PP455 - NON-LINEAR ASSOCIATION BETWEEN BMI AND FUNCTIONAL RECOVERY IN OLDER PATIENTS WITH ISCHEMIC STROKE: A NATIONWIDE INPATIENT DATABASE STUDY

PP455

NON-LINEAR ASSOCIATION BETWEEN BMI AND FUNCTIONAL RECOVERY IN OLDER PATIENTS WITH ISCHEMIC STROKE: A NATIONWIDE INPATIENT DATABASE STUDY

A. Saito^1,2,*, Y. Ishida³, T. Inoue², F. Kawase⁴, K. Murotani^5,6, K. Maeda^7,8

¹Department of Nutrition, Yoshida Hospital, Nagaoka, ²Graduate School of Health and Welfare, Niigata, ³Department of Nutrition, Aichi Medical University Hospital, Nagakute, ⁴Department of Sports and Health Sciences, Tohoku University, Sendai, ⁵School of Medical Technology, Kurume University, ⁶Biostatistics Center, Kurume University, Kurume, ⁷Nutrition Therapy Support Center, Aichi Medical University Hospital, Nagakute, ⁸Department of Geriatric Medicine, National Center for Geriatrics and Gerontology, Obu, Japan

Rationale: Functional recovery is a key outcome in patients with acute ischemic stroke, but its relationship with body mass index (BMI) remains inconsistent. Because the Barthel Index (BI) can quantitatively measure changes in activities of daily living, this study investigated the association between BMI at admission and BI change, with a focus on non-linear relationships.

Methods: This retrospective study used a nationwide inpatient database in Japan (2014–2024). Patients aged 65–99 years with first ischemic stroke were included. The outcomes were 30-day in-hospital mortality and BI change. Patients who died during hospitalization or had a BI score of 100 at admission were excluded from the BI analysis. Fine–Gray and linear regression models were used, adjusted for age, sex, Charlson Comorbidity Index, number of medications, level of consciousness, and BI at admission. BMI was treated as a continuous variable, and non-linear associations were evaluated using restricted cubic splines.

Results: A total of 174,053 patients were included (mean age 79.9 ± 8.0 years; 44.9% female). BMI showed an L-shaped association with 30-day in-hospital mortality. The risk increased progressively below a BMI of 23.0 kg/m², reaching a subdistribution hazard ratio of 1.75 [95% confidence interval (CI) 1.39–2.21] at a BMI of 18.5 kg/m², while the association plateaued above 23.0 kg/m². BI change showed an inverted U-shaped association, with a peak at a BMI of 24.8 kg/m² (β = 29.71 [95% CI 29.36–30.06]). This pattern was consistent in sex-stratified analyses, with peaks at BMI 25.0 kg/m² in men (β = 39.23 [95% CI 38.74–39.71]) and BMI 24.4 kg/m² in women (β = 26.38 [95% CI 25.91–26.85]).

Conclusion: BMI at admission was non-linearly associated with mortality and functional recovery. These findings highlight the potential importance of BMI in prognostic assessment of ischemic stroke.

Disclosure of Interest: None declared