Foundations of Economic Evaluation

The language of economic evaluation is built on a set of precise concepts that allow analysts to compare the costs and outcomes of public‑health interventions in a systematic and transparent way. Mastery of these terms is essential for designing, conducting, and interpreting studies that inform resource‑allocation decisions. Below is an extensive catalogue of the most frequently encountered vocabulary, each explained with definition, example, practical use, and common challenges.

Economic evaluation is the comparative analysis of alternative courses of action in terms of both their costs and consequences. It provides a framework for assessing whether a health intervention delivers value for money. For instance, a city health department may compare a new smoking‑cessation program with an existing one by estimating the total expenditure required and the number of quitters achieved. The primary challenge lies in ensuring that all relevant costs and outcomes are captured consistently across alternatives, which often demands careful delineation of the study perspective and time horizon.

Cost refers to the monetary value of resources consumed in delivering an intervention. Costs can be divided into direct, indirect, and intangible categories. Direct costs include items such as medication, staff salaries, and equipment, while indirect costs capture productivity losses due to illness. Intangible costs, though harder to quantify, represent pain, suffering, or loss of dignity. A practical application is the calculation of the total cost of a vaccination campaign, which adds the price of doses, cold‑chain logistics, and the time health workers spend administering vaccines. A common obstacle is assigning a monetary value to intangible costs, which may require the use of willingness‑to‑pay surveys or other valuation techniques.

Benefit denotes the positive outcomes generated by an intervention. In health economics, benefits are usually expressed in health‑related units such as life‑years saved, cases averted, or improvements in quality of life. For example, a mass deworming program may be described in terms of the reduction in infection prevalence and the associated gain in school attendance. Translating benefits into monetary terms is challenging because it often requires assumptions about the value of a statistical life or the use of quality‑adjusted measures.

Perspective defines whose costs and benefits are considered in the analysis. The most common perspectives are the societal perspective, the health‑system perspective, and the payer perspective. A societal analysis includes all costs and benefits, irrespective of who incurs them, while a health‑system analysis restricts consideration to costs borne by the public health service. If a private insurer evaluates a new diabetes management program, it would adopt the payer perspective, focusing on reimbursements and patient co‑payments. Selecting an inappropriate perspective can lead to biased results, especially when external costs such as productivity losses are substantial.

Time horizon is the period over which costs and outcomes are measured. Short‑term horizons may miss long‑lasting effects, whereas very long horizons increase uncertainty. In a cost‑effectiveness study of a childhood vaccination, a lifetime horizon is often used to capture the full stream of disease‑prevented cases and associated cost savings. Determining an appropriate horizon requires balancing completeness against the reliability of long‑term projections.

Discounting adjusts future costs and benefits to present values, reflecting the preference for immediate over delayed outcomes. The standard discount rate in many health‑economic guidelines is 3 % per annum for both costs and health effects. For example, a program that yields health gains ten years after implementation will have those gains reduced by the discount factor before being compared with immediate costs. A frequent challenge is the selection of the discount rate; different rates can substantially alter the cost‑effectiveness ratio, especially for interventions with long‑term benefits.

Cost‑minimisation analysis (CMA) is applied when the outcomes of the alternatives are proven to be equivalent, allowing the focus to be solely on identifying the least costly option. An example is the comparison of two generic antihypertensive drugs that have identical efficacy and safety profiles; the analysis would simply compare procurement prices. The limitation of CMA is that it can only be used when equivalence is convincingly demonstrated, which is rarely the case in public‑health interventions.

Cost‑effectiveness analysis (CEA) compares the relative costs and outcomes measured in natural units, such as cases prevented, life‑years gained, or symptom‑free days. A classic CEA might evaluate a new tuberculosis screening strategy by calculating the cost per additional case detected compared with the standard approach. Practical application requires the selection of an appropriate effectiveness measure that reflects the health goal of interest. One challenge is that natural‑unit outcomes are not directly comparable across different disease areas, limiting the ability to prioritize interventions on a common scale.

