Ask Mallard

Economic evaluation

An economic evaluation compares two or more courses of action on their costs and their health consequences together, so that a benefit can be weighed against what it costs to obtain.
NHMRC EVIDENCE LEVEL · 2009Outside the NHMRC hierarchy
Why this level?This hierarchy places a design by the kind of question it answers: intervention, diagnostic accuracy, prognosis, aetiology and screening intervention. This design answers a different kind of question, so no level on this scale describes it. That is a statement about the scale, not about the quality of the evidence such a study can produce. NHMRC 2009 source

An economic evaluation is not one of NHMRC's five research areas. What it can support rests on the model structure, the perspective, the time horizon and the discount rate, none of which a design label captures. CHEERS is the reporting standard.

What question does this help answer?

Not whether an intervention works, which a trial answers, but whether it is worth adopting given a finite budget. The two questions have different answers surprisingly often.

The main types

Cost-effectiveness analysis expresses consequences in a natural unit, such as cost per case prevented. Cost-utility analysis expresses them in quality-adjusted life years, which allows comparison across unrelated conditions and is what most reimbursement bodies expect. Cost-benefit analysis puts everything in money. Budget impact analysis is a separate question again: what adopting this would cost the payer in total, which affordability depends on and cost-effectiveness does not.

The choices that drive the answer

  • The PERSPECTIVE, which decides whose costs count: a health-system perspective excludes patient time and productivity, a societal one includes them.
  • The TIME HORIZON, which must be long enough to capture the important differences in cost and outcome, often lifetime.
  • The DISCOUNT RATE applied to future costs and benefits, stated and varied.
  • The comparator, which must be what is actually done now rather than a straw man.
  • The source of effectiveness estimates, and whether they were transported from a population unlike the one being modelled.

The ICER and what it hides

The incremental cost-effectiveness ratio is the extra cost divided by the extra benefit. As a ratio it behaves badly: a small denominator makes it explode, and a negative numerator and a negative denominator produce the same ratio as two positives while meaning something quite different. Reporting incremental cost and incremental effect separately, alongside the ratio, avoids most of that.

The cost-effectiveness plane

The cost-effectiveness plane, with difference in effect across and difference in cost up, divided into four quadrants. A dashed threshold line runs through the origin, and a cloud of uncertain estimates sits in the more effective, more costly quadrant, straddling the threshold.The cost-effectiveness plane: where the new option sits, with its uncertaintyWillingness-to-pay thresholdMore effective, more costlyMore effective, less costlyLess effective, less costlyLess effective, more costlyDifference in costDifference in effectA single ratio hides where the uncertainty falls: here the estimates straddle the threshold. Perspective,time horizon and discount rate can move the whole cloud.
Conceptual diagram. It shows the structure of the idea, not data from any study.

Uncertainty is a distribution

Every input is estimated, so the result is a cloud rather than a point. Probabilistic sensitivity analysis propagates the input distributions through the model and reports the result on the cost-effectiveness plane, usually with an acceptability curve showing the probability that each option is cost-effective at each willingness-to-pay threshold. One-way sensitivity analyses on the structural choices, especially perspective, horizon and discount rate, belong beside it.

Why Mallard may recommend it

Mallard may point here when a description asks whether an intervention is worth its cost, or mentions cost per QALY, budget impact or value for money.

How it is sized

Mallard does not size an economic evaluation. Where the evaluation is carried alongside a trial, the trial's sample size is usually driven by the clinical outcome and is rarely adequate for the cost outcome, because costs are strongly skewed by a few expensive patients. Say so explicitly rather than implying the economic result inherits the trial's power.

What to watch out for

The result depends heavily on the perspective, the time horizon and the discount rate, none of which are statistical choices, and changing any of them can reverse the conclusion. Uncertainty must be presented as a distribution rather than a point estimate of the incremental cost-effectiveness ratio.

Clinical example

Question: Is a new anticoagulant worth its price compared with the current standard? Design: Model lifetime costs and quality-adjusted life years from a health-system perspective at a stated discount rate, run a probabilistic sensitivity analysis, and present the acceptability curve rather than a single ICER.

Recent studies that used this
Searching PubMed…
Related Learn pages
Sources & further reading
  1. Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) Statement: Updated Reporting Guidance for Health Economic Evaluations (opens in a new tab) · Value Health · doi 10.1016/j.jval.2021.11.1351
  2. Recommendations for Conduct, Methodological Practices, and Reporting of Cost-effectiveness Analyses: Second Panel on Cost-Effectiveness in Health and Medicine (opens in a new tab) · JAMA · doi 10.1001/jama.2016.12195
CHEERS 2022 is a reporting guideline and governs disclosure, not method. The Second Panel report is the methodological reference behind the reference-case recommendations, including the two required perspectives.
Last source review: 2026-09-06
AI-generated educational content
Mallard uses reputable methodological and reporting sources, but this material may contain errors or omit important context. Verify important methodological decisions with the cited sources and, when appropriate, a qualified research methodologist or statistician.