Cost-Effectiveness Analysis for Policy
- Introduced by
- Health-economics and program-evaluation tradition (Drummond et al.; Gold et al.)
- Year
- 2015
- Type
- Economic evaluation comparing cost per unit of effect
- Family
- Operations research / economic evaluation
What does it do?
Cost-effectiveness analysis (CEA) is an economic evaluation that compares competing policies or programs by their cost relative to a single, common measure of effect — lives saved, cases averted, years of education gained, or quality-adjusted life years (QALYs).
Rather than valuing outcomes in money, CEA expresses results as an incremental cost-effectiveness ratio (ICER): the extra cost of one option per extra unit of outcome it delivers compared with the next-best alternative. Codified in standard references such as Drummond and colleagues' Methods for the Economic Evaluation of Health Care Programmes and the US Panel's Cost-Effectiveness in Health and Medicine, CEA is the dominant appraisal tool for health and increasingly for other public programs with a shared outcome metric.
The intuition behind the method
Public budgets are fixed, so the real question is rarely 'is this program good?' but 'which program gives us the most for the money?'. Cost-effectiveness analysis answers it without having to put a price tag on a life or a school year, which is often the sticking point of cost-benefit analysis. Instead it asks: for each extra unit of the outcome we care about, how much does this option cost compared with the alternative? If vaccinating costs $500 per life-year gained and a screening program costs $50,000 per life-year gained, the comparison is stark even though no one had to declare what a life-year is worth in dollars. Decision-makers then judge those ratios against what they are willing to pay.
History
Cost-effectiveness analysis developed in the post-war defence and public-sector planning era as an alternative to monetising all outcomes, and migrated strongly into health economics from the 1970s. The 1996 US Panel on Cost-Effectiveness in Health and Medicine, convened by the Public Health Service and reported in the volume edited by Gold and colleagues, set influential methodological standards and recommended the societal perspective and QALY outcome. Drummond and colleagues' textbook, now in its fourth edition, became the global reference, and agencies such as the UK's NICE institutionalised CEA with explicit cost-per-QALY thresholds, cementing it as the backbone of evidence-based resource allocation.
When to use it
Use cost-effectiveness analysis when comparing programs that share a single, well-defined outcome and decision-makers want the most effect per unit of cost without monetising the outcome itself. It is the standard tool in health technology assessment and is increasingly applied to education, social and environmental programs with a common effect measure. It is most appropriate when the outcome of interest is one-dimensional and comparable across options. It is unsuitable when programs produce fundamentally different kinds of benefits that cannot be reduced to one metric (where cost-benefit or multi-criteria analysis is needed), when the absolute worth of the outcome must be established rather than relative value, or when the budget and threshold are entirely undefined.
Limitations and considerations
- Can only compare programs that share the same outcome measure; it cannot weigh heterogeneous benefits against one another.
- Reduces value to a single dimension, ignoring other consequences (equity, broader welfare) unless captured separately.
- Requires a willingness-to-pay threshold to reach a decision, and that threshold is often implicit, contested or absent.
- Results are sensitive to the chosen perspective, time horizon, discount rate and costing boundaries, which can be set inconsistently.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Avoids the controversial step of placing a monetary value on outcomes such as lives or health, sidestepping a major objection to cost-benefit analysis.
- Produces an intuitive, comparable metric — cost per unit of effect — that supports prioritisation across competing programs under a fixed budget.
- Well-established methods and reporting standards (e.g., CHEERS) give it transparency and credibility in health and beyond.
- Probabilistic sensitivity analysis provides a rigorous account of decision uncertainty rather than a single deterministic answer.
Applications
- Health technology assessment deciding whether to reimburse drugs, devices and procedures on a cost-per-QALY basis.
- Public-health priority setting across vaccination, screening and prevention programs.
- Comparing education interventions on a cost-per-additional-learning-outcome basis.
- Appraising social programs where a common outcome (e.g., cost per crime averted or per person housed) can be defined.
- International development decisions ranking interventions by cost per death averted or per DALY averted.
How it works
How it works
The analysis specifies the policy options to be compared — always including the relevant comparator, often current practice or no intervention — and a single common measure of effectiveness that captures what the programs are meant to achieve. The choice of outcome is decisive: it must be shared by all options compared and must reflect the policy's true objective, since CEA can only rank options on this one dimension.
