Every attempt in the record arrived at its decisive moment carrying half a ledger. Vermont could show its legislature a payroll tax and not what the tax would buy. Colorado could show voters ten percent of their income and not what they would get for it. California watched support fall twenty-three points the instant a cost became visible, with $37 billion in projected savings sitting unread on the other side. And Britain, having won the argument the others lost, was told by its own analysis that improving primary care was not worth the money — by an instrument built to evaluate products.
None of that is an accident of politics. It is the predictable output of a measurement system, and the system can be described precisely.
Three levels of analysis
Not all cost analysis is the same, and the differences are not academic. They determine what a number can actually tell you, and what it can be used to decide.
What a disease costs. Diabetes is $413 billion a year. Back pain is the leading cause of disability on Earth. This is the level nearly all public conversation operates at — including, until this primer, every page in this section. It establishes the stakes and prescribes nothing. It can alarm you. It cannot tell you what to do.
Costs and benefits both converted into dollars. Does the intervention return more money than it consumes? Useful in some settings, but it requires assigning a dollar value to health itself — and ultimately to a human life. That is a deeply uncomfortable exercise, which is one reason it is used less in health care than people assume.
Cost per unit of health gained. Not "does it pay for itself" but "what does it cost to produce one more year of healthy life?" This is the level where real decisions get made, and it is the dominant method in health economics worldwide. It is also the level almost nobody outside the field talks about — and the one that could change this conversation entirely.
What a QALY is
The third level needs a common unit of health, and the one the field settled on is the quality-adjusted life year, or QALY. One QALY is one year of life in full health. A year lived at roughly half of full function counts as about half a QALY. It is an attempt to capture, in a single number, both how long someone lives and how well.
Divide the extra cost of an intervention by the extra QALYs it produces, and you get an incremental cost-effectiveness ratio — the cost per QALY gained. It is a genuinely useful instrument. It lets you compare a cancer drug against a hip replacement against a screening program, in a common currency of health, and ask the plain question: what are we getting for the money?
For decades, US analyses anchored on $50,000 per QALY as the benchmark of what a year of healthy life was worth paying for. That figure had little theoretical or empirical foundation — it traces back to the cost of dialysis decades ago and simply persisted. It has since drifted upward toward $100,000 and $150,000. The lesson is not that the method is worthless. It is that the number carries more authority than its origins deserve, and it should be read as a tool for thinking, not a verdict handed down.
Where the instrument is aimed
Here is where this stops being a methods lesson and becomes the case this primer exists to make. The cost-per-QALY instrument works. It is applied constantly. Look at what it is applied to:
Every one of these is a discrete, packageable intervention inserted into primary care. A test. A device. A screening protocol. A monitoring technology. The instrument is pointed, almost without exception, at things that arrive as products.
What it is almost never pointed at is the thing this entire site is about: the delivery architecture itself. The continuity. The relationship. The arc of care held over time. Nobody is running the ratio on a person having a real primary care relationship versus a person cycling through episodic encounters, and asking what a year of healthy life costs under each.
The experiment that was run, and what it proved
One large-scale attempt to improve primary care was put through this analysis, and the result deserves far more attention than it received.
The United Kingdom's Quality and Outcomes Framework was the largest primary-care pay-for-performance programme in the world — an enormous, well-funded effort to improve care by attaching financial incentives to measured targets. When researchers modelled its cost-effectiveness, it came out at roughly £49,362 per QALY: well above the threshold the UK uses to decide what is worth funding. The published conclusion was blunt — continuing the programme was not cost-effective, and the country should either redesign it or pursue something else.
This is not evidence that primary care fails the test. It is evidence that bolting incentives onto an unchanged delivery structure fails the test. The QOF did not rebuild how care was delivered. It layered measurement and payment on top of the existing architecture and hoped behaviour would follow. It is the most expensive demonstration available that you cannot buy your way to better delivery without changing the delivery. That is this framework's core case, tested at national scale, at a cost of billions — and confirmed.
The blind spot
Now the deeper problem, and the reason this primer matters more than any single figure in this section. It begins with a question the field almost never asks.
Where does health care actually exist?
Not where is it paid for. Not where is it recorded, or manufactured, or billed. Where does it exist? There is only one answer. It exists in the relationship between a person and their care team — in that space and nowhere else. Everything else in the entire apparatus is there to support that relationship, or ought to be. The building, the imaging suite, the drug, the billing department: infrastructure in service of the encounter, or overhead pretending to be care.
The relationship is where the history is taken and the pattern is seen. Where trust makes honesty possible, and honesty makes accurate diagnosis possible. Where the plan is built, adjusted, and held over time. Every therapy that works, works through that relationship; no drug administers itself to a stranger. The relationship is the cornerstone of any good therapy, and it is where the arc of care exists — because the arc is a relationship, extended over time.
Now look at what the instrument counts. QALYs are assigned to interventions. A drug is credited with the health gain measured in its trial. A screening protocol is credited with the cancers it catches. But consider everything that had to happen for that drug to produce its QALYs. Someone had to know the patient well enough to notice. Someone had to be trusted enough to be told the truth. Someone had to catch the problem early enough for the treatment to work, get the patient to it at the right moment, and have them arrive in a condition to benefit from it.
