Groningen Activity Restriction Scale

1. Abstract

The Groningen Activity Restriction Scale (GARS) is a standardized, patient-reported outcome measure developed to assess functional disability and restrictions in personal autonomy. Originally formulated in the Netherlands by Gertrudis I. J. M. Kempen, D. M. Doeglas, and Th. P. B. M. Suurmeijer in the early 1990s, the instrument captures an individual’s competence in executing vital daily tasks across two primary domains: Activities of Daily Living (ADL), centered on basic self-care, and Instrumental or Household Activities of Daily Living (IADL/HDL), centered on domestic autonomy. The standard version comprises 18 items (11 ADL items and 7 IADL/HDL items) administered via a multiple-choice response format, most commonly evaluated across four ordinal response levels (GARS-4), though dichotomous (GARS-2) and trichotomous (GARS-3) variants have been calibrated for specific epidemiological applications. Grounded psychometrically in Non-parametric Item Response Theory (IRT) and Mokken scale analysis, the GARS possesses robust hierarchical scaling properties, characterized by Loevinger scalability coefficients ($H$) typically exceeding 0.50, demonstrating a strong unidimensional continuum alongside distinct, interpretable sub-dimensions. Cross-national studies, notably through the European Research on Incapacitating Diseases and Social Support (EURIDISS) project, have validated its cross-cultural invariance, high internal consistency (Cronbach’s $\alpha$ frequently between .88 and .95), and pronounced construct validity across geriatric cohorts and patient populations with chronic physical morbidities such as rheumatoid arthritis, stroke, and cardiovascular diseases. The GARS serves as an essential tool in clinical gerontology, rehabilitation medicine, and health services research to evaluate loss of independence, guide institutional care placement, and quantify the allocation of home care resources.

2. Keywords

Groningen Activity Restriction Scale, GARS, activities of daily living, instrumental activities of daily living, functional disability, self-reliance, gerometrics, Mokken scaling, home care utilization, physical limitation, elderly assessment, EURIDISS.

3. Authors

The Groningen Activity Restriction Scale was designed and psychometrically evaluated by a multidisciplinary team of health services researchers and medical sociologists associated with the University of Groningen and Maastricht University in the Netherlands:

• Gertrudis I. J. M. (Rik) Kempen, PhD: Professor of Social Gerontology and Health Services Research at the Care and Public Health Research Institute (CAPHRI), Faculty of Health, Medicine and Life Sciences, Maastricht University, Maastricht, Netherlands.
• D. M. Doeglas, PhD: Researcher in Medical Sociology and Psychometrics, Northern Centre for Healthcare Research (NCH), University of Groningen, Groningen, Netherlands.
• Th. P. B. M. Suurmeijer, PhD: Emeritus Associate Professor of Medical Sociology and Health Services Research, Northern Centre for Healthcare Research (NCH), Department of Health Sciences, University of Groningen, Groningen, Netherlands.

Correspondence regarding the foundational validation studies and the ongoing developmental history of the instrument has historically been routed through the Department of Health Sciences at the University of Groningen and the Department of Health Promotion at Maastricht University.

4. Purpose

The overarching purpose of the Groningen Activity Restriction Scale is to provide a clinically sensitive, theoretically coherent, and psychometrically rigorous measure of self-reported functional limitations in both personal maintenance and domestic management. Developed within the context of population aging and the proliferation of chronic, debilitating medical conditions, the instrument addresses the pressing imperative in healthcare systems to evaluate an individual’s operational autonomy within their natural home environment. Unlike clinical performance batteries that assess maximal motor capability in an artificial, supervised setting, the GARS measures actual, self-perceived operational disability in everyday real-world routines.

In clinical practice, the GARS was specifically designed to evaluate the degree of dependency exhibited by vulnerable individuals—primarily community-dwelling older adults and patients suffering from degenerative, musculoskeletal, neuromuscular, or cardiopulmonary illnesses. A primary applied objective of the instrument is to establish need-based benchmarks for formal home care services, social support interventions, and municipal welfare accommodations. By delineating whether an individual can execute a routine task completely without difficulty, independently with some difficulty, independently with great difficulty, or solely with interpersonal or mechanical assistance, the scale provides clinicians and health policy administrators with an objective profile of unmet functional needs.

