knowing a patient over years is what general practice is built around.
Who trains a GP
In a hospital, much of the teaching happens inside the work: on the ward round, in clinic, at handover. General practice doesn't work like that. The registrar usually has their own list in the room next door, and so does the supervisor. Questions get answered between patients or at the end of a session. Observed consultations, case reviews and debriefs all come out of the same appointment book.
| Hospital team | General practice | |
|---|---|---|
| Where the supervisor is | Working alongside the trainee | In the next room, seeing their own patients |
| When teaching happens | Ward rounds, clinics, handovers | Between patients, or after the session |
| Who else can teach | Other consultants, registrars and fellows | The few GPs in the practice |
| What teaching displaces | Some of the team's time | Booked patient appointments |
That's a strength of the model. A GP learns to own a consultation from the start, in the setting where most people get their care. It also means every hour of teaching is an hour of appointments someone else doesn't get.
More learners for the same teachers
Registrars aren't the only ones in the building. England's plan also aims for every foundation doctor to spend at least four months in general practice by 2030/31 (NHS England, 2023). In New Zealand, every intern must complete a community-based attachment (Medical Council of New Zealand), newly registered doctors who trained overseas work under a supervisor (Medical Council of New Zealand), and physician associates, regulated since 1 October 2026, always work under a doctor's supervision (Medical Council of New Zealand, 2026). Medical students come through too.
A student taking their first history and an experienced doctor learning a new health system need very different things. They still draw on the same few GPs.
What teaching costs a practice
In Australia, someone has done the sums. A study supported by the Royal Australian College of General Practitioners estimated that hosting a full-time registrar in their first GP term for six months left a practice A$52,760 worse off, once supervisor and staff time, the registrar's salary and the income the consulting room would otherwise earn were counted. Rural practices lost more. Seventy-one per cent of supervisors said the teaching allowance was inadequate (Elliott and Laurence).
In the United States, finding practices willing to teach has become hard enough that states now pay for it. Georgia introduced a tax incentive for community preceptors in 2014, and by 2023 seven more states had their own. In a 2016 survey of 141 clerkship directors in the United States and Canada, 62% said finding family medicine teaching sites was difficult or very difficult (American Academy of Family Physicians, 2023).
The teachers are stretched
The people doing this teaching are under pressure of their own. In the United Kingdom, only 52% of trainers could use their allocated training time for training (General Medical Council, 2025). Responding to England's plan, the chair of the Royal College of General Practitioners said trainers need "sufficient resource and protected time" (Pulse, 2023). In an Australian survey of 267 GP supervisors, more than 70% had experienced burnout in the past year and 31% said they would stop supervising within five years (Toukhsati et al., 2024, Australian Journal of General Practice). New Zealand has about 74 GPs per 100,000 people, against 116 in Australia and 122 in Canada (Royal New Zealand College of General Practitioners, 2022), so the same work falls on fewer people.
Most GPs who teach do it because they believe in it. What runs out first is time.
Making a supervisor's time go further
We argued in what pro-human AI means in medical education that the scarce thing in training is expert attention. In general practice, that attention belongs to a handful of GPs with full lists. So the question for anything new in training, artificial intelligence (AI) included, is whether it makes that attention go further or asks for more of it.
What makes it go further is practice the learner can do without the supervisor in the room, with specific feedback on each attempt. A registrar who has worked through more consultations, and had their reasoning questioned each time, turns up to the next supervised session with more to show. If they choose to share that practice, the supervisor can start from where the learner actually is, not from scratch.
The supervisor's observation and judgement don't change. What changes is how much ground the learner has covered before the supervisor's time is spent.
Where Gestalt fits
Gestalt builds consultation practice with formative feedback. A learner works through a realistic consultation and gets specific feedback on how they reasoned and communicated.
It's built for the medical schools and colleges that train doctors. For a student, junior doctor or registrar placed in general practice, it means more practice between supervised sessions without drawing on the supervisor's list. Practice stays private unless the learner chooses to share it, and the training organisation sees a cohort-level picture. Feedback is formative, and every judgement about a learner's progress stays with the people responsible for their training.
Health systems are right to bet on general practice. That bet rests on the GPs who teach, and on how far their time can go.