Validity of noncycloplegic refraction in the assessment of refractive errors: the Tehran Eye Study

Purpose: To determine the sensitivity and specificity of noncycloplegic autorefraction for determining refractive status compared to cycloplegic autorefraction. Methods: The target population was noninstitutionalized citizens of all ages, residing in Tehran in 2002, selected through stratified cluster sampling. From 6497 eligible residents, 70.3% participated in the study, from August to November 2002. Here, we report data on 3501 people over the age of 5 years who had autorefraction with and without cycloplegia (two drops of cyclopentolate 1.0% 5 min apart, with autorefraction 25 min after the second drop). Results: Overall, the sensitivity of noncycloplegic autorefraction for myopia was 99%, but the specificity was only 80.4%. In contrast, the sensitivity for hyperopia was only 47.9%, but the specificity was 99.4%. At all ages, noncycloplegic autorefraction overestimated myopia and underestimated hyperopia. Overestimation of myopia was highest in the 21-30 and 31-40 year groups. Underestimation of hyperopia was high up to the age of 50 (20-40%), but decreased with age, to about 8% after the age of 50, down to almost 0% after 70. The difference in mean spherical equivalent with and without cycloplegia fell from 0.71 dioptres (D) in the 5-10 age group to 0.14D in those over 70. Conclusion: Use of noncycloplegic autorefraction in epidemiological studies leads to considerable errors relative to cycloplegic measurements, except in those over 50-60. The difference between cycloplegic and noncycloplegic measurements varies with age and cycloplegic refractive category, and there is considerable individual variation, ruling out adjusting noncycloplegic measurements to obtain accurate cycloplegic refractions.