Association between physical fitness and myopia among school-entry students: a school-based cross-sectional study

We examined whether school physical-fitness performance was associated with screening-defined current myopia in primary grade 1 and junior middle grade 1 students. We also explored item-level fitness profiles in relation to current myopia status.

We analyzed routine health-screening records collected in 2025 from three schools in Xicheng District, Beijing. Screening-defined current myopia required unaided visual acuity <5.0 together with a non-cycloplegic spherical equivalent ≤ -0.50 D in the same eye. Official fitness scores were recalculated under the National Student Physical Health Standard (2014 revision). Prevalence ratios (PRs) were estimated with modified Poisson regression, and continuous associations were assessed using restricted cubic splines. Sensitivity analyzes incorporated school fixed effects and class-clustered standard errors. XGBoost and SHAP were used only to explore patterns in the observed data.

The analysis included 1,296 students: 642 in primary grade 1 and 654 in junior middle grade 1. Current myopia by the screening definition was identified in 496 students (38.3%), with prevalence of 6.7 and 69.3% in the two grade-stage groups, respectively. In the original adjusted models, students who failed the fitness standard had a higher myopia prevalence than those rated excellent (PR = 1.95, 95% CI: 1.36-2.79). A 1-SD lower motor-fitness score was also associated with higher prevalence (PR = 1.14, 95% CI: 1.03-1.26). After accounting for school and within-class clustering, the fail-versus-excellent estimate remained evident (PR = 1.95, 95% CI: 1.13-3.38); so did the estimate for lower motor fitness in primary grade 1 (PR = 2.05, 95% CI: 1.13-3.69). The remaining estimates pointed in similar directions but were less precise. Internal machine-learning performance was moderate, and item attributions varied by model and grade stage.

Failing the official fitness standard and having lower motor fitness were associated with screening-defined current myopia in this cross-sectional sample. The clearest association was found in primary grade 1. These results cannot establish which condition came first and may reflect non-cycloplegic misclassification, reverse causation, or shared behavioral, familial, developmental, educational, and environmental influences. Fitness measures are not ready for use in myopia prediction or screening.