Health Utilities Index (HUI)

1. Abstract

The Health Utilities Index (HUI) is a family of generic, preference-based multi-attribute health status classification systems and health-related quality of life (HRQoL) measurement instruments. Developed predominantly at McMaster University by George W. Torrance, David H. Feeny, William J. Furlong, and colleagues, the HUI systems—most notably the Health Utilities Index Mark 2 (HUI2) and Health Utilities Index Mark 3 (HUI3)—provide comprehensive, standardized frameworks for describing functional health states and assigning cardinal utility weights to those states. The HUI2 system assesses seven attributes: Sensation, Mobility, Emotion, Cognition, Self-Care, Pain, and Fertility; while the HUI3 assesses eight independent attributes: Vision, Hearing, Speech, Ambulation, Dexterity, Emotion, Cognition, and Pain. Across both systems, each attribute consists of three to six hierarchically ordered, structurally distinct functional levels ranging from completely normal functioning to severe limitation or absolute functional loss.

The instruments employ mathematical multi-attribute utility theory (MAUT) grounded in von Neumann-Morgenstern expected utility theory. Through rigorously determined multi-attribute multiplicative preference functions, categorical responses are transformed into single-attribute utility scores and combined into a universal summary index anchored at 1.00 (perfect health) and 0.00 (death), accommodating states assessed as worse than death (down to -0.34 for HUI2 and -0.36 for HUI3). Extensive empirical evaluations demonstrate that the HUI exhibits robust test-retest reliability (intraclass correlation coefficients frequently ranging from 0.70 to 0.90), high construct validity across diverse clinical cohorts (including oncology, neurology, orthopedics, and pediatric populations), and remarkable responsiveness to clinically meaningful changes in health status over time. As a cornerstone of cost-utility analysis (CUA) and epidemiological surveillance, the HUI plays an indispensable role in generating Quality-Adjusted Life Years (QALYs), clinical trial efficacy evaluations, and healthcare resource allocation worldwide.

2. Keywords

Health Utilities Index, HUI2, HUI3, health-related quality of life, multi-attribute utility theory, cost-utility analysis, standard gamble, psychometrics, quality-adjusted life years, functional health status

3. Authors

The Health Utilities Index was conceptualized, designed, and psychometrically standardized by a multidisciplinary team of health economists, biostatisticians, and clinical researchers at McMaster University, located in Hamilton, Ontario, Canada, and Health Utilities Incorporated (HUInc):

• George W. Torrance, PhD: Professor Emeritus of Health Economics, DeGroote School of Business and Department of Clinical Epidemiology and Biostatistics, McMaster University; a pioneer in economic evaluation of healthcare interventions and foundational architect of health-state utility measurement.
• David H. Feeny, PhD: Professor Emeritus of Economics, McMaster University, and former Senior Investigator at the Center for Health Research, Kaiser Permanente Northwest; extensively published authority on preference measurement, technology assessment, and quality-of-life measurement systems.
• William J. Furlong, MSc: Associate Professor and Senior Health Economist, Department of Clinical Epidemiology and Biostatistics, McMaster University; Co-Founder and Managing Director of Health Utilities Incorporated (HUInc).
• Michael H. Boyle, PhD: Professor Emeritus, Department of Psychiatry and Behavioural Neurosciences and Department of Clinical Epidemiology and Biostatistics, McMaster University; specialized in behavioral epidemiology and child development.
• Ronald D. Barr, MB, ChB, MD: Professor of Pediatrics, Pathology, and Medicine, McMaster University; Chief of Pediatric Hematology/Oncology at McMaster Children’s Hospital, instrumental in adapting the system to pediatric oncology cohorts (HUI2).

Inquiries regarding administrative licensing, algorithmic score generation, user guides, and standardized questionnaires are coordinated through Health Utilities Incorporated (HUInc), Dundas, Ontario, Canada.

4. Purpose

The Health Utilities Index was created to bridge a critical methodological divide in healthcare evaluation: the need for an instrument that simultaneously captures a standardized, granular, clinical description of an individual’s multifaceted health status and produces an empirically defensible, preference-weighted single index of overall health utility. Traditional health-related quality-of-life profiles—such as the SF-36 or disease-specific functional checklists—yield multi-dimensional profiles that cannot be mathematically synthesized into an interval-scaled economic metric without arbitrary weighting assumptions. Conversely, direct holistic valuation techniques like the standard gamble (SG) or time trade-off (TTO) are cognitively burdensome, costly, and practically infeasible for routine use across large clinical trials or national population health surveys.

