Comparison of measurements of refractive errors between the hand-held retinomax and on-table autorefractors in cyclopleged and noncyclopleged children

PURPOSE: To compare the measurement of refractive errors (sphere, cylinder, and axis) between the hand-held Retinomax and on,table Topcon autorefractors in cyclo-pleged and noncyclopleged young children. The average bias and measurement agreement were assessed. DESIGN: Observational cross,sectional study. METHODS: The study included 114 cyclopleged and 156 noncyclopleged young children. The mean difference between the two methods and the 95% limits of agreement were calculated to evaluate the average bias. Two types of analyses were conducted to assess the degree of agreement. First, the proportion of the absolute mean differences was presented in different ranges (less than or equal to0.25, 0.25-0.5, 0.5-0.75, 0.75-1.0, and >1.0 diopters for sphere and cylinder; 0-10, 11-20 and >20 degrees for axis). Second, the paired t test was conducted to evaluate the consistency of two types of measurements. RESULTS: The data by the Retinomax had mild bias (0.59 diopters) toward a lower sphere data under noncycloplegia but no bias under cycloplegia. For cylinder and axis, there was either no bias or clinically acceptable bias (0.02-0.13 diopters for cylinder and 2-7 degrees for axis) regardless of cycloplegia. Besides the sphere data under noncycloplegia, in general 90% of the mean differences of sphere and cylinder were within 0.5 diopters. More than 97% of the difference in axis under cycloplegia and 68% under noncycloplegia were within 20 degrees. After adjusting for mild bias, the paired t test showed very consistent results. CONCLUSIONS: The data by the Retinomax were consistent with those by the Topcon. The Retinomax is a useful instrument to screen refractive errors in young children. (C) 2003 by Elsevier Inc. All rights reserved.