Between-class variation and compositional correlates of physical fitness, and the concordance of composite scores with BMI-defined abnormality: a multilevel analysis of annual surveillance records from one Chinese university

Abstract

Introduction This repeated cross-sectional retrospective study examined between-class variation in university students’ physical-fitness scores and whether annual composite-score trends were concordant with changes in BMI-defined abnormality. Methods Annual surveillance records collected from 2020 to 2024 at one public university in Sichuan Province, China, were analysed. The dataset included 106,461 records from 2,591 class-year units (44,240 male, 62,221 female; mean age 20.67 ± 1.54 years; mean BMI 21.64 ± 3.52 kg/m2). Multilevel models quantified between-class variation and examined individual- and class-compositional correlates, with calendar year entered categorically. A complementary temporal analysis compared annual composite scores with the prevalence of underweight, overweight, or obesity using Chinese national cut-points. Results The null-model intraclass correlation coefficient was 0.178, indicating meaningful between-class clustering. After adjustment, higher within-class BMI, older age, and male sex were associated with lower composite scores. At the class level, higher mean BMI was associated with lower scores, whereas the class-size association was statistically significant but practically negligible. The within-class BMI–fitness association varied across class-year units, although heterogeneity was modest and the association was negative in almost all units. Composite scores increased in 2021, declined in 2022, and increased again in 2023–2024. In contrast, BMI-defined abnormality rose every year, from 23.3% in 2020 to 31.4% in 2024. Discussion University fitness data have a meaningful clustered structure, and improvement in a weighted composite score does not necessarily imply parallel improvement in the BMI distribution. Surveillance reports should therefore account for clustering and present composite scores alongside BMI categories and component-specific outcomes.