Cost‑utility analysis (CUA) extends CEA by incorporating patient preferences for different health states, usually expressed as quality‑adjusted life‑years (QALYs) or disability‑adjusted life‑years (DALYs). A CUA of a flu vaccination program might report the cost per QALY gained. The advantage of CUA is that it provides a common metric for comparing interventions across disease domains. However, collecting reliable utility data can be resource‑intensive, and cultural differences may affect the valuation of health states.

Cost‑benefit analysis (CBA) translates both costs and benefits into monetary terms, enabling the calculation of net monetary benefit or benefit‑cost ratio. For instance, a CBA of a road‑safety campaign could compare the monetary value of lives saved with the program’s expenses. While CBA facilitates direct comparison with other societal projects, placing a dollar value on health outcomes is ethically contentious and methodologically complex.

Incremental cost‑effectiveness ratio (ICER) is the cornerstone metric of CEA and CUA. It is calculated as the difference in costs divided by the difference in effects between two interventions. An ICER of $5,000 per QALY for a new vaccine indicates that each additional QALY costs $5,000 compared with the comparator. The ICER is used by decision‑makers to judge whether an intervention is “cost‑effective” relative to a willingness‑to‑pay threshold. Interpreting ICERs can be problematic when the denominator (incremental effect) is very small, leading to unstable ratios.

Willingness‑to‑pay (WTP) threshold represents the maximum amount a society is prepared to spend for a unit of health gain, such as a QALY. In some countries, a threshold of one to three times the gross domestic product per capita is employed. If the ICER of a new screening test falls below the threshold, it is considered acceptable. Setting an appropriate threshold is challenging because it involves normative judgments about the value of health and the opportunity cost of diverting resources.

Quality‑adjusted life‑year (QALY) combines length of life with health‑related quality of life, assigning a weight between 0 (death) and 1 (full health) to each year lived. A year lived with moderate chronic pain might be valued at 0.7, Resulting in 0.7 QALYs. QALYs enable comparison of interventions that affect both survival and morbidity. The main difficulties lie in measuring utilities accurately and ensuring that the weighting reflects the preferences of the target population.

Disability‑adjusted life‑year (DALY) measures the burden of disease by summing years of life lost due to premature mortality and years lived with disability, weighted by disability severity. A DALY averted by a malaria control program indicates a reduction in the combined mortality and morbidity burden. DALYs are frequently used in global health priority‑setting. Critics argue that the disability weights may not capture cultural variations in the experience of illness.

Utility is a numerical expression of the desirability of a particular health state, usually derived from techniques such as the time‑trade‑off, standard‑gambles, or the EQ‑5D questionnaire. For example, the utility for a mild asthma state might be 0.85. Utilities are essential inputs for QALY calculations. Obtaining valid utility values can be hampered by limited sample sizes, language translation issues, and the cognitive burden on respondents.

Effectiveness denotes the real‑world performance of an intervention, as opposed to efficacy, which is measured under ideal, controlled conditions. A community‑based nutrition program may have an efficacy of 30 % in a clinical trial but an effectiveness of 15 % when rolled out at scale due to implementation barriers. Distinguishing between efficacy and effectiveness is crucial for realistic cost‑effectiveness estimates, yet data on effectiveness are often scarce.

Outcome is any measurable result that reflects the impact of an intervention, ranging from intermediate (e.G., Blood pressure reduction) to final health outcomes (e.G., Mortality). Selecting appropriate outcomes determines the relevance of the analysis to policy makers. A challenge is that some outcomes, such as behavioral change, are difficult to quantify and may require proxy measures.

Decision‑analytic model is a structured representation of the pathways through which costs and outcomes accrue, used when trial data are incomplete or when extrapolation beyond observed periods is needed. Decision‑analytic models can be simple decision trees or more complex state‑transition (Markov) models. They allow the synthesis of diverse data sources and the exploration of alternative scenarios. Model building demands transparency and validation, as hidden assumptions can bias results.

Decision tree is a graphical model that maps out a series of mutually exclusive and exhaustive pathways, each associated with a probability, cost, and outcome. A decision tree could depict the possible outcomes of a rapid HIV test: True positive, false positive, true negative, and false negative, each with its own cost and health impact. Decision trees are easy to construct but become unwieldy when long‑term or recurring events must be modeled.