The HUI resolves this operational dilemma by decoupling the descriptive health classification from the valuation process. Patients, clinicians, or proxies complete a standardized questionnaire documenting functional capacity across key biological, physical, cognitive, and sensory systems. The descriptive health state is then mapped directly onto a multi-attribute utility function derived from rigorous preference elicitation studies administered to general community representative samples. Consequently, the HUI fulfills three interconnected core purposes:

• Clinical Efficacy & Longitudinal Monitoring: In randomized controlled trials (RCTs) and prospective cohort studies, HUI provides sensitive, domain-specific continuous scores for tracking therapeutic benefits, disease progression, toxicities, and rehabilitative gains across individual functional axes (e.g., dexterity, speech, ambulation, cognition, pain).
• Pharmacoeconomic & Cost-Utility Analysis: By generating valid interval-scaled utilities on the conventional zero-to-one scale (where 0.00 represents death and 1.00 represents optimal, unimpaired health), HUI functions as an engine for calculating Quality-Adjusted Life Years (QALYs). These QALY metrics are submitted to regulatory and reimbursement authorities—such as the National Institute for Health and Care Excellence (NICE) in the United Kingdom and the Canadian Agency for Drugs and Technologies in Health (CADTH)—to evaluate the incremental cost-effectiveness ratio (ICER) of emerging medical technologies.
• Population Health Surveillance: Deployed in massive national demographic surveys (e.g., Statistics Canada’s Canadian Community Health Survey and the National Population Health Survey), the HUI quantifies the population-level burden of chronic diseases, socioeconomic health gradients, and long-term consequences of public health interventions.

5. Psychological Construct

The latent psychological and physiological construct assessed by the Health Utilities Index is preference-weighted functional health status, conceptualized as an individual’s intrinsic capacity to perform core human functions free from physical, sensory, cognitive, and affective impairments. The instrument focuses explicitly on functional capacity (within-the-skin ability) rather than performance or environmental accommodations, thereby isolating intrinsic health status from external socio-technical adaptations. This broad construct is operationalized through two prominent structural iterations: HUI2 and HUI3.

Health Utilities Index Mark 2 (HUI2)

Originally designed with an emphasis on childhood cancer survivors, the HUI2 models health along seven key attributes:

• Sensation: Evaluates visual, auditory, and vocal integration across four levels, distinguishing between normal sensory processing, the necessity of sensory prosthetics (e.g., eyeglasses, hearing aids), residual sensory loss despite corrective equipment, and complete sensory deprivation (blind, deaf, or mute).
• Mobility: Assesses gross motor functioning, balance, and locomotion across five levels, ranging from normal agility (walking, bending, running, jumping) to limitations requiring mechanical mobility aids (canes, crutches, wheelchairs), personal caregiver assistance, and total loss of voluntary extremity motor control.
• Emotion: Reflects psychiatric well-being, psychological distress, and internalizing symptoms across five levels, spanning habitual happiness and freedom from worry, through intermittent neurotic symptoms (anxiety, depression, irritability, night terrors), up to extreme affective distress requiring psychiatric institutionalization or hospitalization.
• Cognition: Captures learning efficiency, intellectual processing, and mnemonic retention across four levels, benchmarked against age-appropriate educational attainment and progressing toward profound learning disability or total inability to retain memory.
• Self-Care: Quantifies basic activities of daily living (ADLs)—feeding, washing, dressing, and toileting—across four levels, from complete independence, through mechanical device reliance, to absolute dependency on personal attendants.
• Pain: Measures physical pain intensity, chronicity, and activity interruption across five levels, categorized by the pharmacological potency necessary for relief (from over-the-counter analgesics to prescription narcotics and refractory, unmanageable intractable pain).
• Fertility: Specifically incorporates reproductive physiological capacity across three levels: fully fertile, compromised fecundity, or absolute biological sterility.

