ABILHAND

1. Abstract

The ABILHAND questionnaire is a standardized, patient-reported outcome measure (PROM) developed through modern psychometric theory to evaluate manual ability in everyday bilateral activities. Originating from the work of Massimo Penta, Jean-Louis Thonnard, and Luigi Tesio at the Université Catholique de Louvain, the instrument was conceived to overcome the severe methodological constraints of classical ordinal rating scales in clinical neurorehabilitation and rheumatology. By operationalizing the Rasch measurement model, ABILHAND converts discrete, ordinal observations of personal difficulty into an invariant, linear, interval-level metric of perceived manual ability calibrated in log-odds units (logits).

The scale focuses specifically on perceived manual ability, defined as the capacity to manage daily activities requiring the coordinated engagement of both hands and upper limbs, irrespective of the compensatory strategies, assistive devices, or biological mechanisms employed by the individual. The adult stroke and neurorehabilitation version comprises 23 carefully calibrated bilateral tasks, such as fastening a zipper, buttoning a shirt, opening a screw-top jar, peeling onions, and unwrapping a chocolate bar. Responses are registered via a semi-structured interview protocol across a 3-point rating continuum: “Impossible” (score 0), “Difficult” (score 1), and “Easy” (score 2); activities that have not been attempted or performed over the prior three months are recorded as missing/unknown to prevent speculative bias.

Extensive psychometric investigations across diverse pathological cohorts—including cerebrovascular stroke, rheumatoid arthritis, systemic sclerosis, multiple sclerosis, and peripheral neuropathies—consistently confirm its rigorous unidimensionality, high person separation reliability (typically ranging from 0.88 to 0.94), and exceptional test-retest reproducibility (intraclass correlation coefficients commonly exceeding 0.90). ABILHAND exemplifies the state of the art in item response theory applications, offering clinicians and clinical researchers an objective, responsive, and cross-culturally validated measurement tool that interfaces seamlessly with the International Classification of Functioning, Disability and Health (ICF) activity domain.

2. Keywords

ABILHAND, manual ability, Rasch analysis, item response theory, upper extremity, bilateral activities, stroke rehabilitation, rheumatoid arthritis, patient-reported outcome measures, psychometrics

3. Authors

The ABILHAND measurement system was conceived and developed by an interdisciplinary consortium of researchers specializing in physical medicine, neurophysiology, rehabilitation sciences, and psychometrics at the Université Catholique de Louvain (Brussels and Louvain-la-Neuve, Belgium) in collaboration with European clinical research centers:

• Massimo Penta, PhD: Lead psychometrician and biomedical engineer, Laboratoire de Rééducation et Readaptation, Faculté des Sciences de la Motricité, Université Catholique de Louvain, Belgium. Dr. Penta spearheaded the mathematical formulation and software development that underpinned the Rasch transformation of rehabilitation rating scales.
• Jean-Louis Thonnard, PhD: Professor of Neurophysiology and Physical Medicine, Institute of Neuroscience (IoNS), Faculté des Sciences de la Motricité, Université Catholique de Louvain, Brussels, Belgium. Dr. Thonnard has spent decades directing empirical research into human sensorimotor control, precision grip dynamics, and functional assessment of the human upper limb.
• Luigi Tesio, MD: Professor of Physical Medicine and Rehabilitation, Department of Biomedical Sciences for Health, Universita degli Studi di Milano, and Director of the Department of Neurorehabilitation Sciences, Istituto Auxologico Italiano IRCCS, Milan, Italy. Dr. Tesio is a world-renowned authority on physical and rehabilitation medicine, clinical biomechanics, and the theoretical foundations of Rasch calibration in outcome measurement.

Subsequent psychometric adaptations for specific clinical populations, such as ABILHAND-Kids, were developed by Laurent Vandervelde, Yannick Bleyenheuft, and Jean-Louis Thonnard; the rheumatoid arthritis model was advanced by Patrick Durez and collaborators; and the systemic sclerosis validation was pioneered by Florence Vanderveken and colleagues.