Markov model represents health states and the probabilities of moving between them over discrete time cycles. For chronic disease management, states might include “healthy,” “diseased,” “complication,” and “death.” The Markov framework captures recurring events and time‑dependent transitions, making it suitable for lifetime horizon analyses. However, the Markov assumption of memorylessness (future state depends only on current state) can oversimplify complex disease histories.

Microsimulation extends Markov modelling by simulating individual patient trajectories, allowing heterogeneity in risk factors, treatment histories, and outcomes. A microsimulation of a cancer screening program could assign each simulated person a unique set of characteristics, generating more nuanced estimates of cost‑effectiveness across subpopulations. The main drawback is the computational intensity and the need for detailed individual‑level data.

Sensitivity analysis assesses how changes in input parameters affect the results of an economic evaluation. This analysis highlights which assumptions drive the conclusions and quantifies the robustness of the findings. A deterministic one‑way sensitivity analysis might vary the discount rate from 0 % to 5 % to see its effect on the ICER. Sensitivity analysis is indispensable, yet it can be time‑consuming when many parameters are uncertain.

Probabilistic sensitivity analysis (PSA) assigns probability distributions to uncertain parameters and uses Monte Carlo simulation to propagate this uncertainty through the model, producing a distribution of ICERs. PSA generates cost‑effectiveness acceptability curves that show the probability that an intervention is cost‑effective at different WTP thresholds. PSA requires specification of appropriate distributions and sufficient computational runs to achieve stable results.

Deterministic sensitivity analysis varies one or more parameters systematically without assigning probability distributions. It includes one‑way, two‑way, and scenario analyses. For example, a deterministic analysis could explore the impact of assuming a higher vaccine efficacy (80 % versus 70 %). While easier to implement than PSA, deterministic analysis does not capture the joint uncertainty of all parameters simultaneously.

Scenario analysis examines the impact of alternative structural assumptions, such as using a societal versus a health‑system perspective, or changing the time horizon. A scenario analysis might compare the cost‑effectiveness of a water‑sanitation project when indirect productivity gains are included versus when they are excluded. The challenge lies in selecting plausible scenarios that reflect realistic policy options.

Monte Carlo simulation is the computational engine behind PSA, repeatedly drawing random values from the specified distributions and recalculating the model outcomes. The resulting set of simulated ICERs forms the basis for probabilistic statements about cost‑effectiveness. Monte Carlo methods require careful checking for convergence and sufficient iterations (often 1,000–10,000) to ensure stable estimates.

Net monetary benefit (NMB) converts the health gain into monetary terms using the WTP threshold and subtracts the cost, yielding a single value that can be compared across alternatives. NMB = (WTP × ΔEffect) – ΔCost. An intervention with a positive NMB is considered cost‑effective at the chosen threshold. NMB simplifies statistical testing but depends critically on the selected WTP value.

Net health benefit (NHB) expresses the incremental benefit in health units after adjusting for cost, using the inverse of the WTP threshold. NHB = ΔEffect – (ΔCost / WTP). NHB is useful when the analyst wishes to keep the health metric as the primary outcome. Like NMB, NHB is sensitive to the threshold and may be misleading if the threshold is poorly justified.

Budget impact analysis (BIA) estimates the financial consequences of adopting a new intervention within a specific budgetary context over a short‑ to medium‑term horizon, typically 1–5 years. A BIA for a new HPV vaccine would project the additional expenditures for procurement, storage, and administration, as well as any cost offsets from reduced cervical cancer treatment. BIA complements cost‑effectiveness analysis by providing decision‑makers with information on affordability, but it often requires detailed utilization forecasts that are uncertain.

Affordability captures whether a health system can bear the financial outlay required for an intervention without compromising other services. An intervention may be cost‑effective but unaffordable if its total cost exceeds the available budget. Assessing affordability involves not only the aggregate cost but also the distribution of expenditures over time and across program components.

Equity considerations refer to the distributional impact of an intervention across different population groups, such as socioeconomic status, ethnicity, or geographic location. A cost‑effectiveness analysis that shows a high‑value intervention may still be rejected if it exacerbates health inequities. Incorporating equity often requires additional modelling, such as subgroup analyses or the use of equity‑weighted QALYs, which adds complexity and data demands.

