SF-36v2 Health Survey

1. Abstract

The SF-36v2 Health Survey (36-Item Short Form Health Survey Version 2) represents one of the most widely utilized, empirically scrutinized, and methodologically validated generic patient-reported outcome measures (PROMs) in global clinical research and health services evaluation. Developed by John E. Ware Jr. and colleagues as an advancement of the original Medical Outcomes Study 36-Item Short Form (SF-36), the Version 2 instrument systematically assesses functional health, well-being, and health-related quality of life (HRQoL) across eight multi-item operational domains: Physical Functioning (10 items), Role-Physical (4 items), Bodily Pain (2 items), General Health (5 items), Vitality (4 items), Social Functioning (2 items), Role-Emotional (3 items), and Mental Health (5 items), supplemented by an unscaled single-item measure of perceived health transition over a one-year window. The instrument comprises 36 items calibrated with diverse categorical response structures ranging from 3-point to 6-point polytomous scales.

A central advancement of Version 2 over Version 1 involves the recalibration of the Role-Physical and Role-Emotional subscales from dichotomous (yes/no) choices to 5-level polytomous response categories, substantially attenuating severe ceiling and floor effects, reducing item non-response, and elevating empirical precision. Psychometric evaluations across diverse clinical and general population samples consistently establish high internal consistency reliability, with Cronbach’s alpha coefficients routinely exceeding α = .80 across individual scales and reaching ≥ .90 for the two overarching higher-order constructs: the Physical Component Summary (PCS) and Mental Component Summary (MCS). Structural equation modeling and confirmatory factor analyses replicate an orthogonal, two-dimensional second-order hierarchical construct comprising physical and mental dimensions, demonstrating invariance across demographic strata, chronic disease populations, and international linguistic adaptations.

2. Keywords

SF-36v2 Health Survey, health-related quality of life, psychometrics, Physical Component Summary, Mental Component Summary, patient-reported outcome measures, Medical Outcomes Study, functional status, construct validity, norm-based scoring

3. Authors

The SF-36v2 Health Survey was engineered by an interdisciplinary team of psychometricians, health economists, and medical sociologists led by John E. Ware, Jr., Ph.D. (Department of Health and Social Behavior, Harvard School of Public Health; Health Assessment Lab, Boston, MA; and QualityMetric Incorporated). Key primary co-investigators and methodologists include Mark Kosinski, M.A. (QualityMetric Incorporated, Lincoln, RI), Kimberly K. Snow, M.A., and Jakob E. Dewey, M.S. The conceptual foundations stem from the landmark Medical Outcomes Study directed by Alvin R. Tarlov, John E. Ware, Jr., Allyson Ross Davies, and Cathy Donald Sherbourne under the auspices of the RAND Corporation.

4. Purpose

The fundamental purpose of the SF-36v2 Health Survey is to provide a standardized, psychometrically robust, and comprehensive measure of subjective health status that can be administered efficiently across diverse populations without imposing excessive respondent burden. Generic health profiling tools serve an indispensable function in contemporary clinical research, comparative effectiveness studies, health economics, and population epidemiology. Unlike disease-specific instruments designed exclusively to detect manifestations of particular pathologies (such as oncology-specific or cardiology-specific scales), a generic metric enables direct, cross-condition comparisons of the disease burden inflicted by disparate chronic and acute illnesses.

From an applied clinical perspective, the SF-36v2 fulfills multiple interrelated functions:

• Clinical Trial Efficacy Evaluation: Quantifying patient-reported therapeutic benefits and treatment-related adverse functional impacts within randomized controlled trials (RCTs).
• Health Services and Outcomes Research: Tracking quality of care, institutional performance, and the macro-level longitudinal trajectory of community-dwelling populations.
• Individual Patient Monitoring: Assisting clinicians in outpatient ambulatory and hospital settings to detect insidious functional declines, identify psychiatric comorbidities (e.g., major depressive disorder masked as somatic complaints), and personalize rehabilitative management plans.
• Cost-Utility and Health Economic Modeling: Translating multidimensional HRQoL profiles into preference-based health state utilities (e.g., via the derived Short-Form Six-Dimension or SF-6D algorithm) to inform Quality-Adjusted Life Year (QALY) computations.

