To quantify the impact of the measurement of non-cycloplegic refraction on estimates of refraction and classification of refractive status in young adults.

This cross-sectional analysis included 825 young adults (mean age 26.6±3.3y; 56.7% female) from the Guangzhou Twin Eye Study. All participants underwent both non-cycloplegic and cycloplegic autorefraction. Error in refraction was defined as the difference between cycloplegic and non-cycloplegic spherical equivalent refraction (SER).

Using a cut-off of -0.50 D for non-cycloplegic refraction to diagnose myopia, the prevalence of myopia was overestimated (90.2% vs 87.0%), resulting in the identification of 3.2% (26/825) pseudo-myopes. After increasing the threshold to -0.75 D, the overall prevalence (87.5%) was approximately equivalent to that with cycloplegic refraction, 9 of the 26 pseudo-myopes would have been re-classified correctly as emmetropes, but 13 genuine myopes would have been re-classified incorrectly as emmetropes. Lack of cycloplegia resulted in a significant myopic shift compared to cycloplegic refraction. Notably, 25.6% of participants exhibited a clinically significant myopia refractive shift of ≥0.50 D, and 5.7% demonstrated a large shift of ≥1.00 D.

At a threshold of -0.75 D, the calculated prevalence of myopia is close to that obtained with cycloplegic refraction. However, threshold adjustment does not solve the mis-classification problem, because while it will eliminate some pseudo-myopes, it will also eliminate some genuine myopes. Our study confirms that non-cycloplegic refraction leads to significant measurement errors in young adults, highlighting the necessity of cycloplegia for accurate epidemiologic assessment of myopia prevalence.