For each option, total relevant costs and the level of the effectiveness outcome are measured over an appropriate time horizon, from a stated perspective (e.g., health-system or societal). Future costs and effects are discounted to present value. Defining which costs count — direct program costs, downstream savings, sometimes productivity or carer costs — depends on the perspective and must be applied consistently across options.
Formula
C_k,\ E_k \quad k = 0,1,\dots
Options are compared incrementally. The incremental cost-effectiveness ratio is the difference in cost between an option and its comparator divided by the difference in effect: the additional cost per additional unit of outcome. Options are ranked from least to most effective, dominated options (more costly and less effective) are eliminated, and ICERs are computed along the efficient frontier between the remaining options.
Formula
ICER = \dfrac{C_1 - C_0}{E_1 - E_0}Each ICER is judged against a cost-effectiveness threshold — the maximum a decision-maker is willing to pay per unit of outcome, such as a cost-per-QALY threshold used by a health technology agency. An option whose incremental ratio falls at or below the threshold is considered good value and a candidate for adoption; one above it is judged not worth the additional cost given the budget and the threshold.
Formula
\text{Adopt if } ICER \le \lambda \quad (\lambda = \text{willingness-to-pay})Because costs and effects are uncertain, the analysis tests how conclusions change under alternative assumptions. Deterministic sensitivity analysis varies key parameters one at a time; probabilistic sensitivity analysis assigns distributions to inputs and simulates the ICER, summarising results with cost-effectiveness acceptability curves that show the probability an option is cost-effective at each threshold. This characterises decision uncertainty rather than presenting a single point estimate.
Common pitfalls
- Comparing average cost-effectiveness ratios across options instead of incremental ratios, which can lead to choosing a dominated or inefficient option.
- Failing to discount future costs and effects, or applying inconsistent discount rates, which distorts comparisons of programs with different time profiles.
- Omitting the appropriate comparator or leaving relevant options out of the analysis, so the efficient frontier is mis-specified.
- Reporting a point-estimate ICER without sensitivity or probabilistic analysis, overstating the certainty of the recommendation.
Frequently asked questions
What is the difference between cost-effectiveness analysis and cost-benefit analysis?
Cost-benefit analysis converts all outcomes into money and asks whether monetised benefits exceed costs, producing a net benefit or benefit-cost ratio comparable across very different policies. Cost-effectiveness analysis leaves the outcome in natural units (lives, QALYs, learning gains) and reports cost per unit of that outcome, so it avoids monetising the benefit but can only compare programs sharing the same outcome. CEA is preferred when valuing the outcome in money is contentious; CBA is preferred when benefits are heterogeneous and must be made commensurable.
What is an ICER and how is it interpreted?
The incremental cost-effectiveness ratio is the additional cost of one option divided by its additional effect relative to the next-best comparator — for example, extra dollars per extra QALY. It is interpreted by comparison with a willingness-to-pay threshold: if the ICER is below the threshold, the extra effect is judged worth the extra cost. An option that is both cheaper and more effective 'dominates' and needs no ratio; the ICER matters when an option costs more but also achieves more.
What is a cost-effectiveness acceptability curve?
It is the main output of probabilistic sensitivity analysis. After assigning probability distributions to uncertain cost and effect inputs and simulating many times, the curve plots, for each possible willingness-to-pay threshold, the probability that a given option is the most cost-effective choice. It communicates decision uncertainty directly: instead of a single ICER, it tells decision-makers how confident they can be that an option is good value across the plausible range of thresholds.
Related methods
Sources
Drummond, M. F., Sculpher, M. J., Claxton, K., Stoddart, G. L., & Torrance, G. W. (2015). Methods for the Economic Evaluation of Health Care Programmes (4th ed.). Oxford: Oxford University Press.
ISBN: 9780199665877Copied.Gold, M. R., Siegel, J. E., Russell, L. B., & Weinstein, M. C. (Eds.) (1996). Cost-Effectiveness in Health and Medicine. New York: Oxford University Press.
ISBN: 9780195108248Copied.