4. Purpose

The primary purpose of the ABILHAND scale is to quantify an individual’s perceived manual ability when executing common manual tasks in daily life. Unlike objective, performance-based impairment metrics—such as dynamometric handgrip strength measurements, pinch force assessments, goniometric joint range of motion measurements, or laboratory-based timed pegboard tests (e.g., the Box and Block Test, Nine-Hole Peg Test)—ABILHAND captures the ecological reality of functional hand use from the client’s own perspective. The instrument operates precisely at the “Activity” level of the World Health Organization’s International Classification of Functioning, Disability and Health (ICF), bridging the critical disconnect between anatomical-neurological impairment and authentic participation in society.

Manual ability cannot be directly observed through isolated motor execution because human bimanual coordination is fundamentally adaptive. A patient surviving a devastating stroke or managing severe systemic joint deformities due to rheumatoid arthritis might exhibit profound hemiparesis or articular subluxation, yet successfully navigate their environment using compensatory postures, contralateral limb stabilization, trunk leaning, or adaptive assistive tools. Conversely, an individual with subtle sensory deficits or motor apraxia might demonstrate near-normal muscle strength on a clinical examination table but remain utterly unable to coordinate the fine bi-digital forces necessary to fasten buttons or slice food. ABILHAND measures what the patient actually experiences as possible, difficult, or impossible across this dynamic continuum.

Clinically, ABILHAND serves three vital functions. First, it enables precise baseline diagnostics during occupational and physical therapy intake evaluations, mapping the patient’s unique location along a standardized continuum of functional ability. Second, it serves as a sensitive, linear outcome measure capable of detecting genuine rehabilitation-induced functional gains, distinguishing authentic functional recovery from ordinal scaling artifacts or floor/ceiling distortions. Third, it guides individual goal setting by highlighting which daily manual tasks fall right at the boundary of a patient’s current capabilities—the zone of proximal development in neurorehabilitation—allowing therapists to formulate targeted, client-centered intervention plans.

In research arenas, ABILHAND offers an indispensable methodological advantage: it supplies continuous, interval-scale metric data that comply with the mathematical requirements of parametric statistical hypothesis testing. Clinical trials investigating novel pharmacological agents, constraint-induced movement therapies (CIMT), robotic-assisted arm training, neuromodulation, or surgical interventions can employ ABILHAND logits in structural equation models, repeated-measures analysis of variance, and longitudinal mixed-effects regression without violating statistical assumptions.

5. Psychological and Functional Construct

The construct targeted by the ABILHAND scale is manual ability (French: capacité manuelle). While traditionally categorized under physical medicine and occupational therapy, manual ability is a complex neuro-behavioral and functional construct that synthesizes motor competence, sensory feedback, cognitive planning, spatial orientation, executive problem-solving, and emotional resilience in the face of perceived challenge.

The construct is grounded on three critical operational definitions:

• Unidimensionality of Manual Performance: The construct posits that all targeted manual tasks, regardless of their specific biomechanical gestures (e.g., rotational torque for unscrewing jars, fine bi-digital precision grip for buttoning, lateral pinch for inserting keys, bilateral traction for pulling down zippers), measure a single underlying, latent trait: the general capacity to use one’s hands together to interact with physical objects in daily life.
• Bilateral Nature of Hand Function: Over 85% of meaningful ecological activities performed by adults require bilateral hand coordination. One hand serves as a stabilizing, supportive, or orienting anchor (frequently the nondominant or more affected limb), while the contralateral hand performs active manipulation, rotation, or fine adjustments. ABILHAND explicitly restricts its domain to tasks that intrinsically require both upper extremities, thereby capturing real-world functional synergy rather than unimanual motor capacity.
• Perceived Difficulty Relative to Environmental Constraints: The construct operationalizes ability as the inverse of task difficulty. An individual with high manual ability perceives challenging manual tasks (e.g., threading a needle, peeling apples with a knife) as straightforward, whereas an individual with diminished manual ability encounters severe friction and frustration even when confronting relatively undemanding tasks (e.g., washing one’s hands, turning on a light switch, tearing open a packet of sugar).

