Decision Rule

Decision Rule in Cost-Effectiveness Analysis

In health economic evaluation and health technology assessment, a decision rule is the stated procedure that translates comparative cost and outcome evidence into a conditional recommendation, using dominance, the incremental cost-effectiveness ratio and net benefit. Decision rules make the analytical logic behind a recommendation explicit. The term here does not refer to the rejection rules of statistical hypothesis testing. Uncertainty, budget impact and the wider health technology assessment process all bear on how the result of the rule feeds into the final decision.

A decision rule does not remove judgement or guarantee adoption. The rule must match the decision context, use the relevant threshold or constraint, and distinguish a cost-effectiveness conclusion from a complete reimbursement or coverage decision.

What a decision rule does

A decision rule states how evidence will be interpreted before the result is known. In economic evaluation, the rule commonly compares additional costs with additional health outcomes and applies a stated cost-effectiveness threshold.

The output is a conditional recommendation, not an automatic command. Decision-makers may also need to consider uncertainty, affordability, equity, severity, implementation and other responsibilities within their remit.

• A decision rule identifies the alternatives being compared and the outcome that determines preference.
• A cost-effectiveness decision rule states the threshold or other value used to judge whether additional health is worth the additional cost.
• A decision rule should specify how dominance, uncertainty and ties are handled.
• A decision rule should distinguish the economic-evaluation result from the final institutional recommendation.

The decision question behind the rule

The rule can only be interpreted correctly when the population, alternatives, perspective, outcomes and time horizon are clear. These choices determine which costs and health effects enter the comparison and whose opportunity costs matter.

The comparator is especially important because cost-effectiveness is incremental. A technology may appear attractive against one option but not against the next-best relevant alternative.

• The decision population identifies the people to whom the recommendation would apply.
• The alternatives identify the healthcare options that must be compared.
• The analytical perspective determines which costs and consequences enter the economic evaluation.
• The time horizon should capture all material differences between the alternatives.
• The outcome measure and cost-effectiveness threshold establish how benefits and opportunity costs are valued.

How the rule works for two options

For two options, the decision rule can be expressed with incremental net monetary benefit or incremental net health benefit. Both forms compare the gain in health with the opportunity cost represented by the additional expenditure.

Let ΔE be the incremental health effect, ΔC the incremental cost and λ the cost-effectiveness threshold, with both differences taken as the evaluated option minus the comparator. A positive result favours the evaluated option on cost-effectiveness grounds, zero represents a tie at the stated threshold, and a negative result favours the comparator.

INMB = (λ × ΔE) − ΔC

INHB = ΔE − (ΔC ÷ λ)

Incremental net health benefit equals incremental net monetary benefit divided by λ, so the two forms always give the same decision at any positive threshold. When ΔE > 0, the net-benefit rule is the same as the familiar ICER rule, under which the evaluated option is preferred when ΔC ÷ ΔE does not exceed λ.

• A positive incremental net monetary benefit means the valued incremental health gain exceeds the incremental cost at the stated threshold.
• A zero incremental net monetary benefit means the alternatives are tied under the stated cost-effectiveness rule; the convention of selecting the most effective option whose ICER does not exceed λ settles such a tie in favour of the more effective option.
• A negative incremental net monetary benefit means the evaluated option is not preferred on cost-effectiveness grounds at the stated threshold.
• The threshold must be reported with its currency, price year, health unit, jurisdiction and decision context.

Why an ICER cannot be interpreted by its sign alone

The incremental cost-effectiveness ratio divides incremental cost by incremental effect, but the same sign can arise in different quadrants of the cost-effectiveness plane. A negative ICER may describe either a dominant option or a dominated option, so the directions of cost and effect must be checked before the ratio is interpreted.

Dominance provides the decision before a threshold comparison is needed. Each of the four possible combinations of cost and effect leads to a different conclusion.