Health Utilities Index Mark 3 (HUI3)

Developed to refine sensory domains and accommodate universal adult and pediatric general populations, the HUI3 decomposes sensory functioning into discrete structural channels and excludes fertility, featuring eight mutually independent attributes:

• Vision: Stratifies visual acuity and focal capacity across six levels, testing the dual criteria of reading ordinary newsprint and identifying an acquaintance across the street, both with and without corrective lenses, culminating in complete visual blindness.
• Hearing: Delineates auditory acuity across six levels, evaluating bilateral comprehension in both quiet dyadic settings and noisy group environments (conversations with three or more individuals), with and without hearing amplification aids.
• Speech: Characterizes expressive communication across five levels based on articulatory clarity and comprehensibility to familiar social contacts versus unfamiliar strangers.
• Ambulation: Evaluates lower-limb locomotion and community mobility across six levels, moving from unassisted neighborhood navigation to the requirement of mobility equipment, personal physical support, wheelchair dependency, and complete inability to walk.
• Dexterity: Assesses fine motor coordination and manipulation of objects across six levels, using the structural benchmark of having full use of two hands and ten fingers, independence achieved via adaptive tools, or progressive reliance on personal human assistance.
• Emotion: Focuses on affective outlook, existential interest, and dysphoria across five levels, ranging from happy and interested in life, through progressive sadness and severe unhappiness, to existential despair where life is judged not worthwhile.
• Cognition: Indexes higher-order executive functioning, memory recall, cognitive clarity, and daily problem-solving across six levels, ranging from unimpaired cognitive agility to catastrophic global cognitive failure (inability to remember anything, think clearly, or resolve basic tasks).
• Pain: Measures pain severity and functional interference across five levels, spanning complete absence of pain to excruciating, severe pain disrupting nearly all daily activities.

6. Theoretical Framework

The structural and mathematical foundation of the HUI is rooted in von Neumann-Morgenstern Expected Utility Theory and Multi-Attribute Utility Theory (MAUT) as formulated by Ralph Keeney and Howard Raiffa (1976). Classical psychometric measures generally assume additive or factor-analytic measurement models where ordinal items are summed under the axioms of Classical Test Theory (CTT). While useful for profiling, summative rating scores violate cardinal interval assumptions necessary for economic optimization models: a five-point decrement on a physical mobility scale cannot be assumed equivalent in utility to a five-point decrement on an affective anxiety scale.

To overcome this limitation, the McMaster team operationalized health states as multi-attribute commodity bundles. A complete health state vector $X = (x_1, x_2, dots, x_n)$ represents a specific profile where $x_j$ denotes the functional level attained within attribute $j$. Under MAUT, if the decision-maker’s preferences over these multidimensional states satisfy specific axiomatic conditions—notably preferential independence and utility independence—the joint utility function $u(X)$ can be expressed mathematically in a closed multiplicative functional form.

Mathematical Multiplicative Model

The generalized multi-attribute utility function for both HUI2 and HUI3 is represented as:

$$u(x_1, x_2, dots, x_n) = \frac{1}{K} \left[ \prod_{j=1}^{n} (1 + K k_j u_j(x_j)) – 1 \right]$$

Where:

• $u(x_1, x_2, dots, x_n)$ is the multi-attribute utility of the specified global health state, anchored at 1.00 for full health and 0.00 for death.
• $u_j(x_j)$ is the single-attribute utility value assigned to level $x$ within attribute $j$, scaled such that the highest level equals 1.00 and the lowest level equals 0.00.
• $k_j$ represents the attribute-specific scaling constant (utility weight), reflecting the relative importance of moving from the worst to the best level of attribute $j$ when all other attributes are held at their lowest levels.
• $K$ is a non-zero scaling constant calculated by solving the polynomial identity: $1 + K = \prod_{j=1}^{n} (1 + K k_j)$.

Empirical valuation surveys conducted with representative general population samples used the Standard Gamble (SG)—the axiomatic gold standard for measuring decision-making under uncertainty—complemented by Visual Analogue Scales (VAS). In the standard gamble, respondents choose between remaining in an intermediate chronic impaired health state with certainty or accepting a hypothetical medical intervention that offers a probability $p$ of returning to perfect health and a complementary probability $(1 – p)$ of immediate painless death. The probability $p$ at indifference establishes the cardinal von Neumann-Morgenstern utility score.

Through empirical estimation, the multiplicative constant $K$ was computed as positive for both systems, revealing that attributes exhibit a complementary (non-additive) structure: the utility loss associated with multiple simultaneous impairments is slightly less than the simple additive sum of each individual impairment, realistically reflecting human psychological adaptation and the non-linear compounding of disability.

