Economic Evaluation

Why this matters in health economics

Every health budget is finite, and every pound, euro, rupee, or baht spent on one service is unavailable for another. That is not an accounting inconvenience; it is the moral centre of the field. When a payer funds a new cancer drug, the money comes from somewhere — a diabetes clinic not opened, a mental health post not filled, a maternity ward not refurbished. Economic evaluation exists to make that hidden trade-off explicit, so that decisions about scarce resources are made deliberately rather than by default, drift, or whoever lobbies loudest.

The stakes are simultaneously financial and clinical. A decision to fund a low-value intervention does not merely waste money; it displaces care that would have produced more health for the same spend, so the true cost is measured in the health forgone by other patients. This is why economic evaluation is a safety and equity discipline, not only an efficiency one. Getting it wrong means real people elsewhere in the system go without, usually invisibly and usually the least powerful among them.

For a director, the discipline earns its keep at exactly the moments that feel most political: introducing a new technology, defending a service against cuts, deciding a screening programme, or justifying a prevention budget whose payback lies years away. Economic evaluation will not make those decisions painless, and it should not pretend to. Its promise is narrower and more valuable — to structure the argument honestly, to show what is being given up, and to make the reasoning defensible to a regulator, a treasury, a clinician, and the public. This chapter is the workhorse of Part 2; the modelling that generates the estimates is developed in Chapter 2.2 — Modelling, and the drug-specific pricing questions are the subject of Chapter 2.4 — Pharmacoeconomics.

Core concepts

Economic evaluation compares two or more courses of action by both their costs and their consequences. A study that reports only cost is a costing exercise; a study that reports only outcomes is a clinical one; only when the two are set against each other, for at least two options, does the analysis become a full economic evaluation. The unifying idea beneath every method is opportunity cost — the value of the best alternative given up — because in a fixed budget the real cost of anything is the health it displaces.

The methods differ mainly in how they value consequences. Cost-effectiveness analysis (CEA) measures outcomes in natural clinical units — cases detected, deaths averted, millimetres of mercury of blood pressure lowered — and expresses results as cost per unit of effect. It is intuitive and clinically legible, but it cannot compare across unlike conditions: cost per stroke prevented tells you nothing about whether that beats cost per case of blindness averted.

Cost-utility analysis (CUA) solves that by measuring outcomes in a common currency of health that combines length and quality of life. Its dominant unit is the quality-adjusted life year (QALY), where one year in full health scores 1.0, death scores 0, and states in between are weighted by a utility value. A treatment that gives ten people an extra year at a quality weight of 0.5 produces five QALYs. Because the QALY is generic, CUA lets a payer compare a hip replacement with a cancer therapy on the same scale — which is precisely why bodies such as England's National Institute for Health and Care Excellence (NICE) adopt it as their reference case. Its close cousin, the disability-adjusted life year (DALY), measures health loss rather than health gain, combining years of life lost with years lived with disability; it underpins the Global Burden of Disease work and much World Health Organization (WHO) and World Bank priority-setting, especially in low- and middle-income settings.

Cost-benefit analysis (CBA) goes further and values consequences in money, so that health gains and costs are expressed in the same units and net benefit can, in principle, be compared with spending outside health entirely — roads, schools, defence. Its power is also its discomfort: to run it you must attach a monetary value to health and to life itself, which many find ethically fraught even when unavoidable. Two adjacent framings borrowed from finance and the social sector are return on investment (ROI), the ratio of monetized benefit to cost often used in public-health business cases, and social return on investment (SROI), which widens the lens to include social and wellbeing value to families and communities.

Comparing options produces an incremental cost-effectiveness ratio (ICER): the difference in cost between two options divided by the difference in effect. An ICER is judged against a cost-effectiveness threshold — the maximum a decision-maker will pay for a unit of health. Thresholds vary enormously by country and are contested everywhere: some jurisdictions publish an explicit cost-per-QALY range, others such as Germany's Institute for Quality and Efficiency in Health Care (IQWiG) reject a single threshold in favour of an efficiency frontier within a therapeutic area, and WHO's earlier practice of benchmarking against multiples of gross domestic product (GDP) per capita has since been qualified as too crude. Never quote one country's threshold as if it were universal.

