ABSTRACT
Purpose: To determine repeatability and agreement of the ARK-30 handheld autorefractor with retinoscopy under cycloplegic and noncycloplegic conditions in children. Methods: Three consecutive autorefractor measurements (with and without cycloplegia) and retinoscopy were performed and compared in 30 randomized eyes of 30 children (mean age of 6.7 ± 2.7 years with spherical equivalent [SE] refraction from ?4.01 to +7.38 D) in a cross-section and masked study. Bland–Altman analysis of autorefractor measurements (with and without cycloplegia) and agreement with retinoscopy were calculated with conventional notation (sphere [Sph] and cylinder [Cyl]) and vector notation (SE, J0, and J45coefficients). Results: ARK-30 measurements without cycloplegia were lower than under cycloplegic conditions (Sph: ?0.52 ± 2.37 D vs + 0.86 ± 2.60 D, P < 0.01; Cyl: ?0.83 ± 0.80 D versus ?0.78 ± 0.77 D, P = 0.37; and SE: ?0.94 ± 2.19 D vs + 0.47 ± 2.44 D, P < 0.01, respectively) and statistically different (P < 0.03) from retinoscopy (Shp: +0.83 ± 2.66 D; Cyl: ?0.71 ± 0.87 D; SE: +0.51 ± 2.49 D). Without statistical differences were in J0and J45coefficients. Cyloplegic autorefraction measures were not found to be statistically significantly different to retinoscopy measures. ARK-30 under cycloplegia shows better repeatability with lower limits of agreement (LoA) in Sph (LoA: ?0.66 to +0.69 D), and SE (LoA: ?0.66 to +0.65 D) than without cycloplegia (LoA: ?1.45 to +1.77 D, and ?1.38 to +1.74 D, respectively). Conclusion: Under noncycloplegic conditions, ARK-30 autorefractor has low repeatability and a tendency toward minus over correction in children. However, repeatability and agreement with retinoscopy under cycloplegic conditions allow use of ARK-30 in children to estimate refraction but not to substitute gold standard retinoscopic refraction.