Geriatrics & Gerontology International · Nov 2, 2026

Discordance Between Height‐ and BMI ‐Adjusted AWGS 2025 Definitions of Confirmed Sarcopenia After Stroke

By Yoshimura Y, Matsumoto A, Yoneda K et al.

Aim: The Asian Working Group for Sarcopenia (AWGS) 2025 allows height‐ or body mass index (BMI)‐adjusted muscle mass in diagnosing confirmed sarcopenia. We evaluated classification discordance and associated clinical characteristics after stroke. Methods: This retrospective study included 864 patients undergoing post‐acute stroke rehabilitation. Confirmed sarcopenia was defined using low handgrip strength combined with either height‐ or BMI‐adjusted muscle mass. Patients were classified into four groups: Neither, Height‐only, BMI‐only, and Both. Agreement between definitions was quantified, and the two discordant groups were compared using multivariable logistic regression. Model robustness and discharge outcomes were examined in sensitivity and exploratory analyses, respectively. Results: Discordant classification occurred in 183 patients (21.2%; 95% confidence interval [CI], 18.6%–24.0%), with moderate agreement between definitions ( κ = 0.578). Compared with the Height‐only group, the BMI‐only group had higher BMI, serum albumin, and phase angle. In adjusted analyses of 182 patients, higher albumin (odds ratio [OR] per 0.5 g/dL, 1.71; 95% CI, 1.10–2.64), phase angle (OR per degree, 1.93; 95% CI, 1.23–3.04), and medication count (OR per drug, 1.17; 95% CI, 1.03–1.32) were associated with BMI‐only classification. Firth estimates were similar, but internal validation indicated overfitting. Adjusted discharge outcomes did not clearly differ between discordant groups. Conclusions: Height‐ and BMI‐adjusted AWGS 2025 criteria classified approximately one in five patients differently after stroke. Beyond expected body‐size differences, discordant groups exhibited distinct clinical characteristics, highlighting the importance of considering the indexing method when interpreting sarcopenia classifications.

doi.org/10.1111/ggi.70898PubMed 42971943Biomarkers

AgingResearch.News