The EQ-5D is a generic measure of health-related quality of life developed by the EuroQol Group. It describes a person's health today on five dimensions - mobility, self-care, usual activities, pain or discomfort, and anxiety or depression - each rated at one of three levels (EQ-5D-3L) or five levels (EQ-5D-5L), from no problems to extreme problems. The five ratings form a health-state profile written as a five-digit code, such as 11111 for full health; the 3L can describe 243 distinct states and the 5L 3,125. A second part, the EQ VAS, asks the person to rate their overall health today on a vertical 0 to 100 scale.
It is deliberately generic: the same five dimensions are applied whether the patient has depression, arthritis or heart failure, which is what makes results comparable across conditions and usable in health-economic evaluation. That breadth is also its main trade-off in psychiatry. Only one of the five dimensions addresses mental health, and qualitative work with service users shows the EQ-5D covers little of what they themselves define as quality of life - control and autonomy, self-perception, belonging, hope. It is a tool for measuring the health-related burden of a condition in a common currency, not a symptom or recovery measure.
02 - Origin & purpose
Where it comes from.
The EuroQol Group formed in 1987 as an interdisciplinary team of researchers from five European countries and published the original instrument in 1990 in Health Policy. The goal was a short, standardised, self-completed descriptive system whose health states could be valued by the general public, so that a single index value could feed cost-utility analyses alongside disease-specific measures. Paul Dolan's 1997 UK time-trade-off study of around 3,000 adults produced the first widely used value set and established the convention that states worse than dead take negative values.
The three-level version proved blunt at the healthy end, with many patients reporting no problems at all, so the Group introduced the five-level version in 2009 and published its development in 2011. The 5L reduced ceiling effects and roughly quintupled the number of observed health states in patient samples, and country-specific 5L value sets followed - England in 2018, Denmark in 2021, and a new UK value set in 2026 that NICE adopted as its preferred approach in August 2026. Youth versions (EQ-5D-Y-3L and -5L) exist for ages 8 to 15, and experimental bolt-on dimensions, including cognition and sleep, are in development.
03 - Scoring & cutoffs
How scoring works.
The EQ-5D is not scored by adding up the five ratings. The level numbers are labels, not quantities, and the EuroQol user guide is explicit that they must not be summed. Instead there are three outputs. First, the profile itself: the five-digit health-state code, reported dimension by dimension. Second, the EQ VAS, the person's own 0 to 100 rating of their health today, which is analysed as a plain number. Third, the index value, obtained by looking the profile up in a country value set derived from general-population preference studies: full health scores 1, each problem level subtracts a country-specific decrement, and the worst states fall below zero, which represents states the public judged worse than being dead. The UK 3L value set runs from -0.594 to 1; the 2018 England 5L set from -0.285 to 1; the 2026 UK 5L set bottoms out at -0.567. Index values from different value sets are not interchangeable, so always report which set was used. There are no diagnostic cutoffs, because the EQ-5D is a health-status measure rather than a screening test.
04 - Validation evidence
How well it performs.
Reliability is respectable for a five-item profile. A systematic review of 99 EQ-5D-5L studies across 32 countries found a pooled test-retest intraclass correlation of 0.82 for the index value, with weighted kappas of 0.48 to 0.61 for the individual dimensions; a large web-based general-population study in Italy, the Netherlands and the UK reported ICCs of 0.73 to 0.84 for the 5L index and a lower 0.61 to 0.68 for the EQ VAS. Cronbach's alpha is not a meaningful statistic here, since the five dimensions are single items describing different things rather than indicators of one construct.
In mental health specifically, the 2014 NIHR Health Technology Assessment review by Brazier and colleagues pooled seven datasets (n = 5,748) and concluded that the EQ-5D is valid in common mental disorders, particularly depression and to a lesser extent anxiety, discriminating between severity groups and showing moderate responsiveness. Results in schizophrenia were mixed: the EQ-5D separated severity groups but was not responsive to change. Ceiling effects remain a concern. Across patient groups, 20% of respondents reported full health on the 3L and 16% on the 5L; in general populations the figures are 56% and 49%, so the instrument has little room to detect improvement in people who are already mildly affected.