7. Validity

The Health Utilities Index has undergone extensive international validation across clinical, epidemiologic, and psychometric investigations, demonstrating exceptional structural, convergent, discriminant, and predictive validity.

Construct and Convergent Validity

Construct validity has been verified by evaluating the convergence of HUI single-attribute and overall utility scores against established legacy profiles, including the SF-36, EQ-5D, FACT, and clinical disease markers. In multi-center oncology trials, HUI3 emotion and pain attributes correlate strongly ($r > 0.65$) with the corresponding emotional and physical subscales of the SF-36 and the Beck Depression Inventory (BDI). Studies analyzing stroke recovery report that HUI3 Ambulation and Dexterity scores correlate strongly ($r > 0.75$) with objective performance measures such as the Timed Up and Go (TUG) test, the Barthel Index of Activities of Daily Living, and the Chedoke-McMaster Stroke Assessment.

Discriminant and Known-Groups Validity

HUI instruments reliably discriminate between cohorts with varying disease severities. In clinical investigations involving patients with multiple sclerosis, osteoarthritis, rheumatoid arthritis, chronic obstructive pulmonary disease (COPD), and major depressive disorder, the HUI overall score decreases systematically across escalating clinical stages (e.g., Expanded Disability Status Scale tiers in MS or Gold stages in COPD; $p < 0.001$). Furthermore, the granularity of HUI3’s 8-attribute architecture allows it to capture subtle sensory and motor deficits that remain undetected by simpler instruments like the three-level or five-level EQ-5D, demonstrating superior discriminative efficiency in sensory-impaired populations.

Predictive Validity and Responsiveness

The HUI exhibits substantial longitudinal responsiveness, demonstrated by high standardized response means (SRMs > 0.80) following major surgical interventions, such as total hip or knee arthroplasty, and organ transplantation. Furthermore, baseline HUI utility scores are powerful independent predictors of secondary health outcomes, including post-discharge institutionalization rates, clinical resource consumption, and all-cause mortality over 5- and 10-year follow-up intervals in national population cohorts, even after adjusting for baseline chronologic age and medical comorbidities.

8. Reliability

Because the HUI is grounded in multi-attribute utility theory rather than classical test theory, internal consistency metrics (such as Cronbach’s alpha) are theoretically inappropriate for the aggregate index, as the health dimensions (e.g., vision vs. pain vs. dexterity) are conceptualized as causal indicators of health status rather than parallel items reflecting a single homogeneous internal construct. Consequently, reliability evaluations focus primarily on test-retest reproducibility, inter-rater concordance, and proxy-patient agreement.

• Test-Retest Reliability: In stable patient populations reassessed over intervals ranging from 1 to 4 weeks, the intraclass correlation coefficients (ICCs) for both HUI2 and HUI3 overall utility scores routinely exceed the 0.80 threshold. For instance, in stable rheumatology outpatients, test-retest ICCs have been documented at 0.84 to 0.88 for HUI3. Attribute-specific test-retest reproducibility exhibits weighted kappa coefficients ($\kappa_w$) ranging from 0.68 to 0.92, with sensory and motor domains demonstrating higher stability than the emotion and pain dimensions, which naturally show greater clinical fluctuation.
• Inter-Rater & Administration Mode Concordance: High consistency has been confirmed across alternative administration modes. Comparisons between self-administered paper questionnaires, computer-assisted telephone interviews (CATI), and in-person clinical interviews yield concordance coefficients generally exceeding 0.85, confirming that mode of administration introduces minimal systematic measurement error.
• Proxy-Subject Concordance: Given that HUI is frequently deployed in pediatric, demented, or severely incapacitated clinical samples, proxy reliability has been rigorously examined. Agreement between patients and family caregivers is moderate to high for observable physical, ambulation, and sensory attributes (ICCs: 0.70–0.85), whereas more subjective, internalizing attributes such as emotion and pain exhibit lower, yet clinically acceptable, concordance (ICCs: 0.50–0.68).

9. Factor Analysis

The structural topology of the HUI was established through behavioral decision research and multi-attribute utility axiomatic validation rather than exploratory factor analysis (EFA). Nonetheless, modern psychometric evaluations have subjected the HUI systems to confirmatory structural modeling and item-level analyses to evaluate the independence and structural coherence of its dimensions.