Three further concepts recur throughout. First, discounting: costs and health that fall in the future are worth less than the same amount today, so both streams are discounted to present value at a rate set by national guidance — a choice that bears heavily on prevention, whose costs come first and benefits much later. Second, analytic perspective: whether the evaluation counts only health-system costs, payer costs, or the full societal costs including productivity and unpaid care determines what is in scope and can flip a conclusion. Third, valuing health states: utility weights for the QALY are elicited from populations using instruments such as the EQ-5D descriptive system, and preference-elicitation methods such as time trade-off and the visual analogue scale, while monetizing benefit for CBA relies on stated-preference willingness to pay. These building blocks feed the formal models of Chapter 2.2 — Modelling.

The utility weights that turn a year of life into a QALY do not fall from the sky; they are elicited from people using formal preference methods, and the method changes the number. The visual analogue scale is the simplest: respondents mark a health state on a line from the worst to the best imaginable state. It is quick, but it is a rating exercise rather than a genuine trade-off, so its values do not sit naturally on the QALY's dead-to-full-health scale. The time trade-off, named above, asks how many years in a poorer state a person would surrender to spend fewer years in full health, converting duration into a quality weight. The standard gamble goes further and frames the choice as risk: a respondent weighs a certain impaired state against a treatment offering either full health or immediate death at some probability, and the probability at which they are indifferent becomes the utility. Standard gamble is grounded in expected-utility theory and is often called the classical method, but it is cognitively demanding and people's attitudes to risk can distort the answer, which is why the more tractable time trade-off is now more common in practice. The lesson for a director is that a QALY is only as solid as the elicitation method beneath it, and values produced by different methods should not be pooled uncritically.

Two families of instrument feed these methods. Generic, multi-attribute utility instruments describe health in a fixed set of dimensions and attach a pre-scored value set to every combination of answers: the EQ-5D is the most widely used, but it is not the only one — the SF-6D (derived from the SF-36 health survey) and the Health Utilities Index are established alternatives, and each can return a different utility for the same patient, so the choice of instrument is not neutral. Generic measures buy comparability across conditions — the entire point of the QALY — but can be insensitive to changes that matter intensely within a single disease. Condition-specific measures, by contrast, are tuned to one area — vision, hearing, mental health, cancer symptoms — and detect change the generic instruments miss, but they cannot be compared across conditions and usually yield no utility value directly. Where a study has collected only a condition-specific measure, analysts use "mapping" (also called cross-walking): a statistical function that predicts generic utility values from the condition-specific responses, so that QALYs can be recovered. Mapping is a useful bridge, but it adds a layer of uncertainty, and a mapped utility is weaker evidence than one measured directly.

A less technical question sits underneath all of this: whose preferences should the value set represent? Value sets can be elicited from the general public — the dominant convention, on the reasoning that a publicly funded system spends citizens' money and should reflect citizens' values — or from patients who actually live in the health state, who often rate it as less severe than the public imagines. That choice is a value judgement, not a technicality, and it embeds a whole theory of whose views count (the deeper argument about what health economics chooses to measure and value is developed in Chapter 3.5 — Capabilities). Because instrument, value set, elicitation method, perspective, and discount rate could each be decided differently and would each move the answer, health technology assessment (HTA) bodies fix a "reference case": a stipulated set of methodological choices every submission must follow, so evaluations are comparable with one another even at the cost of some case-by-case tailoring. Underpinning the raw data are patient-reported outcome measures (PROMs) — standardized questionnaires that patients complete themselves about their symptoms, functioning, and quality of life. Increasingly collected in routine care as well as in trials, PROMs are becoming the primary raw material from which outcome values, and ultimately QALYs, are built.

Best practices

1. Frame the decision problem before you choose a method. State precisely what is being compared, for which population, against which comparator, over what time horizon, and from whose perspective. The comparator matters most and is most often fudged: comparing a new intervention against "doing nothing" flatters it when the real alternative is current standard care. A well-specified question — often summarized as population, intervention, comparator, outcome, and setting — is worth more than a sophisticated model built on a vague one.

