The metric that decides what you are allowed to have
Almost every decision about whether a test, a scan or a surveillance schedule gets paid for runs through one number: the QALY, or quality-adjusted life year. A year of life in perfect health scores 1.0. A year lived with a serious impairment scores less. Add the scored years together and you have a currency in which a hip replacement and a chemotherapy regimen can be compared.
That total is then divided into the cost of producing it, across a population, and the resulting ratio is compared with a threshold — conventionally between $50,000 and $150,000 per QALY in the United States. Below the line, fund it. Above the line, do not.
This is a defensible way to spend a fixed public budget, and it is not a stupid idea. A society with finite money and infinite claims on it needs some principle for choosing. The QALY is an honest attempt at one.
But look at what it treats as the substance and what it treats as the modifier. Years are the substance. Quality is only a weight applied to them. A benefit that gives someone back a great deal of themselves for a short time is, in this arithmetic, worth almost nothing — because the thing being counted is duration, and quality merely discounts it. That is not a rounding error in the QALY. It is the QALY working as designed.
It is also why a survivor is told that annual echocardiograms are “not indicated,” and hears a clinical judgement about their heart when what actually happened is that a population ratio crossed a threshold that had nothing to do with them.
The iQALY: quality is the achievement, not the weight
CureRays proposes a companion metric — not a replacement, a competitor. We call it the iQALY: the individual quality-adjusted life year.
The QALY asks: how many quality-weighted years of human life does this produce in a society?
The iQALY asks: how much quality does this return to this person — and how much of what they had lost does it give back? We call the underlying quantity Time-Adjusted Quality, or TAQ. Time still counts, and counts fully. But the return of quality is counted in its own right, at parity with duration, rather than as a coefficient on it.
For one individual:
iQALY = Δq × ( H + R × Href )
The first term inside the bracket is ordinary lived time. The second is the restoration credit: returning a person to themselves is worth something intrinsically, and not merely in proportion to how long they survive to enjoy it.
The four terms
| Term | What it is | Why it is there |
|---|---|---|
| Δq depth | The change in quality on a 0–1 scale — quality gained, or a decrement prevented. | A prevented loss counts exactly as much as a gain. Survivorship care is made almost entirely of prevented losses, and a metric that only counts gains is structurally blind to it. |
| H endurance | The years over which that change actually holds for this person. | Durable benefit is worth more than transient benefit. This term is kept in full — the correction is not to demote time, it is to stop time being the only thing counted. |
| R percent quality returned | Δq divided by the person's quality deficit — the fraction of what they had lost that is handed back. R = 1.0 means fully restored. | This is the term the QALY has no room for. Giving someone back three-quarters of what their disease took is a profound result whether they live forty more years or nine more months. |
| Href restoration horizon | A declared constant, set here at 5 years, converting the dimensionless R into years so it is commensurable with H. | It fixes the exchange rate between restoring quality and extending time. At 5 years a full restoration is worth as much as five years of that same gain — regardless of prognosis. |
The correction this makes: short and profound versus long and shallow
Two patients. One gets a small benefit that lasts a long time. The other gets a large part of themselves back for under a year. Here is what each metric says about them.
| Case | Delta q | H | R | QALY | iQALY |
|---|---|---|---|---|---|
| A. Heart failure prevented in a survivor — a modest decrement avoided, held for 15 years | 0.15 | 15 yr | 1.00 | 2.25 | 3.00 |
| B. A painful, fungating skin cancer treated in someone with 9 months to live — quality 0.45 restored to 0.85 | 0.40 | 0.75 yr | 0.73 | 0.30 | 1.75 |
Under the QALY, patient B's treatment is worth 13% of patient A's. Under the iQALY it is worth 58%. The same nine months of restored comfort, dignity and function — scored one way as a rounding error, scored the other way as most of what medicine is for.
Nobody who has actually stood in front of patient B believes 13% is the right answer. The reason the QALY produces it is not malice or carelessness; it is that the QALY is denominated in years, and patient B has few years to be denominated in. A metric built for allocating a national budget will always do this, and it will always do it to the sickest people in the room.
And a different decision rule
The QALY yields a cost-effectiveness ratio judged against a threshold set by a payer. The iQALY yields an iCER — cost per iQALY — judged against a threshold set by the individual. That is the whole of the freedom argument, and it is structural rather than arithmetic: the number is computed the same careful way, but the person entitled to say whether it is worth it changes.
