Abstract
Purpose :
With increasing emphasis on slowing myopia progression and using axial length to monitor eye growth, normative data for axial elongation in myopic children provide valuable clinical information. Percentile growth curves have been presented but are specific to the population tested and often include myopes and non-myopes, limiting their application. We conducted a meta-analysis to model axial elongation in myopic children with emphasis on the influence of age and race.
Methods :
A comprehensive electronic systematic search was performed using Ovid Medline, EMBASE, Cochrane Central Register of Controlled Trials and the following terms: (myopia OR myopic) AND (child OR children) AND (progression OR longitudinal OR follow-up OR shift) AND axial. There were 79 studies with 105 untreated subpopulations and 203 evaluations of the mean axial change from baseline that met the inclusion criteria. Log(mean rate of axial elongation) was analyzed using a weighted multivariable linear mixed effects meta-analysis model. All collected covariates were tested for significance (age, race and baseline axial length selected). The model included three levels of random effects to account for all variability.
Results :
Only data without missing observations of significant covariates were included in the final model (118 evaluations). The figure plots modeled annual axial elongation (with prediction intervals) by mean age of evaluation and race. Mean axial elongation decreases as age increases (15.0% decrease per year; 95%CI 11.4–18.5%, P <0.0001) and is greater in Asian children (by 27.9%; 95%CI 7.6–52.2%, P <0.01) compared to non-Asians. No other input variables, including baseline axial length, were statistically significant.
Conclusions :
This analysis sets benchmark values for assessing axial elongation and monitoring myopia progression. To our knowledge, this is the first weighted random effects meta-analysis of axial elongation in myopic children. Interpretation is limited by use of aggregated data rather than individual subject data. The large prediction intervals should be borne in mind when interpreting individual rates of axial elongation clinically.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.