Wolf Motor Function Test (WMFT)

Abstract

The Wolf Motor Function Test (WMFT) is a standardized, quantitative, performance-based instrument engineered to assess upper extremity motor ability, functional dexterity, and movement quality among patients recovering from neurological insults, primarily cerebrovascular accidents (stroke) and traumatic brain injury (TBI). Developed originally by Dr. Steven L. Wolf and colleagues in 1989, and subsequently refined into a standardized protocol in 2001, the WMFT serves as the cornerstone outcome measure in neurorehabilitation clinical trials, most notably the landmark Extremity Constraint-Induced Therapy Evaluation (EXCITE) trial. The test comprises 15 timed motor tasks arranged hierarchically from simple, proximal, gross-motor joint movements (e.g., shoulder abduction, elbow extension) to complex, distal, fine-motor integrative tasks (e.g., manipulating a paper clip, turning a key in a lock, folding a towel). In addition to measuring execution speed in seconds (with a standardized ceiling cutoff of 120 seconds per item), qualitative execution is concurrently evaluated across tasks using the 6-point Functional Ability Scale (FAS; scored 0–5). Extensive psychometric evaluations demonstrate remarkable reliability (inter-rater ICC > 0.93; test-retest ICC > 0.90; Cronbach’s alpha > 0.90) and high convergent validity with the Fugl-Meyer Assessment (FMA) and the Action Research Arm Test (ARAT). Modern Rasch analysis and factor-analytic models confirm a primary unidimensional construct of paretic limb functional capacity while revealing a robust two-factor nested architecture distinguishing proximal gross-motor stability from distal fine-motor manipulative coordination.

Keywords

Wolf Motor Function Test, WMFT, upper extremity motor function, stroke rehabilitation, constraint-induced movement therapy, CIMT, neurorehabilitation, hemiparesis, motor recovery, psychometrics, Functional Ability Scale, Rasch analysis, motor control

Authors

The original conceptualization and empirical foundation of the Wolf Motor Function Test were formulated by:

• Steven L. Wolf, PhD, PT, FAPTA, FAHA — Professor, Department of Rehabilitation Medicine, Department of Cell Biology, and Department of Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
• Deborah LeCraw, PT — Department of Rehabilitation Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
• Laurence A. Barton, PT — Center for Rehabilitation Medicine, Emory University Hospital, Atlanta, Georgia, USA.
• Bobbi B. Jann, MD — Department of Rehabilitation Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.

Subsequent psychometric standardization, normative benchmarking, and validation for research initiatives were established by Steven L. Wolf, Pamela A. Catlin, Michael Ellis, Audrey Link Archer, Bryn Morgan, and Aimee Piacentino (2001) in collaboration with the Department of Rehabilitation Medicine at Emory University and the Division of Physical Therapy.

Purpose

The primary clinical and scientific purpose of the Wolf Motor Function Test (WMFT) is to systematically evaluate and track changes in upper extremity motor impairment, movement efficiency, and manual dexterity in individuals presenting with neurological hemiparesis. Originally constructed to quantify the therapeutic benefits of Constraint-Induced Movement Therapy (CIMT)—an intensive behavioral intervention forcing the use of the paretic limb by restraining the unaffected limb—the WMFT provides an objective, performance-based metric that circumvents self-report biases while capturing real-time functional performance under timed constraints.

Neurological hemiparesis following middle cerebral artery infarction or focal brain injury typically results in profound disorganization of corticospinal control, manifested as muscle weakness, hypertonicity, loss of fractionation of movement, and compensatory motor patterns. Traditional clinical neurological examinations (such as manual muscle testing or reflex grading) exhibit limited sensitivity to functional motor restitution and often fail to detect subtle kinematic improvements occurring across prolonged rehabilitation protocols. The WMFT addresses this gap by measuring both the temporal velocity of task execution (WMFT Performance Time) and the biomechanical quality of movement patterns (WMFT Functional Ability Scale).

