EQ-5D-5L Value Set

EQ-5D-5L Value Set in Valuation Studies and NICE Appraisals

The five-level EQ-5D describes a patient's health as one of 3,125 profiles, and those profiles carry no utility until a value set attaches a number to each one. The value set therefore decides how large a QALY gain a trial or model reports, and two scoring routes applied to the same EQ-5D-5L responses can give different incremental cost-effectiveness ratios. England shows how much rests on the choice. From 2017 NICE asked for five-level responses to be scored with three-level values, in 2019 it confirmed that it would not use the first English EQ-5D-5L value set after an independent quality assurance questioned its data, and in August 2026 it adopted a new UK EQ-5D-5L value set for adults. Value sets for the older three-level version follow a different history and are described under the EQ-5D-3L value set.

Scoring 3,125 health states from a small valued sample

The EQ-5D-5L has five dimensions with five response levels each, which gives 3,125 possible health states. No valuation study asks people about all of them. Respondents value a designed subset, and a statistical model fitted to their answers predicts a utility for every state on a scale where 1 is full health and 0 is dead. Hernández Alava, Pudney and Wailoo noted that the time trade-off tasks in the English valuation study covered less than 3% of the 3,125 states, so most published values are model predictions.

The simplest model is additive. Each dimension has a decrement for each of levels 2 to 5, and a state's utility is 1 minus the decrements for the levels it reports.

$$U(s) = 1 − \sum_{d=1}^{5} \beta_{d}(l_{d})$$

where $U(s)$ is the utility of health state $s$, $l_d$ is the level reported on dimension $d$, and $\beta_d(l_d)$ is the decrement the model estimates for that level, which is zero at level 1. With four decrements on each of five dimensions this form has 20 parameters, the size of the hybrid model Devlin and colleagues estimated for England. Some national models add a constant or interaction terms, but the principle is the same. The estimated decrements are the value set, and a value set for another instrument works in the same way.

The EQ-VT valuation protocol

The EuroQol Group wrote a standard protocol for valuing the EQ-5D-5L, known as EQ-VT and released with software for computer-assisted interviews. Stolk and colleagues describe how problems in the first studies led to revisions and to version 2 of the protocol.

The main task is the composite time trade-off. Janssen and colleagues introduced it as a combination of a conventional time trade-off, for states the respondent regards as better than dead, and a lead-time time trade-off, for states regarded as worse than dead. Respondents also complete a set of paired comparisons, a form of discrete choice experiment, in which they choose the better of two EQ-5D-5L states.

Oppe and colleagues describe the experimental design. The composite time trade-off part includes 86 health states with a minimum sample of 1,000 respondents, grouped in blocks that mix states expected to lie at different points on the value scale, and each respondent completes no more than ten time trade-off tasks, excluding warm-up and practice tasks.

Stolk and colleagues report that the first studies, in England, the Netherlands, China, Canada and Spain, produced high rates of inconsistent composite time trade-off responses. Later versions added quality control of interviewer performance during fieldwork, described on the interviewer bias page.

Hybrid models and models using time trade-off data alone

Once the data are collected, the analysts choose a model. The protocol supports hybrid models that combine the time trade-off and choice data in one likelihood. The choice data show which of two states respondents prefer and by how much on a latent scale, and the time trade-off data anchor that scale so that dead equals 0 and full health equals 1. The English set came from a 20-parameter hybrid model of this kind. Other teams model the composite time trade-off data alone.

Because most of the 3,125 values are predictions, the model specification shapes the value set as well as the raw data do. The quality assurance of the English set criticised its statistical analysis as well as its data.

Interim scoring by crosswalk to EQ-5D-3L values

Before national five-level value sets existed, the EuroQol Group offered an interim scoring method. Van Hout and colleagues published a crosswalk that gives each EQ-5D-5L state a value taken from an existing national EQ-5D-3L value set, using data from people who had completed both versions. The crosswalk lets 5L responses be scored, but the values keep the preferences and the scale of the three-level set.

NICE used the same idea. Its position statement of 2017 already asked for five-level responses to be scored with three-level values, and the October 2019 update asked for utility values in reference-case analyses to be calculated by mapping EQ-5D-5L descriptive data onto the EQ-5D-3L value set. A crosswalk or mapping function of this kind is a statistical link between two instruments, whereas a five-level value set rests on the public's own valuations of five-level states. The two can give very different numbers for the same response, because the UK 3L set valued over a third of states as worse than dead.