0.82
TEST-RETEST ICC, 5L INDEX (POOLED)
0.61-0.68
TEST-RETEST ICC, EQ VAS
20% to 16%
FULL-HEALTH CEILING, PATIENTS (3L TO 5L)
0.074
MID, UK 3L INDEX
05 - How it compares
How it compares to the alternatives.
Instrument
Items
Time
When to reach for it
EQ-5D
5 + VAS
~2 min
Five dimensions plus a VAS; produces a preference-based index for QALYs. Licensed by the EuroQol Research Foundation - free for non-commercial use after registration, fee-bearing for commercial and most digital use.
Free, five items, positively worded wellbeing over two weeks. The natural free choice when you want a short, mental-health-relevant wellbeing score rather than a utility value.
WHO functioning and disability across six domains; 12- and 36-item forms. Free for clinical use. Better than EQ-5D for capturing the functional impact of psychiatric illness.
RAND-36
36
5-10 min
Public-domain version of the SF-36 (same 36 items, RAND scoring). Eight health domains including mental health and role-emotional; no fees or permission required.
The proprietary QualityMetric versions; the SF-6D utility index is derived from them. Licence and fees required, so not hosted here.
ReQoL-10
10
~3 min
Recovery-focused quality of life written with mental-health service users; free for publicly funded healthcare via Oxford University Innovation licence. Covers the themes the EQ-5D misses.
06 - When to use it
Right tool, wrong tool.
Reach for it when
-Health-economic evaluation and cost-utility analysis, where a preference-based index value is required - it is NICE's preferred measure of adult health-related quality of life.
-Service-level outcome or PROM programmes that need one generic measure comparable across physical and mental health conditions.
-Depression and, to a lesser extent, anxiety cohorts, where its validity and responsiveness have been demonstrated.
-Non-commercial clinical use in a registered project, where the licence is free.
Reach for something else when
-Measuring symptom severity or recovery in an individual patient - use a condition-specific scale such as the PHQ-9 or GAD-7, or a recovery measure such as ReQoL.
-Psychosis, where the EQ-5D discriminates severity but has not been shown to detect change.
-Mildly affected or well populations, where ceiling effects leave little room to improve.
-Any commercial product, app or website that has not obtained a EuroQol licence - the instrument may not be reproduced, adapted or made available digitally without written consent.
07 - Confidence & precision
Reading the score with care.
Because the index is a lookup rather than a sum, its precision depends on the value set and on how reliably the person picks the same levels. With test-retest ICCs around 0.8 and a general-population standard deviation of about 0.2 on the 5L index, the standard error of measurement is roughly 0.08 to 0.09, so differences of about 0.1 between two administrations of the same patient can plausibly be noise. Walters and Brazier's pooled anchor-based analysis across eight longitudinal studies put the minimal important difference for the UK 3L index at a mean of 0.074 (range -0.011 to 0.140), a figure that is widely used but varies with baseline health and treatment type. For the EQ VAS, anchor-based MIDs in the region of 7 to 10 points are typically cited, although published estimates range from under 1 to over 20. Report the value set, and treat index changes smaller than about 0.07 and VAS changes smaller than about 7 points as within uncertainty.
08 - Limitations
What it cannot tell you.
Only one of five dimensions concerns mental health, so the EQ-5D under-represents the burden of psychiatric illness relative to physical illness, and it covers little of what service users themselves describe as quality of life. Ceiling effects are large, especially on the 3L, and worst on mildly affected people. Index values depend on the country value set, so they are not comparable across sets, and the index is derived from public preferences rather than the patient's own weighting. Responsiveness is moderate in depression and poor in schizophrenia. Attempts to map PHQ-9, GAD-7 or HADS scores onto EQ-5D values perform poorly, so utilities cannot be reliably recovered from symptom scales. Finally, the instrument is trademarked and licensed by the EuroQol Research Foundation: it cannot be reproduced or hosted digitally without permission, and commercial or digital use carries a fee, which is why the questionnaire does not appear on this page.
09 - Licensing, explained
How licensing works.
The EQ-5D is managed by the EuroQol Group and requires registration and a licence, so we cannot host it. Free function and wellbeing measures are being added to the library.
See how Aisel removes friction where it costs most. A 20-minute walkthrough tailored to your clinic.
We value your privacy
We use cookies to analyse site usage and improve your experience. Analytics and embedded media (e.g. YouTube) only load if you accept. Read our cookie policy.