How to evaluate impact of public longevity infrastructure?
To evaluate public longevity infrastructure, decide what outcome matters — healthy life expectancy and the gap between lifespan and healthspan, broken down by how deprived an area is — then use a study design that can tell the programme's effect from background trends (phased randomised rollouts, comparisons with similar regions, time-series around a law), check that the programme actually reached people before judging its outcomes, report equity as a headline result, model costs honestly as cost-effective rather than cost-saving, and pre-register the evaluation so null results get published.
Evaluating public longevity infrastructure means answering one question — did the population receive something that added healthy years, and can that be attributed to the programme rather than to trend — and the method is ranked here by how much each step contributes to an honest answer. First, choose outcomes that matter and can be measured: healthy life expectancy and the morbidity gap, disaggregated by deprivation decile, with intermediate outcomes (smoking prevalence, blood-pressure control, coverage) that move sooner. Second, use a design that can attribute: randomised or stepped-wedge rollout where a programme is phased anyway, difference-in-differences and synthetic controls for national policies with comparison jurisdictions, interrupted time series for laws with a date — because before-and-after comparisons attribute secular trends to whatever was announced. Third, measure delivery and coverage before outcomes, since a programme that reached 8% of the eligible population has not been tested, only announced. Fourth, evaluate equity explicitly, reporting every outcome by deprivation, because a 19–20-year healthy-life gap between deciles is where impact is won or lost and national averages hide it. Fifth, model cost-effectiveness at a stated discount rate from a stated perspective, claiming cost-effective rather than cost-saving. Sixth, pre-register the evaluation and publish null results, or the literature's publication bias — the reason median prevention ROI figures overstate — continues. The UK's healthy-years mission failed its own evaluation by the simplest metric, healthy life expectancy, which fell to a record low; Singapore's Healthier SG is evaluable because it has an enrolment metric to check first.
- The outcome is healthy years by deprivation decile; everything else is intermediate.
- Before-and-after is not evaluation; use designs that can attribute — stepped-wedge, difference-in-differences, synthetic controls, interrupted time series.
- Coverage comes before outcome: an unreached population cannot show an effect.
- Equity is a primary outcome, not a subgroup analysis.
- Pre-register and publish nulls; the field's ROI literature shows what happens otherwise.
The evaluation method, ranked by contribution to an honest answer
Ranked on: how much each step contributes to attributing a change in healthy years to the programme rather than to trend, and how often its omission has produced a false claim of success.
| # | Option | Verdict | Grade |
|---|---|---|---|
| 1 | 1. Choose outcomes that matter, disaggregated | Healthy years by deprivation decile, plus intermediates that move sooner | GRADE AEstablished |
| 2 | 2. Use a design that can attribute | Before-and-after attributes trend to the announcement | GRADE AEstablished |
| 3 | 3. Measure delivery and coverage first | An unreached population cannot show an effect | GRADE AEstablished |
| 4 | 4. Evaluate equity as a primary outcome | The 20-year gap is where impact is won or lost | GRADE AEstablished |
| 5 | 5. Model cost-effectiveness honestly | Stated perspective, stated discount rate, no 'savings' | GRADE BPromising |
| 6 | 6. Pre-register and publish null results | The cure for the literature's publication bias | GRADE BPromising |
- 01
1. Choose outcomes that matter, disaggregated
GRADE AEstablishedHealthy years by deprivation decile, plus intermediates that move soonerPrimary: healthy life expectancy and the morbidity gap (life expectancy minus healthy life expectancy), by deprivation decile and by area. Intermediate: smoking prevalence, share of hypertensives at target, vaccination and screening coverage, falls, hospital admissions for ambulatory-sensitive conditions. Outcomes chosen after the fact are the first evaluation trap.
