Cost-Utility Analysis

This page explains how cost-utility analysis fits within economic evaluation, how time and health-state utility are combined into quality-adjusted life years, and how costs and QALYs are compared incrementally. It also explains how resource use, costs and utility evidence are measured, how uncertainty affects the result, what QALYs may leave out, and why good value for money does not necessarily mean that an intervention is affordable.

Cost-utility analysis is commonly shortened to CUA. It allows health gains from different diseases and interventions to be compared using a shared outcome measure, but its conclusions still depend on the population, comparator, perspective, time horizon, evidence, assumptions and institutional decision context.

How CUA fits within cost-effectiveness analysis

CUA is commonly treated as a specialised form of cost-effectiveness analysis. Both methods compare incremental costs with incremental health outcomes, but CUA specifically uses a preference-weighted outcome such as the quality-adjusted life year.

A conventional cost-effectiveness analysis might report cost per infection prevented, cost per hospital admission avoided or cost per life-year gained. CUA instead usually reports cost per QALY gained, allowing a wider range of health effects and clinical areas to be compared using the same outcome measure.

CUA is also different from cost-benefit analysis, which expresses benefits in monetary terms. A QALY is a health outcome, not a monetary value.

Which choices define a cost-utility analysis

A CUA is shaped by several choices made before the calculations begin. These choices determine which costs and health effects enter the analysis and whether its results are relevant to the healthcare decision being studied. They must be reported clearly rather than hidden in the model.

1. Define the decision problem and population.
2. Select the relevant alternatives and comparators.
3. Choose the analytical perspective.
4. Select an adequate time horizon.
5. Identify and measure relevant resource use.
6. Assign appropriate unit costs and a common price year.
7. Select appropriate health-state utility evidence.
8. Combine utility and time to estimate QALYs.
9. Apply discounting consistently to future costs and outcomes.
10. Calculate incremental costs and incremental QALYs.
11. Examine dominance before interpreting a ratio.
12. Calculate the incremental cost-utility ratio and net benefit where appropriate.
13. Characterise parameter, structural and methodological uncertainty.
14. Examine relevant population heterogeneity and distributional considerations.
15. Interpret and report the findings within the wider decision context.

Measuring resource use and costs

The costing component should begin with the analytical perspective because the perspective determines which resources and costs are relevant. A healthcare-system perspective may include treatment, monitoring, adverse-event and downstream healthcare costs. A broader societal perspective may also include patient time, caregiver time, travel or productivity effects when permitted by the applicable reference case.

Resource quantities and unit costs should be kept separate wherever possible:

Cost of a resource = Quantity used × Unit cost

Resource-use quantities may come from clinical trials, observational data, administrative records, surveys, expert elicitation or economic models. Unit costs may come from national tariffs, published cost schedules, institutional accounting systems, procurement data or costing studies.

For each important cost input, the analysis should report:

• The resource being measured.
• The quantity and unit of measurement.
• The population and setting.
• The data source.
• The unit cost and valuation method.
• The currency and price year.
• Any inflation or currency adjustment.
• Any assumptions used to address missing or unavailable data.
• Whether the cost is included or excluded under the selected perspective.
• The uncertainty assigned to the input.

Resource quantities and prices should not be combined into unexplained totals. Charges, reimbursement amounts and market prices should not be treated as automatically equal to opportunity costs. Implementation, monitoring, adverse-event and downstream costs should be included when they are relevant, and costs that appear in more than one category should not be double counted.

How QALYs are calculated

A quality-adjusted life year combines the length of time spent in a health state with a preference-based value assigned to that state.

For a period with constant utility:

QALYs = Utility × Time

Across multiple periods:

Total QALYs = Σ(Utilityₜ × Timeₜ)

If utility changes during a period, the analyst may divide the period into smaller intervals or use an appropriate area-under-the-curve method. The timing convention and any assumptions about change between observations should be stated.

Utility is usually anchored at 1 for full health and 0 for dead, although some value sets assign values below zero to health states considered worse than dead. A utility value is not money and should not be interpreted as a percentage of health.

How time horizon and discounting change results

The time horizon determines how long costs and health outcomes are followed. It should be long enough to capture all material differences between the alternatives, including effects on survival, disease progression, treatment duration, later complications and downstream resource use.

Future costs and QALYs may require discounting. The applicable rates and timing conventions depend on the jurisdiction and reference case. A single discount rate should not be presented as universally correct.

Discounted QALYs may be calculated as:

Discounted QALYsₜ = QALYsₜ ÷ (1 + r)ᵗ

Discounted costs may be calculated as:

Discounted costsₜ = Costsₜ ÷ (1 + r)ᵗ

Here, r is the applicable annual discount rate and t represents timing under the stated convention.

Analysts should explain whether costs and outcomes use the same or different discount rates, how time zero is handled and whether alternative discounting assumptions materially change the result.

Choosing utility evidence

Utility evidence should fit the population, condition, health states, intervention effects and decision jurisdiction.

Analyses should report the:

• Descriptive instrument.
• Instrument version.
• Value set.
• Valuation method.
• Respondent population.
• Administration mode.
• Timing of measurement.
• Treatment of missing data.
• Source of the utility estimates.