Structural Independence Verification

A central postulate of MAUT is mutual preferential independence: the trade-off rate between any two attributes must remain invariant regardless of the fixed levels of the remaining attributes. During the foundational validation studies at McMaster University, structured lottery assessments and trade-off experiments confirmed that the structural attributes of both HUI2 and HUI3 satisfy preferential and utility independence criteria within acceptable empirical margins. Structural equation modeling (SEM) confirms that the eight dimensions of HUI3 operate as structurally distinct functional components without high residual collinearity (inter-attribute correlations $r$ typically range between 0.15 and 0.45).

Confirmatory Factor and Measurement Models

When subjected to higher-order confirmatory factor analysis (CFA), the HUI3 attributes consistently conform to a bi-dimensional higher-order architecture separating Physical/Sensory Capacity (Vision, Hearing, Speech, Ambulation, Dexterity) from Psychosocial/Somatic Well-Being (Emotion, Pain, Cognition). Fit indices for this generalized two-factor structural representation yield robust statistical parameters:

• Comparative Fit Index (CFI) > 0.95
• Tucker-Lewis Index (TLI) > 0.94
• Root Mean Square Error of Approximation (RMSEA) < 0.05 (90% CI: 0.038–0.059)
• Standardized Root Mean Square Residual (SRMR) < 0.04

Item response theory (IRT) analyses, specifically graded response modeling, confirm that within each attribute, the hierarchical response steps exhibit ordered threshold parameters ($b_k$) without item reversals, verifying that moving from lower to higher functional levels represents an unambiguous worsening of impairment along the latent disability trajectory.

10. Instrument / Measurement Tool

The Health Utilities Index family encompasses self-administered questionnaires, proxy-response questionnaires, and interviewer-administered surveys tailored for both 1-week and 4-week recall windows, as well as current “usual” health status. The primary classification parameters and operational scoring rules are detailed below.

Structural Attributes and Level Counts

• HUI Mark 2 (HUI2): 7 attributes defining 24,000 unique health states:
  • Sensation (4 levels)
  • Mobility (5 levels)
  • Emotion (5 levels)
  • Cognition (4 levels)
  • Self-Care (4 levels)
  • Pain (5 levels)
  • Fertility (3 levels)
• HUI Mark 3 (HUI3): 8 attributes defining 972,000 unique health states:
  • Vision (6 levels)
  • Hearing (6 levels)
  • Speech (5 levels)
  • Ambulation (6 levels)
  • Dexterity (6 levels)
  • Emotion (5 levels)
  • Cognition (6 levels)
  • Pain (5 levels)

Multi-Attribute Scoring Formulas

Once a respondent completes the questionnaire, an algorithm maps the selected items to specific attribute levels. Each level corresponds to an empirically determined single-attribute utility score ($u_j(x_j)$). The overall multi-attribute utility score ($u$) is then calculated using the following mathematical formulas:

HUI2 Multi-Attribute Utility Function:

$$u_{\text{HUI2}} = 1.06 \times (u_{\text{Sens}} \times u_{\text{Mob}} \times u_{\text{Emot}} \times u_{\text{Cogn}} \times u_{\text{Self}} \times u_{\text{Pain}} \times u_{\text{Fert}}) – 0.06$$

HUI3 Multi-Attribute Utility Function:

$$u_{\text{HUI3}} = 1.371 \times (u_{\text{Vis}} \times u_{\text{Hear}} \times u_{\text{Spch}} \times u_{\text{Amb}} \times u_{\text{Dext}} \times u_{\text{Emot}} \times u_{\text{Cogn}} \times u_{\text{Pain}}) – 0.371$$

Single-Attribute Utility Coefficients (HUI3 Scoring Table)

The single-attribute utility values ($u_j$) applied in the HUI3 formula are detailed below across each level:

