ICECAP-A

ICECAP-A in Economic Evaluation of Health and Social Care

ICECAP-A asks adults what they are able to do and be in their lives, and the measure treats health as only one of the influences on that capability. For a health economist the measure matters when an intervention changes more than health-related quality of life, as with social care, public health programmes, addiction treatment or support for carers, where a health-only instrument such as the EQ-5D can miss much of the benefit. The UK index values were estimated with best-worst scaling, so an ICECAP-A score sits on a capability scale anchored at full capability and is not a utility anchored at dead. An ICECAP-A score therefore cannot be dropped into a QALY calculation without further assumptions.

How the ICECAP-A was developed from interviews with adults

Al-Janabi, Flynn and Coast built the measure from qualitative work with the general adult population, with the stated aim of producing an outcome measure for economic evaluation. They ran in-depth, participant-led interviews with 36 adults in the UK to identify the attributes of capability wellbeing, and then 18 repeat semi-structured interviews to turn those attributes into a descriptive system. Informants were selected purposively to vary in socio-economic status, age, sex, ethnicity and health.

The developers describe the result as a departure from traditional health economics outcome measures, because health status enters as one influence on a broader capability wellbeing and is not the thing being measured. The work belongs to a programme of capability measures from the same group, which includes the earlier ICECAP-O for older people. The name is expanded in two ways in the literature. The UK valuation paper calls it the Investigating Choice Experiments Capability Measure for Adults, while later work uses the ICEpop Capability Measure for Adults.

Five attributes with four levels each

The descriptive system has five attributes, named stability, attachment, achievement, autonomy and enjoyment. Each attribute is a single item with four response options, so the questionnaire is short enough to add to a trial alongside a health measure.

Four levels on each of five attributes give the number of different capability states the instrument can describe.

$$4 \times 4 \times 4 \times 4 \times 4 = 1024$$

where each 4 is the number of response levels on one attribute and 1024 is the number of distinct ICECAP-A states.

A state is written as a five-digit code, one digit per attribute. The developers' scoring guide codes full capability on an attribute as level 4, which is the reverse of the EQ-5D convention, where level 1 means no problems. State 44444 therefore describes full capability on every attribute, and a code such as 43211 mixes full capability on one attribute with the lowest level on two others.

UK index values estimated with best-worst scaling

Flynn and colleagues valued the ICECAP-A in a best-worst scaling study of the profile case, in which respondents see one complete state at a time and pick the best and the worst attribute level within it. A main effects plan plus its foldover was used to estimate a weight for each of the four levels of all five attributes, and the survey was administered to 413 randomly sampled people in the UK. Scale-adjusted latent class analysis identified two preference classes and two scale classes, which separates differences in what people value from differences in how consistently they choose.

The resulting UK index values give attachment and stability around 22 per cent of the value space each, and autonomy, achievement and enjoyment around 18 per cent each. The index value for a state is the sum of the values of its five levels.

$$I = \sum_{k=1}^{5} v_k(l_k)$$

where $I$ is the ICECAP-A index value for a state, $k$ indexes the five attributes, $l_k$ is the level reported on attribute $k$ and $v_k(l_k)$ is the UK value attached to that level of that attribute.

In the developers' scoring guide the level 4 values of the five attributes sum to 1.000, so full capability scores 1, and the worked example state 43211 scores 0.513. The developers' frequently asked questions describe the index values as anchored on a full capability and no capability scale, with 1 for full capability and 0 for no capability. A main effects design estimates no interactions between attributes, so the index assumes that the value of a level on one attribute does not depend on the levels of the others.

Why an ICECAP-A score is not a QALY weight

Best-worst scaling yields values that are relative to one another. Flynn and colleagues note that one proposed route to placing such values on a scale where death is zero is to rescale them with time trade-off tasks. The ICECAP-A index is instead read on a capability scale, where 1 is full capability and 0 is no capability, and full capability is not the same as full health.

The developers call the product of the index and time Years of Full Capability (YFC) and state that a YFC is not equivalent to a QALY, because QALY measures are anchored on a perfect health and dead scale. Death enters through its measured effect on capability, so a person who dies has zero years of full capability after death. A YFC result cannot enter a cost-utility analysis or be judged against a cost per QALY threshold. In a comparative review of the literature on capability instruments, which included 10 economic evaluations, Helter and colleagues found no consensus on the most appropriate way to use them, with discussion of capability-adjusted life years, years of capability equivalence and the choice between maximising capability and securing sufficient capability.