2. Fix the perspective before you count a single cost. Whether you take a health-system, payer, or societal perspective decides which costs and benefits are in scope, and it can reverse a conclusion. A workplace mental health programme may look poor from a narrow health-payer view and excellent once lost productivity is counted. Agree the perspective with decision-makers up front, state it prominently, and if you report a societal view, report the health-system view alongside it so each audience can see their own budget.

3. Choose the evaluation type to fit the decision, not fashion. Use CEA when the comparison is within one clinical area and a natural outcome unit suffices; use CUA when you must compare across conditions or when quality of life is central; use CBA or ROI when the decision competes with spending outside health or must be argued to a treasury in monetary terms. Matching method to question prevents both false precision and false comparability.

4. Measure outcomes with a validated instrument, and say how you valued them. A QALY is only as credible as the utility weights behind it, and those weights depend on which instrument you used, whose preferences you elicited, and by what method — time trade-off, standard gamble, or a mapped descriptive system such as the EQ-5D. Different instruments and value sets can move an ICER materially. State the source of your values explicitly rather than burying it, because a reviewer cannot judge a result whose valuation basis is hidden.

5. Count costs completely and consistently with the perspective. Identify, measure, and value every relevant resource — staff time, drugs, devices, facilities, and, under a societal perspective, patients' and carers' time and productivity. The commonest error is silent omission: leaving out downstream costs an intervention avoids or creates. Use consistent unit costs and a clear price year, and never mix perspectives within a single comparison.

6. Discount future costs and health, and test the rate. Because a health year or a pound arriving in a decade is worth less than one today, discount both streams to present value using the rate your national guidance specifies. The rate is not neutral: a higher rate penalizes prevention and long-horizon programmes whose payoff is distant, so always show how the conclusion moves if the rate changes. Report undiscounted results too, so the raw shape of costs and benefits stays visible.

7. Report the ICER, but interpret it against the right threshold and its own uncertainty. An ICER is a point estimate surrounded by uncertainty, and its meaning depends entirely on the decision-maker's threshold. Present it with the uncertainty analysis that Chapter 2.2 — Modelling develops — confidence regions, cost-effectiveness acceptability curves — and name the threshold you are judging it against, its source, and its contestedness. A result reported as a bare number without a threshold and a spread is not yet an answer.

8. Treat the threshold as a contested policy choice, never a law of nature. Cost-effectiveness thresholds differ across countries by an order of magnitude and are argued over within each. Some are set by convention, some empirically estimated from the health displaced at the margin, and some — as at IQWiG — rejected altogether in favour of therapeutic-area frontiers. State which logic your threshold follows, and never import one jurisdiction's figure into another as though health were priced the same everywhere.

9. Make opportunity cost concrete, not rhetorical. The point of the whole exercise is to show what is displaced. Where you can, say what the marginal spending would otherwise have bought — the services at the edge of the budget that a new commitment pushes out. An empirically grounded threshold is one attempt to represent that displaced health; naming it turns "we cannot afford everything" from a slogan into a specific, defensible trade-off (the system-level version of this is the subject of Chapter 3.3 — Rationing).

10. Run and report sensitivity and uncertainty analysis as a matter of course. Every evaluation rests on assumptions — effect sizes, costs, utilities, the discount rate, the time horizon. Vary the influential ones and show whether the conclusion holds; a result that flips under a plausible change of assumption is a fragile basis for a decision. Distinguish parameter uncertainty (handled probabilistically) from structural and methodological choices (handled by scenario analysis), and be honest about which drives your result.

11. Surface the distributional and equity consequences alongside the average. A favourable average ICER can conceal that the gains accrue to the already-healthy while the costs fall on the disadvantaged. Report who gains and who bears the cost, not only the aggregate, and consider equity weighting or a distributional analysis where the stakes warrant it. Efficiency and equity can pull in opposite directions, and hiding the tension does not resolve it (developed in Chapter 3.4 — Equity).

12. State the limits of the QALY rather than defending it as complete. The QALY is a workhorse, not a scripture: it can under-weight what matters at the end of life, in carers' experience, and in dimensions of wellbeing beyond health. Acknowledging this openly strengthens an evaluation, and where capability or broader wellbeing is central, complementary measures may be needed (see Chapter 3.5 — Capabilities). Present the QALY result as one important input to judgement, not a verdict that ends it.

