Health-Related Quality of Life

Health-Related Quality of Life in Health Economic Evaluation

Health-related quality of life, shortened to HRQoL or HRQL, is the part of a person's quality of life that illness and health care can change, and in health economics it supplies the quality weight in the quality-adjusted life year. Economic evaluation needs it because many treatments change how people feel and function more than how long they live. Measuring it involves separate choices about the instrument, the respondent, the timing and the handling of missing responses. Each choice can move a cost per QALY result, and NICE sets out preferred methods in detail so that evaluations are consistent.

Where health ends and quality of life begins

Wilson and Cleary proposed five levels of patient outcome, running from biological and physiological factors through symptoms, functioning and general health perceptions to overall quality of life, and observed that each level is harder to define and measure than the one before. HRQoL covers the middle of that chain, where disease and treatment are felt in daily life.

The boundary is contested. Karimi and Brazier argue that many definitions fail to separate HRQoL from health or from quality of life as a whole, and that many so-called HRQoL questionnaires in fact measure self-perceived health status. They suggest defining HRQoL as the way health is empirically estimated to affect quality of life, or reserving the term for the utility associated with a health state.

The US Food and Drug Administration guidance on patient-reported outcome measures of December 2009 describes HRQL as a multidomain concept representing the patient's general perception of the effect of illness and treatment on physical, psychological and social aspects of life. It adds that claiming a meaningful improvement in HRQL implies that every important domain was measured, that a general improvement was shown and that no domain showed a decrement.

Instruments that describe health-related quality of life

HRQoL is observed through questionnaires, usually a form of patient-reported outcome. Patrick and Deyo held that generic measures are needed to compare outcomes across populations and interventions, particularly for cost-effectiveness studies, while disease-specific measures assess the concerns of a diagnostic group and may be more sensitive to small changes that matter to clinicians or patients. Guyatt, Feeny and Patrick added a distinction by purpose, between discriminative instruments that measure differences between people at one time and evaluative instruments that measure change within people over time.

Profiles and preference-based indexes

A profile reports a separate score for each domain and leaves the reader to weigh them. A preference-based index describes a health state with a classification system and converts it to a single number using a value set estimated from stated preferences. Only the second kind yields a health state utility value on the QALY scale, where full health is 1 and dead is 0.

The EQ-5D-5L, the version NICE now prefers for adults, is one example. According to the EuroQol Group, its descriptive system has five dimensions (mobility, self-care, usual activities, pain or discomfort, and anxiety or depression), each with five levels of severity, together with a visual analogue scale on which patients rate their own health. The same responses can yield different utilities under different value sets, as the health state valuation page explains.

Who completes the questionnaire

The default respondent is the patient. The NICE manual (PMG36, section 4.3.3) states that HRQoL should be measured directly by patients and that, when this is not possible, measurements should come from the person who acts as their carer rather than from healthcare professionals. For children and young people, section 4.3.15 asks evaluations to report who completed the measures and the age of the children.

A proxy answer records a different observation from self-report. Pickard and Knight separate a proxy-patient perspective, in which the proxy answers as they think the patient would, from a proxy-proxy perspective, in which the proxy gives their own view. The FDA guidance discourages proxy-reported outcome measures in trials, particularly for symptoms that only the patient can know.

Carers can also be measured in their own right. PMG36 (section 4.3.17) asks evaluations to consider carers' health effects when relevant, with evidence that the condition substantially affects the carer's HRQoL, a topic covered on the caregiver quality of life page.

Measurement properties behind a trustworthy score

An instrument earns its place through evidence about how it behaves in the population at hand. Guyatt, Feeny and Patrick wrote that HRQoL instruments must be valid and must have a high ratio of signal to noise, which they call reliability for discriminative instruments and responsiveness for evaluative ones, and that scores must be interpretable. A ceiling effect, where many respondents report the best possible state, hides improvement, and the EuroQol Group states that the five-level EQ-5D was introduced in 2009 partly to reduce ceiling effects compared with the three-level version.