The theoretical rationale for the transition from the legacy SF-36 to the SF-36v2 centered on solving acute measurement limitations identified during widespread international utilization of Version 1. In Version 1, the two role functioning domains (Role-Physical and Role-Emotional) relied on dichotomous “Yes/No” response formats, creating substantial floor effects in severely disabled populations and pervasive ceiling effects in ambulatory cohorts. Version 2 addressed these structural deficits by replacing dichotomous options with 5-point polytomous frequency ratings (“All of the time” to “None of the time”), streamlining item wording, eliminating ambiguous colloquial phrasing, and introducing standardized norm-based scoring (T-score metric: Mean = 50, Standard Deviation = 10) anchored to updated general population norms.

5. Psychological Construct

The SF-36v2 conceptualizes health-related quality of life as a multidimensional construct encompassing broad subjective representations of well-being, personal functional capacity, and illness burden. Grounded in the World Health Organization’s tripartite definition of health as a state of complete physical, mental, and social well-being (and not merely the absence of disease or infirmity), the instrument captures eight primary domains clustered beneath two higher-order dimensions:

Physical Health Subdomains

• Physical Functioning (PF, 10 items): Measures the degree to which an individual’s current health status limits performance across a hierarchical continuum of physical behaviors. These span basic activities of daily living (ADLs)—such as independent bathing and dressing—through intermediate mobility tasks (walking 100 yards, climbing one flight of stairs, bending, kneeling, carrying groceries) to demanding, high-energy vigorous endeavors (running, lifting heavy objects, strenuous sports).
• Role Limitations due to Physical Health (RP, 4 items): Assesses the behavioral impact of physical health problems on execution of vocational, academic, and domestic responsibilities over the preceding four weeks. It gauges quantitative cutbacks in time spent, diminished task accomplishment, restrictions in the range or type of work performed, and qualitative performance deficits (e.g., requiring excessive physical effort).
• Bodily Pain (BP, 2 items): Captures two interrelated dimensions of the pain experience over the prior four-week period: the sensory intensity/magnitude of bodily discomfort (ranging from none to very severe) and the degree to which that pain interferes with normal occupational activities and domestic housework.
• General Health Perceptions (GH, 5 items): Evaluates an individual’s personal, cognitive evaluation of their overall health status. This scale incorporates a single-item global self-rating (“Excellent” to “Poor”), alongside validated psychometric item batteries that rate susceptibility to illness, parity with peer health status, internal expectations of health decline, and current holistic excellence of physiological well-being.

Mental Health Subdomains

• Vitality (VT, 4 items): Represents a bipolar energy-fatigue continuum capturing subjective stamina and vigor versus physical exhaustion and lethargy over the preceding four weeks. Items measure the frequency with which a person felt “full of life” and had “a lot of energy” versus feeling “worn out” and “tired.”
• Social Functioning (SF, 2 items): Evaluates the quantity and quality of interpersonal engagement, examining the extent to which physical health or emotional distress disrupts normal social connections with family members, friends, neighbors, and broader community groups over the prior month.
• Role Limitations due to Emotional Problems (RE, 3 items): Measures the disruptive spillover of affective disturbances (such as clinical depression or generalized anxiety) into occupational and daily life roles. Items quantify time reductions, lower overall accomplishment, and decreases in the care, concentration, or precision with which vocational or daily tasks are completed.
• Mental Health (MH, 5 items): Comprises four major clinical dimensions of psychological well-being and affective distress over the preceding four weeks: depressive symptomatology (e.g., feeling downhearted, blue, or so down in the dumps that nothing could cheer the person up), generalized anxiety/nervousness, behavioral and emotional control, and generalized positive affect (e.g., feeling peaceful, calm, and happy).

Additionally, the Reported Health Transition (HT, 1 item) evaluates perceived temporal trajectories in systemic health status over the course of the preceding year. Although intentionally excluded from the mathematical calculation of the eight scale scores and two higher-order summary components, this unscaled index provides vital clinical insight into acute versus chronic dynamic shifts.