• When the evaluated option is more effective and less costly, it dominates the comparator. The same holds when it is equally costly and more effective, or equally effective and less costly.
• When the evaluated option is more effective and more costly, it is preferred when its ICER does not exceed λ, which is the same condition as a non-negative incremental net monetary benefit because ΔE > 0.
• When the evaluated option is less effective and less costly, the ICER measures the saving per unit of health forgone, so the option is preferred only when that ratio exceeds λ. Net benefit gives the same answer without reversing the inequality.
• When the evaluated option is less effective and more costly, it is dominated by the comparator.

How to compare several alternatives

Several alternatives should be compared together rather than through an arbitrary collection of pairwise ratios. Fully incremental analysis removes inefficient options and identifies the cost-effectiveness frontier before the remaining alternatives are judged against the threshold.

The order of analysis matters because each non-dominated option is compared with the next-less-costly option on the frontier. Expected net benefit offers an equivalent way to identify the preferred option when the same threshold, perspective and evidence apply to all alternatives, because the option with the highest expected net monetary benefit is the one the frontier rule selects.

1. Order the alternatives from lowest to highest expected cost.
2. Remove any alternative that is dominated because another option costs no more and produces no less health, with at least one strict improvement.
3. Calculate sequential incremental cost-effectiveness ratios for the remaining alternatives.
4. Remove any alternative subject to extended dominance, shown by an ICER higher than that of the next more effective alternative, because a combination of other options produces health more efficiently.
5. Recalculate the frontier after each removal.
6. Select the most effective alternative on the frontier whose ICER does not exceed the stated threshold, which is also the alternative with the highest expected net monetary benefit.

A worked two-option example

The figures in this example are illustrative. A new programme produces 0.4 additional quality-adjusted life-years (QALYs) per person and costs an additional £8,000 compared with current care. At an illustrative threshold of £30,000 per QALY, the incremental net monetary benefit is £4,000.

The positive incremental net monetary benefit means the programme is cost-effective under the stated assumptions and threshold. It does not by itself establish that the programme is affordable or that an HTA body should recommend unrestricted adoption.

INMB = (£30,000 × 0.4) − £8,000 = £4,000

The ICER rule reaches the same conclusion. The ICER is £8,000 ÷ 0.4 = £20,000 per QALY, which is below £30,000 with ΔE > 0, and the incremental net health benefit is 0.4 − (£8,000 ÷ £30,000), or about 0.13 QALYs per person, which equals £4,000 ÷ £30,000. The incremental net monetary benefit stays non-negative at any threshold of £20,000 per QALY or more.

• The worked example uses illustrative values and does not represent a real technology or institutional decision.
• The worked example's conclusion changes if the incremental cost, incremental effect or threshold changes enough to make incremental net monetary benefit zero or negative.
• The worked example requires separate budget-impact evidence to assess the total financial consequences of adoption.

How uncertainty changes the interpretation

Expected net benefit is the appropriate basis for choosing the option that maximises expected value under the stated model and threshold. On this decision-theoretic view, the adoption decision rests on expected net benefit whether or not the difference is statistically significant, and uncertainty bears on whether more evidence should be acquired. The probability that an option is cost-effective answers a different question, namely how often that option has the highest net benefit across uncertainty draws.

An option can have the highest expected net benefit without having the highest probability of being cost-effective, for example when the net-benefit distribution is skewed. Decision uncertainty may justify further research when additional evidence could change the preferred choice and the value of resolving uncertainty exceeds the cost of research.

• Expected net benefit identifies the option with the greatest average net benefit across the modelled uncertainty.
• The probability of cost-effectiveness describes decision uncertainty and is not itself the decision rule.
• A cost-effectiveness acceptability curve should not replace expected net benefit when selecting the preferred alternative.
• Value-of-information analysis can assess whether reducing uncertainty may be worthwhile.

How the decision rule relates to budget impact and HTA

Cost-effectiveness analysis and budget impact analysis answer related but different questions. The cost-effectiveness decision rule asks whether the additional health is worth the opportunity cost, whereas budget impact analysis estimates the change in spending for a defined budget holder over a stated period.