To preserve construct integrity, the administration instructions demand that the patient rate their ability regardless of the limb they use predominantly, the biomechanical strategies adopted, or the assistive aids utilized. The construct is intentionally invariant to biological compensation: if an individual manages to uncap a tube of toothpaste using their teeth or stabilization between knees and hands with ease, the activity is scored as “Easy” because their functional manual autonomy in that activity is preserved. The construct measures functional task autonomy rather than kinematic normality.

6. Theoretical Framework

The structural and conceptual architecture of the ABILHAND questionnaire is grounded in Item Response Theory (IRT), specifically the measurement framework established by Danish mathematician Georg Rasch (1960). In traditional classical test theory (CTT), clinical assessment tools typically rely on summing ordinal rating scores (e.g., Likert items) to generate a raw composite score. However, modern psychometricians have established that summing ordinal categories introduces profound mathematical errors: ordinal categories possess unequal interval distances, raw scores are inherently sample-dependent and item-dependent, and the measurement scale suffers from unpredictable floor and ceiling distortions.

The Rasch model provides the mathematical foundation necessary to transform bounded ordinal observations into an objective, invariant metric. The formulation applied in ABILHAND is the polytomous Rating Scale Model (Andrich, 1978) or the Partial Credit Model (Masters, 1982). Under this formulation, the probability $P_{nijk}$ that a person $n$ with latent manual ability $\beta_n$ responds in category $k$ (rather than $k-1$) on task item $i$ having difficulty $\delta_i$ and category threshold $\tau_k$ is expressed as:

ln [ P_nik / P_ni(k-1) ] = β_n − δ_i − τ_k

This mathematical formulation embodies several foundational scientific assumptions:

• Specific Objectivity: The comparison between two patients’ manual abilities is completely independent of the particular subset of calibrated items administered. Reciprocally, the calibration of item difficulties is independent of the distribution of ability within the sampled patient population.
• Conjoint Additivity: Person ability and item difficulty are mapped onto the exact same linear continuum, measured in identical mathematical units called logits (logarithmic odds ratios). A logit of 0 is traditionally centered on the mean difficulty of the item set.
• Item Invariance Across Subgroups: A fundamental postulate of the Rasch paradigm is that an item must preserve identical calibration across demographic sub-strata, such as sex, age cohorts, etiology of neurological damage, and dominant limb involvement. This invariance is empirically tested through the analysis of Differential Item Functioning (DIF).

By rooting the instrument in Rasch measurement theory, the creators ensured that ABILHAND functions analogously to a physical measurement device (such as a thermometer or meter stick). When a client advances from -1.0 logit to +0.5 logits following two months of intensive constraint-induced movement therapy, that change represents an identical, verifiable magnitude of functional gain regardless of whether the baseline measurement was taken at -3.0 logits or 0.0 logits.

7. Validity

The construct, criterion, convergent, and discriminant validity of the ABILHAND scale have been thoroughly scrutinized across multiple decades of empirical research worldwide.

Construct and Internal Validity

Construct validity in a Rasch framework is primarily demonstrated by the fit of observational data to the mathematical expectations of the Rasch model. In the definitive stroke calibration study conducted by Penta et al. (2001) comprising 113 stroke survivors, all 23 items demonstrated excellent fit to the unidimensional continuum. Item goodness-of-fit was verified using normalized mean square residuals (Infit and Outfit statistics). For all retained items, Infit and Outfit Mean Square (MnSq) statistics consistently fell within the established quality boundaries of 0.70 to 1.30, confirming that the responses were neither unpredictably erratic (noise) nor artificially redundant (overfit).

Convergent and Concurrent Validity

Extensive studies have demonstrated strong, statistically significant correlations between ABILHAND logit scores and established measures of upper extremity motor performance, physical impairment, and activities of daily living (ADL):