From a research perspective, the instrument facilitates multi-site comparative clinical trials by offering standardized test materials, exact patient positioning constraints, objective item termination cutoffs (120 seconds), and precise functional scoring definitions. In clinical settings, the WMFT enables physical therapists, occupational therapists, and rehabilitation physicians to:

• Establish baseline motor capacity across discrete single-joint and multi-joint synergies.
• Identify specific kinematic bottlenecks (e.g., adequate proximal shoulder abduction coupled with complete failure of distal finger extension or thumb opposition).
• Formulate targeted therapeutic regimens aimed at reversing learned non-use.
• Monitor longitudinal neuroplastic adaptations resulting from pharmacological, robotic, neuromuscular electrical stimulation, or task-oriented behavioral therapies.

Psychological Construct

The central psychological and behavioral construct captured by the WMFT is upper extremity functional motor competence, which encompasses motor programming, executive-motor planning, sensorimotor integration, and kinematic coordination in goal-directed tasks. Unlike purely biomechanical assessments of isolated torque generation, the WMFT views motor execution through an ecological, task-oriented lens.

1. Proximal Motor Execution and Dynamic Postural Stabilization

Tasks 1 through 6 assess gross motor control mediated predominantly by proximal musculature—the trapezius, deltoid, pectoralis major, biceps brachii, and triceps brachii. These items evaluate the participant’s ability to overcome flexor synergy dominance (which classically locks the hemiparetic arm in internal rotation, adduction, and elbow flexion). For instance, Task 1 (Forearm to table) and Task 2 (Forearm to box) demand deliberate shoulder abduction and external rotation against gravity. Task 3 and Task 4 (Extend elbow to side, without and with weight) probe the capacity to isolate triceps activation while suppressing involuntary co-contraction of the antagonistic flexor muscle groups. This reflects the neurological construct of motor fractionation—the ability to actuate a single joint independently without obligate mass synergies.

2. Multi-Joint Reaching and Sensorimotor Coordination

Tasks 7 and 8 (Reach and retrieve, Lift can) bridge gross proximal positioning with distal preparatory shaping. Reaching requires the smooth, coordinated coupling of shoulder flexion, elbow extension, and wrist stabilization. The psychological construct here includes feedforward anticipatory postural adjustments and visual-proprioceptive calibration. To lift a 12-ounce beverage can close to the lips, the patient must not only grasp the cylindrical object without crushing or slipping, but also orchestrate elbow flexion and forearm supination while modulating hand aperture size, exemplifying functional sensorimotor integration.

3. Distal Fine-Motor Dexterity and Manipulative Control

Tasks 9 through 13 (Lift pencil, Pick up paper clip, Stack three checkers, Flip three cards, Turning the key in lock) recruit the fine motor corticomotoneuronal system, which relies on monosynaptic projections from the primary motor cortex to spinal motor neurons controlling intrinsic hand muscles. These tasks operationalize manual dexterity through distinct grip configurations:

• Tripod Grasp (3-Jaw Chuck): Evaluated in Task 9 (Lift pencil), necessitating radial finger-thumb opposition.
• Pincer Grasp: Required in Task 10 (Paper clip) and Task 12 (Flipping cards), testing terminal pad-to-pad precision between the index finger and thumb.
• Rotational Forearm-Wrist Torque: Assessed in Task 13 (Key in lock), demanding stabilized pinch combined with active, isolated pronation-supination.

4. Complex Integrative Bimanual and Functional Sequencing

Tasks 14 and 15 (Fold towel, Lift basket) evaluate advanced functional sequencing. Task 14 requires stabilizing an unstructured textile, initiating bilateral or unilateral folding planes, and performing repetitive grasping and smoothing. Task 15 transitions the patient from an unconstrained seated posture to an active upright standing posture, integrating lower extremity weight-bearing stability, core equilibrium, and bilateral/unilateral lifting ergonomics under gravity.

Theoretical Framework

The Wolf Motor Function Test is grounded in three converging paradigms within contemporary motor neuroscience: the Learned Non-Use Hypothesis, the Dynamic Systems Theory of Motor Control, and the neurobiological doctrine of Activity-Dependent Neuroplasticity.

1. The Learned Non-Use Hypothesis

Developed extensively by Edward Taub, the concept of learned non-use posits that following unilateral cortical injury, initial attempts to mobilize the affected limb are met with failure, awkwardness, or pain. These failed attempts act as a negative reinforcement schedule, driving the central nervous system to rely exclusively on the intact, contralateral upper extremity. Over weeks and months, this behavioral compensation suppresses the neural representation of the affected limb within the motor cortex, producing a deficit far more severe than what is dictated by structural tissue destruction alone. The WMFT was created directly as an experimental apparatus to determine whether forced-use paradigms (CIMT) could extinguish this learned non-use by demonstrating progressive chronometric and kinematic gains across an array of standardized, structured daily movement challenges.