Questions to discuss with your team

1. From whose perspective are we actually making this decision, and who is paying for the health we are about to displace? Perspective is the most consequential choice in an evaluation and the one teams most often leave implicit. A narrow health-payer view keeps the analysis tractable and matches most reimbursement rules, but it writes out costs that fall on patients, carers, employers, and other public budgets — and a programme that fails on the narrow view can succeed handsomely on a societal one. The honest tension is that the perspective which makes your case strongest is rarely the one your budget-holder is accountable for. A good discussion names the primary perspective explicitly, agrees whether to report a societal view alongside it, and is candid about the fact that whatever you fund displaces care elsewhere in the system — so the question is not only "is this good value?" but "good value measured against what, for whom?"

2. What threshold are we judging value against, where did it come from, and are we willing to defend it in public? Teams reach for a cost-per-QALY figure as though it were a fixed exchange rate, when in reality thresholds are conventions, estimates, or outright rejected depending on the jurisdiction. This question forces the group to say whether their threshold is set by national guidance, estimated from the health displaced at the margin, borrowed from another country, or simply assumed — and each of those has very different authority. The deeper tension is that a threshold is a statement about how much a year of someone's health is worth, which is politically and ethically loaded however technical it looks. An honest answer states the threshold's source and logic, admits its contestedness, distinguishes it from the practice of other systems, and is prepared to explain to a clinician or a citizen why the line sits where it does rather than pretending the number fell from the sky.

3. Where in this analysis are the assumptions doing the real work, and would the decision survive if we are wrong about them? Every evaluation compresses uncertainty into tidy point estimates, and the danger is mistaking a precise-looking ICER for a robust one. This question sends the team to the parameters and structural choices the result is most sensitive to — the effect size extrapolated beyond the trial, the discount rate applied to distant prevention benefits, the utility values chosen, the time horizon. The tension is between the decision-maker's wish for a single clear answer and the analyst's duty to show the answer's fragility. An honest response identifies the two or three assumptions that could flip the conclusion, reports how the result moves across their plausible range, and states plainly whether the recommendation holds throughout or only under favourable assumptions — because a decision that depends on the optimistic end of every range is a decision resting on hope.

4. What are we really comparing this against — and is that the thing patients would otherwise receive? The comparator is the most quietly decisive choice in any evaluation, because an intervention's value is defined entirely by what it displaces, yet teams routinely compare against "no treatment" or a placebo when the real-world alternative is an established standard of care. This question forces the group to name the counterfactual explicitly and to check it against clinical reality rather than trial convenience. The tension is that a weak comparator flatters the new option and makes the numbers easier, while the honest comparator — current best practice, perhaps already cheap and effective — is a far harder benchmark to beat. An honest answer states what the intervention would actually replace in this setting, justifies that choice, and resists the temptation to widen the gap by picking a comparator no clinician would defend. It also asks whether the comparator differs across the population, since what one district already does may not be what another does at all.

5. Who gains and who bears the cost once we look past the average, and could a "cost-effective" decision still widen inequality? A single favourable ICER can conceal that the health gains flow to the already-advantaged while the costs, or the displaced services, fall on the worst-off — and averaging over a population hides exactly the distribution a payer with equity duties is meant to watch. This question moves the team from the mean to the spread: which groups take up the intervention, which are missed, and where does the opportunity cost land. The genuine tension is that efficiency and equity can pull in opposite directions, so the most cost-effective option on average may be the one that entrenches an existing gradient. An honest answer reports who benefits and who pays rather than the aggregate alone, considers whether equity weighting or a distributional analysis is warranted, and is candid that resolving the tension is a value judgement for decision-makers, not something the arithmetic settles (developed in Chapter 3.4 — Equity).