NICE uses the same properties to test whether the EQ-5D fits a condition. Under PMG36 (section 4.3.10), a case against it needs qualitative evidence that it lacks content validity because key dimensions of health are missing, supported by evidence from a synthesis of peer-reviewed literature that it performs poorly on construct validity and responsiveness in the patient population.

Reading a change in scores

A statistically significant change in an HRQoL score may still be too small to matter. Jaeschke, Singer and Guyatt anchored score changes to patients' global ratings of change and, in three studies of questionnaires on dyspnoea, fatigue and emotional function in chronic heart and lung disease, found the minimal clinically important difference to be a mean change of about 0.5 per item on a seven-point scale. The figure belongs to those instruments, not to other scales.

Adaptation can also move the yardstick. Sprangers and Schwartz describe response shift as a change in the meaning of a person's self-evaluation of quality of life, following changes in internal standards, values or conceptualisation as people adapt to chronic or life-threatening illness. Stable scores after a serious diagnosis may therefore reflect adaptation rather than an absence of effect.

Turning measured health into QALYs in NICE appraisal

Cost-utility analysis needs utilities attached to modelled health states. PMG36 (section 4.3.2) describes deriving them in two elements, measuring HRQoL in people in the relevant health state and valuing it according to preferences relative to other states, usually perfect health and death. Section 4.3.4 requires public preferences from a representative sample of the UK population using a choice-based method, and section 4.3.5 notes that different measurement methods produce utility values that cannot always be compared.

PMG36 (section 4.3.6) prefers the EQ-5D for adults, and NICE's interim methods statement PMG51, published on 27 August 2026, makes the EQ-5D-5L descriptive system and the UK EQ-5D-5L value set the preferred approach for adult reference-case utilities in new topics.

Under PMG36 (section 4.3.9), a mapping function from another HRQoL measure is a departure from the reference case that must be justified and tested in sensitivity analysis. The FDA guidance draws the line differently, discouraging the use in trials of preference weights developed for resource allocation unless their relationship to the trial population is known and found adequate. The same HRQoL data can therefore serve as a profile for a regulator and as a utility for an HTA body.

Missing questionnaires and a worked example

Repeated questionnaires go missing when patients are too unwell, drop out or die. Fielding, Fayers and Ramsay examined five clinical trials and found evidence against the assumption that data were missing completely at random in three of them. When missingness depends on the unobserved health of the non-responder, the data are missing not at random and a complete-case mean is biased.

$$\bar{u} = \pi_R \, \bar{u}_R + (1 - \pi_R) \, \bar{u}_M$$

where $\bar{u}$ is the mean utility for everyone in the arm, $\pi_R$ is the proportion who returned the questionnaire, $\bar{u}_R$ is the mean among responders and $\bar{u}_M$ is the mean among non-responders. The bias of the complete-case mean is therefore $(1 - \pi_R)(\bar{u}_R - \bar{u}_M)$.

All numbers in this example are illustrative. A trial randomises 200 patients to each arm and records EQ-5D-5L utilities at 12 months. On the new treatment, 140 patients return the questionnaire with a mean of 0.78, while the 60 who do not are sicker, with a mean that would have been 0.58. On the comparator, 180 return it with a mean of 0.74 and the 20 who do not would have had a mean of 0.60.

1. Complete-case difference. Using responders only, the difference is 0.78 − 0.74 = 0.04 in favour of the new treatment.

2. Full-arm means. For the new treatment, $\pi_R$ = 140/200 = 0.70, and 0.70 × 0.78 = 0.546 while 0.30 × 0.58 = 0.174, so the mean is 0.546 + 0.174 = 0.720. For the comparator, $\pi_R$ = 180/200 = 0.90, and 0.90 × 0.74 = 0.666 while 0.10 × 0.60 = 0.060, so the mean is 0.666 + 0.060 = 0.726.

3. True difference. The full-arm difference is 0.720 − 0.726 = −0.006, a slight disadvantage for the new treatment.

4. Source of the gap. The bias is 0.30 × (0.78 − 0.58) = 0.060 in the new arm and 0.10 × (0.74 − 0.60) = 0.014 in the comparator, so the difference is overstated by 0.060 − 0.014 = 0.046, matching 0.04 − (−0.006).

