Abstract
PURPOSE: The purpose of this study was to generate childhood and adolescent refractive centile curves for East Asian and Western populations to facilitate evidence-based myopia control. METHODS: Spherical equivalent refractive (SER) error from 70,022 eyes of participants (aged = 6-20 years) from 9 large population-based studies were included (40,132 European/North American/Australian, and 29,890 East Asian). Both cycloplegic (n = 36,344) and non-cycloplegic refraction (n = 47,314), with some studies using both techniques. A new method of adjusting the distribution of cycloplegic and non-cycloplegic refractive data was developed and validated. RESULTS: Large regional differences were found in refractive centiles, with the centiles in East Asians being more myopic than the corresponding centiles in Western children and adolescents after age 8 years. The magnitude of these regional differences among the 50th, 75th, and 90th myopia centiles increased with increasing age until stabilizing at approximately ages 16 to 18 years in female patients, higher centiles (i.e. more myopic eyes) showing less stabilization in male subjects. Multi-Gaussian models fit raw SER distributions better than Box-Cox Power Exponential (BCPE) models, particularly for more highly skewed SER distributions. Using a multivariable linear regression model, estimated cycloplegic SER distributions from non-cycloplegic SER were not significantly different from the measured distribution (Cramér-von Mises test, all P ≥ 0.41). Separate cycloplegic and non-cycloplegic refractive centiles were developed for both regions. CONCLUSIONS: Population-based centile analysis offers the prospect of estimating the risk of myopia development in individuals and monitoring the treatment efficacy of myopia control interventions. This meta-analysis shows the need for age-, sex-, and region-specific reference curves.