Development and validation of predictive models based on the first-month axial length change for the long-term efficacy of orthokeratology in myopic children

Aims: To identify simple and practical predictors and develop predictive models for the long-term efficacy of orthokeratology (ortho-k) treatment in myopic children.

Methods: The training and external validation cohorts were composed of children with myopia who received ortho-k treatment from two separate prospective studies conducted at Zhongshan Ophthalmic Centre, respectively. Baseline variables and the first-month change in axial length (AL) were entered into logistic regression to develop predictive models to assess the 1-year efficacy of ortho-k treatment on retarding AL elongation. Nomograms and an online calculator were used to present the final models, and receiver operating characteristic (ROC) curves, calibration curves and decision curve analysis (DCA) were used to assess model performance in both the training and external validation cohorts.

Results: The training and external validation cohorts comprised 150 and 88 children with myopia, respectively. Age and the first-month AL change were included in the final predictive model for excessive myopia progression with area under the ROC curves (AUC) above 0.84. Only the first-month AL change was included in the final predictive model for rapid myopia progression with AUC above 0.85. All the discrimination, calibration and clinical net benefit performed excellently in the training and external validation cohorts. The models were integrated into an open-source online risk calculator.

Conclusions: The first-month AL change and age are simple and practical predictors for the long-term efficacy of ortho-k treatment in children with myopia. Predicting long-term ortho-k efficacy after the first month will provide eye care practitioners with guidance to evaluate subsequent outcomes and adjust follow-up plans.