5. Interpretation. If a model held the 12-month values for a full year, the complete-case figures would credit the new treatment with 0.04 × 1 = 0.04 QALYs per patient, while the full arms imply 0.006 QALYs fewer. Non-responder means are never observed in practice, so the analysis must state an assumption about them and test others, as the missing data page describes.

Neighbouring concepts

Several terms overlap with HRQoL. Health status describes health itself, while HRQoL concerns its effect on a person's life, although Karimi and Brazier show that many instruments blur the two. A patient-reported outcome is a mode of report that can measure HRQoL or narrower things such as a single symptom, and a wellbeing or capability instrument aims at an outcome broader than health.

Limitations and common misreadings

Most errors come from treating one measurement choice as if it were the concept. Each misreading below can change an incremental QALY estimate.

1. Treating a profile score as a utility. A condition-specific domain score cannot be multiplied by time to give QALYs without a value set or a validated mapping.
2. Assuming instruments are interchangeable. Different methods produce different utility values, so switching instruments between arms distorts comparisons.
3. Reading proxy answers as self-report. A carer's or clinician's answer is another perspective, and the respondent should be reported.
4. Ignoring missing questionnaires. Complete-case means overstate HRQoL when the sickest patients stop responding.
5. Expecting the QALY to capture all of quality of life. The descriptive system sets which dimensions count, and anything outside it is not valued.

Sources

• EuroQol Group. EQ-5D-5L (web page, accessed 1 November 2026). https://euroqol.org/information-and-support/euroqol-instruments/eq-5d-5l/
• Fielding S, Fayers PM, Ramsay CR. Investigating the missing data mechanism in quality of life outcomes: a comparison of approaches. Health and Quality of Life Outcomes. 2009;7:57. https://doi.org/10.1186/1477-7525-7-57
• Food and Drug Administration. Guidance for Industry. Patient-Reported Outcome Measures: Use in Medical Product Development to Support Labeling Claims. December 2009. https://www.fda.gov/media/77832/download
• Guyatt GH, Feeny DH, Patrick DL. Measuring health-related quality of life. Annals of Internal Medicine. 1993;118(8):622-629. https://doi.org/10.7326/0003-4819-118-8-199304150-00009
• Jaeschke R, Singer J, Guyatt GH. Measurement of health status: ascertaining the minimal clinically important difference. Controlled Clinical Trials. 1989;10(4):407-415. https://doi.org/10.1016/0197-2456(89)90005-6
• Karimi M, Brazier J. Health, health-related quality of life, and quality of life: what is the difference? PharmacoEconomics. 2016;34:645-649. https://doi.org/10.1007/s40273-016-0389-9
• 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. Sections 4.3.2 to 4.3.6, 4.3.9, 4.3.10, 4.3.15 and 4.3.17. https://www.nice.org.uk/process/pmg36/chapter/economic-evaluation-2
• National Institute for Health and Care Excellence. Interim methods statement: implementing the EQ-5D-5L value set (PMG51). Published 27 August 2026. Sections 1.7 and 1.17. https://www.nice.org.uk/process/pmg51
• Patrick DL, Deyo RA. Generic and disease-specific measures in assessing health status and quality of life. Medical Care. 1989;27(3 Suppl):S217-S232. https://doi.org/10.1097/00005650-198903001-00018
• Pickard AS, Knight SJ. Proxy evaluation of health-related quality of life: a conceptual framework for understanding multiple proxy perspectives. Medical Care. 2005;43(5):493-499. https://doi.org/10.1097/01.mlr.0000160419.27642.a8
• Sprangers MA, Schwartz CE. Integrating response shift into health-related quality of life research: a theoretical model. Social Science and Medicine. 1999;48(11):1507-1515. https://doi.org/10.1016/S0277-9536(99)00045-3
• Wilson IB, Cleary PD. Linking clinical variables with health-related quality of life: a conceptual model of patient outcomes. JAMA. 1995;273(1):59-65. https://doi.org/10.1001/jama.1995.03520250075037