Distributional cost‑effectiveness analysis (DCEA) extends conventional CEA by explicitly accounting for how costs and health benefits are distributed across equity‑relevant groups. A DCEA of a maternal‑health program might reveal that while the overall ICER is favorable, the benefits accrue mainly to urban women, leaving rural women underserved. Implementing DCEA demands disaggregated data and ethical judgments about the weight given to equity.

Ethical considerations encompass the moral judgments involved in valuing health, prioritising interventions, and allocating scarce resources. Issues such as the use of QALYs, the inclusion of productivity losses, and the treatment of vulnerable populations raise ethical questions. Economic evaluations must be transparent about the ethical assumptions embedded in the analysis to allow stakeholders to critique them.

Data sources for economic evaluation include primary data collection (e.G., Micro‑costing studies, patient surveys), secondary data (e.G., Administrative databases, published literature), and systematic reviews. Choosing appropriate data sources affects the credibility and relevance of the analysis. Primary data provide context‑specific detail but are costly, whereas secondary data are readily available but may lack relevance to the target setting.

Primary data collection involves gathering new information directly from the field, such as measuring the time spent by health workers on a program activity. This approach yields high‑quality, context‑specific cost data. However, it is resource‑intensive, may suffer from measurement bias, and often requires ethical approvals and informed consent.

Secondary data are obtained from existing sources, such as national health accounts, published cost studies, or electronic health records. Secondary data can accelerate the evaluation process and reduce costs, but they may not reflect local price variations, inflation adjustments, or specific program characteristics. Validation of secondary data against local benchmarks is essential.

Systematic review aggregates evidence on effectiveness, costs, or utilities from multiple studies, applying a transparent and reproducible methodology. Conducting a systematic review of the effectiveness of a tobacco‑control policy can provide a robust estimate of the pooled effect size for use in a CEA. The major difficulty lies in heterogeneity across studies, which may require meta‑analysis techniques to synthesize.

Meta‑analysis statistically combines results from independent studies to produce a pooled estimate of effect size, often expressed as a relative risk or odds ratio. Meta‑analysis can increase precision and allow subgroup analyses. Nevertheless, publication bias, varying study quality, and differing outcome definitions can compromise the validity of the pooled estimate.

Health technology assessment (HTA) is a multidisciplinary process that evaluates the clinical, economic, ethical, and social implications of health technologies. HTA reports often include cost‑effectiveness analyses as a core component. The use of HTA ensures that policy decisions are evidence‑based, but the process can be lengthy and may be constrained by limited data availability.

Payer denotes the entity that finances health services, such as an insurance company, government agency, or employer. The payer’s perspective determines which costs are relevant to the analysis. For example, a private insurer evaluating a chronic‑disease management program will focus on reimbursable expenses and potential reductions in claim costs. Understanding payer incentives is critical for aligning the evaluation with decision‑maker priorities.

Stakeholder includes all parties with an interest in the outcome of the evaluation, such as patients, clinicians, policymakers, and industry representatives. Engaging stakeholders early can improve the relevance of the research question, ensure appropriate outcome selection, and facilitate uptake of results. However, balancing divergent interests may be challenging and may introduce bias if not managed transparently.

Intervention is any action, program, or policy intended to improve health outcomes. Interventions can range from vaccination campaigns to health‑education workshops. Precise definition of the intervention, including its components, intensity, and delivery mode, is vital for accurate costing and outcome measurement.

Comparator refers to the alternative against which the intervention is assessed, which may be “no intervention,” “standard care,” or another active program. Selecting an appropriate comparator is essential for internal validity; a poorly chosen comparator can overstate benefits or underestimate costs. In many public‑health evaluations, the comparator is the current practice within the health system.

Control is a specific type of comparator used in experimental designs, wherein participants receive no active treatment or a placebo. Control groups help isolate the effect of the intervention. In pragmatic public‑health trials, ethical considerations may limit the use of a true control, leading to the adoption of an “usual care” comparator instead.

Baseline denotes the starting point for costs and outcomes before the intervention is implemented. Baseline data provide the reference against which incremental changes are measured. Accurate baseline estimation is often hindered by data gaps, recall bias, or changes in disease epidemiology over time.