- 02
2. Use a design that can attribute
GRADE AEstablishedBefore-and-after attributes trend to the announcementWhere a programme is rolled out in phases anyway, randomise the order (stepped-wedge) — the most powerful and least used design in public health. For national laws, interrupted time series with a comparison series; for regional programmes, difference-in-differences or synthetic controls built from comparable jurisdictions. Simple before-and-after comparison is how a falling smoking trend gets credited to whichever strategy was current.
- 03
3. Measure delivery and coverage first
GRADE AEstablishedAn unreached population cannot show an effectEnrolment, invitations sent and answered, programme capacity against eligible population, coverage by decile. A falls-prevention programme reaching 8% of eligible older adults has not tested the evidence; it has under-delivered it. Coverage is the first published number, and the one the UK's mission never had.
- 04
4. Evaluate equity as a primary outcome
GRADE AEstablishedThe 20-year gap is where impact is won or lostReport every outcome by deprivation decile and by ethnicity where available, and pre-specify the gap as a primary endpoint. The healthy-life-expectancy gap between deciles in England is roughly double the life-expectancy gap; a programme that improves the average by widening the gap has failed the population it was for.
- 05
5. Model cost-effectiveness honestly
GRADE BPromisingStated perspective, stated discount rate, no 'savings'Cost per healthy life-year or QALY from the perspective of the budget that paid, at a conventional discount rate (3–3.5%), with sensitivity analysis. Claim cost-effective, not cost-saving: prevention raises lifetime medical costs because people live longer. Quoting a societal 14:1 to a health treasury is the second evaluation trap.
- 06
6. Pre-register and publish null results
GRADE BPromisingThe cure for the literature's publication biasRegister the outcomes, design and analysis before the programme starts; publish whatever the answer is. The prevention ROI literature's median of 14.3:1 is inflated by exactly the positive-results bias that unregistered, in-house evaluations produce. A strategy evaluated only by its authors will succeed on paper.
Evaluation traps, and the honest alternative
How target-led strategies claim success, and how to evaluate instead
| Trap | What it does | Honest alternative |
|---|---|---|
| Milestone reporting | Counts documents published and boards convened as progress | Report coverage and outcomes, not activities |
| Before-and-after | Credits secular trend (falling smoking, rising screening) to the strategy | Comparison series, difference-in-differences, synthetic control |
| National averages only | Hides a widening deprivation gap behind a stable mean | Every outcome by decile; the gap as a primary endpoint |
| Outcome switching | Replaces healthy life expectancy with 'awareness' when HLE falls | Pre-registered primary outcome |
| Societal ROI to a health budget | Promises savings that appear in other budgets or never | Cost-effectiveness from the paying budget's perspective |
| Evaluating the pilot, scaling the announcement | Pilot results at 80% coverage; national rollout at 10% | Coverage-adjusted effect; evaluate the rollout |
| In-house evaluation | The author marks the homework | Independent, pre-registered, published |
Frequently asked questions
How do I evaluate the impact of public longevity infrastructure?
In six ranked steps: choose outcomes that matter — healthy life expectancy and the morbidity gap by deprivation decile, with intermediate measures; use a design that can attribute, such as stepped-wedge rollout, difference-in-differences, synthetic controls or interrupted time series; measure delivery and coverage first; evaluate equity as a primary outcome; model cost-effectiveness from the paying budget's perspective at a stated discount rate; and pre-register with an independent evaluator who publishes null results.
What is the best outcome measure for longevity infrastructure?
Healthy life expectancy, disaggregated by deprivation decile and area, alongside the morbidity gap (life expectancy minus healthy life expectancy). They are published, hard to game, and measure the thing the infrastructure exists for. Intermediate outcomes — smoking prevalence, hypertension control, coverage, falls — move sooner and should be reported alongside, not instead.
Why is before-and-after comparison not enough?
Because most outcomes have secular trends: smoking falls, screening rises, mortality drifts. A before-and-after comparison attributes the trend to whatever strategy was current. Designs with a comparison — stepped-wedge randomised rollout, difference-in-differences against similar jurisdictions, synthetic controls, interrupted time series with a control series — separate the programme's effect from the trend.
