This work found that an extra hour spent outside every day during childhood was associated with a hyperopic shift of +0.53D to +0.94D, and an extra hour spent reading every day was inversely associated with a myopic shift of -0.44D to -0.88D. Photo: Artem Kniaz on Unsplash.

The current increase seen in myopia development worldwide has been attributed at least in part to lifestyle risk factors; however, the known variables—chiefly, the extent of outdoor time and near vision work in a child’s life—explain little about wide variations in refractive error recorded in the population. Consequently, researchers from the UK wanted to investigate how so-called “instrumental variable” (IV) methods of analysis might explain these factors as they relate to myopia. This method is designed to avoid reverse causation, decrease confounding bias and to gauge lifestyle risk factor effect sizes. An instrumental variable is one that isn’t directly related to the outcome but is a factor in the actions or exposure being studied.

 Included in the analysis were three myopia risk factors: time outdoors, time reading and sleep duration. These were assessed in participants of the Avon Longitudinal Study of Parents and Children, with a sample of 2,302 kids aged 15 years old and a longitudinal sample of 3,086 kids who were followed from seven to 15 years old. Seven IVs were jointly considered, including dog ownership, cat ownership, bedtime variability, birth order, polygenic scores quantifying genetic predisposition to spend additional time outdoors, years in full-time education and time spent asleep overnight.

The study researchers found that effect sizes of risk factors were fourfold to ninefold higher when using the IV analyses when compared with conventional regression analyses. With IV analyses, one extra hour spent outside every day during childhood was associated with a hyperopic shift of +0.53D to +0.94D, and an extra hour spent reading every day was inversely associated with a myopic shift of -0.44D to -0.88D. There was inconsistent evidence to suggest that sleep duration influenced refractive error.

The authors of the study discuss in their paper their four main findings, the first being that risk factor effect sizes were consistently underestimated in all three risk factors when using standard observational analysis methods. They explain that this underestimation likely occurred because of bias from unmeasured confounders or residual confounding.

They also discuss how only patchy evidence was found supporting a relationship between sleep duration and myopia development. The authors speculate that the complex pattern of relationships between sleep duration and other risk factors of myopia may contribute to a spurious association seen in prior reports. 

Third on their list of findings was their confirmed genetic contribution to sleep duration in adulthood being distinct from in childhood. The investigators emphasize that due to this finding, valid Mendelian randomization studies analyzing the effects of differences in children’s sleep duration on other traits like myopia won’t be possible until single nucleotide polymorphisms have been identified to be strongly associated with sleep duration in childhood.

Finally, they highlight that genetic predisposition to educational attainment and genetic predisposition to spend time outdoors were both associated with child sleep duration. Put succinctly, the authors write that “these findings are indicative of widespread (horizontal) pleiotropy and suggest that IV and MR studies of sleep duration, educational attainment and time outdoors may inadvertently capture the effects of multiple exposures, resulting in the underestimation or overestimation of causal effects of the index exposure.”