Maximising capability or securing sufficient capability

The capability approach gives two broad decision rules. A maximising rule adds up capability gains across people and over time, in the same way that QALYs add up health. A sufficiency rule asks how many people reach a normatively sufficient level of capability wellbeing and for how long, and gives no credit for gains above that level.

Mitchell, Roberts, Barton and Coast drew on the multidimensional poverty and health economics literatures to set out methods for economic evaluation with an objective of sufficient capability, and developed a Years of Sufficient Capability (YSC) measure of capability wellbeing over time, which the developers describe as counting changes in capability relative to a threshold deemed sufficient. Their illustration, in patients with osteoarthritis undergoing joint replacement, used ICECAP-O and not ICECAP-A. Goranitis and colleagues then used ICECAP-A in a methodological case study of two psychological interventions for drug addiction compared with usual care, based on pilot trial data. They examined the effect of widening the evaluative space beyond health to capability wellbeing and of replacing health maximisation with the egalitarian aim of sufficient capability for everyone, and measured both years of full capability and years of sufficient capability with ICECAP-A.

The choice of rule is a value judgement about equity. A sufficiency rule concentrates attention on people below the threshold, so the same ICECAP-A data can rank interventions differently under the two rules. The developers report that the sufficient level of capability on ICECAP-A, and how much the UK general public would pay for a YFC and a YSC, have been estimated in UK studies by Kinghorn and colleagues. NICE has not adopted a threshold for either.

NICE guidance and the use of capability measures in decisions

NICE's guideline manual, Developing NICE guidelines (PMG20, last updated 23 October 2025), sets out a reference case that depends on the decision problem. In table 7.1, the column for interventions with a social care focus names capability or social care-related quality of life measures, such as ICECAP or ASCOT, for interventions that result in both health and capability or social care outcomes. The same chapter says NICE will consider using social care QALYs if they are validated, and counts benefits for people using services and their carers when the focus is social care. The manual now notes that an interim methods statement on the EQ-5D-5L takes precedence over it and that section 7.4 (which holds table 7.1), section 7.7 and section 7.9 will be updated, so this table is under revision.

Use in practice has been limited. Zhang and colleagues reviewed NICE social care and public health guidelines published before 26 August 2025 and interviewed five experts. ICECAP appeared in 1 per cent of eligible guidelines and ASCOT in 4 per cent, and neither measure significantly affected committee decision making. The experts pointed to underfunding, structural differences between social care and health, scarce training and resources, the absence of established willingness-to-pay thresholds, and multiple instrument versions with differing value sets.

Worked example of a social care intervention with illustrative capability scores

All numbers in this example are illustrative and undiscounted. A community support programme for adults with long-term conditions is compared with usual care over 2 years in 100 people per arm. The programme adds GBP 1,800 per person. Mean ICECAP-A index values are 0.76 with the programme and 0.70 with usual care, and EQ-5D utilities differ by 0.02 a year.

1. Years of Full Capability. The index gain is 0.76 − 0.70 = 0.06 a year, and 0.06 × 2 = 0.12 YFC per person.

$$\frac{1800}{0.12} = 15000$$

where 1800 is the incremental cost in GBP and 0.12 is the YFC gain per person, giving GBP 15,000 per YFC.

2. Years at or above a sufficiency threshold. With an illustrative threshold of 0.80, 45 of the 100 programme participants and 30 of the 100 usual care participants score at or above the threshold throughout. The difference is 15 people, and 15 × 2 = 30 person-years of sufficient capability per 100 people, or 0.30 per person. This simplified count credits only whole years spent at or above the threshold and is not the published YSC method.

$$\frac{1800}{0.30} = 6000$$

where 0.30 is the gain in years at or above the threshold per person, giving GBP 6,000 per year of sufficient capability.

3. The health-only view. The QALY gain is 0.02 × 2 = 0.04 per person, so the cost per QALY is 1,800 ÷ 0.04 = 45,000, or GBP 45,000 per QALY.

Against an illustrative threshold of GBP 30,000 per QALY the programme would look poor value on health alone, while the capability results show a benefit that the EQ-5D did not capture. UK estimates of willingness to pay for a YFC and a YSC exist, but NICE has not adopted a threshold for either, so the analysis can show the trade-off without settling whether GBP 15,000 or GBP 6,000 is worth paying. A cost-consequence analysis that reports the QALY and capability results side by side is one way to present such evidence to a committee.

