Abstract
Purpose: :
To develop a predictive model for the onset of myopia during childhood.
Methods: :
Non-myopic children enrolled in the Collaborative Longitudinal Evaluation of Ethnicity and Refractive Error (CLEERE) Study in the first grade who had at least one additional visit are reported here (n=1999). Incident myopia by the eighth grade was defined as at least -0.75 D of myopia in both meridians by cycloplegic autorefraction. Candidate first grade predictors included spherical equivalent refractive error; number of myopic parents; diopter-hours of near work; hours of sports/outdoor activity; axial length; lens thickness; corneal power; Gullstrand lens power; AC/A ratio; relative peripheral refraction; accommodative lag; and astigmatism. We fitted a discrete time survival analysis model with age, gender, and ethnicity entered as covariates, and the log of the hazard odds was computed for each candidate predictor. The best, final model was the model with the lowest average deviance comprised by significant predictors.
Results: :
The baseline data in first grade (average ± sd) were: age: 6.74 ± 0.45 years; ethnicity: 48% White, 23% Hispanic, 14% African-American, 13% Asian, 2% Native-American; 51% female; spherical equivalent: 0.88 ± 0.75 D; 55% with at least one myopic parent; 28.05 ± 20.66 D-hrs/week; sport/outdoor activity: 9.14 ± 7.78 hours/week; axial length: 22.56 ± 0.69 mm; lens thickness: 3.54 ± 0.16 mm; corneal power: 43.63 ± 1.44 D; Gullstrand lens power: 21.29 ± 1.46 D; AC/A ratio: 3.44 ± 1.96 pd/D; relative peripheral refraction: -0.71 ± 0.91 D; lag: 1.09 ± 0.81 D; astigmatism J0: -0.02 ± 0.38 D; J45: 0.15 ± 0.30 D. The best model predicting myopic onset by the eighth grade incorporated (OR; p-value): AC/A ratio (1.1; 0.0073), axial length (3.17; <0.0001), corneal power (1.55; <0.0001), astigmatism (both J0 [0.46; <0.0001] and J45 [0.48; 0.0084]), and spherical equivalent (7.14; <0.0001).
Conclusions: :
The onset of myopia in childhood can be predicted with solely clinically measurable variables assessed at school entry. Future refinements to the model will include older baseline age, age of onset, and metrics of predictive model performance.
Keywords: myopia • refractive error development • refraction