In health services and epidemiological research, the GARS functions as a dependent or explanatory variable across a broad range of investigative designs. It enables researchers to map the natural history of functional decline, monitor post-operative or rehabilitative recovery trajectories, investigate the disabling impacts of chronic multimorbidity, and evaluate the cost-effectiveness of geriatric healthcare programs. The deliberate separation and simultaneous integration of personal self-care (ADL) and instrumental household functioning (HDL) allows investigators to detect early-stage frailty—which typically manifests first in complex household activities—long before severe functional dependency compromises basic self-care activities such as feeding or personal hygiene.

5. Psychological Construct

The core construct captured by the Groningen Activity Restriction Scale is activity restriction, operationalized as the subjective perception of difficulty and dependence encountered when performing tasks central to autonomous community living. This construct resides at the intersection of psychomotor competence, psychological self-efficacy, and socio-environmental press, embodying how physical impairment translates into everyday behavioral disruption. The GARS operationalizes functional disability across two deeply interrelated yet functionally distinct sub-dimensions:

1. Basic Activities of Daily Living (ADL)

This subscale encompasses 11 items assessing fundamental bodily upkeep and primary functional mobility. These items reflect survival-level, self-oriented behaviors that develop early in human ontogeny and are typically lost in the advanced stages of physical or cognitive deterioration. The construct spans:

• Self-Care Competence: Tasks such as washing hands and face, washing the entire body, brushing hair or shaving, brushing teeth, and getting dressed and undressed.
• Basic Axial and Lower-Limb Mobility: Dynamic postural transitions, including getting into and out of bed, standing up from a seated chair position, and executing personal mobility within the interior environment (walking indoors).
• Nutritional Ingestion and Elimination: Eating and drinking independently, as well as managing personal hygiene routines while using the toilet.

Difficulty in this domain reflects severe functional compromise, representing a direct breakdown of basic somatic self-sufficiency and signifying an acute vulnerability to institutionalization.

2. Instrumental and Household Activities of Daily Living (IADL/HDL)

This subscale comprises 7 items assessing domestic, context-dependent activities requiring a higher degree of cognitive organization, environmental interaction, physical stamina, and complex motor coordination. The construct encompasses:

• Culinary Preparation: Preparing breakfast or lunch, and cooking a complete warm dinner.
• Environmental Cleaning and Upkeep: Executing light domestic chores (dusting, washing dishes) alongside heavy household labor (scrubbing floors, turning mattresses, window cleaning).
• Garment Maintenance: Washing clothes and ironing clothing.
• Community Interface: Venturing out of the house into the neighborhood, including negotiating stairs, curbs, and transporting purchases (shopping).

Because household management tasks demand broader physical exertion and cognitive planning, restrictions in the HDL subscale generally precede deficits in basic ADLs. Consequently, the GARS measures a continuum that spans from incipient vulnerability to total physical dependency.

6. Theoretical Framework

The conceptual architecture of the GARS is grounded in classical sociomedical models of disability, specifically Saad Nagi’s Disablement Scheme (1965, 1991), the World Health Organization’s International Classification of Impairments, Disabilities, and Handicaps (ICIDH) (1980), and its contemporary evolution, the International Classification of Functioning, Disability and Health (ICF) (2001). Furthermore, the construct incorporates the environmental press paradigm articulated in Lawton and Nahemow’s ecological model of aging (1973).

Under Nagi’s framework, an active pathology leads to anatomical or structural impairment, which in turn induces functional limitation in generic organ systems or physiological tasks (e.g., reduced muscular torque, restricted joint range of motion). When these limitations impede the execution of normative, socially defined roles and essential tasks of independent living within a given physical environment, disability emerges. The GARS measures disability directly at the interface of functional limitation and environmental demand. It avoids confounding impairment (e.g., knee flexion angle) with disability (e.g., inability to stand up from a chair), anchoring its diagnostic gaze strictly on the behavioral consequence of disease within domestic life.

From a psychometric perspective, the GARS is theoretically anchored in Mokken Scale Analysis (MSA), a non-parametric latent trait model within the family of Item Response Theory. Traditional classical test theory assumes linear relationships and invariant standard errors of measurement across the score continuum. In contrast, Mokken’s model of Monotone Homogeneity assumes that all items measure a single underlying latent continuum ($\theta$, representing functional capacity) and that the item response functions are monotonically non-decreasing. Crucially, the GARS was designed to satisfy Mokken’s more stringent model of Double Monotonicity, meaning that the item response functions do not intersect across varying levels of the latent trait. This mathematical property guarantees an invariant, cumulative hierarchy of item difficulty: if an individual is restricted in an easy item (such as washing one’s face), the model predicts with high probability that they will also experience difficulty in more challenging items (such as washing the whole body or executing heavy household chores). This hierarchical scaling provides an empirically justified total score representing an invariant progression of functional restriction.