6. What is the QALY not capturing here, and are we treating it as one input to judgement or as the verdict? The quality-adjusted life year is a workhorse, but it can under-weight what matters at the end of life, in carers' experience, and in dimensions of wellbeing that sit outside health, and the chosen instrument may simply be blind to the change this intervention produces. This question asks the team to name, before they defend the number, what the measure leaves out and whether that omission is material to this decision. The tension is that the QALY's great strength — a single comparable currency across conditions — is also what forces heterogeneous human experience through one narrow gate. An honest answer states the QALY's limits openly rather than defending it as complete, brings a condition-specific or capability measure where health alone does not capture what is at stake (see Chapter 3.5 — Capabilities), and presents the cost-utility result as one important input to a judgement that people, not formulas, must make.

In practice: a health economics example

The National Health Fund of the fictional upper-middle-income country of Costa Verde is deciding whether to introduce a national bowel-cancer screening programme. The clinical case is promising: a faecal immunochemical test posted to everyone aged 55 to 74 every two years, with colonoscopy for positive results, could detect cancers earlier and prevent some through the removal of pre-cancerous polyps. The programme is not cheap — postage and laboratory testing at population scale, a large expansion of colonoscopy capacity, and the management of complications — and the Fund's budget is fixed. Its analysts are asked whether the programme is worth funding, and against what.

They begin by framing the decision rather than the arithmetic. The population is the eligible age band; the comparator is current practice, in which most cancers present symptomatically at a later stage; the outcome is best measured in QALYs, because the programme trades near-term costs and a small risk of colonoscopy harm against future life-years gained and quality of life protected. The perspective is set as health-system first, with a societal analysis reported alongside, because screening spares productivity losses and family care that a payer-only view would ignore. The time horizon is lifetime, since the benefits of a cancer prevented accrue for decades — which immediately makes discounting decisive.

The analysts build the evaluation as a cost-utility analysis feeding into the modelling of Chapter 2.2 — Modelling, projecting costs and QALYs for the screened and unscreened cohorts over their remaining lifetimes. The headline result is an incremental cost-effectiveness ratio: the extra cost of screening divided by the extra QALYs it produces. On the health-system perspective the ICER lands in a range the Fund considers acceptable by its own published cost-per-QALY guidance — but the analysts refuse to stop at the point estimate. Probabilistic sensitivity analysis shows the result is robust to uncertainty in test accuracy and cancer incidence, but highly sensitive to two things: the discount rate applied to the distant life-years, and the assumed uptake of screening invitations. At a higher discount rate the future benefits shrink and the ICER worsens sharply; at low uptake the fixed programme costs are spread over too few detected cancers and value collapses.

This turns the analysis from an arithmetic exercise into a decision brief. The analysts make the opportunity cost concrete: at the Fund's threshold, the money required would otherwise fund roughly the equivalent marginal expansion of the existing diabetes and cardiovascular services, and they say so, so the board is choosing between named alternatives rather than approving a good idea in the abstract. They flag an equity dimension the average conceals — uptake of postal screening tends to be lowest in the poorest districts, so a programme that is cost-effective on average could widen the very inequalities the Fund is mandated to narrow unless outreach is funded (the tension developed in Chapter 3.4 — Equity). Their recommendation is therefore conditional: fund the programme, but ring-fence a targeted uptake budget for underserved districts and commit to reviewing real-world uptake before national roll-out. The evaluation did not make the decision, and it did not pretend to; it made the trade-offs, the fragilities, and the losers visible enough for the board to decide with its eyes open.

Four sector lenses

Startup

A digital health start-up typically meets economic evaluation as an evidence hurdle it must clear to be reimbursed — a payer or HTA body asking not "does it work?" but "does it deliver enough health per pound to displace something else?" Early ventures rarely have lifetime outcome data, so they lean on modelled evaluation and surrogate endpoints, which raises the bar for transparency about assumptions. The pragmatic move is to design evaluation in from the first pilot: capture resource use and a validated outcome measure such as the EQ-5D alongside clinical data, choose the perspective the target payer uses, and avoid the classic trap of comparing against nothing rather than against real standard care. A credible early ICER, honestly bounded, opens more doors than an inflated one.

