Cost-Effectiveness

Cost-Effectiveness Judged Against a Threshold

Every pound a health system spends on one treatment is a pound it cannot spend on other care, so the question behind cost-effectiveness is whether the health bought by an option outweighs the health that the same money would have produced elsewhere. That question only has an answer within a set of alternatives, which is why a cheap intervention can fail the test and an expensive one can pass it. The judgement runs from incremental costs and effects, through dominance and the cost-effectiveness plane, to the incremental cost-effectiveness ratio (ICER) set against a cost-effectiveness threshold and the equivalent net benefit rule, and it rests on the opportunity-cost basis of the threshold, which NICE applies under PMG36 in technology appraisal. The method that produces the numbers is cost-effectiveness analysis, and ICER here means the ratio, not the US Institute for Clinical and Economic Review, which shares the abbreviation.

A judgement made within a choice set

Cost-effectiveness belongs to an option within a particular choice set rather than to the option on its own. The standard textbook framing of economic evaluation is a comparison of alternative courses of action in terms of both their costs and their consequences, so the verdict depends on the comparators, the patients, the time horizon and the perspective. The same medicine can be cost-effective against best supportive care and not against an established generic, or in a high-risk group and not in a low-risk one, and a change in price can move it across the line.

Under a fixed budget, funding only cost-effective options is a route to allocative efficiency, in which the budget is spread so that no reallocation would produce more health. The criterion counts a QALY the same whoever gains it, so it says nothing about how health is distributed; distributional cost-effectiveness analysis addresses that question separately.

Several ideas are often mistaken for cost-effectiveness. A low price says nothing about value until the health gained is set against it, and a cost saving that loses health has to be large enough to justify the loss. Affordability, the total spending a decision commits a budget to, is a separate question for budget impact analysis. An average cost-effectiveness ratio for a single option also misses the point, because the decision concerns the extra cost of moving from one option to another; the cost per QALY page sets out that contrast in more detail.

Incremental costs and effects decide the question

The comparison rests on differences in expected cost and expected health effect between options. For an option 1 compared with a comparator 0, these two differences are the starting point for every later step. They are written as follows.

$$\Delta C = C_1 - C_0 \qquad \Delta E = E_1 - E_0$$

where $\Delta C$ is the incremental cost, $C_1$ and $C_0$ are the expected costs of option 1 and the comparator, $\Delta E$ is the incremental effect, and $E_1$ and $E_0$ are their expected health effects, usually in quality-adjusted life years (QALYs).

The signs of $\Delta C$ and $\Delta E$ place the comparison in one of four quadrants of the plane. An option that costs less and produces more health dominates its comparator, and one that costs more and produces less is dominated, as the dominance page explains; neither case needs health to be valued. The other two quadrants involve a trade-off, more health for more money or less health for a saving, and only these need a threshold. With more than two options, those that survive strict dominance and extended dominance form the cost-effectiveness frontier, and each is compared with the next less effective option on it in a fully incremental analysis. NICE's PMG36 manual asks for results in this form, with dominated and extendedly dominated technologies removed, and requires each ICER to be the ratio of expected additional total cost to expected additional QALYs.

The ICER and the threshold rule

The ICER expresses a trade-off as the price paid for each extra unit of health. Comparing that price with the value placed on a unit of health turns the ratio into a decision. In the north-east quadrant the rule is as follows.

$$ICER = \frac{\Delta C}{\Delta E} \le \lambda \quad \text{when} \quad \Delta E > 0$$

where $ICER$ is the incremental cost per unit of health gained, $\lambda$ is the threshold, the amount regarded as worth paying for one unit of health (for example, pounds per QALY), and $\Delta C$ and $\Delta E$ are as defined above.

In the south-west quadrant the rule reverses. The underlying condition is that the health gained, valued at $\lambda$, exceeds the extra cost, so $\lambda \Delta E - \Delta C > 0$, which gives $\Delta C < \lambda \Delta E$. Dividing by a negative $\Delta E$ reverses the inequality, so a cheaper, less effective option is acceptable only when $\Delta C / \Delta E > \lambda$, meaning the saving per unit of health given up exceeds the threshold. A ratio of £10,000 per QALY is a good purchase in the north-east quadrant and a poor bargain in the south-west, so an ICER should always be reported with the signs of its components. NICE's PMG36 manual states that for technologies giving less health benefit at lower cost, the usual levels of £25,000 to £35,000 per QALY should be considered.

