Extended dominance condition for three ordered options

Subtracts the sequential ICER of the next more effective option from the sequential ICER of the middle of three consecutive options, ordered by expected effect. The middle option is extendedly dominated when the difference is greater than zero.

Signature

D_M = (C_M - C_L) / (E_M - E_L) - (C_H - C_M) / (E_H - E_M)
Inputs
InputsDefinitionUnit
C_MExpected cost of the intermediate option being testedcurrency per defined population or person
C_LExpected cost of the next less effective option still in the comparisoncurrency per defined population or person
E_MExpected health effect of the intermediate option being testedhealth-outcome unit per defined population or person
E_LExpected health effect of the next less effective optionhealth-outcome unit per defined population or person
C_HExpected cost of the next more effective option still in the comparisoncurrency per defined population or person
E_HExpected health effect of the next more effective optionhealth-outcome unit per defined population or person
Output
D_MSequential ICER of M against L minus the sequential ICER of H against M; M is extendedly dominated when D_M is greater than 0currency per unit of health effect

Function

Extended dominance function

Maps the expected costs and effects of options that survive the dominance screen, ordered by effect, to a flag for each intermediate option that a blend of its neighbours would beat.

Try this function

Implementations

  • Excel

    Flag an extendedly dominated middle option

    Excel calculates D_M, the sequential ICER into the middle option minus the sequential ICER out of it, using named cells for the three adjacent options, and flags the middle option when D_M is greater than 0.

    =IF((MiddleCost-LowerCost)/(MiddleEffect-LowerEffect)-(HigherCost-MiddleCost)/(HigherEffect-MiddleEffect)>0,"Extendedly dominated","Retain")

Assumptions

  • Dominated options removed first

    The three options remain after dominated options, including ties, have been removed, so expected cost and expected effect both rise from L to M to H and both denominators are positive.

  • Adjacent options, recalculated after each removal

    L, M and H are consecutive in the current ordering by expected effect. After any removal, adjacency changes, the sequential ICERs are recalculated and the check is repeated until the ICERs rise with effect.

  • Equal sequential ratios

    When D_M equals 0, the two sequential ICERs are equal and M lies on the straight line joining L and H, so it is not treated as extendedly dominated.

Worked examples

  • Middle option extendedly dominated

    Three options cost £13,000, £18,000 and £22,000 and produce 4.3, 4.5 and 4.8 QALYs. The sequential ICER into the middle option is £25,000 per QALY and the next is about £13,333 per QALY, so D_M is about £11,667 per QALY. Because D_M is greater than 0, the middle option is extendedly dominated. The figures are illustrative and match options B, C and D in the article's four-option example.

    C_L = 13000; E_L = 4.3; C_M = 18000; E_M = 4.5; C_H = 22000; E_H = 4.8; D_M = 11666.67
  • Frontier after recalculation

    With option C removed, A, B and D are consecutive. A costs £10,000 for 4.0 QALYs, B £13,000 for 4.3 and D £22,000 for 4.8. The ICER from A to B is £10,000 per QALY and from B to D is £18,000, so D_M is minus £8,000 per QALY. Because D_M is not greater than 0, B is retained and A, B and D form the frontier.

    C_L = 10000; E_L = 4.0; C_M = 13000; E_M = 4.3; C_H = 22000; E_H = 4.8; D_M = -8000

Common errors

  • Not recalculating after a removal

    Keeping ICERs computed before a removal leaves comparisons with an option that is no longer adjacent. In the example, the ICER for D is recalculated against B, £18,000 per QALY, rather than kept at about £13,333 against C.

  • Testing before dominated options are removed

    Sequential ICERs calculated while a dominated option remains can be negative or misleading, so the extended dominance check gives unreliable results.

  • Relying on a common baseline

    An option can have an acceptable ICER against a common baseline and still be extendedly dominated, because the baseline comparison ignores the next option on the frontier.

Sources

  • Extended dominance and mixed strategies

    Cantor SB. Cost-effectiveness analysis, extended dominance, and ethics: a quantitative assessment. Medical Decision Making. 1994;14(3):259-265.

    View source →

  • NICE manual on extended dominance

    National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). Published 31 January 2022, last updated 31 March 2026. Chapter 4 Economic evaluation, section 4.10.8 (extendedly dominated technologies, where a combination of 2 or more other technologies would be more cost effective, removed from the fully incremental analysis).

    View source →

  • Extended dominance in practice

    Postma MJ, de Vries R, Welte R, Edmunds WJ. Health economic methodology illustrated with recent work on Chlamydia screening: the concept of extended dominance. Sexually Transmitted Infections. 2008;84(2):152-154.

    View source →

Canonical Identity

Stable URI · Machine-readable · Resolvable · CC BY 4.0