• Fugl-Meyer Assessment (Upper Extremity): ABILHAND demonstrates robust positive correlations with the upper extremity subscale of the Fugl-Meyer Motor Assessment ($r = 0.65$ to $0.82$, $p < .001$), confirming that greater motor recovery and reduced synergistic constraint directly foster perceived manual competence in bilateral contexts.
• Action Research Arm Test (ARAT): Significant positive associations are consistently documented between ABILHAND and the ARAT ($r = 0.68$ to $0.79$, $p < .001$), indicating that fine grasping, gripping, and pinch capabilities assessed in laboratory simulations directly translate into self-reported daily bilateral mastery.
• Functional Independence Measure (FIM) and Barthel Index: ABILHAND correlates moderately to highly with self-care and functional motor subscores of the FIM ($r = 0.58$ to $0.74$) and Barthel Index ($r = 0.55$ to $0.70$), while demonstrating weak or negligible correlations with cognitive and sphincter management subscales, confirming sharp convergent validity alongside divergence from non-motor domains.
• Rheumatoid Arthritis Measures: In patients with rheumatoid arthritis (Penta et al., 1998; Durez et al., 2007), ABILHAND logit scores correlated strongly with the Health Assessment Questionnaire (HAQ) disability index ($r = -0.76$ to $-0.85$, $p < .001$), grip strength, and disease activity parameters.

Discriminant and Known-Groups Validity

ABILHAND demonstrates exemplary discriminant validity. In clinical cohort studies, the instrument successfully differentiates between healthy older adults and patients presenting with unilateral upper limb impairment following ischemic or hemorrhagic stroke ($p < .001$). Furthermore, the instrument effectively discriminates between distinct clinical subgroups categorized by the severity of motor paralysis (e.g., mild hemiparesis versus severe flaccid plegia) and stages of manual dexterity classification.

8. Reliability

The reliability of the ABILHAND questionnaire has been established using both classical psychometric indices and Rasch-specific separation statistics across multiple international clinical trials and epidemiological cohorts.

Person and Item Separation Reliability

Within the Rasch measurement framework, classical reliability metrics such as Cronbach’s alpha are largely replaced by the Person Separation Index ($G_p$) and Person Separation Reliability ($R_p$). While Cronbach’s alpha assumes an ordinal raw-score scale and normal data distribution, Rasch reliability quantifies the precision of person calibration across the linear logit scale relative to measurement error:

• Person Separation Reliability ($R_p$): Across validation cohorts in adult stroke (Penta et al., 2001; Vandervelde et al., 2008), $R_p$ values consistently range between 0.88 and 0.94. This exceptional reliability level indicates that the 23 items successfully divide the patient continuum into at least four to five distinct, statistically reliable strata of manual ability.
• Item Separation Reliability ($R_i$): The item reliability index regularly exceeds 0.95 to 0.98, establishing that the hierarchical calibration of the 23 task difficulties along the latent metric is exceptionally robust and will reproduce identical hierarchical orderings across independent clinical samples.
• Classical Internal Consistency: When evaluated via classical test theory, ABILHAND exhibits extraordinary internal consistency, with Cronbach’s alpha coefficients consistently reported between 0.92 and 0.96 across diverse neurological and rheumatological samples.

Test-Retest Reliability and Reproducibility

Temporal stability has been verified across test-retest interval periods ranging from 7 days to 4 weeks in stable chronic stroke survivors and patients with stable connective tissue disorders:

• Intraclass Correlation Coefficient (ICC): Test-retest reproducibility for the interval logit scores demonstrates outstanding stability, with two-way mixed-effects ICC coefficients typically documented at 0.91 to 0.96 ($95%\text{ CI } [0.87, 0.98]$).
• Standard Error of Measurement (SEM) & Smallest Detectable Change (SDC): The standard error of measurement across the logit continuum is typically reported around 0.25 to 0.35 logits. The resulting smallest detectable change (SDC at the individual level with $95%$ confidence) approximates 0.70 to 0.85 logits. Clinicians can confidently interpret an observed individual improvement greater than 0.80 logits as true biological and functional progress beyond measurement noise.

9. Factor Analysis and Dimensionality Verification

The verification of strict unidimensionality is the non-negotiable prerequisite for constructing a valid Rasch measurement scale. Because factor analysis on raw ordinal categories yields spurious difficulty factors and distorted eigenvalues, modern validation studies of ABILHAND employ specialized psychometric procedures: Principal Component Analysis of Rasch Residuals (PCAR) combined with Confirmatory Factor Analysis (CFA) tailored for polytomous categorical indicators.

