A woman in her seventies with heart failure gains three kilos of fluid over a week. Her ankles swell. She sleeps propped on pillows. If someone checks on her on day three, the fix is a change in her diuretic dose and a phone call two days later. If nobody does, she arrives in an emergency department on day eight, breathless, and spends five nights on a ward.
That is the whole case for primary care as a hospital bed. The clinic visit on day three does the job of the ward on day eight, at a fraction of the cost. Health economists have a name for the admissions it prevents: ambulatory care sensitive conditions. The OECD tracks them precisely because hospital stays for asthma, COPD and congestive heart failure are a readable signal of whether primary care is working.
Almost everyone in health policy agrees with this argument. That is part of the problem. It is usually made in a form that is true clinically and wrong financially.
Here is the version I think is closer to right: primary care prevents admissions, but it only saves money when someone is willing to take the freed-up hospital capacity out of the system. Without that second step, the empty bed rarely stays empty.
The bed that refills itself
In 1961 the health services researcher Milton Roemer observed that hospital beds, once built, tend to get used. The phrase that stuck is Roemer’s law: a bed built is a bed filled. It has been argued over for sixty years, and it is not an iron law, but anyone who has run a hospital budget recognises the pressure behind it.
A hospital’s costs are mostly fixed. Staff, buildings, equipment and debt do not shrink when admissions fall by 5%. If the hospital is paid per admission, as it is under most activity-based and DRG systems, a drop in avoidable admissions is a drop in revenue with almost no drop in cost. Managers respond as any rational manager would. Elective lists get worked down faster. Thresholds for admitting borderline cases drift lower. Length of stay creeps up at the margin.
None of this is fraud. It is what a fixed-cost institution does when its income depends on volume. But it means that primary care investment can succeed on its own terms, fewer heart failure crises, while total spending barely moves.
Germany: plenty of access, plenty of beds
Germany is the uncomfortable test case. Access to ambulatory doctors is generous by international standards. Yet the OECD’s Health at a Glance 2023 reported that, among 32 countries with comparable data, aggregated avoidable admissions for these chronic conditions were highest in Türkiye, Germany and the United States in 2021. The same year Germany had about 7.8 hospital beds per 1,000 people, one of the highest ratios in the OECD.
A 2014 analysis of German regional data in Deutsches Ärzteblatt International found that the number of local hospital beds was positively correlated with rates of ambulatory care sensitive hospitalisations. More beds, more of the admissions that primary care is supposed to prevent.
The point is not that Germany’s GPs are failing. It is that supply exerts its own pull. A country can have a reasonable front door and still route people through the most expensive room in the house if that room is sitting there, staffed and paid for.
The United States shows the other half
The American problem runs the other way. According to the Milbank Memorial Fund’s 2024 primary care scorecard, primary care’s share of total US health spending fell from 5.4% in 2012 to 4.7% in 2021. The OECD average share for primary care, measured on a somewhat different basis, was 13% of health spending in 2021.
So the US underfunds the front door and pays handsomely for the back. It is also on the OECD list of highest avoidable admissions. The two facts are connected, but the fix is not simply “spend more on primary care.” A US health system that doubled primary care funding while leaving hospital capacity, pricing and payment untouched would probably get better care and a bigger bill. That might be worth it. It should just not be sold as a saving.
What has to change for the saving to be real
If primary care is going to function as a cheaper bed, the hospital side of the ledger has to move too. That means decisions most health ministries find politically painful:
- Paying hospitals in ways that do not reward avoidable volume. Global budgets, capitation for defined populations, or blended payments where a hospital gains when its catchment area needs it less.
- Retiring capacity deliberately. Converting wards, not just leaving them understaffed. Beds that are closed on paper and reopened in winter do not count.
- Moving money, not just patients. If the savings from avoided admissions stay inside the hospital’s budget, primary care will always be funded from new money, which is exactly the money finance ministries are least willing to give.
Each of these runs into a real constituency. A hospital closure is local, visible and felt by the people who work there. A prevented admission is invisible. Nobody holds a press conference for the heart failure patient who stayed home.
A worked example: where the saving goes
Take a hypothetical district of 300,000 people. The ministry funds a programme of nurse-led heart failure and COPD follow-up in its general practices: phone checks after discharge, a weight and symptom diary, fast access to a clinician when something changes. Suppose it works as well as its designers hoped, and emergency admissions for those two conditions fall noticeably over two years.
Follow the money.
The primary care programme is a new cost, paid every year. The district hospital sees fewer heart failure admissions, but its medical wards do not go quiet. Patients who had been waiting on trolleys in the emergency department get a bed sooner. The elective surgical list, which had been cancelled every winter because medical patients spilled into surgical beds, runs more often. Length of stay for frail patients drifts up a little, because the pressure to discharge by noon has eased.
