The limitations of centile curves for evaluating myopic eye growth
Mark A. Bullimore, Xu Cheng, Noel A. Brennan
Abstract
SIGNIFICANCE: Pediatric growth charts are widely used to track height and weight. Recently, axial length growth charts have been developed. Unfortunately, they underestimate the rate of normal myopic eye growth, making it challenging to evaluate the benefits of myopia control interventions, due to the conflation of myopes and nonmyopes. PURPOSE: The aim is to assess the value of axial length centile curves in the management of childhood myopia. METHODS: Papers reporting centile curves were identified by searching PubMed. For comparison, axial length values for a representative selection of baseline values (21 to 24 mm at 6 years) were calculated as a function of age and ethnicity using published meta-analyses of myopic and emmetropic eye growth data. RESULTS: Six published centile curves, largely based on cross-sectional data, were identified: three from European populations, two from China, and one from India. The trajectory of the emmetropic eye growth model generally tracks the European and Indian centile curves at lower centiles. This is not the case for the Chinese centile curves, likely due to the significant numbers of myopic children even at lower centiles. In contrast, the trajectory of the myopic eye growth model is steeper than that of the centile curves, even at higher centiles. This suggests that the higher centiles contain substantial numbers of nonmyopic children. Only in the centile curves for Chinese children, who have a higher prevalence of myopia, do they approach myopic eye growth, and then only for older children and at higher centiles. CONCLUSIONS: Centile curves do not accurately represent myopic eye growth, are not the best tool to monitor myopia progression and treatment, do not accurately represent growth in incident myopes, and are not the best way to predict myopia onset. Separate centile curves for myopic eyes do not alleviate the problem because of incident myopia. Annualized growth models may provide a better approach to assessing axial elongation relative to population norms.