2. Dynamic Systems Theory

Under the Dynamic Systems Theory championed by Nikolai Bernstein and Esther Thelen, movement is not simply pre-programmed by an invariant central motor program; rather, it emerges organically from the self-organizing interactions among the central nervous system, musculoskeletal constraints, the physical properties of the environment, and the demands of the specific motor goal. The WMFT embodies this theory by testing movement across varied physical configurations—manipulating objects with varying friction coefficients (plastic-coated paper clip vs. wooden checkers), differing mass profiles (1 lb cuff weights vs. empty towels), and varied vertical workspace elevations (table surface vs. 10-inch box).

3. Activity-Dependent Neuroplasticity

The theoretical framework of the WMFT assumes that motor recovery is mediated by cortical remap and corticospinal reorganization driven by repetitive, salient, task-oriented practice. Neuroimaging studies (such as functional magnetic resonance imaging and transcranial magnetic stimulation) reveal that improvements in WMFT performance times directly correspond with an expansion of the motor map of the paretic hand in the peri-infarct cortex and recruitment of the supplementary motor area (SMA) and premotor cortex. By evaluating tasks along a clear continuum of neuroanatomical complexity (proximal subcortical/rubrospinal-influenced movements transitioning to fine pyramidal/corticospinal terminal grasping), the WMFT serves as an external index of underlying synaptic plasticity and motor unit recruitment.

Validity

The measurement properties of the WMFT have been subjected to extensive psychometric scrutiny across acute, subacute, and chronic stroke cohorts, as well as traumatic brain injury populations.

Construct and Discriminant Validity

The WMFT exhibits superior construct validity, consistently discriminating between levels of stroke severity, paretic versus non-paretic limbs, and healthy age-matched control subjects. In the validation study by Wolf et al. (2001), the mean performance time for non-dominant upper extremities of healthy adults across all tasks was less than 1.5 seconds per task, with zero tasks reaching the 120-second threshold. In contrast, individuals with mild-to-moderate hemiparesis exhibited average performance times ranging from 2.5 to over 40 seconds per task, with severely impaired patients hitting the 120-second truncation ceiling on multiple fine-motor distal tasks.

Construct validity is further evidenced by the hierarchical difficulty structure of the test: execution times and functional failure rates scale systematically from proximal shoulder abduction tasks (Tasks 1–2) to distal fine-motor manipulation items (Tasks 9–13). Known-groups validity studies confirm that the WMFT differentiates patients categorized by different motor stages according to Brunnstrom stages and Chedoke-McMaster Stroke Assessment classifications.

Convergent and Criterion-Related Validity

Convergent validity has been established through simultaneous cross-sectional comparisons with gold-standard impairment and activity-limitation metrics:

• Fugl-Meyer Assessment (FMA – Upper Extremity): WMFT Functional Ability Scale (FAS) scores demonstrate robust, statistically significant positive correlations with upper extremity FMA motor scores (Pearson r = 0.78 to 0.89; Spearman rho = 0.82 to 0.91). WMFT log-transformed performance times show strong negative correlations with FMA scores (r = -0.68 to -0.84), confirming that faster motor completion coincides with reduced motor impairment.
• Action Research Arm Test (ARAT): Correlations between the WMFT FAS and the ARAT are remarkably high (ranging from r = 0.81 to 0.94), establishing that the WMFT effectively measures grasp, grip, pinch, and gross upper extremity movements.
• Motor Activity Log (MAL): The WMFT correlates moderately to strongly with real-world spontaneous limb use measured via the MAL Amount of Use (AOU, r = 0.63 to 0.74) and Quality of Movement (QOM, r = 0.65 to 0.76) scales, validating that laboratory-based performance capacity translates into daily functional independence.
• Stroke Impact Scale (SIS): The physical domain of the SIS correlates significantly (r = 0.60 to 0.72) with WMFT functional scores, affirming broader impacts on stroke-related health status.