Incremental describes the difference between two alternatives in terms of cost or effect. Incremental analysis focuses on the additional resources required and the additional health gains achieved. The incremental approach is central to the calculation of ICERs, but it can be misleading when one option dominates (i.E., Is both cheaper and more effective).

Marginal cost or effect refers to the change associated with a small (often infinitesimal) increase in the level of an intervention. Marginal analysis is useful for determining the optimal scale of a program. For example, the marginal cost of adding an extra vaccination site may be lower than the average cost per site due to economies of scale. Estimating marginal values requires detailed cost functions.

Average cost is the total cost divided by the number of units (e.G., Cost per patient). Average cost is easy to compute but may conceal variations across subpopulations or over time. Decision‑makers often need to compare average cost to marginal cost to assess scaling decisions.

Average effect is the total health gain divided by the number of participants or units. Like average cost, it provides a simple summary measure but may hide heterogeneity. For instance, the average reduction in blood pressure across a study population may not reflect the larger benefit experienced by high‑risk individuals.

Cost per unit expresses cost in terms of a specific output, such as cost per vaccine dose administered or cost per screened individual. This metric facilitates budgeting and resource planning. However, it does not account for downstream health benefits, which is why it is often complemented by cost‑effectiveness measures.

Cost per case calculates the expense required to prevent or treat a single case of disease. A cost‑per‑case analysis of a malaria prophylaxis program might reveal that each averted case costs $200. This metric is useful for program managers but may be misleading if the intervention also yields broader community benefits.

Cost per life‑year saved measures the expense needed to gain one additional year of life. It is frequently used in CEA when mortality reduction is the primary outcome. For example, a cancer screening program might have a cost per life‑year saved of $4,500. The measure does not capture quality of life, which is why QALYs are often preferred.

Health economics is the discipline that studies how scarce resources are allocated to improve health outcomes. It provides the theoretical foundation for economic evaluation, including concepts such as opportunity cost, marginal analysis, and welfare economics. Health economists must balance methodological rigor with policy relevance.

Health outcomes encompass any measurable change in health status resulting from an intervention, including mortality, morbidity, functional status, and quality of life. Selecting appropriate health outcomes determines the relevance of the evaluation to stakeholders. Challenges include measuring outcomes that are rare or occur long after the intervention.

Effectiveness measure is the specific metric used to quantify the health benefit, such as cases averted, hospital admissions prevented, or QALYs gained. The choice of effectiveness measure should align with the intervention’s objective. Misalignment can lead to misleading conclusions about cost‑effectiveness.

Health state describes a particular condition or level of health, often defined by clinical criteria or utility values. In a Markov model, health states might include “asymptomatic infection,” “symptomatic infection,” and “recovered.” Accurate definition of health states is critical for valid transition probabilities and cost assignments.

Transition probability is the likelihood of moving from one health state to another during a model cycle. These probabilities are derived from epidemiological data, clinical trials, or expert opinion. Inaccurate transition probabilities can substantially bias model outputs, and they often require sensitivity analysis to assess uncertainty.

Cycle length is the time interval over which transitions are evaluated in a state‑transition model. Common cycle lengths are one year, six months, or one month, depending on the disease’s natural history. Choosing an inappropriate cycle length can either oversimplify rapid events or unnecessarily complicate slow‑progressing conditions.

Half‑cycle correction adjusts for the assumption that events occur, on average, halfway through each cycle, improving the accuracy of cost and outcome estimates. Failure to apply the correction can over‑ or underestimate total costs, especially when cycle lengths are long relative to the timing of events.

Discount rate is the percentage used to convert future costs and benefits into present values. The rate may differ for costs and health outcomes in some jurisdictions. Selecting a rate that reflects societal time preference is essential, yet empirical justification is often limited, leading to debate over the appropriate level.

Inflation adjustment updates historical cost data to a common price year using consumer price indices or health‑specific inflation rates. Without adjustment, cost comparisons across years are misleading. Inflation adjustment requires reliable price indices and careful documentation of the base year.

Exchange rate conversion is necessary when costs are reported in foreign currencies. Purchasing power parity (PPP) adjustments may provide a more accurate comparison of real resource use across countries. However, exchange‑rate fluctuations can introduce volatility into cost estimates, requiring sensitivity analysis.

