Background <p>Long-term caregiving burden can lead to multiple health issues among caregivers of disabled elderly patients, including impaired physical functioning, psychological distress, and restricted social roles. The World Health Organization (WHO) defined it as “third-party disability”, referring to the sustained negative impacts on caregivers who provide long-term care for patients. These negative impacts include impairments in physiological functions, social participation, and environmental adaptation. Based on that, this study developed and validated a specialized assessment tool to evaluate third-party disability among caregivers of elderly patients with post-stroke disability, aiming to provide scientific and precise measurement criteria for clinical practice.</p> Methods <p>This study is comprised of two phases. In the first phase, scale items were identified through literature review and semi-structured interviews. The content of items was revised by expert review and pre-surveys to generate a preliminary scale. In the second phase, data from 251 family caregivers of elderly patients with post-stroke disability were used. We conducted item analysis and exploratory factor analysis (EFA) to refine scale items, k-means cluster analysis and discriminant analysis to refine scale score levels. Furthermore, we completed the confirmatory factor analysis (CFA) and reliability analysis of the refined scale with data from another 398 family caregivers.</p> Results <p>The final developed scale consists of four dimensions (totally 26 items), which are mental functioning, physical functioning, daily living, and environmental adaptation. The scale demonstrates strong content validity, with item-level content validity indices (I-CVIs) ranging from 0.79 to 1.00 and a scale-level content validity index (S-CVI) of 0.93. The CFA demonstrates that the scale model fits well and has good factorial validity. The average variance extracted (AVE) of the scale model ranges from 0.57 to 0.70. Meanwhile, the correlation coefficients between different dimensions are all less than the square roots of their respective AVEs, indicating the scale model has good convergent validity and discriminant validity. The scale also demonstrates robust reliability, with Cronbach’s α coefficient of 0.91, split-half reliability of 0.80, and test-retest reliability (assessed with a two-week interval) of 0.96.</p> Conclusions <p>While the final developed scale demonstrates strong reliability and validity, further refinement is still necessary for the scale to perfectly serve as a reference for constructing targeted intervention measures.</p>