Factors associated with U15-to-U17 roster reappearance in Chinese female youth basketball players: a retrospective longitudinal cohort study

Abstract Background Relative age effects (RAE) are well documented in youth basketball, but whether age-related patterns at U15 are associated with later roster reappearance in multi-birth-year competitions remains unclear. This study examined factors associated with U17 roster reappearance among Chinese female youth basketball players, with particular attention to within-year relative age and birth year. Methods This retrospective longitudinal cohort included 439 players from 37 teams in the 2023 National U15 Women’s Basketball Championship, matched against the publicly available 2025 U17 roster. Roster reappearance was defined as presence on that roster. Birth-quarter distribution was assessed against equal-quarter and population-based sensitivity references. Sequentially adjusted logistic regression included continuous relative age, birth year, anthropometric characteristics, playing position, minutes per appearance, and mGmSc/min. Team-cluster-robust and other sensitivity analyses were performed. Results Eighty-five players reappeared at U17 (19.4%, 95% CI 15.9–23.3%). The U15 birth-quarter distribution was non-uniform (χ²(3) = 19.943, p < 0.001, Cohen’s w = 0.213), with a similar result using the population-based reference (χ²(3) = 19.395, p < 0.001). In the fully adjusted model ( n = 435), continuous relative age was not clearly associated with roster reappearance (OR = 1.040 per 30 days later, 95% CI 0.973–1.112, p = 0.244), and birth year showed no clear overall association ( p = 0.773). Greater height was associated with higher odds of roster reappearance (OR = 1.080 per 1 cm, 95% CI 1.027–1.136, p = 0.003), as were greater minutes per appearance (OR = 1.264 per 5 min, 95% CI 1.030–1.552, p = 0.025). The latter association was attenuated after additional adjustment for games played (OR = 1.217, 95% CI 0.981–1.509, p = 0.074). No clear adjusted association was observed for mGmSc/min. Model performance was modest (Nagelkerke R²=0.094; AUC = 0.684). Conclusions The U15 cohort showed a birth-quarter distribution consistent with an RAE in roster representation, but neither within-year relative age nor birth year showed a clear association with subsequent U17 roster reappearance. Height showed the most consistent positive association, whereas the association with competition exposure was less robust after accounting for games played. These findings concern U17 roster reappearance in national competition records and do not establish ongoing membership within the same talent-development pathway.