Abstract
Purpose :
Accurate epidemiological modelling requires differentiation of the effects of age, sex, and birth cohort. The repeat cross-sectional design and raw data availability of the National Health and Nutrition Examination Studies (NHANES) of the US and South Korea provide an opportunity to investigate these effects on refractive error in contrasting countries.
Methods :
We analyzed the age- and sex-specific mean, variance, skewness and kurtosis of refractive error distributions, plus prevalence of refractive error at four cut-points (≤-5D, ≤-0.5D, ≥+1D, ≥+3D) on the earliest (US = 1999, Korea = 2008) and latest (US = 2007, Korea = 2018) available refractive data.
Results :
The Figures show high and low myopia prevalence by age in 1999 (US) and 2008 (Korea). They illustrate a familiar shape, with prevalence increasing until 25 years of age then decreasing until the mid-60s before a cataract-related rise. The birth cohort effect is demonstrated in the Figures by the labels and scaled vertical arrows. Teenage myopia prevalence appears to be consistently rising, e.g. myopia in 37.9% of US boys born in 1983, rising to 41.3% of US boys born in 1991. However, the 25-year-old myopia prevalence peak appears to be stabilizing in Korea, with myopia in 80.7% of those born in 1983 compared to 79.6% of those born in 1990. Prevalence is rising significantly in middle-age groups of both countries – e.g. myopia in 52.7% of 45-year-old Korean women born in 1963, rising to 73.3% in 45-year-old Korean women born in 1973 – turning the 25-year-old peak into a plateau. The older-age, cataract-related increase in myopia appears to be decreasing over time in Korea, particularly in males – potentially indicating improved cataract surgical rate.
Conclusions :
Isolated, single-time-point cross-sectional, age-specific refractive error prevalence data is common. However, longitudinal models are difficult to construct without differentiating birth cohort effects. There appears to be significant cohort effects with age- and sex-specific aspects in these contrasting countries, even when measured over relatively short periods. These results reinforce the need for sex-disaggregated data and further exploration of birth-cohort effects in refractive error distributions.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.