What follows is a complete worked example, because a metric that cannot be computed is a slogan.
Six surveillance and survivorship tests the QALY systematically undervalues
| Test | What it catches or prevents | Why the QALY scores it poorly |
|---|---|---|
| Serial echocardiography with global longitudinal strain after anthracyclines or chest radiation | Cardiac dysfunction while it is still asymptomatic and reversible, when stopping or modifying therapy still works. | Most people detected would have survived either way. The benefit is a heart-failure decade avoided — quality returned, almost no years added — so it lands near zero. |
| DXA bone densitometry during aromatase-inhibitor or androgen-deprivation therapy | Treatment-induced bone loss before the first fragility fracture. | A prevented hip fracture is a prevented collapse in function and independence. It rarely shows as a survival gain, and surveillance intervals are commonly restricted on that basis. |
| Audiometry after platinum chemotherapy, especially in children | Ototoxic hearing loss early enough to modify dose or fit amplification. | Hearing loss does not shorten life. In a child it reshapes language, schooling and earnings for sixty years — and none of that is a life-year. |
| Neurocognitive assessment after cranial irradiation | Deficits in memory, processing speed and executive function that respond to rehabilitation and accommodation. | Purely a quality endpoint, and one the standard instruments capture poorly. Often treated as assessment rather than surveillance. |
| Early lymphoedema detection (bioimpedance) after axillary surgery or radiation | Subclinical fluid accumulation, when compression still reverses it, rather than after chronic swelling is established. | Prevents a permanent, disfiguring, function-limiting condition and adds no time. Frequently restricted to clinically evident disease — that is, to after the window has closed. |
| Molecular residual disease (ctDNA) surveillance | Recurrence months before imaging, when intervention can be smaller and gentler. | Commonly considered investigational. Where it changes the intensity of salvage rather than the fact of survival, its whole value is quality — and quality alone does not clear a threshold. |
Worked example: how often should a survivor have an echocardiogram?
A 45-year-old survivor completed anthracycline chemotherapy and radiotherapy involving the chest. Both carry a long-term risk of cardiac dysfunction, which can be detected before it becomes symptomatic and, once detected, can often be slowed. The clinical question is cadence: annually, every three years, every five?
Read this before the numbers. Every figure below is a declared illustrative parameter, chosen to demonstrate the arithmetic of the metric. None of it is an epidemiological estimate, a clinical recommendation, or a claim about any real patient's risk. Real surveillance intervals are set by your own cardiologist and oncologist against your actual exposure, dose and cardiac history, and against published guidance from bodies such as the European Society of Cardiology and the Children's Oncology Group. What is being demonstrated here is the shape of the reasoning, not these numbers.
Declared parameters
| Parameter | Value |
|---|---|
| Cohort and horizon | 1,000 survivors, 20 years |
| Risk of clinically significant cardiac dysfunction | 8% over 20 years |
| Caught before symptoms — every 5 years vs annually | 30% vs 70% of cases |
| Progression prevented when caught early | 40% of those caught |
| Quality decrement of symptomatic heart failure (Δq) | 0.15 |
| Years lived with that decrement (H) | 15 |
| Restoration horizon (Href) | 5 years |
| Cost per echocardiogram | $500 |
What the QALY says
Comparing annual surveillance against every-five-years, across the whole cohort:
| Step | Result |
|---|---|
| Cases arising over 20 years | 80 |
| Caught pre-symptomatically — q5y vs annual | 24 vs 56 (a difference of 32) |
| Progressions prevented by the extra scanning | 12.8 people |
| Quality-adjusted years saved per prevented case | 0.15 × 15 = 2.25 |
| Cohort benefit | 28.8 QALY |
| Cohort cost, at 16 extra echoes each | $8,000,000 |
| Cost-effectiveness ratio | $277,778 per QALY |
| The same, discounted at 3% a year as standard practice requires | $343,693 per QALY |
Against a $100,000 threshold, annual echocardiography is roughly three times too expensive. The population metric returns a clear answer: stretch the interval. A coverage policy built on that number tells a survivor yearly imaging is not indicated — and the survivor hears it as a statement about their heart.