Predictive and Ecological Validity

Baseline WMFT scores possess pronounced predictive validity for long-term functional recovery. In the multi-site EXCITE trial (Wolf et al., 2006), baseline performance times and FAS scores on the WMFT predicted functional limb retention and continued spontaneous real-world arm usage up to 24 months post-intervention. Furthermore, the test shows minimal floor and ceiling effects across mild-to-moderate stroke populations, although severe floor effects are encountered when testing completely flaccid or severely hemiparetic patients incapable of any active finger extension.

Reliability

The WMFT possesses exceptional psychometric reliability, established across both real-time in-person evaluations and blinded videotaped observational ratings.

Inter-Rater Reliability

Inter-rater reliability metrics are consistently high when raters are trained using standardized operational protocols. In the 2001 standardization study by Wolf and colleagues involving multiple trained physical and occupational therapists scoring stroke patients via videotape:

• WMFT Performance Time: Intraclass Correlation Coefficients (ICC, model 2,1) ranged from 0.93 to 0.99 for individual timed tasks, with the aggregate mean time ICC reaching an astounding 0.97 to 0.99.
• WMFT Functional Ability Scale (FAS): Inter-rater ICC values for individual task quality ratings ranged from 0.86 to 0.97, yielding an overall FAS composite reliability ICC of 0.93 to 0.96.
• Kendall’s coefficient of concordance (W) across clinical raters has routinely exceeded 0.90, demonstrating uniformity in qualitative rank ordering.

Test-Retest and Intra-Rater Reliability

Repeated evaluations across stable chronic stroke subjects (measured 7 to 14 days apart without intervening therapy) confirm substantial stability over time:

• Test-Retest Reliability: ICC values for total performance time range from 0.90 to 0.96 across chronic cohorts. For the Functional Ability Scale, test-retest ICCs range from 0.92 to 0.98.
• Intra-Rater Reliability: Repeated scoring of videotaped administrations by the same clinician yielded intra-class correlation coefficients exceeding 0.95 for performance time and 0.94 for qualitative functional ratings.

Internal Consistency

The 15-item instrument displays elevated internal consistency. Chronometric analysis yields Cronbach’s alpha values between 0.88 and 0.92. When evaluating the 15 items under the Functional Ability Scale, Cronbach’s alpha systematically ranges from 0.92 to 0.95. These figures demonstrate that while the items measure distinct kinematic demands, they align cohesively along a unified continuum of upper extremity motor competence without presenting redundant collinearity.

Factor Analysis

Both classical exploratory/confirmatory factor analysis (EFA/CFA) and modern Item Response Theory (IRT) paradigms have been conducted on the WMFT to interrogate its structural dimensionality.

Exploratory and Confirmatory Factor Structure

Early psychometric investigations supported a primary single-factor model representing general upper extremity motor capability. However, subsequent structural equation modeling and principal component analyses (e.g., Morris et al., 2001; Lin et al., 2009) have demonstrated that a two-factor oblique model offers superior statistical and conceptual fit across diverse clinical samples:

• Factor 1: Proximal Gross Motor Control (Items 1–6, 15). These items load heavily (standardized loadings: 0.72–0.89) onto a factor defined by glenohumeral elevation, scapulothoracic stabilization, gross elbow excursion, and whole-limb postural lifting. Variance in this factor reflects recovery of broad rubrospinal and reticulospinal gross-motor pathways.
• Factor 2: Distal Fine Motor Dexterity and Manipulation (Items 7–14). These items load strongly (standardized loadings: 0.68–0.91) onto a distinct manipulative factor demanding fractionation of the fingers, thumb opposition, pincer grasp, and coordinated forearm supination/pronation. This factor depends heavily on monosynaptic corticospinal projections.

In confirmatory factor analysis (CFA), the two-factor model yields excellent goodness-of-fit indices (Comparative Fit Index [CFI] > 0.95; Tucker-Lewis Index [TLI] > 0.94; Root Mean Square Error of Approximation [RMSEA] < 0.07), outperforming the strictly unidimensional model while showing a strong inter-factor correlation (r ≈ 0.75–0.82), which justifies computing a single aggregate composite score when necessary.