ICECAP-A beside ICECAP-O, EQ-5D and ASCOT

ICECAP-A and ICECAP-O come from the same research group and both rest on the capability approach. ICECAP-A is the version for the general adult population and ICECAP-O the version for older people, so a study has to choose between them, and their scores should not be pooled as if they came from one instrument.

The EQ-5D describes health problems and its value sets produce utilities anchored at full health and dead. In 452 adults with knee pain, Keeley and colleagues found a mean ICECAP-A index of 0.88 (standard deviation 0.12) and a moderate correlation of 0.49 with the EQ-5D-3L index, which fits the view that the two measures overlap without measuring the same thing. ASCOT measures social care-related quality of life, which is closer to the outcomes a care service sets out to change, while ICECAP-A covers capability across a whole life. Subjective wellbeing measures ask how people feel, whereas ICECAP-A asks what they are able to do, which matters when someone chooses not to use a capability they have. Caregiver quality of life measures address carers specifically, and ICECAP-A is not carer-specific.

Common misreadings of ICECAP-A scores

The commonest error is to treat the index as a utility and multiply it by life years to report QALYs. The zero point is the lowest capability state, not death, and the top is full capability, not full health, so the product is Years of Full Capability, not QALYs.

A second error is to read level numbers the wrong way round. Level 4 is full capability in ICECAP-A, so code written for the EQ-5D convention, where level 1 is the best level, will reverse the scoring. Summing raw level numbers is a third error, because the attributes carry different weights in the UK index values.

Two further cautions concern the sample and the scale. The UK index values reflect the preferences of a UK general population sample, so analyses elsewhere should ask whether a local tariff exists. The mean of 0.88 in the knee pain sample shows that many respondents score near the top of the scale, which leaves less room to detect improvement and raises the risk of a ceiling effect in populations without serious capability limits.

Sources

• Al-Janabi H, Flynn TN, Coast J. Development of a self-report measure of capability wellbeing for adults: the ICECAP-A. Quality of Life Research. 2012;21(1):167-176. doi:10.1007/s11136-011-9927-2
• Flynn TN, Huynh E, Peters TJ, Al-Janabi H, Clemens S, Moody A, Coast J. Scoring the ICECAP-A capability instrument. Estimation of a UK general population tariff. Health Economics. 2015;24(3):258-269. doi:10.1002/hec.3014
• Goranitis I, Coast J, Day E, Copello A, Freemantle N, Frew E. Maximizing health or sufficient capability in economic evaluation? A methodological experiment of treatment for drug addiction. Medical Decision Making. 2017;37(5):498-511. doi:10.1177/0272989X16678844
• Helter TM, Coast J, Łaszewska A, et al. Capability instruments in economic evaluations of health-related interventions: a comparative review of the literature. Quality of Life Research. 2020;29(6):1433-1464. doi:10.1007/s11136-019-02393-5
• Keeley T, Coast J, Nicholls E, et al. An analysis of the complementarity of ICECAP-A and EQ-5D-3L in an adult population of patients with knee pain. Health and Quality of Life Outcomes. 2016. doi:10.1186/s12955-016-0430-x
• Mitchell PM, Roberts TE, Barton PM, Coast J. Assessing sufficient capability: a new approach to economic evaluation. Social Science and Medicine. 2015;139:71-79. doi:10.1016/j.socscimed.2015.06.037
• National Institute for Health and Care Excellence. Developing NICE guidelines: the manual (PMG20), chapter 7 Incorporating economic evaluation. Last updated 23 October 2025. https://www.nice.org.uk/process/pmg20/chapter/incorporating-economic-evaluation
• University of Bristol. ICECAP frequently asked questions. https://www.bristol.ac.uk/population-health-sciences/projects/icecap/frequently-asked-questions/
• University of Bristol. Using ICECAP in decision making. https://www.bristol.ac.uk/population-health-sciences/projects/icecap/using-icecap-in-decision-making/
• University of Bristol. ICECAP-A scoring guide (developers' calculation document for the UK index values). https://www.bristol.ac.uk/media-library/sites/social-community-medicine/images/centres/heb/icecap-a-scoring.docx
• Zhang Z, Saygın Avşar T, Cooper S, Dietz J. Adult Social Care Outcomes Toolkit and ICEpop Capability Measure in decision making: a review of NICE social care and public health guidelines. Value in Health. 2026;29(1):108-118. doi:10.1016/j.jval.2025.08.021