At the end of the two years, the district has better heart failure care, shorter corridor waits and more hip replacements done. Every one of those is a genuine gain. What it does not have is lower spending. The hospital’s staff, buildings and overheads are the same. Its income, if it is paid by activity, may even have risen, because the elective work it now fits in is better paid than the medical admissions it lost.
Then a finance official asks where the promised saving went. Nobody can point to it, because it was converted into activity. The primary care programme is labelled as a cost pressure and becomes a candidate for the next round of cuts.
That sequence is not a failure of primary care. It is a failure to decide, in advance, what the freed capacity was for.
The strongest counterargument
The best objection to my argument goes like this: in systems where hospitals are chronically overloaded, a freed bed is never wasted. If patients are waiting on trolleys and elective lists stretch into years, the refilling of beds is not supplier-induced demand. It is unmet need finally being met. Closing capacity in that situation would be perverse.
This objection is partly right, and it matters. In a system running hot, primary care’s return is not measured in money saved but in access gained. The heart failure patient who stays home frees a bed for someone who genuinely needs one. That is a real benefit and it should be counted.
But it changes the pitch, and the pitch matters for what happens next. A ministry that justifies primary care investment on the grounds that it will relieve hospital pressure is making an access argument. It should be judged on access measures: waiting times, emergency department delays, cancelled operations. A ministry that justifies it on cost savings is making a fiscal argument, and will be judged on spending. Mixing the two is how good programmes get cut. They deliver the access gain and get blamed for missing the fiscal target.
The objection also has a limit. Not every system is running hot everywhere. Germany’s combination of high bed numbers and high avoidable admissions suggests a different dynamic: capacity that exists partly because it exists, sustaining admission patterns that a smaller hospital sector would not. In those settings, the refill is less about unmet need and more about fixed costs looking for revenue.
Second-order effects nobody budgets for
Shifting care toward primary care also moves workload and risk, and those moves have consequences of their own.
First, primary care absorbs complexity it was not staffed for. Keeping a frail patient with heart failure at home means more contacts, more medication changes and more judgement calls in general practice. If practices are paid per registered patient without adjustment, the practice that does this well takes on work without extra income. Over time the best practices burn out their staff, and the programme’s effect fades.
Second, hospitals adapt. Some respond to falling admissions by moving into community services themselves, running outreach clinics or virtual wards. That can be good integration. It can also be a way for the hospital to recapture the activity and the budget that was meant to shift outward, with the hospital’s cost base attached.
Third, measurement drifts. Once avoidable admissions become a target, coding changes. An admission for heart failure can be recorded as a short-stay observation, or as a different primary diagnosis. The indicator improves faster than the reality. Anyone evaluating a primary care programme should look at total emergency activity and mortality, not only the flagged conditions.
What this means for each actor
- For ministries: decide explicitly whether a primary care investment is an access policy or a savings policy, and set the success measures to match. If it is meant to save money, name the hospital capacity that will be retired and when.
- For hospital leaders: a payment system that rewards you for filling beds will set you against every prevention programme in your catchment area. Pushing for population-based or blended budgets is in your long-term interest, even if it feels like giving up income.
- For funders and development partners: stop financing primary care pilots whose business case rests on hospital savings that nobody has authority to capture. Fund the hospital-side reform alongside.
- For clinicians: avoided admissions are only visible if someone records them. General practices that track patients kept out of hospital, and why, give ministries the evidence they need to protect the programme.
What to watch next
Three signals will show whether a system is serious about primary care as a substitute rather than an addition. The first is whether hospital bed numbers or hospital budgets are ever reduced in a region after primary care is strengthened there. The second is whether payment reform reaches hospitals, not just general practice. The third is whether evaluations report total system spending rather than cherry-picked admission rates. When all three appear together, the cheapest bed will finally be allowed to replace the expensive one.
The honest pitch
I think the primary care argument has been weakened by overselling. When ministers are told that investing in general practice will lower hospital costs, and then hospital costs do not fall, the lesson they draw is that primary care was a nice idea that did not deliver. The better lesson is that it delivered clinically and the savings were absorbed elsewhere, by design.
The honest pitch goes like this. Strong primary care will prevent a meaningful share of admissions for chronic conditions. It will make people’s lives better in ways that never show up in a budget. And it will save money only if the system is willing to shrink something else. Without that last commitment, a ministry is not buying a cheaper bed. It is buying a second one.
Primary care really is the cheapest hospital bed there is. It just has to replace the expensive one rather than sit next to it.
Have you seen a health system actually take hospital capacity out after strengthening primary care? What made it politically possible?