Rasch Analysis and Item Hierarchies

Application of the Rasch measurement model (partial credit model) to the WMFT Functional Ability Scale provides vital insight into item calibration and task difficulty ordering. Rasch analysis confirms an ordered item-difficulty continuum:

• Least Difficult (Lowest Infit/Outfit Logit Calibrations): Task 1 (Forearm to table), Task 5 (Hand to table), and Task 3 (Extend elbow to side).
• Moderately Difficult: Task 2 (Forearm to box), Task 6 (Hand to box), Task 8 (Lift can), Task 14 (Fold towel).
• Most Difficult (Highest Logit Calibrations): Task 9 (Lift pencil), Task 10 (Pick up paper clip), Task 12 (Flip three cards), and Task 13 (Turning key in lock).

Rasch infit and outfit Mean Square (MnSq) statistics for the 15 items largely fall within the acceptable psychometric range of 0.7 to 1.3, confirming that the tasks define a cohesive, unidimensional latent measurement ruler without significant item underfit or distortion.

Instrument / Measurement Tool

Administration Guidelines and Standardized Setup

The Wolf Motor Function Test requires a rigorously controlled physical testing station to prevent environmental confounders. The standardization manual mandates the following specifications:

• Seating & Workspace: Armless, straight-backed wooden or metal chair (seat height 18 inches / 45.7 cm) positioned facing a smooth desk/table of standard height (29 inches / 73.5 cm; 54 inches wide by 30 inches deep).
• Visual Positioning Template: A paper or vinyl template is taped flush to the table top, outlining exact starting locations for the patient’s trunk, hands, and test apparatuses (e.g., target boxes, soda can, key apparatus).
• Standardized Elevation Boxes: A cardboard box measuring 10 inches (25.4 cm) in height (representing approximate shoulder height for an average adult seated at the desk). Supplemental boxes of 8 inches (20.3 cm) and 6 inches (15.2 cm) must be available for shorter individuals.
• Standard Test Implements:
  • One 1-lb cuff-weight with Velcro strap.
  • Sandbag / friction-controlled resistance apparatus.
  • Unopened 12-oz soft drink can (weight approximately 392 grams).
  • 7-inch (17.78 cm) standard wooden pencil featuring 6 flat sides.
  • 2-inch (5.08 cm) vinyl/plastic-coated colored paper clip.
  • Three standard checkers (interlocking plastic draughts).
  • Three standard 3 × 5 inch (7.62 × 12.7 cm) index cards.
  • Custom lock and key board set at a 45-degree angle (tumbler mechanism calibrated to allow exactly 180 degrees of rotation; 90 degrees on either side of midline).
  • Standard cotton dish towel: 25 inches (63.5 cm) × 15 inches (38.1 cm).
  • Plastic or wicker tote basket with handles (15 inches tall, 8.5 inches wide, 14 inches long).
  • Handheld stopwatch (measuring to 0.01 seconds) and talcum/baby powder to reduce table surface friction.

Testing and Scoring Protocol

The WMFT is administered bilaterally, typically assessing the less-affected (unimpaired) limb first to familiarize the patient with operational demands, followed by the paretic (involved) limb. All tasks are initiated upon a standardized verbal countdown (“Ready, Set, Go!”).

• Test Type: Performance-based, objective observational clinician-administered motor examination.
• Number of Items: 15 distinct functional motor tasks (Items 1–15). Note: Extended research variants sometimes incorporate two grip/strength dynamometer items (Tasks 7 and 14 in a 17-item version), but the primary clinical standard consists of the 15 performance-timed tasks detailed herein.
• Response Scale 1: Performance Time (Velocity):
  • Measured continuously via stopwatch in seconds.
  • Timing commences immediately at the “Go” prompt (or first initiated physical movement) and ceases the instant task completion criteria are achieved.
  • Ceiling / Truncation Cutoff: If a task cannot be completed within 120 seconds, timing ceases and a score of 120.00 seconds is recorded.
  • Data transformation: Due to positive skewness caused by 120-second truncation, clinical trials compute log-transformed times: $\text{LogTime} = \ln(\text{Time} + 0.1)$ or report median completion speeds.
• Response Scale 2: Functional Ability Scale (FAS – Quality of Movement):
  • Concurrently rated for each task on an ordinal 6-point (0 to 5) Likert-type scale:
  • 0 = Does not attempt: The involved upper extremity does not participate, or shows complete inability to engage with the object.
  • 1 = Minimal attempt: Involved limb attempts to move, but movement is influenced entirely by synergy patterns, produces major compensatory trunk posturing, or fails to accomplish any segment of the goal.
  • 2 = Partial completion / Synergistic: Task is accomplished partially or completed with marked synergistic posturing, pronounced tremor, severe dysmetria, or extreme compensatory trunk flexion.
  • 3 = Accomplished with noticeable effort: Task is completed, but movement is characterized by mild synergistic co-contractions, slowness, hesitation, or minor compensatory posturing.
  • 4 = Near-normal completion: Movement approaches normal quality and velocity, with only minor deviations in trajectory, subtle incoordination, or slight clumsiness.
  • 5 = Normal movement: Task performance is completely smooth, fluent, coordinated, and kinematically indistinguishable from the unaffected contralateral limb.

