Purpose To evaluate the intra‐ and inter‐examiner repeatability of cycloplegic retinoscopy in young children aged 4–5 years old. Methods Examiner 1 refracted all children in the first sample (n = 108); firstly with masked loose lenses, then using unmasked loose lenses (intra‐examiner repeatability). Examiners 1 and 2 refracted all children in the second sample (n = 97) using unmasked loose lenses, blind to the child's refractive error, presence/magnitude of habitual spectacle correction and to each other's findings (inter‐examiner repeatability). Refractions were performed on one eye chosen at random. Mean differences, 95% limits of agreement (LOAs) and confidence intervals were calculated for intra‐ and inter‐examiner repeatability of sphere, cylinder and spherical equivalent refraction (SER). Results Participants had a wide range of refractive errors (−1.50DS to +7.25DS; ≥4.50DC). Mean differences (95% LOAs) were small for both intra‐ and inter‐examiner repeatability [Intra: Sphere 0.00D (−0.85, +0.85D), Cylinder −0.03D (−0.68, +0.62D), SER −0.06D (−0.90, +0.78D); Inter: Sphere −0.08D (−0.92, +0.76D), Cylinder −0.08D (−0.75, +0.59D), SER −0.13D (−0.95, +0.69D). A statistically significant proportional bias was present for intra‐examiner repeatability of cylinder (ρ = 0.20, p = 0.04) and SER measurement (ρ = 0.19, p = 0.049). Proportional bias was not present for any other measure (p > 0.12). Examiners agreed on cylinder axis within ±20° in 71% of refractions where astigmatism of −0.75D or higher was present. 80% of intra‐ and inter‐examiner measures fell within ±0.50D for spherical and cylindrical components. Conclusions Differences of ±1.00D and ±0.75D or more for spherical and cylindrical measures respectively can be considered significant when performing cycloplegic retinoscopy on young children.
- refractive error