Of the 1386 nonmyopic children at baseline after cycloplegia, newly developed myopia occurred in 271 participants over 2 years of follow-up. At baseline, there were 54 participants with myopia after cycloplegia. At the 1-year follow-up, the number of participants with myopia increased to 165 after cycloplegia, indicating 111 children became myopic by the end of 1 year. At 2 years, there were 325 children with myopia after cycloplegia, indicating 160 children newly became myopic over the second year of the follow-up. In analyses of the potential determinants of incident myopia, 111 children were included for myopia onset over 1 year and 271 over 2 years. Interestingly, nonmyopic children were stratified to hyperopia and emmetropia for sub-sample analysis because of the significant association of baseline SE with the incident myopia. Among children with baseline hyperopia, the incidence of myopia was significantly lower with baseline HAT ≥0.5 D (9.7%,
n = 949), compared to children with baseline HAT <0.5 D (24.3%,
n = 288,
P < 0.001). Multivariable logistic regression result found that children who had less baseline SE (odds ratio [OR] = 0.05, 95% confidence interval [CI] = 0.03 to 0.10,
P < 0.0001), decreased baseline HAT (OR = 0.43, 95% CI = 0.28 to 0.67,
P < 0.001) and lower 1-year change in HAT (OR = 0.36, 95% CI = 0.23 to 0.54,
P < 0.0001) had elevated risk in developing myopia over 2 years. There was no collinearity among baseline SE (VIF1.11), baseline HAT (VIF2.84) and 1-year change in HAT (VIF2.66). Among children with baseline emmetropia, there was no significant difference in the incidence of myopia between emmetropic children with baseline HAT ≥0.5 D (67.2%,
n = 58) and <0.5 D (78.0%,
n = 91,
P = 0.14). Multivariable logistic regression result found that only baseline SE (OR = 0.16, 95% CI = 0.02 to 0.92,
P = 0.050) and lower 1-year change in HAT (OR = 0.56, 95% CI = 0.30 to 0.92,
P = 0.043) were in borderline significantly associated with the incidence of myopia over 2 years (
Supplementary Table S1).