An honest accounting of what actually changes
Here we have to be careful, because it would be easy to cheat. Take the cohort benefit, divide by the number of people, divide that into one person's cost, and you get exactly the same ratio back: $8,000 ÷ 0.0288 = $277,778. Per-person and per-cohort arithmetic are identical. Any argument that pretends otherwise is sleight of hand, and the iQALY is not a trick for making expensive things look cheap.
Four things — and only these four — genuinely move the number:
- You are not the cohort mean. The 8% figure averages over people with trivial exposure and people with high cumulative anthracycline dose, substantial cardiac radiation dose, hypertension and diabetes. The average is a fact about the population; it is not a fact about you. This is the largest effect by far.
- Quality returned counts on its own. The restoration credit, R × Href, adds value that does not depend on how long the person survives. It is the term that stops a short prognosis from erasing a real benefit.
- Duration is not discounted. Standard practice shrinks a benefit arriving in fifteen years to about 60% of face value, which systematically penalises anything whose payoff is decades out — precisely screening and survivorship. The iQALY does not apply that discount.
- The threshold belongs to you. No arithmetic here at all. A payer's $100,000 line is a budget constraint, and a budget constraint is not a statement about what your health is worth to you.
What the iQALY says
Preventing the decrement returns all of what would have been lost, so R = 1.0 and the realised benefit is 0.15 × (15 + 5) = 3.00 iQALY. This individual's own cost for the extra scanning is $8,000 — about $400 a year.
| Whose view | Expected iQALY | Cost per iQALY |
|---|---|---|
| The person in whom progression is actually prevented | 3.00 | $2,667 |
| An average-risk person deciding in advance (1.28% chance) | 0.038 | $208,333 |
| A high-risk person deciding in advance, at 3× the cohort mean | 0.115 | $69,444 |
The same test the population metric prices at $277,778 per QALY prices at $69,444 per iQALY for the individual who actually carries the risk — inside the very threshold the population used to refuse it. Nothing was fudged. The population number was an average over a mixed group; this person is not in the middle of that group, and stratification is a legitimate arithmetic operation rather than a rhetorical one.
And the average-risk person, at $208,333? That is $400 a year for roughly a one-in-eighty chance of not developing heart failure. Reasonable people will decide that differently, and that is the point. It is a question about how someone spends their own money on their own body, and it has no population-level right answer.
The cadence calculation
The useful output is not a verdict but a schedule. Each step is priced against the next-less-intensive one, for the high-risk individual — which is how the marginal value of one more scan reveals itself.
| Cadence | Echoes in 20 years | Cost | Expected iQALY | Incremental cost per iQALY |
|---|---|---|---|---|
| Every 5 years | 4 | $2,000 | 0.086 | $23,148 (against no surveillance) |
| Every 3 years | 6 | $3,000 | 0.138 | $19,290 — the marginal bargain |
| Every 2 years | 10 | $5,000 | 0.167 | $69,444 |
| Every year | 20 | $10,000 | 0.202 | $144,676 |
This result is what makes the metric credible, because the iQALY does not simply say “more tests.” Moving from five-yearly to three-yearly is the best purchase on the table: $1,000 buys a 60% increase in expected benefit. Going from three-yearly to two-yearly costs three and a half times as much per unit of benefit. Going to annual costs seven and a half times as much.
A metric that always recommended maximum intensity would be a marketing instrument, not a measurement. This one identifies an optimum and then states plainly what each further increment costs, so the individual can decide whether they want it. Under these illustrative parameters, three-yearly is where a high-risk survivor gets the most quality per dollar — and annual surveillance remains a real option at a stated price, rather than something quietly withheld.
Optimising the outlier moves the average
There is a second-order argument here that matters more than the first.
Population medicine treats the average as the target and the outlier as noise. But averages are not causes; they are summaries. The only way a population mean has ever risen is that individuals inside it got better outcomes, one at a time. A system that optimises for the mean will decline every intervention whose benefit is concentrated in a minority — and so will never produce the upper tail that would have pulled the mean upward.
Optimising the outlier is therefore not selfishness dressed as philosophy. It is the mechanism. Individually-focused care that lets a high-risk survivor buy the surveillance their own biology warrants produces the upper tail; the upper tail becomes the evidence base; the evidence base moves the guideline. The population benefits downstream of individual optimisation, not instead of it.
This is what CureRays Radiation Medicine® means by Screen. Treat. Survive. Repeat.™ The loop is built around one person's trajectory rather than a cohort's, and the iQALY is the arithmetic that makes that design accountable instead of merely aspirational.