Principal Component Analysis of Rasch Residuals (PCAR)

Following the estimation of the primary Rasch dimension (the manual ability construct), the variance unexplained by the measurement model (the residuals) is extracted and subjected to principal component analysis. If the items conform to strict unidimensionality, the residuals must represent random, uncorrelated white noise without meaningful systematic structure:

• Empirical Variance Explained by the Measure: The primary latent trait of manual ability typically explains between 58% and 68% of the total variance in response observations, far exceeding the standard psychometric threshold of 50%.
• Eigenvalue of the First Contrast in Residuals: Across empirical studies, the first secondary component extracted from the residuals systematically exhibits an eigenvalue below 2.0 (typically ranging between 1.4 and 1.8), with unexplained variance in the first contrast accounting for less than 5% of the total variance. This definitively proves the absence of a secondary meaningful dimension or subscale.

Local Item Independence and Residual Correlations

Local independence requires that, once the common latent trait of manual ability is controlled for, responses to any given task are completely independent of responses to any other task. Pairwise residual correlation matrices are examined to ensure that no two items share excessive unexplained variance. Across the 23 items of the adult stroke calibration, residual correlation coefficients rarely exceed 0.20, with none crossing the critical threshold of 0.30, confirming that items do not possess redundant content or contextual chaining effects.

Differential Item Functioning (DIF)

Extensive factor invariance evaluations have been conducted to determine whether item difficulty calibrations shift significantly across demographic and clinical covariates. Analyses of variance on standardized residuals across participant sex, age categories (<65 vs. ≥65 years), hemiparetic side (left vs. right hemisphere lesion), and dominant hand involvement revealed no pervasive, statistically meaningful uniform or non-uniform Differential Item Functioning (DIF), demonstrating the universal structural stability of the scale across diverse clinical presentations.

10. Instrument / Measurement Tool

The ABILHAND measurement system is structured as a standardized, semi-structured clinical interview administered by an occupational therapist, physical therapist, neuropsychologist, or trained clinical researcher.

• Assessment Classification: Patient-Reported Outcome Measure (PROM) / Clinician-Administered Structured Interview.
• Target Demographics: Adult patients experiencing functional upper limb impairments resulting from cerebrovascular accidents (ischemic and hemorrhagic stroke), traumatic brain injury, multiple sclerosis, rheumatoid arthritis, systemic sclerosis, peripheral neuropathies, or geriatric frailty.
• Assessed Anatomical Region: Bilateral Upper Extremities (cooperative interaction between hands and arms).
• Total Item Count: 23 calibrated bilateral manual activities in the standard adult stroke instrument (alternative versions include 27 items for rheumatoid arthritis and 21 items in the ABILHAND-Kids adaptation for pediatric cerebral palsy).
• Item Randomization Protocol: The 23 task items are presented to the patient in a randomized order to prevent systematic anchoring, fatigue bias, or learning trajectory distortions during the interview. The administrator utilizes a pre-printed randomized record form.
• Observation Window: The client evaluates their difficulty in performing each task as experienced within the preceding 3 months.
• Response Categories: A 3-level ordinal response scale is utilized:
  • 0 = Impossible: The patient cannot perform the activity under any circumstances, even when using alternative grips, personal compensatory strategies, or regular non-specialized household aids; performance requires total human assistance.
  • 1 = Difficult: The patient can successfully accomplish the activity autonomously, but with noticeable difficulty, requiring unusual physical effort, extended time, compensatory posturing, awkward stabilization, or minor assistive adaptations.
  • 2 = Easy: The patient executes the task comfortably without abnormal difficulty, hesitation, pain, or functional strain, essentially in the same manner as prior to the onset of the impairment.
  • Unknown / Question Mark (“?”): If an activity has not been attempted or performed at all during the preceding 3 months (e.g., because someone else always handles that domestic chore, or because the patient had no opportunity to perform it), it is coded as missing (“?”) rather than guessed, ensuring that only genuine experiential data enter the calibration.
• Scoring and Transformation Methodology:
  • Raw ordinal scores cannot be simply summed to yield a mathematically valid interval score. Instead, the completion pattern of 0s, 1s, 2s, and missing codes is inputted into dedicated Rasch transformation software or the validated online conversion system.
  • The algorithm estimates the patient’s location along the logit metric (β), complete with an associated standard error of measurement (SE).
  • For routine clinical documentation, logit scores can be linearly transformed into a user-friendly normalized scale ranging from 0 to 100, where higher scores consistently denote superior manual ability.