7. Validity

The validity of the GARS has been examined extensively through rigorous psychometric testing across diverse clinical cohorts and cross-cultural populations.

Construct and Structural Validity

Structural validity was originally established using non-parametric IRT models. In the foundational validation studies by Kempen, Suurmeijer, and Doeglas, the cumulative scalability of the GARS was demonstrated through Loevinger’s scalability coefficients ($H$). For the overall 18-item scale, the total $H$ coefficient consistently surpassed the critical psychometric threshold of .40, frequently achieving values between .50 and .60, which signifies a highly scalable, robust hierarchical instrument. Individual item scalability coefficients ($H_i$) regularly exceed .45, verifying that every constituent item contributes uniquely and substantially to the underlying construct of activity restriction without violating invariant item ordering.

Convergent and Criterion Validity

Convergent validity has been repeatedly corroborated through strong statistical correlations with established external measures of physical function, clinical disease activity, and physiological parameters:

• Health Assessment Questionnaire (HAQ): In cohorts of patients with rheumatoid arthritis, GARS scores demonstrate robust correlations with the disability index of the Stanford Health Assessment Questionnaire, with Pearson and Spearman correlation coefficients consistently falling in the $r = .75$ to $.86$ range ($p < .001$).
• Short Form Health Survey (SF-36): The GARS correlates strongly and negatively with the Physical Functioning (PF) subscale of the SF-36 ($r = -.70$ to $-.82$), while displaying significantly lower correlations with the Mental Health (MH) and Role-Emotional (RE) subscales ($r = -.25$ to $-.38$), providing rigorous evidence of both convergent and divergent/discriminant validity.
• Barthel Index: Among geriatric stroke rehabilitation patients, the ADL component of the GARS shows high convergent alignment with the observer-rated Barthel Index ($r = -.78$ to $-.85$), confirming that self-reported assessments on the GARS parallel objective clinical evaluations of motor competence.
• Clinical and Objective Biomarkers: GARS scores correlate significantly with objective physical performance measures, including grip strength measured via dynamometry ($r = -.45$ to $-.58$), timed chair-stand tests, walking velocity over standardized distances, and joint tenderness scores (Ritchie Articular Index).

Predictive and Ecological Validity

Longitudinal investigations have demonstrated that elevated baseline GARS scores strongly predict critical downstream clinical outcomes, including increased demand for municipal home care services, informal caregiver burden, emergency hospitalization, institutionalization in nursing homes, and elevated all-cause mortality over 3- and 5-year follow-up intervals. These predictive relationships remain robust even after controlling for chronological age, gender, and comorbidity indices.

Cross-Cultural Validity

Within the multi-center European Research on Incapacitating Diseases and Social Support (EURIDISS) project, the cross-cultural equivalence of the GARS was evaluated across cohorts in the Netherlands, the United Kingdom, France, and Norway. Utilizing structural equation modeling and differential item functioning (DIF) analyses, the researchers demonstrated that the dimensional architecture, item loadings, and hierarchical difficulty patterns remained stable across international contexts, confirming its utility for cross-national comparative research.

8. Reliability

The reliability of the Groningen Activity Restriction Scale has been demonstrated across diverse patient groups, testing formats, and cultural settings.

Internal Consistency

Internal consistency estimates for both the full instrument and its constituent subscales are exceptionally high:

• Overall 18-Item Scale: Cronbach’s alpha ($\alpha$) typically ranges from .92 to .96 among community-dwelling older adults and chronic illness populations.
• ADL Subscale (11 items): Chronbach’s alpha values consistently fall between .88 and .93.
• IADL/HDL Subscale (7 items): Internal consistency metrics for this subscale range from .86 to .91.
• Mokken Rho ($rho$): Because the GARS satisfies non-parametric IRT assumptions, internal consistency has also been evaluated via Mokken’s reliability coefficient $rho$ (analogous to the IRT separation reliability), yielding coefficients consistently $ge .90$.

Test-Retest Stability

In stable clinical cohorts evaluated across test-retest intervals spanning 14 to 28 days, the GARS demonstrates high temporal reproducibility:

• The Intraclass Correlation Coefficient (ICC) for the overall scale score regularly ranges between .85 and .93 under standard self-administration conditions.
• The ADL subscale exhibits test-retest stability coefficients exceeding $\text{ICC} = .88$.
• The IADL/HDL subscale demonstrates test-retest stability between $\text{ICC} = .83$ and $.89$, reflecting minor fluctuations attributable to variations in household environmental demands.