Net benefit as the same rule without a ratio

Ratios are awkward because their sign is ambiguous, they are undefined when the effect difference is zero, and their uncertainty is hard to summarise. Stinnett and Mullahy proposed net health benefits as an alternative framework for analysing uncertainty in cost-effectiveness analysis, avoiding some problems of ratio-based methods, and the monetary version follows the same logic. For each option the net benefits are defined as follows.

$$NMB_i = \lambda E_i - C_i \qquad NHB_i = E_i - \frac{C_i}{\lambda}$$

where $NMB_i$ is the net monetary benefit of option $i$, $NHB_i$ is its net health benefit, $E_i$ and $C_i$ are its expected health effect and cost, and $\lambda$ is the threshold. Between two options the difference is as follows.

$$\Delta NMB = \lambda \Delta E - \Delta C = \lambda \, \Delta NHB$$

where $\Delta NMB$ and $\Delta NHB$ are the incremental net monetary and net health benefit of option 1 over the comparator. Option 1 is cost-effective when $\Delta NMB$ is positive, which for a positive $\Delta E$ matches an ICER below $\lambda$ and for a negative $\Delta E$ matches an ICER above $\lambda$. The rule works in every quadrant without special cases, and with many options the one with the highest net benefit is the one a fully incremental analysis selects. NICE's manual asks for expected net health benefits, at values of £25,000 and £35,000 per QALY gained, alongside ICERs, and allows net monetary benefits as well; the net benefit page covers the measure in depth.

Why the threshold stands for health forgone elsewhere

Under a fixed budget, money spent on a new intervention is no longer available for other care, so the health it buys has to be set against the health it displaces. This is the opportunity cost of the decision, and on this view the threshold is the cost at which the system currently produces an extra QALY at the margin. An option with an ICER above that level displaces more health than it adds, so population health falls even though its own patients gain.

NICE's PMG36 manual takes this position. Section 6.3.1 states that, given the fixed NHS budget, the appropriate maximum acceptable ICER is that of the opportunity cost of programmes displaced by new, more costly technologies, that NICE lacks the information to define it precisely and so uses a range, and that cost-effectiveness is necessary but not the only basis for decisions. An empirical estimate for the English NHS by Claxton and colleagues put the central threshold at £12,936 per QALY (2008 expenditure, 2008 to 2010 mortality), with a probability of 0.89 that it lies below £20,000 per QALY; the threshold page discusses this work and the demand-side alternative based on willingness to pay.

Worked example with four options, one dominated and one extendedly dominated

The figures below are illustrative. Usual care (U) and three new options (A, B and C) are mutually exclusive, with expected costs and QALYs per patient, already discounted.

1. Remove dominated options. C costs £1,600 more than B and gives 0.10 fewer QALYs, so B dominates C and C is removed. Against usual care alone, C would have shown an ICER of £26,000 per QALY (£15,600 for 0.60 QALYs), inside the NICE range, which shows how a comparison with usual care only can mislead.

2. Calculate sequential ICERs. Moving from U to A gives the following.

$$ICER_{A,U} = \frac{13500 - 10000}{5.10 - 5.00} = \frac{3500}{0.10} = 35000$$

where $ICER_{A,U}$ is the cost in pounds per QALY gained by moving from U to A. Moving from A to B gives the following.

$$ICER_{B,A} = \frac{24000 - 13500}{5.70 - 5.10} = \frac{10500}{0.60} = 17500$$

where $ICER_{B,A}$ is the cost per QALY gained by moving from A to B.

3. Check for extended dominance. The ratios fall from £35,000 to £17,500 per QALY as effect rises, so A is extendedly dominated. Assuming the options can be divided among patients at constant per-patient cost and effect, giving B to one patient in seven and usual care to the other six yields 5.00 + 0.70/7 = 5.10 QALYs on average, the same as A, at an average cost of £10,000 + £14,000/7 = £12,000, which is £1,500 less than A. A is removed.

4. Recalculate and decide. The frontier is now U and B, giving the following.

$$ICER_{B,U} = \frac{24000 - 10000}{5.70 - 5.00} = \frac{14000}{0.70} = 20000$$

where $ICER_{B,U}$ is the cost per QALY gained by moving from U to B. At an illustrative threshold of £25,000 per QALY, B is cost-effective; at any threshold below £20,000, usual care would be preferred.