Purchasing power parity (PPP) standardises costs by accounting for differences in price levels between countries, allowing for more meaningful international comparisons. PPP is especially useful in global health evaluations where interventions are compared across low‑, middle‑, and high‑income settings. Accurate PPP data may be unavailable for some regions, limiting its applicability.

Opportunity cost represents the value of the best alternative foregone when resources are allocated to a particular intervention. In public health, the opportunity cost of funding a new vaccination program might be the foregone investment in water sanitation. Quantifying opportunity cost is conceptually straightforward but empirically difficult, as it requires a clear definition of the next best alternative.

Resource allocation is the process of distributing limited financial, human, and material assets among competing health programmes. Economic evaluation provides evidence to guide allocation decisions, aiming to maximise health gains per unit of resource spent. Policy makers must balance efficiency with equity, political feasibility, and ethical considerations.

Disease burden quantifies the impact of a disease on a population, typically expressed in DALYs or QALYs lost. Burden estimates help prioritise interventions by highlighting high‑impact conditions. Accurate burden measurement depends on reliable epidemiological data, which may be lacking for emerging or neglected diseases.

Prevalence is the proportion of a population that has a particular disease at a specific point in time. Prevalence data are essential for estimating the number of individuals who could benefit from a screening or treatment programme. Prevalence may be underestimated in settings with limited diagnostic capacity.

Incidence measures the number of new cases that develop in a defined period. Incidence is crucial for modelling the spread of infectious diseases and for evaluating preventive interventions. Incidence data can be unstable in small populations, requiring aggregation or smoothing techniques.

Morbidity refers to the presence of disease or health impairment, encompassing both the frequency and severity of illness. Morbidity measures are used to calculate DALYs and to assess the benefits of interventions that reduce symptom burden. Capturing morbidity accurately often requires patient‑reported outcome measures.

Mortality denotes death rates within a population. Mortality is a primary outcome in many cost‑effectiveness analyses, especially for life‑saving interventions. Mortality data are generally reliable, but attributing cause‑specific deaths can be problematic in settings with weak vital registration systems.

Prevalence vs incidence distinguishes between the total number of existing cases (prevalence) and the number of new cases (incidence). Both metrics are relevant for economic evaluation: Prevalence informs the size of the target population, while incidence informs the potential impact of preventive measures. Confusing the two can lead to mis‑estimation of intervention reach and cost.

Health impact captures the change in health status attributable to an intervention, often expressed in QALYs, DALYs, or cases averted. Quantifying health impact enables comparison across diverse interventions. However, health impact estimates rely on assumptions about causal pathways and may be sensitive to model structure.

Health gain is the positive change in health resulting from an intervention, such as additional years of healthy life. Health gain is the numerator in most cost‑effectiveness ratios. Measuring health gain accurately requires valid outcome data and appropriate utility weights.

Health loss represents the negative health change that would occur without the intervention, often estimated through counterfactual scenarios. Accounting for health loss is essential for calculating the net benefit of a programme. Counterfactual modelling can be complex, particularly when natural disease progression is uncertain.

Cost offset occurs when an intervention reduces the need for other health‑care services, generating savings that partially or fully compensate for its own cost. For example, a smoking‑cessation program may lower future hospital admissions for respiratory disease, creating a cost offset. Identifying all relevant offsets is challenging, as they may be dispersed across different budget lines.

Cost saving describes a situation where the intervention’s total costs are lower than those of the comparator, after accounting for offsets. A cost‑saving intervention is attractive because it improves health while reducing expenditures. Nevertheless, cost‑saving claims must be robust to sensitivity analysis, as small changes in assumptions can overturn the conclusion.

Cost neutral indicates that the intervention’s costs are roughly equal to the costs of the comparator after considering offsets. Cost‑neutral programmes may still be justified if they generate important health gains or address equity concerns. Demonstrating true cost neutrality requires comprehensive accounting of all cost components.

Incremental analysis focuses on the differences between two alternatives, rather than on absolute costs or effects. Incremental analysis is the basis for ICER calculation and for determining dominance relationships. It requires careful alignment of costing methods to ensure comparability.

