Prognostic performance of baseline spherical equivalent and axial length for progression to high myopia in children: a retrospective cohort study

To evaluate baseline spherical equivalent (SE) and axial length (AL) as prognostic markers for progression to high myopia in myopic children and assess the incremental value of short-term changes. This retrospective cohort study included 2,129 myopic eyes of 1,309 children with baseline manifest refraction and axial-length biometry obtained within 90 days of each other. Marginal Cox proportional-hazards models were fitted for combined high myopia (SE < − 6.00 D or AL > 26.00 mm), refractive high myopia (SE < − 6.00 D), and axial high myopia (AL > 26.00 mm). Annualized SE and AL changes were evaluated in exploratory predictor-anchored landmark analyses using the >3-month and >6-month change windows in 829–996 eyes, depending on the outcome. The study assessed comparative prognostic performance rather than an individualized prediction model. During follow-up, combined, refractive, and axial high myopia developed in 168 eyes (7.9%), 123 eyes (5.8%), and 100 eyes (4.7%), respectively. For the combined outcome, the median follow-up duration was 1.04 years among eyes without combined high myopia and 1.87 years among eyes that developed combined high myopia.The model including age, baseline SE, and baseline AL yielded C-indices of 0.943, 0.952, and 0.972 for the three outcomes, respectively. Among the individual baseline markers, baseline SE and AL discriminated comparably for combined high myopia (difference in C-index, 0.015; 95% CI, − 0.036 to 0.066; P = 0.565), whereas SE showed higher discrimination for refractive high myopia and AL for axial high myopia (both P < 0.001). The 3-year cumulative incidence of combined high myopia was 32.7% among eyes with baseline SE ≤ − 2.38 D and 36.5% among those with baseline AL > 24.14 mm. In exploratory landmark analyses, adding annualized AL change to the baseline model improved continuous net reclassification for the combined and refractive outcomes in both windows and for the axial outcome in the > 6-month window, whereas annualized SE change improved it only for the combined outcome in the > 6-month window; improvements in the other incremental measures were inconsistent. Baseline SE and AL were strong and complementary prognostic markers. The high discrimination partly reflects the definitional relationship between these baseline markers and the threshold-based outcomes. Adding longitudinal AL change to baseline age, SE, and AL may provide additional prognostic information, although its clinical relevance requires prospective validation.