Permissions & Fee and Test Year

The Wolf Motor Function Test was first introduced in 1989 by Dr. Steven L. Wolf and colleagues at Emory University School of Medicine. Its definitive standardized protocol, instructional manuals, and psychometric profiles were published in the journal Stroke in 2001 (Wolf et al., 2001).

Licensing and Accessibility: The WMFT is maintained in the public domain for clinical and academic research purposes. No royalty fees or proprietary purchase costs are mandated for downloading the test sheets, manual guidelines, or templates. The standardized testing guidelines, video training manuals, and item checklists have been broadly disseminated by the National Institutes of Health (NIH), the EXCITE Trial operational archives, and state rehabilitation repositories (e.g., the Queensland Health Compendium of Clinical Measures). Clinicians and research investigators are permitted to construct or assemble their own testing kits using standard commercial components that strictly adhere to the published metric dimensions.

References

• Lin, K. C., Chuang, L. L., Wu, C. Y., Hsieh, Y. W., & Chang, W. Y. (2009). Responsiveness and validity of three upper extremity outcome measures in stroke rehabilitation: The Action Research Arm Test, the Fugl-Meyer Assessment, and the Wolf Motor Function Test. Archives of Physical Medicine and Rehabilitation, 90(8), 1286–1295. https://doi.org/10.1016/j.apmr.2009.01.031
• Morris, D. M., Uswatte, G., Crago, J. E., Cook, E. W., & Taub, E. (2001). The reliability of the Wolf Motor Function Test for assessing upper extremity function following stroke. Archives of Physical Medicine and Rehabilitation, 82(6), 750–755. https://doi.org/10.1053/apmr.2001.23183
• Taub, E., Miller, N. E., Novack, T. A., Cook, E. W., Fleming, W. C., Nepomuceno, C. S., Connell, J. S., & Crago, J. E. (1993). Technique to improve chronic motor deficit after stroke. Archives of Physical Medicine and Rehabilitation, 74(4), 347–354. https://pubmed.ncbi.nlm.nih.gov/8466415/
• Winstein, C. J., Miller, J. P., Blanton, S., Taub, E., Uswatte, G., Morris, D., Nichols, D., & Wolf, S. (2003). Methods for a multisite randomized trial to investigate the effect of constraint-induced movement therapy in improving upper extremity function among adults recovering from a cerebrovascular stroke. Neurorehabilitation and Neural Repair, 17(3), 137–152. https://doi.org/10.1177/0888439003255811
• Wolf, S. L., Catlin, P. A., Ellis, M., Archer, A. L., Morgan, B., & Piacentino, A. (2001). Assessing Wolf Motor Function Test as outcome measure for research in patients after stroke. Stroke, 32(7), 1635–1639. https://doi.org/10.1161/01.str.32.7.1635
• Wolf, S. L., Lecraw, D. E., Barton, L. A., & Jann, B. B. (1989). Forced use of hemiplegic upper extremities to reverse the effect of learned nonuse among chronic stroke and head-injured patients. Experimental Neurology, 104(2), 125–132. https://doi.org/10.1016/1004-8464(89)90115-6
• Wolf, S. L., Winstein, C. J., Miller, J. P., Taub, E., Uswatte, G., Morris, D., Giuliani, C., Light, K. E., & Nichols-Larsen, D. (2006). Effect of constraint-induced movement therapy on upper extremity function 3 to 9 months after stroke: The EXCITE randomized clinical trial. JAMA, 296(17), 2095–2104. https://doi.org/10.1001/jama.296.17.2095

Items of the Scale