Abstract
Purpose :
There is rapid eye growth in early life and children who will become myopic have longer axial lengths, culminating in failure of emmetropization and childhood-onset myopia. Using a life course epidemiology approach we examined multiple candidate myopia risk factors during critical periods of childhood development and ocular growth.
Methods :
TEDS is a longitudinal birth cohort studied from a neurodevelopmental perspective. Subjective refractive error was obtained on a subset (n=1991, median age 16.7 years, 42% male, 25.9% myopic). Candidate myopia risk factors were evaluated in five life stages: preconception; perinatal; pre-school (≤ 4 years); childhood (≤ 11 years); adolescence (≤ 17 years). Univariable and multivariable (adjusted for age, sex and factors significant at any earlier life stage) logistic regression models for myopia (≤ -0.75 D) at each life stage were constructed. Variance explained and predictive ability was assessed using area under the receiver operator curve (AUROC) statistics.
Results :
Consistent associations were observed with socioeconomic factors, educational attainment, reading enjoyment and cognitive variables, particularly verbal cognition, at multiple points over the life course. In the final multivariable model the following factors remained significant: maternal education (myopia odds ratio (OR) 1.33, 95% CI 1.11-1.59) at preconception, fertility treatment (OR 0.63, 95% CI 0.43 - 0.92) and summer birth (1.93, 95% CI 1.28 - 2.90) from the perinatal period, no factors during the preschool or childhood life stages, and hours spent playing computer games (OR 1.03, 95% CI 1.01-1.06) during adolescence. The total variance explained by this model was 6.9% (p<0.001) and the AUROC was 0.71 (95% CI 0.68 - 0.74).
Conclusions :
Ocular growth trajectory is associated with multiple factors during childhood development in a modern day British cohort. We replicate known predictors of myopia, such as maternal education, and identify novel associations including computer gaming (likely on video game consoles) and fertility treatment. Twin modeling on this dataset suggested 14% of refractive error variance was due to environmental factors and models examined in this study explain up to 7%.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.