Inter-Rater and Proxy Agreement

When evaluated in proxy-reporting formats (e.g., comparing self-reports from frail older adults with evaluations provided by primary informal caregivers or visiting community nurses), weighted kappa ($\kappa_w$) values for individual items vary from .62 to .81, and overall score ICCs range between .78 and .86. This demonstrates strong concordance, confirming the reliability of proxy administrations when patient cognitive impairment precludes self-report.

9. Factor Analysis

The latent dimensionality of the Groningen Activity Restriction Scale has been scrutinized using both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA), alongside non-parametric item response approaches.

Exploratory Factor Analysis (EFA)

Initial principal components analyses and maximum likelihood exploratory factor extractions with oblique (Promax or Oblimin) rotation consistently yield a distinct two-factor solution explaining between 58% and 68% of the total variance:

• Factor 1: Basic Self-Care / Activities of Daily Living (ADL): Comprising items 1 through 11. Salient factor loadings for these items typically range from .56 to .88. Items such as ‘washing your body’, ‘dressing and undressing’, and ‘getting in and out of bed’ load heavily onto this dimension.
• Factor 2: Instrumental / Household Activities of Daily Living (HDL): Comprising items 12 through 18. Factor loadings range from .61 to .89. Items including ‘heavy domestic work’, ‘washing clothes’, and ‘preparing a meal’ load substantially onto this domestic management component.
• Factor Inter-correlation: The correlation between the two factors is substantial, typically ranging from $r = .65$ to $.78$, indicating that both domains are manifestations of an overarching second-order physical disability construct.

Confirmatory Factor Analysis (CFA)

Subsequent structural equation modeling has tested competing structural specifications: a strictly unidimensional model, a two-factor uncorrelated model, a two-factor correlated model, and a hierarchical bi-factor model:

• Two-Factor Correlated Model: Demonstrates the best fit across both older adult cohorts and chronic disease populations. Representative model fit indices include: Comparative Fit Index ($\text{CFI}$) = .96 to .98; Tucker-Lewis Index ($\text{TLI}$) = .95 to .97; Root Mean Square Error of Approximation ($\text{RMSEA}$) = .045 to .058 (90% CI [.039, .064]); and Standardized Root Mean Square Residual ($\text{SRMR}$) = .038 to .044.
• Unidimensional Model: While the single-factor model demonstrates acceptable fit in some homogeneous cohorts with severe functional deficits ($\text{CFI} \approx .91$, $\text{RMSEA} \approx .080$), the two-factor specification provides a statistically superior fit ($Delta chi^2, p < .001$), supporting the conceptual distinction between personal care and household maintenance.

Mokken Scaling Dimensionality

Automated Item Selection Procedures (AISP) in Mokken scale analysis demonstrate that at standard threshold settings ($c = .30$), all 18 items load onto a single cumulative scale ($H > .45$). However, when applying higher partition thresholds ($c = .45$ to $.50$), the AISP neatly bifurcates the item pool into its two constituent clinical subscales: the 11 ADL items and the 7 HDL items. This psychometric duality validates using both a unified single composite score and distinct subscale metrics depending on the clinical or research objective.

10. Instrument / Measurement Tool

The Groningen Activity Restriction Scale is structured as follows:

• Instrument Type: Patient-Reported Outcome Measure (PROM); also validated for face-to-face clinical interviews and proxy/informant assessment.
• Target Population: Adults and older adults (aged 18 and older, predominantly utilized in populations aged 60+), chronically ill individuals, rehabilitation patients, and recipients of home care services.
• Administration Modality: Self-administered paper-and-pencil questionnaire, digital/tablet-based online interface, or structured telephone/in-person interviewer administration.
• Administration Time: Approximately 5 to 10 minutes to complete the full 18-item inventory.
• Number of Items: 18 items partitioned into two functional subscales:
  • ADL Subscale: 11 items focusing on personal self-care, bodily hygiene, and basic indoor mobility.
  • IADL / HDL Subscale: 7 items focusing on meal preparation, domestic cleaning, garment upkeep, and community mobility.
• Response Formats:
  • GARS-4 (Standard 4-Category Format):
    • 1 = Yes, I can do this fully independently without any difficulty
    • 2 = Yes, I can do this fully independently but with some difficulty
    • 3 = Yes, I can do this fully independently but with great difficulty
    • 4 = No, I cannot do this independently / only with help from others
  • GARS-3 (Trichotomous Format): 1 = Fully independently without difficulty; 2 = Independently with difficulty; 3 = Only with help.
  • GARS-2 (Dichotomous Format): 1 = Independently; 2 = Only with help.
• Scoring Procedures:
  • Total Score Range (GARS-4): 18 to 72 points. Higher scores indicate greater functional restriction and higher dependency.
  • ADL Subscale Range: 11 to 44 points.
  • IADL / HDL Subscale Range: 7 to 28 points.
  • Missing Data Imputation: If fewer than 2 items are omitted across the total scale, personal mean imputation based on completed items within the respective subscale is psychometrically acceptable. If $ge 3$ items are missing, the overall scale score is considered invalid.

