# Reliability and validity of a newly developed Action Research Arm Test for upper limb function assessment in patients with stroke: A comparison with the conventional version

## TL;DR

A new version of the Action Research Arm Test was developed and validated for assessing upper limb function in stroke patients, showing reliability and validity comparable to the original test.

## Contribution

The new ARAT version was validated for psychometric properties and compared to the conventional ARAT in stroke patients.

## Key findings

- The new ARAT showed excellent intra-rater and inter-rater reliability with ICC values close to 1.0.
- Convergent validity was very strong with FMA-UE and BBT, and moderate-to-strong with other measures.
- Bland-Altman analysis confirmed clinically acceptable agreement between the new and conventional ARAT.

## Abstract

Following the discontinuation of the conventional Action Research Arm Test (ARAT) import in Japan, a domestically manufactured version with identical assessment protocols but modified equipment was developed. We aimed to evaluate the psychometric properties of the newly developed ARAT and compare it with the conventional ARAT in patients with stroke.

This single-center cross-sectional study enrolled 64 patients with stroke and hemiparesis who met predefined inclusion criteria. Participants were randomly allocated to a reliability validation (n = 33) or criterion validity validation (n = 31) group. The reliability group underwent duplicate assessments with the new ARAT at 15-min intervals, with video recording for independent inter-rater evaluation. The validity group received the new and conventional ARAT assessments in randomized order. Convergent validity was assessed using Spearman correlation coefficients with Fugl-Meyer Assessment of the Upper Extremity (FMA-UE), Box and Block Test (BBT), grip strength, Motor Activity Log (MAL), and Jikei Assessment Scale for Motor Impairment in Daily Living (JASMID).

Intra-rater reliability demonstrated excellent agreement (intraclass correlation coefficient [ICC]: 0.997–1.000, 95% confidence interval [CI]: 0.995–1.000) across all subscales and total scores. Inter-rater reliability showed equally excellent agreement (ICC: 0.979–0.999, 95% CI: 0.963–0.999). Bland–Altman analysis revealed limits of agreement within clinically acceptable ranges for all measures. The new ARAT demonstrated very strong convergent validity with FMA-UE (ρ = 0.934, p < 0.001) and BBT (ρ = 0.917, p < 0.001), and moderate-to-strong correlations with grip strength (ρ = 0.683, p < 0.001), MAL subscales (ρ = 0.610–0.666, p < 0.001), and JASMID subscales (ρ = 0.806–0.808, p < 0.001).

The new ARAT demonstrates measurement properties equivalent to the conventional version with excellent reliability and strong criterion-related validity. Its robust convergent validity with established upper limb assessments supports its clinical utility for comprehensive stroke rehabilitation evaluation.