All of that happened in the relationship. And the relationship is credited with nothing. It generates no QALYs of its own. It does not appear in the ratio. It is not a line in the model.
The instrument is measuring everywhere except the one place where health care exists. The arc produces the health; the product takes the score.
The work that does not announce itself
This is not an accounting quirk. It is the reason horizon-chasing keeps winning.
When the scoreboard counts only what arrives as a product, the money flows to products — the next drug, the next device, the next algorithm. Each is measurable. Each is packageable. Each comes with a trial, a number, and a sales force. Meanwhile the thing that actually produces health has no product to sell, no trial to cite, and no line on the scoreboard.
That thing is ordinary, continuous, attentive care, delivered by people who know you, over enough time to matter. It is unglamorous. It does not announce itself. Nobody wins a prize for the diabetes that never developed, the back that recovered fully, the crisis that never came. Its successes are invisible by their nature — they are absences.
And it works. That is the point, and it deserves to be said without hedging: this is not a theory awaiting discovery. It is not proprietary, it is not mysterious, and it requires no breakthrough. It requires doing it — consistently, over years, with the relationship held intact. That is the barrier to entry, and it is the only one.
But because it cannot be packaged and sold, it does not get measured. Because it does not get measured, it does not get funded. Because it does not get funded, it does not get built. The execution gap this framework has named from the beginning is not merely a failure of will or of policy. It is engineered into the instrument that decides what is worth doing.
Compared to whom
The method deserves a harder criticism than it usually receives, and it goes to the foundation.
To call a year of life "quality-adjusted," you need a reference point — some picture of full health against which a given year is discounted. So ask the obvious question: full health compared to whom?
The unstated answer is something like an able-bodied adult. But that standard dissolves the moment it is examined. Able-bodied at what age? Measured against a twenty-five-year-old, almost no fifty-year-old qualifies — which means a healthy, thriving, fully capable person in their sixth decade gets scored as living a discounted year. That is absurd on its face, and it reveals the benchmark for what it is: a phantom, invoked constantly and defined nowhere.
The consequence is not merely philosophical. By construction, a year lived with a disability scores below a full year, so the arithmetic quietly treats a disabled person's year as worth less than someone else's. Most people reject that the instant it is said plainly, and the concern has been serious enough that the use of QALY thresholds in US federal coverage decisions has been legally constrained.
But the assumption underneath is the deeper error: that physical function is a proxy for capacity. It is not, and the world demonstrates this constantly. Consider that same fifty-year-old at work. They may not move as fast as they did at twenty-five. They also carry twenty-five years of judgment, pattern recognition, and hard-won efficiency — they know what matters and what does not, and they frequently accomplish more in less time because of it. Their capacity is not lower. It is differently composed, and in many settings it is plainly higher. A metric that scores them down for the loss of physical peak is measuring the wrong thing.
Health is the relationship between a person and their own potential — assessed both subjectively, in how they experience their life, and objectively, in what they can actually do. Recovery is measured against that person's own baseline and their own ceiling, not against a mythical standard human.
This is emphatically not a rejection of objective measurement. Objective outcome measures matter, and this framework insists on them — capacity that cannot be demonstrated is not capacity. It is a rejection of the wrong comparator. Measure what a person can do against what they could do, and against what they could do before, and the number means something. Measure them against a phantom twenty-five-year-old and the number mostly reflects how far they are from being someone else.
What should be measured
So aim the instrument at the right thing.
The comparison that would settle this is not exotic and it is not out of reach. A cohort in continuous, arc-oriented primary care, against a matched cohort in standard episodic care. Quality of life tracked over time rather than events tallied after the fact. Both subjective capacity and objective outcome measures, because health requires both. And a horizon long enough for the arc to actually complete — which for chronic conditions means years, not a quarter.
Here is what such a study would have to be able to see, because it is what the arc produces:
- Recovery past the assumed ceiling — patients told they had reached their limit who kept improving, for years, past the point where the standard path stopped looking.
- Conditions intercepted in the silent window — the prediabetic stretch, the drifting blood pressure — where the cheapest intervention is also the most effective, and where episodic care simply is not present to catch them.
- Function restored rather than symptoms suppressed, so the same condition is not paid for again every month, indefinitely.
- Patients who moved from dependence to genuine independence and stopped needing care — not discharged, but no longer requiring it.
None of these register as products. Not one of them can be packaged, licensed, or sold. Every one of them is health, produced.
And the expectation should be stated plainly, in advance: continuous care may well cost more per person per year, at least at first. The case has never been that good care is cheap. The case is that the health it produces per dollar has never been counted — and a system that refuses to count it will go on funding whatever photographs well.
Cost-of-illness data can only alarm. Cost per year of healthy life can justify. And as long as the relationship — the only place health care exists — stays invisible to the instrument, every case for building better primary care will be a case made from principle. Principle loses to a spreadsheet, every time.
The path forward is not to abandon the measurement. It is to point it at the right thing. Aim the instrument at the arc, at continuity, at the space between a person and the people who care for them — and let the number fall where it falls. This framework is willing to be measured. It is asking to be.