Background To assess predictive value of short-term choroidal changes for future myopic shift in children.

Methods 577 eyes of 289 primary school children were prospectively followed for 2 years. Cycloplegic refractions at baseline, 1 year and 2 years, and choroidal measurements by optical coherence tomography at baseline and 3 months, were used for analyses. Myopic shift was defined as refraction change of at least −0.50 dioptre/year, at 2 years compared with baseline.

Results 228 participants (455 eyes) completed 2-year follow-up. Approximately 37.6% of 311 initially non-myopic eyes and 73.6% of 144 initially myopic eyes developed a myopic shift. Notably, at 3 months greater reductions were found in initially myopic eyes with myopic shift, than in those without myopic shift—in choroidal thickness (ChT), luminal area (LA), stromal area (SA) and total choroidal area (TCA), but no significant differences in any choroidal parameters were observed between non-myopic eyes, with and without myopic shift. Multivariable analyses showed that in myopic eyes, each percentage increase in ChT, LA, SA and TCA was associated with reduced odds of myopic shift (all p<0.001). Similar associations were observed in non-myopic eyes, with smaller effects than in myopic eyes. Adding a 3-month percentage change of each choroidal parameter to a basic model including age, gender, parental myopia and baseline refraction significantly improved the predictive performance in myopic eyes (area under the receiver operating characteristic curves increasing from 0.650 to approximately 0.800, all p<0.05), but not in non-myopic eyes.

Conclusion Short-term choroidal changes could act as early indicators for future myopic shift in children.