The England value set and the quality assurance dispute

Devlin and colleagues published an EQ-5D-5L value set for England in 2018, using the EuroQol protocol with 996 respondents and a 20-parameter hybrid model. Values ran down to minus 0.285 for the worst state, and the highest value below full health, 0.950, belonged to states 11211 and 21111. Only 5.1% of the 3,125 states were valued as worse than dead, compared with over a third under the UK EQ-5D-3L value set.

An independent quality assurance followed. Hernández Alava, Pudney and Wailoo, at the University of Sheffield, carried it out as an EEPRU project on validating the value set, with access to the data from all 996 participants and the code for the published models. They found that 47% of respondents valued more than 20% of the states inconsistently, about double the three-level rate. They concluded that the valuation data and the statistical analysis behind the English set had significant limitations and recommended considering a new programme of development that included new data collection. Van Hout and colleagues replied in the same issue of the journal that many of the points were misleading, reflected misunderstandings or were irrelevant.

NICE's position statement (ECD16) was first published in November 2017, before the quality assurance. Its October 2019 update explained that the quality assurance and reports from four independent experts raised concerns about the quality and reliability of the valuation data and the methods used to model them, and NICE therefore chose not to use the Devlin value set and kept mapping onto 3L values in the reference case.

The UK value set and NICE's 2026 methods

Rowen and colleagues published a United Kingdom value set for the EQ-5D-5L in Value in Health in 2026, and NICE cites it as the set that fills the earlier gap. NICE's interim methods statement (PMG51), published on 27 August 2026, replaces ECD16 and applies to technology appraisals, highly specialised technologies, HealthTech evaluations and NICE guidelines.

The statement makes the EQ-5D-5L descriptive system and the UK value set NICE's preferred approach for measuring and valuing health-related quality of life in adults, and says the UK set should be used to derive all adult utility values in reference-case analyses. The same set is used for general population utilities, including those behind the absolute and proportional QALY shortfall used for the severity modifier. Utility values derived with the 3L value set should not appear in sensitivity or scenario analyses, and all 3L data should be mapped to 5L utility values, which may first require mapping onto the 3L value set. Any mapping model has to be described, its choice justified and its fit tested, with sensitivity analyses across alternative mapping models. NICE does not recommend a specific measure of health-related quality of life for children and young people.

NICE's questions and answers page says the change applies to technology appraisals and highly specialised technologies whose invitation to participate was issued after the statement was published, and that published negative NICE guidance will not routinely be considered for review solely to incorporate the EQ-5D-5L value set. The technology appraisal manual (PMG36) was last updated on 13 October 2026, a change to the section on varying the funding requirement. Its notice on the EQ-5D-5L still says that the interim statement takes precedence over the manual until the manual is updated later in the year, and it lists sections 4.2.2 (table 4.1), 4.3.6, 4.3.8, 4.3.9, 4.3.12 (with figure 4.1) and 4.3.16 for revision. Until then, the older wording in those sections does not describe NICE's current method.

Illustrative worked example of two scoring routes

All numbers in this example are illustrative and do not come from any published value set. A trial records patients in EQ-5D-5L state 34232 at baseline and in state 12121 after treatment. The gain lasts one year, with no discounting, and the treatment adds GBP 8,800 to costs. The dimensions are numbered 1 to 5 in the order of the state code.

1. Utilities under an illustrative five-level value set. The invented decrements are 0.08 for level 3 on dimension 1, 0.12 for level 4 and 0.03 for level 2 on dimension 2, 0.04 for level 2 on dimension 3, 0.10 for level 3 and 0.05 for level 2 on dimension 4, and 0.06 for level 2 on dimension 5.

$$U(34232) = 1 − (0.08 + 0.12 + 0.04 + 0.10 + 0.06) = 0.60 \qquad U(12121) = 1 − (0.03 + 0.05) = 0.92$$

where each term in brackets is the decrement for the level reported on one dimension. The utility gain is 0.92 minus 0.60, which is 0.32.

2. Utilities under an illustrative mapping onto three-level values. Suppose the mapping gives 0.40 for the baseline state and 0.84 for the follow-up state, a gain of 0.84 minus 0.40, which is 0.44. The mapped baseline is lower, in line with the UK three-level set valuing far more states as worse than dead than the English five-level set.

3. QALYs and cost per QALY. Over one year the QALY gain is 0.32 × 1 = 0.32 under the five-level set and 0.44 × 1 = 0.44 under the mapping.

$$\text{ICER}_{5L} = \frac{8800}{0.32} = 27500 \qquad \text{ICER}_{map} = \frac{8800}{0.44} = 20000$$

where 8,800 is the incremental cost in GBP and the denominators are the QALY gains under each scoring route.

