Distinguishing myopia intervention candidates at premyopia stage in children: a nomogram based on primary refractive error screening parameters

Abstract Background: Hyperopia reserves of premyopia could be used to identify myopia intervention candidates for Chinese children. Primary refractive error screening parameters are commonly employed in clinical and community settings before cycloplegic assessment of hyperopia reserves; however, their utility in distinguishing intervention candidates at the premyopia stage remains underexplored. This study aimed to develop a nomogram based on these routinely measured parameters to support clinical decision-making for early myopia prevention. Methods: Pediatric patients (aged 4–17 years) from two medical centers in China were enrolled in this retrospective cohort study. A predictive model for the candidates of myopia intervention was developed using logistic regression with multiple imputations. The model included the following primary screening parameters: age, gender, uncorrected visual acuity (UCVA), average corneal curvature (ACC), non-cycloplegic spherical equivalent refraction (NCSER), axial length (AL), and the axial length to average corneal radius of curvature (AL/ACRC) ratio. The efficacy of the model was assessed using the area under the receiver operating characteristic (ROC) curve, calibration curves, and decision curve analysis (DCA). R was employed to conduct all statistical analyses. Results: A total of 1006 participants (507 females, 499 boys) were enrolled, with 87.4% demonstrating CSER ≤ + 0.75D. In multivariate logistic regression, UCVA, NCSER, AL, and ALTOACRC were identified as independent predictors. These predictors were incorporated into a nomogram to predict the candidates of myopia intervention. The nomogram exhibited exceptional discrimination in the derivation set (AUC = 0.971, 95% CI: 0.957–0.984), whereas in the external validation set, the AUC was 0.921 (95% CI: 0.866–0.976) when a cutoff of 0.851 in derivation set was employed. Calibration was verified through the calibration curve and Hosmer-Lemeshow tests (P = 0.99 and P = 0.96, respectively), and the decision curve analysis demonstrated robust clinical utility for threshold probabilities of 0.10–1.00 in the derivation set and 0.20–1.00 in the external validation set. Conclusion: The nomogram derived from the parameters of primary refractive error screening has the potential to preliminarily predict myopia intervention candidates at the premyopia stage, thereby facilitating clinical decision-making in the context of early myopia prevention.