Health technology assessment may use both results alongside clinical evidence, equity, severity, feasibility and other considerations. A positive incremental net benefit can therefore support a recommendation without guaranteeing adoption, unrestricted access or immediate implementation.

The NICE manual for technology appraisal (PMG36, section 6.3, last updated 13 October 2026) shows how an institution can apply a threshold as a range rather than a single cut-off. NICE's committee does not use a precise maximum acceptable ICER, and below a most plausible ICER of £25,000 per QALY gained its decision is normally based on the cost-effectiveness estimate. As the ICER rises from £25,000 to £35,000 per QALY gained, the committee makes explicit reference to the factors listed in section 6.3.5, such as the degree of certainty around the ICER and uncaptured benefits, and above £35,000 it needs an increasingly stronger case to recommend the technology.

• Economic evaluation informs the decision rule by providing comparative estimates of costs and consequences.
• Budget impact analysis informs the wider decision by estimating affordability for a defined budget holder.
• Health technology assessment uses the decision rule within a broader appraisal process rather than treating it as the complete decision.
• Equity, severity or implementation considerations should be stated explicitly when they modify the final recommendation.

Common decision-rule errors and safeguards

Decision-rule errors often come from interpreting a ratio without its quadrant, comparing the wrong alternatives or treating one institution's threshold as universal. These mistakes can reverse the apparent conclusion or overstate what the economic evidence establishes.

The safeguard is to state the decision context and apply the analytical steps in a reproducible order. When judgement changes the recommendation, the reason should be visible rather than hidden inside the calculation.

• Interpreting every negative ICER as favourable can confuse dominance with being dominated.
• Comparing every option only with a common baseline can miss strict or extended dominance in a multi-option analysis.
• Selecting the option with the highest probability of cost-effectiveness can differ from selecting the option with the highest expected net benefit.
• Changing the threshold after seeing the result can introduce outcome-driven reasoning unless the change is a declared scenario analysis.
• Calling an option cost-effective without naming the comparator, perspective, threshold and uncertainty can make the statement impossible to interpret.
• Treating cost-effectiveness as affordability can overlook the total spending consequences measured by budget impact analysis.

What a decision-rule analysis should report

A transparent report should let a reader reconstruct the decision rule and understand the limits of its conclusion. The report should separate model outputs from the institutional or deliberative judgement that follows them.

The following information makes the result auditable. Jurisdiction-specific values should include the guidance version and date that governed the analysis.

• The population, alternatives, perspective, time horizon and outcome measure used in the economic evaluation.
• Incremental costs and incremental effects for every relevant comparison.
• Dominance and extended-dominance checks for analyses with several alternatives.
• The threshold, with its currency, price year, health unit, jurisdiction and source.
• Incremental net monetary benefit or incremental net health benefit, with an explanation of its sign.
• Expected net benefit and decision uncertainty for all relevant alternatives.
• Budget impact, reported separately from the cost-effectiveness conclusion.
• Any equity, severity, implementation or institutional judgement that changes the recommendation.
• Scenario analyses, reported without replacing the prespecified base-case decision rule.

Sources

• Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation. Oxford: Oxford University Press; 2006.
• Claxton K. The irrelevance of inference: a decision-making approach to the stochastic evaluation of health care technologies. Journal of Health Economics. 1999;18(3):341-364.
• Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. 4th ed. Oxford: Oxford University Press; 2015.
• Fenwick E, Claxton K, Sculpher M. Representing uncertainty: the role of cost-effectiveness acceptability curves. Health Economics. 2001;10(8):779-787.
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). London: NICE; published 31 January 2022, last updated 13 October 2026. Section 6.3.
• Stinnett AA, Mullahy J. Net health benefits: a new framework for the analysis of uncertainty in cost-effectiveness analysis. Medical Decision Making. 1998;18(2 Suppl):S68-S80.
• Sullivan SD, Mauskopf JA, Augustovski F, Jaime Caro J, Lee KM, Minchin M, et al. Budget impact analysis: principles of good practice. Report of the ISPOR 2012 Budget Impact Analysis Good Practice II Task Force. Value in Health. 2014;17(1):5-14.