Small business

A small but established provider — a group practice, a single clinic, a care home, or a niche device supplier — meets economic evaluation less as a submission it authors and more as a case it must read and answer. Unlike a start-up, it has real operating data: known unit costs, actual utilization, and a track record it can point to rather than model. Its constraint is capacity, not credibility — it rarely has a health economist on staff — so the practical task is to use the perspective and evidence standards its commissioner already applies, and to marshal its own cost and outcome data to show value on those terms. When it bids for a contract or defends a service against a purchaser's disinvestment review, a modest, well-evidenced cost-per-outcome argument grounded in real caseload beats a sophisticated model it cannot resource or sustain.

Enterprise

A large provider, insurer, or hospital group runs economic evaluation as a repeatable internal function, not a one-off study. The questions are portfolio-scale — which services to expand, which to disinvest from, where a new technology earns its place — and the discipline is consistency: a shared reference case, agreed perspective, common unit costs, and a standing threshold so that this year's decisions are comparable with last year's. At enterprise scale the harder problem is often disinvestment, because releasing resource from low-value activity is politically harder than funding something new, yet it is where the opportunity-cost logic bites most. The organizations that do this well treat evaluation as governance, not advocacy, and separate the analysts from the sponsors of the services being appraised.

Government

A ministry or national payer uses economic evaluation as the backbone of accountable rationing across an entire population, usually through an HTA process. Its evaluations must withstand legal challenge, public scrutiny, and industry lobbying, so method, perspective, and threshold are published and defended rather than assumed. Government also carries responsibilities a firm does not: distributional fairness, statutory equity duties, and the credibility of the whole system's promise that decisions rest on evidence and not influence. The comparative landscape shows there is no single right answer — NICE, IQWiG, WHO-CHOICE, and the United States' Institute for Clinical and Economic Review each embody a different, defensible balance of method and values — and a government's task is to choose a coherent stance and apply it consistently.

Common failure modes

• The wrong comparator. Evaluating an intervention against "no treatment" or placebo when the real-world alternative is current standard care, which flatters the new option. Fix: always compare against what the intervention would actually replace, and justify the comparator explicitly.

• Hidden or shifting perspective. Counting costs from one standpoint and benefits from another, or leaving the perspective unstated so scope is chosen conveniently. Fix: declare the perspective up front, keep costs and benefits consistent with it, and report the health-system view even when leading with a societal one.

• Treating the threshold as universal. Importing one country's cost-per-QALY figure into another, or presenting a threshold as an objective fact rather than a contested policy choice. Fix: name the threshold's source and logic, and never transplant it across systems.

• Point estimates without uncertainty. Reporting a single ICER as though it were precise, with no sensitivity or probabilistic analysis. Fix: always accompany the ICER with the analysis of uncertainty and name the assumptions that could reverse the conclusion.

• Ignoring discounting or mishandling it for prevention. Failing to discount, or applying a rate that quietly destroys the case for long-horizon prevention without saying so. Fix: discount both streams to national guidance, test the rate, and report undiscounted results alongside.

• Averages that bury equity. Reporting a favourable mean while gains flow to the advantaged and costs to the disadvantaged. Fix: report who gains and who pays, and consider distributional or equity-weighted analysis where the stakes warrant.

• An outcome measure blind to the condition. Choosing a generic utility instrument that cannot detect the changes the intervention actually produces, so a genuine benefit registers as little or no QALY gain. Fix: check the instrument is sensitive in the relevant condition, and where it is known to be insensitive, add a condition-specific measure — mapped to utilities if a QALY is needed — rather than concluding the intervention does nothing.

• Defending the QALY as complete. Presenting a cost-utility result as a final verdict and dismissing what the QALY misses at end of life, in carers, or in wellbeing. Fix: state the QALY's limits and bring complementary measures where they matter.

Maturity model

Checklist

• State the decision problem: population, intervention, real-world comparator, outcome, setting, and time horizon.
• Declare the analytic perspective (health-system, payer, or societal) before counting any cost, and report the health-system view even if leading with a societal one.
• Choose the evaluation type (CEA, CUA, CBA, ROI/SROI) to fit the decision, and justify the choice.
• Value outcomes with a validated instrument and state the value set and elicitation method used.
• Identify, measure, and value all relevant costs consistently with the perspective, with a stated price year.
• Discount future costs and health to present value per national guidance, test the rate, and report undiscounted results too.
• Report the ICER with its uncertainty, and name the threshold, its source, and its contested status.
• Never transplant one country's threshold into another as if it were universal.
• Run probabilistic and scenario sensitivity analysis and name the assumptions that could flip the conclusion.
• Make the opportunity cost concrete: say what the spending would otherwise have funded.
• Report who gains and who bears the cost, not only the average, and flag equity consequences.
• State the limits of the QALY and bring complementary measures where health alone does not capture what matters.