11. Permissions & Fee and Test Year

• Test Year of Publication: 1990 (initial developmental Dutch reports and validation studies by Kempen, Doeglas, & Suurmeijer; formal psychometric English dissemination in 1993 and 1996).
• Licensing and Copyright: The scale was established as an open-access academic research tool funded by university and national research grant programs in the Netherlands. The copyright is held by the original authors and the Northern Centre for Healthcare Research (NCH), University of Groningen / Maastricht University.
• Fee: The instrument is available free of charge for non-commercial academic research, public health evaluations, epidemiological investigations, and individual clinical practice.
• Permissions: Formal permission is generally not required for non-profit scholarly investigations, provided that the foundational methodological papers are appropriately cited. Commercial organizations, pharmaceutical clinical trials, and proprietary healthcare software vendors seeking to integrate the instrument into digital platforms should contact the original developers or the relevant university knowledge transfer offices for formal clearance.

12. References

• Doeglas, D. M., Suurmeijer, T. P. B. M., Briançon, S., Moum, T., Duchêne, B., & van den Heuvel, W. J. A. (1995). An international study on functional disability in patients with rheumatoid arthritis: The construction of a subscale for the Groningen Activity Restriction Scale. Journal of Clinical Epidemiology, 48(5), 725–733. https://doi.org/10.1016/0895-4356(94)00171-A
• Kempen, G. I. J. M., Doeglas, D. M., & Suurmeijer, T. P. B. M. (1993). The Groningen Activity Restriction Scale (GARS): Manual. Northern Centre for Healthcare Research (NCH), University of Groningen.
• Kempen, G. I. J. M., & Suurmeijer, T. P. B. M. (1990). The development of a hierarchical polychotomous ADL-IADL scale for noninstitutionalized frail elderly. The Gerontologist, 30(4), 497–502. https://doi.org/10.1093/geront/30.4.497
• Kempen, G. I. J. M., Miedema, I., Ormel, J., & Molenaar, W. (1996). The assessment of disability with the Groningen Activity Restriction Scale: Conceptual framework and psychometric properties. Social Science & Medicine, 43(11), 1601–1610. https://doi.org/10.1016/0277-9536(96)00057-3
• Kempen, G. I. J. M., Myers, A. M., & Powell, L. E. (1995). Cross-cultural adaptation of psychometric instruments in aging research: The Groningen Activity Restriction Scale and the Activities-specific Balance Confidence Scale. Canadian Journal on Aging / La Revue Canadienne du Vieillissement, 14(S1), 154–168. https://doi.org/10.1017/S071498080000569X
• Lawton, M. P., & Brody, E. M. (1969). Assessment of older people: Self-maintaining and instrumental activities of daily living. The Gerontologist, 9(3_Part_1), 179–186. https://doi.org/10.1093/geront/9.3_Part_1.179
• Mokken, R. J. (1971). A theory and procedure of scale analysis: With applications in political research. Walter de Gruyter. https://doi.org/10.1515/9783110813203
• Nagi, S. Z. (1991). Disability concepts revisited: Implications for prevention. In A. M. Pope & A. R. Tarlov (Eds.), Disability in America: Toward a national agenda for prevention (pp. 309–327). National Academy Press.
• Suurmeijer, T. P. B. M., Doeglas, D. M., Moum, T., Briançon, S., Krol, B., Sanderman, R., Guillemin, F., Bjelle, A., & van den Heuvel, W. J. A. (1994). The Groningen Activity Restriction Scale for measuring disability: Its utility in international comparisons. American Journal of Public Health, 84(8), 1270–1273. https://doi.org/10.2105/AJPH.84.8.1270
• Verbrugge, L. M., & Jette, A. M. (1994). The disablement process. Social Science & Medicine, 38(1), 1–14. https://doi.org/10.1016/0277-9536(94)90294-1
• World Health Organization. (2001). International Classification of Functioning, Disability and Health (ICF). World Health Organization.

13. Items of the Scale