6. Theoretical Framework

The architectural framework of the SF-36v2 is rooted in classical psychometric theory, operationalized health economics, and the conceptual models developed during the Medical Outcomes Study (MOS) in the late 1980s. Early psychometric theory predominantly treated physical health and mental health either as isolated somatic variables or as singular unidimensional axes of global impairment. The MOS theoretical paradigm revolutionized this viewpoint by postulating that health constitutes an interdependent, hierarchical, bifactorial continuum spanning both physical and psychological poles of human existence.

Under the MOS model, health constructs are systematically arrayed along two axes:

1. The content domain (physical health vs. mental health).
2. The operational level of measurement (internal subjective evaluations vs. outward behavioral limitations in primary functional roles).

At the structural base of the model reside fundamental physiological and affective experiences: physical sensations (e.g., bodily pain, stamina, acute depressive or anxious affect) and subjective cognitive evaluations (e.g., general health perceptions). At the mid-tier reside behavioral manifestations: physical movements, self-care routines, and social interactions. At the highest level of behavioral manifestation reside complex role performances: fulfilling societal, occupational, and family responsibilities without physical or emotional disruption.

A central theoretical assumption of the SF-36v2 framework is the distinction between generic and disease-specific functional health. Generic models presuppose that irrespective of the etiology of an illness—whether an autoimmune disease like rheumatoid arthritis, a metabolic condition like diabetes mellitus, or a psychiatric illness like major depressive disorder—all human conditions share a common behavioral and experiential denominator. Patients suffering from physical or emotional dysfunctions manifest decrements across predictable axes: they walk less, accomplish less work, experience altered energy levels, suffer pain, and withdraw socially. By creating a psychometrically stable, standardized measurement model calibrated to general population norms, the SF-36v2 operationalizes this universal conceptual architecture.

7. Validity

The construct, criterion, convergent, and discriminant validity of the SF-36v2 have been verified across hundreds of empirical studies spanning diverse clinical, demographic, and cross-cultural cohorts.

Convergent and Discriminant Validity

Multitrait-multimethod correlation matrices demonstrate that the items of the SF-36v2 meet the classical criteria for construct validity with high empirical precision:

• Item-Scale Convergent Validity: Correlations between individual items and their hypothesized parent scales (corrected for item-overlap) routinely exceed the conventional psychometric threshold of $r ge .40$, generally clustering between $r = .55$ and $r = .85$.
• Discriminant Validity: Individual items correlate significantly higher with their assigned construct than with competing subscales. In comprehensive validation cohorts analyzed by Ware, Kosinski, and Bjorner, item-scale success rates for discriminant validity approached 98% to 100% across all eight subscales.

Criterion and Predictive Validity

The predictive utility of the SF-36v2 has been verified against an array of objective clinical markers, biological variables, and epidemiological milestones:

• Mortality and Hospitalization: Longitudinal analyses indicate that the Physical Component Summary (PCS) and Mental Component Summary (MCS) independently predict five-year mortality, unexpected emergency department utilization, and hospital readmissions in elderly, cardiovascular, and end-stage renal disease (ESRD) cohorts, even after adjusting for comprehensive sociodemographic factors and clinical comorbidity indices (e.g., Charlson Comorbidity Index).
• Clinical Convergence: Physical Functioning scores correlate robustly ($r = .60$ to $.78$) with objective physiological indicators, such as the six-minute walk test (6MWT), maximum oxygen consumption ($VO_2$ max), and timed up-and-go (TUG) metrics.
• Psychiatric Convergence: The Mental Health (MH) and MCS scales demonstrate robust diagnostic accuracy in discriminating between individuals with and without DSM-classified clinical depression or anxiety disorders, with areas under the receiver operating characteristic curve (ROC-AUC) frequently exceeding $.85$.

8. Reliability

Extensive psychometric investigations across international normative samples, clinical trial datasets, and diverse patient registries corroborate the high reliability of the SF-36v2 across all eight subscales and its two summary measures.