Utility estimates from different instruments, versions, value sets or countries are not automatically interchangeable.

Values may come directly from a preference-based measure or, when direct evidence is unavailable, from a transparent mapping model. Mapping should identify the source and target instruments, estimation population, model specification, predictive performance and uncertainty. It should not conceal weak overlap between the source measure and the health dimensions represented by the target instrument.

Caregiver health effects should be included only when relevant to the decision problem and permitted by the analytical perspective. Analysts should distinguish effects on caregivers’ own health from effects already represented in patient outcomes or costs to avoid double counting.

Comparing the costs and QALYs of two alternatives

CUA examines what changes when one healthcare alternative replaces another. The analysis therefore focuses on incremental costs and incremental QALYs rather than the total or average results for either alternative alone. The comparator must be relevant to the actual decision.

For an intervention compared with the relevant comparator:

ΔC = Cintervention − Ccomparator

ΔQALY = QALYintervention − QALYcomparator

When ratio interpretation is appropriate, the incremental cost-utility ratio (ICUR), also called the incremental cost-effectiveness ratio (ICER), is:

ICUR = ΔC ÷ ΔQALY

The result is normally expressed as a monetary amount per QALY gained. Costs and QALYs should use consistent populations, time horizons, discounting conventions and model assumptions.

Read the incremental position before the ratio

The signs of incremental cost and incremental QALYs determine what the ratio means.

Lower cost and more QALYs

The intervention costs less and produces more health. It dominates the comparator.

Higher cost and fewer QALYs

The intervention costs more and produces less health. It is dominated by the comparator.

Higher cost and more QALYs

The intervention produces additional health at additional cost. Interpretation depends on the incremental cost per QALY, the applicable threshold and the wider decision context.

Lower cost and fewer QALYs

The intervention saves resources but produces fewer QALYs. Interpretation depends on whether the savings justify the health forgone under the applicable decision rule.

A negative ICUR is ambiguous because it can represent either dominance or being dominated. If incremental QALYs equal zero, no ratio should be calculated, and the incremental cost and equal expected QALYs should be reported directly.

When several alternatives are compared, order them appropriately and assess strict and extended dominance before interpreting sequential incremental ratios. An isolated pairwise ratio may not identify the efficient set of alternatives.

Turning incremental QALYs into a decision measure

A decision threshold can be used to express the value assigned to an additional QALY. Incremental net monetary benefit combines the incremental health gain and incremental cost into one measure at that threshold.

At threshold λ:

INMB = (λ × ΔQALY) − ΔC

A positive incremental net monetary benefit favours the intervention on expected value-for-money grounds at the stated threshold. A negative result favours the comparator under the same assumptions.

Net benefit often provides a clearer basis for comparing multiple alternatives and analysing uncertainty than relying on ratios alone. The threshold remains a decision parameter, not an intrinsic monetary price of a QALY.

Worked example: cost per QALY gained

The following figures are illustrative. Current care has discounted costs of £20,000 and produces 6.60 discounted QALYs. A new intervention has discounted costs of £28,000 and produces 7.10 discounted QALYs.

ΔC = £28,000 − £20,000 = £8,000

ΔQALY = 7.10 − 6.60 = 0.50 QALYs

ICUR = £8,000 ÷ 0.50 = £16,000 per QALY gained

At an illustrative threshold of £20,000 per QALY:

INMB = (£20,000 × 0.50) − £8,000 = £2,000

The result favours the intervention on expected value-for-money grounds at that threshold, subject to uncertainty and the wider decision context. It does not demonstrate that the intervention is affordable or require its adoption.

Assumptions and model interpretation

A CUA may rely on assumptions about treatment duration, adherence, disease progression, survival, utility values, adverse events, resource use, unit costs, extrapolation and treatment effects beyond the observed evidence.

Assumptions should be:

• Stated explicitly.
• Supported by evidence or a documented rationale.
• Applied consistently across alternatives.
• Tested when they could materially affect the result.
• Distinguished from directly observed data.
• Reviewed for possible structural bias.
• Linked to the population and setting in which they are used.

A precise numerical result does not mean that the evidence or assumptions are equally precise.

How uncertainty affects the result

CUA should assess uncertainty in costs, resource use, utilities, survival, mapping, extrapolation, time horizon, discounting and structural assumptions.

Relevant approaches may include:

• Scenario analysis.
• One-way sensitivity analysis.
• Multiway sensitivity analysis.
• Probabilistic sensitivity analysis.
• Cost-effectiveness acceptability curves.
• Threshold or switching-value analysis.
• Value-of-information analysis.

Expected costs, expected QALYs and expected net benefit should be distinguished from the probability that an intervention is cost effective. A high probability of cost effectiveness is not necessarily the same as the highest expected net benefit.

CUA results should not be transferred automatically between jurisdictions. Different value sets, prices, clinical pathways, comparators, thresholds and institutional decision rules can materially change the result.

Heterogeneity and distribution

Average results can conceal important differences between population groups. Analysts should consider whether costs, QALYs or cost effectiveness differ across clinically or socially relevant subgroups.