5. Confirm with net benefit. At £25,000 per QALY the net monetary benefits are £115,000 for U, £114,000 for A, £118,500 for B and £114,400 for C. At £35,000 they are £165,000, £165,000, £175,500 and £170,400; A only draws level with usual care there because £35,000 is exactly its ICER against usual care. B is highest at both thresholds. In health units at £25,000 per QALY the comparison is as follows.

$$\Delta NHB_{B,U} = 0.70 - \frac{14000}{25000} = 0.70 - 0.56 = 0.14$$

where $\Delta NHB_{B,U}$ is the net health benefit of B over usual care in QALYs, 0.70 is the QALY gain and 0.56 is the health forgone elsewhere by spending an extra £14,000. B, the most costly option still standing, is the cost-effective choice. A, though cheaper, could never be chosen outright, because it would need a threshold above £35,000 to beat usual care and below £17,500 to beat B.

How NICE and other bodies apply the idea

Health technology assessment bodies turn the concept into procedure. NICE's PMG36 manual for technology appraisal and highly specialised technologies guidance uses cost-utility analysis to determine whether differences in expected costs can be justified by changes in expected health effects in QALYs, and justifies this by NICE's focus on maximising health gains from a fixed NHS and personal social services budget.

Since the update of 31 March 2026 the manual's decision rules use a range of £25,000 to £35,000 per QALY gained, replacing the £20,000 to £30,000 range described in NICE's news article of 1 December 2025; for highly specialised technologies, a single level of £100,000 per QALY gained takes the place of both figures. The cost-effectiveness threshold page covers how the committee applies the range and the decision-making modifiers. Other bodies, such as Australia's Pharmaceutical Benefits Advisory Committee in its guidelines (version 5.0), likewise ask for the incremental cost, incremental effectiveness and ICER.

Assumptions and choices that change the verdict

A cost-effectiveness verdict carries every assumption of the analysis behind it. The comparator set is the most powerful choice, as option C shows. The health measure matters too, and NICE's manual notes that the QALY rests on assumptions such as constant proportional trade-off and additive independence between health states. Time horizon, discounting, the costs counted under the chosen perspective and the threshold itself all shift the result.

The decision uses expected values, and those expectations are uncertain. NICE's manual prefers a cost-effectiveness estimate derived from a probabilistic analysis where possible, and asks for the probability that a technology is cost-effective at £25,000 to £35,000 per QALY together with the error probability that it is not. These usually come from a probabilistic sensitivity analysis, and the manual lists the cost-effectiveness acceptability curve among appropriate ways of presenting that uncertainty. A high ICER with little uncertainty and a low ICER with wide uncertainty carry different risks of a wrong decision.

Common misreadings of "cost-effective"

Most errors come from reading the label without the comparison behind it. Treating cost-effective as cheap overlooks options, such as B, that cost more and are worth it. Reporting an ICER against usual care alone, when other active options exist, can make a dominated option look acceptable, and a negative ICER can come either from a dominant option or from a dominated one.

A threshold is also not an automatic approval line. NICE's manual states that the committee does not use a precise maximum acceptable ICER above which a technology is automatically not cost-effective, and other factors carry more weight as the ICER nears the top of the range.

Sources

• 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.
• 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. Sections 4.2.14 to 4.2.16, 4.7.12, 4.7.15, 4.10.8 and 6.3.1 to 6.3.9.
• National Institute for Health and Care Excellence. Changes to NICE's cost-effectiveness thresholds confirmed. News article, 1 December 2025.
• Claxton K, Martin S, Soares M, Rice N, Spackman E, Hinde S, Devlin N, Smith PC, Sculpher M. Methods for the estimation of the National Institute for Health and Care Excellence cost-effectiveness threshold. Health Technology Assessment. 2015;19(14):1-503, v-vi. doi:10.3310/hta19140
• 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. doi:10.1177/0272989X98018002S09
• Pharmaceutical Benefits Advisory Committee. Guidelines for preparing a submission to the Pharmaceutical Benefits Advisory Committee, version 5.0. Canberra: Australian Government; 2016. Section 3A.8.
• Institute for Clinical and Economic Review. 2023 Value Assessment Framework. Updated 25 September 2023.