Internal Consistency Reliability

In standard normative and disease cohorts, Cronbach’s alpha coefficients consistently surpass the established psychometric benchmark of α = .70 for group-level comparisons, with the majority surpassing α = .80, making them suitable for individual-level clinical decision-making:

• Physical Functioning (PF): α = .92 to .95
• Role-Physical (RP): α = .88 to .92 (a notable improvement over Version 1, which typically yielded α ≈ .80 to .84)
• Bodily Pain (BP): α = .82 to .89
• General Health (GH): α = .78 to .85
• Vitality (VT): α = .84 to .88
• Social Functioning (SF): α = .82 to .86
• Role-Emotional (RE): α = .85 to .90 (substantially elevated compared to Version 1’s α ≈ .75 to .81)
• Mental Health (MH): α = .83 to .88

For the higher-order component summaries, composite reliability coefficients routinely exceed $r = .90$ (typically .92 to .94 for the Physical Component Summary and .88 to .92 for the Mental Component Summary), confirming minimal measurement error.

Test-Retest Stability

In stable clinical and healthy cohorts assessed across two-week to one-month intervals, intraclass correlation coefficients (ICC) consistently span from $.75$ to $.90$, indicating that the instrument is resistant to temporal fluctuation when underlying physiological or psychological states remain stable.

9. Factor Analysis

The internal structural validity of the SF-36v2 has been subjected to rigorous exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) across numerous international investigations.

Second-Order Two-Factor Model

Both orthogonal and oblique exploratory factor analyses with varimax rotation reliably reproduce the classic two-factor higher-order structure that underlies the creation of the PCS and MCS scores:

• Factor 1 (Physical Dimension): Highest principal loadings emerge from Physical Functioning ($lambda > .80$), Role-Physical ($lambda > .75$), and Bodily Pain ($lambda > .70$), with moderate secondary contributions from General Health ($\lambda &\approx; .55$ to $.65$).
• Factor 2 (Mental Dimension): Dominated by high factor loadings from Mental Health ($lambda > .80$), Role-Emotional ($lambda > .75$), Social Functioning ($lambda > .65$), and Vitality ($lambda > .60$).
• Cross-Loading Scales: General Health, Vitality, and Social Functioning exhibit substantial, conceptually consistent cross-loadings across both higher-order factors. Vitality and Social Functioning represent bridge constructs connecting physical fatigue with affective malaise, whereas General Health synthesizes somatic disease with psychological evaluation.

Confirmatory Factor Analysis Fit Indices

Structural equation modeling evaluating the hypothesized correlated two-factor second-order structure yields robust goodness-of-fit indices across large-scale population studies (e.g., $N > 5,000$):

• Comparative Fit Index (CFI): $.94$ to $.97$
• Tucker-Lewis Index (TLI): $.93$ to $.96$
• Root Mean Square Error of Approximation (RMSEA): $.045$ to $.062$ (with 90% confidence intervals remaining firmly beneath the $.08$ cutoff for acceptable model fit)
• Standardized Root Mean Square Residual (SRMR): $.038$ to $.052$

Measurement invariance testing across sexes, age groups, and diverse cultural populations confirms strict metric and scalar invariance, justifying direct mean comparisons across broad demographic segments.

10. Instrument / Measurement Tool

• Test Type: Generic Patient-Reported Outcome Measure (PROM) / Self-Administered Multi-Item Health Survey.
• Administration Format: Standardized paper-and-pencil questionnaire, clinician- or interviewer-assisted script, telephone administration, or validated computerized adaptive testing (CAT) and electronic PROM (ePROM) platforms.
• Target Population: Adults (aged 18 years and older); specialized adolescent adaptations exist for youth aged 14–17.
• Administration Duration: Approximately 5 to 10 minutes for complete self-administration.
• Total Item Count: 36 items (35 contributing directly to the eight profiled scales, plus 1 unscaled health transition item).
• Response Formats: Varies by section: Item 1: 5-point scale (1=Excellent, 2=Very good, 3=Good, 4=Fair, 5=Poor); Item 2: 5-point scale (1=Much better now than one year ago to 5=Much worse now than one year ago); Items 3a-3j: 3-point scale (1=Yes, limited a lot; 2=Yes, limited a little; 3=No, not limited at all); Items 4a-4d & 5a-5c: 5-point scale (1=All of the time to 5=None of the time); Item 6: 5-point scale (1=Not at all to 5=Extremely); Items 7 & 8: 5-point / 6-point scales for pain severity and interference; Items 9a-9i: 5-point scale (1=All of the time to 5=None of the time); Item 10: 5-point scale (1=All of the time to 5=None of the time); Items 11a-11d: 5-point scale (1=Definitely true to 5=Definitely false).
• Scoring and Transformation Procedures: Scoring involves a standardized three-step sequence:
  1. Data Cleaning and Item Recoding: Ten negatively worded items (e.g., items measuring pain intensity, fatigue, nervous affect, health deterioration) are reverse-scored so that higher numerical values consistently denote superior health, greater vitality, or absence of functional limitation.
  2. Raw Scale Summation: Scale scores are generated by summing the individual item scores within each defined domain: Physical Functioning (PF), Role-Physical (RP), Bodily Pain (BP), General Health (GH), Vitality (VT), Social Functioning (SF), Role-Emotional (RE), and Mental Health (MH).
  3. Norm-Based Standardization (T-Score Transformation): Scale sums are transformed using linear algorithms anchored to representative national general population norms (typically the 1998 US General Population reference sample), yielding a standardized mean of 50 and a standard deviation of 10 for each domain.
  4. Component Summary Scores: The Physical Component Summary (PCS) and Mental Component Summary (MCS) are derived by multiplying each of the eight standardized scale scores by their corresponding factor score coefficients (derived via orthogonal principal components analysis), summing the products, and standardizing the resulting composite scores to the same $T$-score metric (Mean = 50, SD = 10).