What is a stepped-wedge evaluation?
A design in which a programme that will be rolled out to all areas anyway is rolled out in randomised order, so each area serves as a control until it receives the programme. It costs nothing extra, is ethically straightforward because everyone eventually gets the programme, and is the most powerful attribution design available to a government — and among the least used.
Why measure coverage before outcomes?
Because a programme that did not reach people cannot have changed their health. Falls-prevention exercise has high-certainty trial evidence and reaches a small fraction of eligible older adults; evaluating its population outcome without its coverage would conclude the evidence is wrong when the delivery is missing. Coverage by deprivation decile is the first number to publish.
How did the UK's healthy-years mission fail its evaluation?
By its own primary metric: it set a target of at least five extra healthy years by 2035, never established a coverage or delivery measure, withdrew the policy page in March 2023, and the ONS reported in February 2026 that healthy life expectancy had fallen to its lowest since the series began. A milestone-reported strategy with no attribution design and no coverage data produced a record-low outcome and no explanation.
Keep reading
- Public health and policy
The data and the cautionary cases.
- What public longevity infrastructure strategies improve population health outcomes?
The strategies with recorded outcomes.
- Evidence based public longevity infrastructure programs for governments.
The programmes worth evaluating.
- How we grade evidence
The grading behind the site's rankings.
More in Public health & policy
- What is public longevity infrastructure and why it matters?
Public longevity infrastructure defined — the laws, budgets, services and built environment a government uses to extend healthy life — and its components ranked by evidence: tobacco and alcohol policy, vaccination and screening delivery, hypertension control, primary and pharmacy care, air quality and active-travel design, falls prevention, and longevity research — with why the morbidity gap makes it matter now.
- How to invest in public longevity infrastructure projects?
How private capital can invest in public longevity infrastructure, ranked by instrument and project type: municipal and sovereign health bonds, social and health impact bonds, public-private partnerships for primary-care and diagnostic facilities, listed healthcare-infrastructure and REIT exposure, and impact funds — with what returns are realistic, what 'longevity dividend' figures actually mean, and the questions to ask before committing.
- Which public longevity infrastructure solutions offer best ROI?
Public longevity infrastructure solutions ranked on return per healthy life-year with the evidence caveats stated: tobacco and alcohol taxation, salt and sugar policy, vaccination delivery, hypertension control, cessation services, falls-prevention exercise, screening programmes, active-travel infrastructure, and longevity clinics or research institutes — with why ROI figures vary from negative to 30:1 and what a treasury should actually expect.
- How can governments fund public longevity infrastructure effectively?
Funding mechanisms for public longevity infrastructure ranked on whether money reaches delivery and survives budget cycles: earmarked health taxes, legally reserved prevention shares (EU4Health's 20%), ring-fenced prevention budgets, outcome-linked payments, general taxation with performance frameworks, and one-off capital programmes — with the evidence from the EU, UK and Singapore on what holds and what evaporates.
- What public longevity infrastructure strategies improve population health outcomes?
Public longevity strategies ranked on recorded population-health outcomes: fiscal and regulatory control of tobacco, alcohol and diet; universal primary care with enrolment (Singapore's Healthier SG model); organised vaccination and screening with call-recall; hypertension control at scale; deprivation-targeted delivery to close the 20-year healthy-life gap; healthy-ageing services; and target-led strategies without delivery — with what each has actually changed.
- What public longevity infrastructure models attract private investors?
Public longevity infrastructure models ranked on attracting private capital without distorting the health objective: availability-payment PPPs for primary-care and diagnostic facilities, earmarked-revenue bonds, outcome-based contracts and impact bonds, blended-finance funds with public first-loss capital, and concession models for healthy-ageing services — with what investors actually need, what governments must protect, and the models that go wrong.