11. Permissions, Licensing, and History

The ABILHAND questionnaire was originally developed in 1998 by Massimo Penta, Jean-Louis Thonnard, and Luigi Tesio at the Université Catholique de Louvain (UCLouvain) in Brussels, Belgium. The seminal validation study in rheumatoid arthritis was published in 1998, followed by the landmark adult stroke calibration published in 2001.

Licensing and Academic Access:

• The ABILHAND scale, including its underlying Rasch calibration matrices, conversion algorithms, specific item inventories, and software interfaces, is protected under international copyright owned by the Université Catholique de Louvain.
• Non-Commercial Clinical and Academic Research Use: The scale is accessible free of charge for individual clinical practice, non-profit academic investigations, and educational training. Researchers and clinicians can access administration manuals, score entry sheets, and online Rasch conversion calculators via the dedicated academic web platform maintained by UCLouvain (formerly hosted through Rehab-Scales / Louvain Bionics).
• Commercial and Funded Clinical Trial Use: Commercial entities, pharmaceutical corporations, contract research organizations (CROs), and sponsored medical device clinical trials wishing to incorporate ABILHAND into electronic clinical outcome assessment (eCOA) systems or proprietary trial platforms must obtain formal licensing permission and contractual authorization from the intellectual property and technology transfer office at UCLouvain.

12. References

• Andrich, D. (1978). A rating formulation for ordered response categories. Psychometrika, 43(4), 561–573. https://doi.org/10.1007/BF02293814
• Arnould, C., Vandervelde, L., Batcho, C. S., Bogaerts, K., & Thonnard, J. L. (2012). The ABILHAND questionnaire: Validity and responsiveness in children with cerebral palsy. Physical Therapy, 92(5), 724–733. https://doi.org/10.2522/ptj.20110190
• Durez, P., Fraselle, V., Houssiau, F., & Thonnard, J. L. (2007). Validation of the ABILHAND questionnaire as a measure of manual ability in patients with rheumatoid arthritis. Annals of the Rheumatic Diseases, 66(8), 1098–1105. https://doi.org/10.1136/ard.2006.061218
• Masters, G. N. (1982). A Rasch model for partial credit scoring. Psychometrika, 47(2), 149–174. https://doi.org/10.1007/BF02296272
• Penta, M., Tesio, L., Arnould, C., Zancan, A., & Thonnard, J. L. (2001). The ABILHAND questionnaire as a measure of manual ability in chronic stroke patients: Rasch-based validation and relationship to impairment. Stroke, 32(7), 1627–1634. https://doi.org/10.1161/01.str.32.7.1627
• Penta, M., Thonnard, J. L., & Tesio, L. (1998). ABILHAND: A Rasch-built measure of manual ability, with application to rheumatoid arthritis. Archives of Physical Medicine and Rehabilitation, 79(9), 1038–1045. https://doi.org/10.1016/S0003-9993(98)90167-8
• Rasch, G. (1960). Probabilistic models for some intelligence and attainment tests. Danish Institute for Educational Research.
• Vandervelde, L., Van den Bergh, P. Y., Renders, A., & Thonnard, J. L. (2008). ABILHAND-Kids: A measure of manual ability in children with cerebral palsy. Annals of Readaptation and Physical Medicine, 51(4), 246–254. https://doi.org/10.1016/j.annrm.2008.02.001
• Vanderveken, F., Houssiau, F., Thonnard, J. L., & Durez, P. (2016). Assessment of manual ability in systemic sclerosis: Psychometric validation of the ABILHAND questionnaire. Arthritis Care & Research, 68(11), 1709–1716. https://doi.org/10.1002/acr.22879

13. Items of the Scale