11. Permissions & Fee and Test Year

• Publication History: Version 1 was introduced in 1992 (Ware & Sherbourne). The upgraded SF-36v2 Health Survey was officially released in 1996–2000 following psychometric refinements and the integration of updated normative datasets.
• Copyright & Intellectual Property: The SF-36v2 is a proprietary instrument protected by international copyright laws. The trademark and distribution rights are held and managed by QualityMetric Incorporated (an Optum company, Johnston, RI, USA).
• Licensing and Royalty Fees: The tool is not in the public domain. Researchers, clinical organizations, commercial sponsors, and healthcare systems must obtain a formal license agreement from QualityMetric or authorized distributors prior to clinical or academic deployment. Licensing fees are determined on a tiered schedule dependent upon study design, commercial vs. non-commercial non-profit funding, mode of administration, and translation requirements. Academic and student discounts are frequently available upon application.

12. References

• Ware, J. E., Jr., & Sherbourne, C. D. (1992). The MOS 36-item short-form health survey (SF-36): I. Conceptual framework and item selection. Medical Care, 30(6), 473–483. https://doi.org/10.1097/00005650-199206000-00002
• Ware, J. E., Jr., Kosinski, M., & Dewey, J. E. (2000). How to score Version 2 of the SF-36 Health Survey. QualityMetric Incorporated.
• Ware, J. E., Jr., Snow, K. K., & Kosinski, M. (2000). SF-36 Health Survey: Manual and interpretation guide. QualityMetric Incorporated.
• Ware, J. E., Jr., Kosinski, M., Bjorner, J. B., Turner-Bowker, D. M., Gandek, B., & Maruish, M. E. (2007). User’s manual for the SF-36v2 Health Survey (2nd ed.). QualityMetric Incorporated.
• Maruish, M. E. (Ed.). (2011). User’s manual for the SF-36v2 Health Survey (3rd ed.). QualityMetric Incorporated.
• McHorney, C. A., Ware, J. E., Jr., & Raczek, A. E. (1993). The MOS 36-Item Short-Form Health Survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Medical Care, 31(3), 247–263. https://doi.org/10.1097/00005650-199303000-00006
• Tarlov, A. R., Ware, J. E., Jr., Greenfield, S., Nelson, E. C., Perrin, E., & Zubkoff, M. (1989). The Medical Outcomes Study: An approach to evaluating the results of medical care. JAMA, 262(7), 925–930. https://doi.org/10.1001/jama.1989.03430070073033
• Gandek, B., Ware, J. E., Jr., Aaronson, N. K., Alonso, J., Apolone, G., Bjorner, J., Brazier, J., Bullinger, M., Kaasa, S., Leplege, A., & Sullivan, M. (1998). Cross-validation of item selection and scoring for the SF-12 Health Survey in nine countries: Results from the IQOLA Project. Journal of Clinical Epidemiology, 51(11), 1171–1178. https://doi.org/10.1016/S0895-4356(98)00109-7

13. Items of the Scale (Questionnaire)