Incremental cost is the additional expenditure required to implement the intervention compared with the comparator. It includes any extra resources, such as training, equipment, or consumables. Accurate estimation of incremental cost may be impeded by shared costs that need to be apportioned.

Incremental effect is the additional health benefit achieved by the intervention relative to the comparator. It can be expressed in cases averted, QALYs gained, or other relevant units. Small incremental effects can produce large ICERs, making the intervention appear less attractive despite modest absolute benefits.

Dominance occurs when one alternative is both less costly and more effective than another; the dominated option is eliminated from further consideration. For example, a low‑cost, high‑impact vaccination strategy dominates a more expensive, less effective one. Detecting dominance simplifies decision‑making but requires precise cost and effect estimates.

Extended dominance happens when a combination of two alternatives provides a better cost‑effectiveness profile than a third option, rendering the third option inefficient. Extended dominance is identified by comparing the ICERs of successive options sorted by increasing cost. Recognising extended dominance helps avoid suboptimal choices that appear attractive when examined in isolation.

Cost‑effectiveness frontier is a graphical representation of the set of non‑dominated interventions, showing the trade‑off between cost and effect. Interventions lying on the frontier are considered efficient. Plotting the frontier aids visual comparison but requires consistent scaling of costs and effects.

Cost‑effectiveness plane depicts the incremental cost on the vertical axis and incremental effect on the horizontal axis, dividing the plane into four quadrants that illustrate the direction of the ICER. The plane assists in interpreting whether an intervention is more effective and more costly, less effective and less costly, or falls into the dominated quadrants. The plane is a useful teaching tool but may oversimplify multidimensional uncertainty.

Threshold analysis explores the value of a parameter at which the decision about cost‑effectiveness would change. For example, a threshold analysis might determine the vaccine price at which the ICER equals the WTP threshold. Threshold analysis is valuable for price negotiations but can be sensitive to other model assumptions.

Value of information (VOI) quantifies the benefit of obtaining additional data to reduce decision uncertainty. The expected value of perfect information (EVPI) estimates the maximum amount a decision‑maker should be willing to pay for perfect information. VOI analysis helps prioritise future research. Implementing VOI requires sophisticated modelling and may be computationally demanding.

Expected value of perfect information (EVPI) is the average monetary gain that would result if all uncertain parameters were known with certainty. A high EVPI indicates that further research could be worthwhile. Calculating EVPI involves integrating over the distribution of uncertain parameters and comparing the expected net benefit of the optimal decision with the current decision. The main challenge is the need for extensive Monte Carlo simulations.

Expected value of partial perfect information (EVPPI) isolates the value of eliminating uncertainty in a specific subset of parameters, such as the effectiveness of an intervention. EVPPI helps identify which parameters most influence the decision, guiding targeted data collection. EVPPI calculations are more complex than EVPI and often require specialised software.

Cost‑effectiveness acceptability curve (CEAC) displays the probability that an intervention is cost‑effective across a range of WTP thresholds, derived from PSA results. CEACs provide decision‑makers with a visual summary of uncertainty. However, they do not convey information about the magnitude of health gains, and overlapping curves can be difficult to interpret.

Cost‑effectiveness acceptability frontier (CEAF) extends the CEAC by showing the probability that each of several interventions is the most cost‑effective option at each WTP threshold. The CEAF assists in selecting the optimal programme when multiple alternatives exist. The frontier can be unstable when the differences in net benefit among alternatives are small.

Willingness‑to‑pay distribution reflects the variation in how different individuals or groups value health gains, acknowledging that a single threshold may not capture societal preferences. Modelling a distribution of WTP values can improve equity analysis but increases model complexity and data requirements.

Health policy refers to the set of decisions, plans, and actions undertaken to achieve specific health goals within a society. Economic evaluation informs health‑policy development by providing evidence on the relative efficiency of competing options. Translating evaluation results into policy can be hindered by political, cultural, and institutional factors.

Program evaluation assesses the design, implementation, and outcomes of a health programme, often incorporating both process and economic dimensions. Economic evaluation is a component of comprehensive program evaluation, linking cost data with health impact. Conducting rigorous program evaluation demands multidisciplinary expertise and stakeholder collaboration.

Program implementation involves the operationalisation of an intervention, including training, logistics, and monitoring.