The same patients and the same costs give GBP 20,000 per QALY under the mapping and GBP 27,500 per QALY under the five-level set. At an illustrative threshold of GBP 25,000 per QALY the scoring route alone would decide the result. A switch of value set can move an ICER in either direction, depending on where patients start and which dimensions improve, which is why NICE specifies one set for the reference case and rules out the other in sensitivity analyses.

How the EQ-5D-5L value set relates to nearby concepts

A value set is the general idea of weights for an instrument, and a country-specific value set is one estimated from a single country's general public. A value set for the EQ-5D-5L is the five-level member of that family, and the sets discussed here are national sets for England and the United Kingdom. The EQ-5D-5L itself is the descriptive system, which records problems on each dimension and carries no values until a set is applied.

The EQ-5D-3L value set scores the three-level instrument with 243 states, and its values sit on a different scale, so they are not mixed with five-level values unless mapped first. A crosswalk or mapping function predicts one instrument's values from another's responses, while a five-level value set comes from direct valuation of five-level states. Each value the set produces is a health state utility value that feeds a QALY.

Common misreadings of EQ-5D-5L value sets

The English dispute concerned the valuation data and the model behind one value set, not the five-level descriptive system. NICE kept accepting EQ-5D-5L responses throughout and only changed how they were scored.

A value of 0.950 for the mildest English states does not mean the instrument cannot show smaller problems. The value 0.950 is what the English model gives to states 11211 and 21111, each with a single level 2 response, and another value set can give the same state a different number. The lowest value, minus 0.285 in England, is also specific to the set, so comparisons of utilities across studies need the value set named alongside the instrument.

Reading PMG36 alone can mislead in late 2026. The sections listed for revision had not been rewritten by the October 2026 update, and PMG51 takes precedence until they are. Reports that name the value set alongside the instrument let readers judge results, because the model behind a set shapes most of the 3,125 values that were never valued directly.

Sources

• Devlin NJ, Shah KK, Feng Y, Mulhern B, van Hout B. Valuing health-related quality of life: an EQ-5D-5L value set for England. Health Economics. 2018;27(1):7-22. doi:10.1002/hec.3564
• Hernández Alava M, Pudney S, Wailoo A. The EQ-5D-5L value set for England: findings of a quality assurance program. Value in Health. 2020;23(5):642-648. doi:10.1016/j.jval.2019.10.017
• Janssen BMF, Oppe M, Versteegh MM, Stolk EA. Introducing the composite time trade-off: a test of feasibility and face validity. European Journal of Health Economics. 2013;14(Suppl 1):5-13. doi:10.1007/s10198-013-0503-2
• National Institute for Health and Care Excellence. Use of the EQ-5D-5L value set for England: NICE position statement (ECD16). Published November 2017, updated October 2019; replaced by PMG51. https://www.nice.org.uk/corporate/ecd16
• National Institute for Health and Care Excellence. Interim methods statement: implementing the EQ-5D-5L value set (PMG51). Published 27 August 2026. https://www.nice.org.uk/process/pmg51/chapter/statement-on-using-the-eq-5d-5l-value-set
• National Institute for Health and Care Excellence. The EQ-5D-5L, questions and answers on the interim methods statement (PMG51). 2026. https://www.nice.org.uk/what-nice-does/faqs/the-eq-5d-5l
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). Published 31 January 2022, last updated 13 October 2026. https://www.nice.org.uk/process/pmg36
• Oppe M, Norman R, Yang Z, van Hout B. Experimental design for the valuation of the EQ-5D-5L. In: Devlin N, Roudijk B, Ludwig K, editors. Value Sets for EQ-5D-5L: A Compendium, Comparative Review and User Guide. Cham: Springer; 2022. https://www.ncbi.nlm.nih.gov/books/NBK589299/
• Rowen D, et al. A United Kingdom value set for the EQ-5D-5L. Value in Health. 2026;29(5):858-869. doi:10.1016/j.jval.2026.03.008
• Stolk E, Ludwig K, Rand K, van Hout B, Ramos-Goñi JM. Overview, update, and lessons learned from the international EQ-5D-5L valuation work: version 2 of the EQ-5D-5L valuation protocol. Value in Health. 2019;22(1):23-30. https://pubmed.ncbi.nlm.nih.gov/30661630/
• van Hout B, Janssen MF, Feng YS, et al. Interim scoring for the EQ-5D-5L: mapping the EQ-5D-5L to EQ-5D-3L value sets. Value in Health. 2012;15(5):708-715.
• van Hout B, Mulhern B, Feng Y, Shah K, Devlin N. The EQ-5D-5L value set for England: response to the "quality assurance". Value in Health. 2020;23(5):649-655. doi:10.1016/j.jval.2019.10.013