Subgroup analysis should be based on a defensible decision question and credible evidence. It should not be created solely by searching retrospectively for favourable results.

Conventional CUA does not automatically show who gains, who loses or whether the distribution of health is considered fair. Equity, severity, unmet need and effects on disadvantaged groups may require additional analysis or explicit deliberation.

What QALYs may leave out

QALYs provide a common measure of health gain but may not fully represent every consequence relevant to a decision. Examples include:

• Non-health benefits.
• Caregiver effects.
• Financial protection.
• Process value.
• Severity.
• Equity.
• Patient experience.
• Implementation consequences.
• Effects falling outside the selected perspective.
• Health dimensions that a generic instrument does not capture well.

These consequences should be reported separately or incorporated using an explicitly justified complementary approach rather than silently treated as having zero value.

Why good value does not guarantee affordability

CUA examines comparative value by asking whether additional QALYs justify additional costs under a stated decision rule. Budget impact analysis asks how adoption could change total expenditure for a particular budget holder over a specified planning period. The two analyses answer related but different questions.

An intervention can have a positive incremental net monetary benefit and still be difficult to afford. This can happen when the eligible population is large, uptake is rapid, treatment costs occur early or the available budget is constrained.

Budget impact analysis therefore complements CUA by adding information about population size, uptake, treatment mix, expenditure timing and total financial consequences. A favourable cost-per-QALY result does not independently establish affordability or guarantee adoption.

What CUA can and cannot decide

CUA provides structured evidence about comparative costs and preference-weighted health outcomes. It can help decision-makers compare alternatives and consider value for money, but it does not incorporate every factor relevant to healthcare coverage, reimbursement or implementation.

• CUA does not determine whether an intervention is affordable.
• CUA does not determine whether health gains are distributed fairly.
• CUA does not replace evidence about clinical effectiveness, safety or evidence quality.
• CUA does not determine whether implementation is organisationally feasible.
• CUA does not establish that every QALY should receive identical priority.
• CUA does not make a result automatically transferable between jurisdictions.
• CUA does not make the final coverage, reimbursement or adoption decision.

Health technology assessment may use CUA alongside clinical, budgetary, ethical, organisational, equity and implementation evidence. The economic result informs that wider process but does not replace it.

Common mistakes and how to avoid them

A cost-utility model can produce precise calculations while still giving a misleading answer. Common problems involve inappropriate utility evidence, inconsistent timing, incorrect incremental analysis or claims that go beyond what the result can support.

• Treating a QALY as a monetary amount or percentage of health.
• Calculating cost per QALY from average rather than incremental results.
• Interpreting a negative ICUR without examining the incremental position.
• Failing to assess strict or extended dominance when multiple alternatives exist.
• Treating the threshold as an intrinsic price of a QALY.
• Assuming that a favourable result establishes affordability.
• Mixing utility instruments, versions or value sets without justification.
• Combining costs from inconsistent currencies or price years.
• Hiding resource quantities and unit costs inside unexplained totals.
• Omitting relevant implementation, adverse-event or downstream costs.
• Double counting patient, caregiver or productivity effects.
• Using a time horizon that omits important future differences.
• Applying discount rates without identifying the jurisdiction or reference case.
• Ignoring uncertainty in utility estimates, costs, survival or extrapolation.
• Treating a high probability of cost effectiveness as the decision itself.
• Transferring results between countries or settings without assessing relevance.

CUA conclusions should remain tied to the population, alternatives, perspective, time horizon, utility evidence, costs, assumptions, uncertainty, threshold and institutional context used in the analysis.

What should be reported

A CUA should report the decision problem, population, alternatives, comparator, perspective, time horizon, costing year, currency, resource-use sources, unit-cost sources, discount rates, utility instrument and version, value set, valuation population, utility source, QALY calculation, missing-data methods, incremental results, dominance assessment, net benefit, uncertainty, heterogeneity, caregiver effects, limitations and decision context.

Sources

• Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes, 4th edition. Oxford University Press. 2015.
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). Published 2022; updated 2026.
• Wailoo AJ, Hernandez-Alava M, Manca A, Mejia A, Ray J, Crawford B, et al. Mapping to estimate health-state utility from non-preference-based outcome measures: an ISPOR Good Practices for Outcomes Research Task Force report. Value in Health. 2017;20(1):18-27.
• Fenwick E, Claxton K, Sculpher M. Representing uncertainty: the role of cost-effectiveness acceptability curves. Health Economics. 2001;10(8):779-787.
• Briggs AH, Weinstein MC, Fenwick EAL, Karnon J, Sculpher MJ, Paltiel AD. Model parameter estimation and uncertainty analysis: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force Working Group-6. Medical Decision Making. 2012;32(5):722-732.
• Cookson R, Mirelman AJ, Griffin S, Asaria M, Dawkins B, Norheim OF, et al. Using cost-effectiveness analysis to address health equity concerns. Value in Health. 2017;20(2):206-212.
• Husereau D, Drummond M, Augustovski F, de Bekker-Grob E, Briggs AH, Carswell C, et al. Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) statement: updated reporting guidance for health economic evaluations. BMJ. 2022;376:e067975.