Reliability and validity of the 15-item Action Research Arm Test with a clinically-informed decision rule.

Abstract

To address the time burden of the widely used 19-item Action Research Arm Test (ARAT), a 15-item version using a decision rule has been proposed. However, the reliability and criterion-related validity of this decision rule have not been formally evaluated. Existing frameworks also do not specify the optimal direction of item administration. This study examined the reliability and validity of the 15-item ARAT using a decision rule and a task direction protocol based on Brunnstrom Recovery Stages. In this cross-sectional study, 30 patients with post-stroke hemiparesis were assessed using the 19-item ARAT and the Fugl-Meyer Assessment (FMA). The 15-item ARAT dataset was derived from the 19-item version. Reliability with and without the decision rule was assessed using intraclass correlation coefficients (ICC), item-level agreement <i>via</i> weighted κ, and criterion-related validity <i>via</i> Spearman's ρ. The decision rules reduced the total number of items from 450 to 226 across all participants. The ICC was 0.997 (95% CI: 0.993-0.999). Item-level agreement was high (κ = 0.88-1.00), although lower for item 6 (κ = 0.68). Correlations with the FMA (ρ = 0.72) and the 19-item ARAT (ρ = 0.96) were strong. This decision rule enhances efficiency without compromising reliability or validity.