Internal Consistency, Construct Validity and Incremental Explanatory Power of the Action Observation Sheet‐B ( AOS ‐B) for Dementia Caregiver Burden and Clinical Severity

BACKGROUND: Although the behavioral Action Observation Sheet-B (AOS-B) addresses the clinical limitations, patient anxiety and caregiver biases of traditional dementia assessments, evidence regarding its incremental explanatory power remains insufficient. Using real-world clinical data, this study evaluated the 47-item AOS-B's internal consistency, validity and independent contribution to explaining caregiver burden. METHODS: We retrospectively analysed data from 107 patients with different dementia subtypes. Internal consistency was assessed using Cronbach's α. Validity was examined using Pearson's correlations between scores on the AOS-B and established scales (the Mini-Mental State Examination-Japanese, Hasegawa Dementia Rating Scale-Revised, Clinical Dementia Rating and the 8-item Zarit Caregiver Burden Interview-Japanese version). Hierarchical multiple regression was used to evaluate the AOS-B's independent association with caregiver burden beyond cognitive scores; receiver operating characteristic analysis was used to determine the optimal AOS-B cutoff for identifying high-risk caregiver burden (Zarit Caregiver Burden Interview score ≥ 13). Receiver operating characteristic analysis yielded an area under the curve of 0.750 and 65.0 as a potential cutoff (sensitivity: 77.8%, specificity: 67.6%) for identifying high-risk caregiver burden. AOS-B scores did not differ significantly among dementia subtypes, though statistical power was limited by the subgroup sizes. CONCLUSIONS: The AOS-B demonstrated high internal consistency and construct validity as a non-invasive tool that complements standardised cognitive tests by capturing practical caregiving challenges. Using the 65.0-point threshold with existing caregiver burden scales may help clinicians better identify specific intervention targets and optimise resource allocation in routine practice.