Abstract
PURPOSE: To quantify the value of using prior changes in spherical equivalent refractive error (SER) and axial length (AL) to predict future myopia progression. METHODS: For this post hoc analysis of a randomized controlled trial, we used data from children ages 5 to 12 years with SER -1.00 to -6.00 D who had been randomized 2:1 to 0.01% atropine or placebo eye drops for 24 months. Multivariable linear regression evaluated the association of baseline-to-12-month change in SER and AL versus 12-to-24-month change while controlling for age and SER or AL at 12 months. Treatment groups were pooled for analyses; sensitivity analyses were conducted using only the placebo group. RESULTS: Among 187 children, 136 (73%) with complete data were included. For predicting a 0.50-D-or-more SER increase of myopia in the second 12 months based on observing a 0.50-D-or-more increase of myopia in the first 12 months, the positive predictive value was 42% (19 of 45; 95% confidence interval [CI], 29%-57%). Greater baseline-to-12-month SER change was weakly associated with greater 12-to-24-month SER change (0.20 D per additional 1.00 D; 95% CI, 0.02 to 0.39; P = 0.03; partial R2 = 0.03). The 95% prediction interval half-width for 12-to-24-month change was ±0.66 D with prior change versus ±0.67 D without (difference = 0.01 D; 95% CI, -0.05 to 0.07). Analyses of AL and sensitivity analyses limited to the placebo group were qualitatively similar. CONCLUSIONS: Changes in SER and AL of children in the prior 12 months were poor predictors of future myopia progression, limiting their usefulness for clinical decision-making or selecting participants for clinical trials.