The mission of CureRays SSS
Screening. Surveillance. Survivorship. The purpose of the CureRays SSS programme, and of the Institute’s research around it, is to validate the iQALY — and to state its limitations honestly. Both halves are the mission. A metric advanced only by the people it flatters is a marketing instrument, and the QALY’s own authority came from decades of exactly this scrutiny.
The reason to do that work is not to abolish the QALY. It is to put a competitor in the field: a second, defensible justification for payment, so that a coverage decision has somewhere to go when the population metric returns an answer that is wrong for the person in the room. More choice, more options, and quality that counts on its own — not quality that only scores when it happens to be coupled to quantity.
What this demands in return: better prognosis
This is the honest cost of the proposal, and we would rather state it than have it pointed out. The moment quality returned becomes the unit, how long that quality holds for this person stops being a footnote and becomes the calculation. The endurance term H is a prognosis. A metric built on individual benefit is only as good as individual prediction.
So the iQALY makes accurate prognostication more paramount, not less. Population cost-effectiveness can average its way past a bad estimate of remaining time; an individual decision cannot. Anyone advancing this framework inherits the obligation to improve prognostic estimation, and to be explicit about its uncertainty when it is poor.
Two probabilities a survivorship clinic should be able to produce
A survivorship clinic that follows patients for years, records toxicity as it emerges, and tracks what prevents recurrence is generating exactly the data this metric needs. Properly instrumented, such a clinic should be able to estimate, for an individual:
- P(quality decline) — the probability that this person, with this treatment history and these comorbidities, sustains a specific decrement in function, and how large, and when.
- P(quantity retention) — the probability that this person retains the years the population curve implies, given what is actually known about them rather than about their diagnosis.
Those two numbers are Δq and H, estimated for a person instead of assumed for a cohort. That is the empirical programme, and it is a long one. We are describing what such a clinic should be able to do, not reporting what ours has done. No outcome from any CureRays cohort appears on this page, and none will until it has been reviewed and published.
A pathway: first appeal, then approval, then standard of care
A new metric does not arrive by replacing the old one. We propose the iQALY advance through three stages, each a higher evidentiary bar than the last:
- A pathway of first appeal. When a population-based determination denies an individual, the iQALY becomes the structured instrument for appealing it — a declared Δq, H and R with the assumptions on the page, rather than a letter of hardship. Appeal is the right first venue precisely because it is already the place where an individual argues against an average.
- A recognised basis for approval. If appeals reliably identify the same kinds of people — and if follow-up shows those people did realise the quality that was argued for — the iQALY earns standing as a prospective justification rather than a retrospective plea.
- A standard of care for when the QALY is overruled. The end state is not that the iQALY wins. It is a defined, auditable rule for when a QALY should be overruled by an iQALY — and, just as importantly, when it should not.
What that would change is the unit of decision. Coverage would be individualised the way care already is: assessed patient by patient rather than population by population, optimising quality and quantity for the person, instead of optimising an average and accepting whatever it does to the people at its edges.
None of this is established. It is a proposal, and the work of validating it — including the possibility that it fails — is the mission stated above.
What the iQALY does not do
A metric claiming to solve everything would deserve suspicion, so here are the limits, stated plainly.
It does not allocate a public budget. If everyone applies their own threshold to a common pot, the pot fails. The iQALY governs decisions an individual makes about their own care and their own money; the QALY still governs collective allocation, and it should. These are two different questions, and the mistake being corrected is answering the first with the tool built for the second.
It does not price harm. More surveillance means more incidental findings, more downstream testing, more anxiety, and occasionally injury from investigating something that was never going to matter. A complete calculation must subtract those decrements — they are quality losses in the same units. The worked example omits them for clarity, which makes it incomplete, and we would rather say so than leave it out quietly.
Href is a value judgement, not a measurement. Setting the restoration horizon at 5 years fixes how much a full restoration is worth against lived time. Set it at 2 and the metric drifts back toward the QALY; set it at 10 and short-prognosis treatment dominates everything. It is the most consequential number on this page and it cannot be derived from data — only argued for.
It is a proposal, not a standard. The iQALY is ours. It has not been validated, adopted by any body, or tested against outcomes. We publish the formula and the arithmetic precisely so it can be argued with.
Have your own numbers run
Your risk stratum, not the cohort average, is what should set your surveillance schedule. Talk with our team about building yours.