Key sources

• Michael Drummond, Mark Sculpher, Karl Claxton, Greg Stoddart & George Torrance, Methods for the Economic Evaluation of Health Care Programmes — the standard reference text on evaluation methods.
• Marthe Gold, Joanna Siegel, Louise Russell & Milton Weinstein (eds.), Cost-Effectiveness in Health and Medicine — the US Panel on Cost-Effectiveness in Health and Medicine's methodological reference.
• John Brazier, Julie Ratcliffe, Joshua Salomon & Aki Tsuchiya, Measuring and Valuing Health Benefits for Economic Evaluation — the standard reference on health-state valuation, utility instruments, and elicitation methods.
• EuroQol Group — the EQ-5D instrument, its descriptive systems, and country value sets — https://euroqol.org
• NICE methods guidance (the manual for health technology evaluations) — an exemplar of a national reference case, perspective, and threshold — https://www.nice.org.uk/process/pmg36
• IQWiG general methods — Germany's efficiency-frontier approach as a contrasting institutional stance — https://www.iqwig.de
• WHO-CHOICE (CHOosing Interventions that are Cost-Effective) — WHO guidance on generalised cost-effectiveness analysis and DALYs.
• Institute for Clinical and Economic Review (ICER) value assessment framework — a United States exemplar.
• GOV.UK, Health economics: a guide for public health teams — practitioner ROI and cost-effectiveness tools — https://www.gov.uk/guidance/health-economics-a-guide-for-public-health-teams
• Economics Network, Health Economics for Teachers — economic evaluation module — https://economicsnetwork.ac.uk/health/teachers

References

1. Cost-effectiveness analysis — Wikipedia — https://en.wikipedia.org/wiki/Cost-effectiveness_analysis
2. Cost–utility analysis — Wikipedia — https://en.wikipedia.org/wiki/Cost-utility_analysis
3. Cost–benefit analysis — Wikipedia — https://en.wikipedia.org/wiki/Cost-benefit_analysis
4. Quality-adjusted life year — Wikipedia — https://en.wikipedia.org/wiki/Quality-adjusted_life_year
5. Disability-adjusted life year — Wikipedia — https://en.wikipedia.org/wiki/Disability-adjusted_life_year
6. Incremental cost-effectiveness ratio — Wikipedia — https://en.wikipedia.org/wiki/Incremental_cost-effectiveness_ratio
7. Opportunity cost — Wikipedia — https://en.wikipedia.org/wiki/Opportunity_cost
8. Discounting — Wikipedia — https://en.wikipedia.org/wiki/Discounting
9. EQ-5D — Wikipedia — https://en.wikipedia.org/wiki/EQ-5D
10. Time trade-off — Wikipedia — https://en.wikipedia.org/wiki/Time_trade-off
11. Willingness to pay — Wikipedia — https://en.wikipedia.org/wiki/Willingness_to_pay
12. Patient-reported outcome — Wikipedia — https://en.wikipedia.org/wiki/Patient-reported_outcome
13. Michael Drummond et al., Methods for the Economic Evaluation of Health Care Programmes — Oxford University Press.
14. John Brazier, Julie Ratcliffe, Joshua Salomon & Aki Tsuchiya, Measuring and Valuing Health Benefits for Economic Evaluation — Oxford University Press.
15. EuroQol Group — the EQ-5D instrument and value sets — https://euroqol.org
16. NICE — Health technology evaluations: the manual (PMG36) — National Institute for Health and Care Excellence — https://www.nice.org.uk/process/pmg36
17. GOV.UK — Health economics: a guide for public health teams — UK Government — https://www.gov.uk/guidance/health-economics-a-guide-for-public-health-teams
18. Economics Network — Health Economics for Teachers — https://economicsnetwork.ac.uk/health/teachers