• Vision: Level 1 = 1.00; Level 2 = 0.98; Level 3 = 0.89; Level 4 = 0.80; Level 5 = 0.61; Level 6 = 0.00
• Hearing: Level 1 = 1.00; Level 2 = 0.95; Level 3 = 0.89; Level 4 = 0.80; Level 5 = 0.70; Level 6 = 0.00
• Speech: Level 1 = 1.00; Level 2 = 0.94; Level 3 = 0.89; Level 4 = 0.66; Level 5 = 0.00
• Ambulation: Level 1 = 1.00; Level 2 = 0.93; Level 3 = 0.86; Level 4 = 0.73; Level 5 = 0.43; Level 6 = 0.00
• Dexterity: Level 1 = 1.00; Level 2 = 0.95; Level 3 = 0.88; Level 4 = 0.76; Level 5 = 0.65; Level 6 = 0.00
• Emotion: Level 1 = 1.00; Level 2 = 0.95; Level 3 = 0.85; Level 4 = 0.64; Level 5 = 0.00
• Cognition: Level 1 = 1.00; Level 2 = 0.92; Level 3 = 0.88; Level 4 = 0.70; Level 5 = 0.42; Level 6 = 0.00
• Pain: Level 1 = 1.00; Level 2 = 0.96; Level 3 = 0.90; Level 4 = 0.77; Level 5 = 0.00

11. Permissions & Fee and Test Year

The initial conceptualization of the Health Utilities Index began with HUI Mark 1 (HUI1) in the late 1970s and early 1980s, primarily applied to evaluate neonatal intensive care outcomes. The HUI2 system was formalized and published in 1996 (Torrance et al.), followed by the standardized community-calibrated HUI3 system in 1999 and comprehensive psychometric overviews in 2001 and 2003 (Furlong et al.; Horsman et al.).

Copyright & Intellectual Property: The Health Utilities Index (HUI®) is a registered trademark of Health Utilities Incorporated (HUInc). All rights, including questionnaires, scoring manuals, scoring software, and proprietary algorithms, are owned by HUInc and McMaster University.

Licensing and Fees:

• Academic & Not-for-Profit Research: Use of the HUI by academic researchers, publicly funded clinicians, or graduate students typically requires the execution of a formal User Agreement. Modest administrative or materials fees may be assessed to support distribution, user guides, and algorithmic quality assurance.
• Commercial & Pharmaceutical Industry Applications: Commercial studies, including sponsored clinical trials, pharmaceutical outcomes research, and proprietary healthcare projects, are subject to commercial licensing fees calculated per protocol, language, and subject volume. Users must obtain explicit written authorization before reproducing or integrating HUI questions into electronic clinical outcome assessment (eCOA) platforms.
• Inquiries: Formal licensing requests, questionnaire acquisition, and scoring code services are managed through Health Utilities Incorporated (www.healthutilities.com).

12. References

• Feeny, D., Furlong, W., Boyle, M., & Torrance, G. W. (1995). Multi-attribute health status classification systems: Health Utilities Index. PharmacoEconomics, 7(6), 490–502. https://doi.org/10.2165/00019053-199507060-00004
• Feeny, D. H., Torrance, G. W., & Furlong, W. J. (1996). Health Utilities Index. In B. Spilker (Ed.), Quality of life and pharmacoeconomics in clinical trials (2nd ed., pp. 239–252). Lippincott-Raven Publishers.
• Feeny, D., Furlong, W., Torrance, G. W., Goldsmith, C. H., Zhu, Z., DePauw, S., Denton, M., & Boyle, M. (2002). Multiattribute and single-attribute utility functions for the Health Utilities Index Mark 3 system. Medical Care, 40(2), 113–128. https://doi.org/10.1097/00005650-200202000-00006
• Furlong, W. J., Feeny, D. H., Torrance, G. W., & Barr, R. D. (2001). The Health Utilities Index (HUI®) system for assessing health-related quality of life in clinical studies (CHEPA Working Paper Series # 01-02). McMaster University Centre for Health Economics and Policy Analysis.
• Horsman, J., Furlong, W., Feeny, D., & Torrance, G. (2003). The Health Utilities Index (HUI®): Concepts, measurement properties and applications. Health and Quality of Life Outcomes, 1, Article 54. https://doi.org/10.1186/1477-7525-1-54
• Keeney, R. L., & Raiffa, H. (1976). Decisions with multiple objectives: Preferences and value tradeoffs. John Wiley & Sons.
• McDowell, I. (2006). Measuring health: A guide to rating scales and questionnaires (3rd ed.). Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195165678.001.0001
• Torrance, G. W., Feeny, D. H., Furlong, W. J., Barr, R. D., Zhang, Y., & Wang, Q. (1996). Multi-attribute preference functions for a comprehensive health status classification system: Health Utilities Index Mark 2. Medical Care, 34(7), 702–722. https://doi.org/10.1097/00005650-199607000-00004

13. Items of the Scale (Questionnaire)