Predictive Validity of Infant Acuity Measurements: A Five Year Follow–up

Abstract

Abstract: : Purpose: In a previous publication we reported an evaluation of methods for assessing visual function in a large cohort of infants at four and eight months of age using Teller Acuity Cards, Transient (TVEP) and Sweep Visual Evoked Potentials (SVEP). Correlations among the various methods of visual acuity estimation in infants were uniformly low. In this study the same cohort returned at five years of age to evaluate predictive validity and test reliability from infancy to early childhood. Methods: To determine the reliability of Teller Acuity, TVEP and SVEP and to evaluate the predictability of these estimates to Bailey–Lovie visual acuity and stereo acuity at five years old, 127 subjects returned for repeat testing. Reliability was assessed by correlation as well as by Cohen’s Kappa. For the latter analyses Teller Acuity, TVEP (amplitudes/latency), SVEP (best sweep/average of 5 sweeps) infant data were split into two groups, below the mean and above the mean then compared to 5 year old test data above and below mean values. Test–retest (infancy to 5 YO) data were assessed by correlation and step–wise regression to see which tests best predicted Bailey–Lovie visual acuity and stereo acuity at five years old. Results: Median binocular visual acuity was 20/21 ranging from 20/16 to 20/110 with only 13 subjects (10%) reporting vision worse than 20/30. No correlation of any test from 4/8 months to 5 YO Bailey–Lovie visual acuity was significant. Only 4 month TVEP amplitude was significantly related to stereo acuity at 5 YO. The coefficient of variation was lowest for TVEP latency 9% and highest for Teller Acuity 57%. In analyzing test–retest infant/child reliability by correlation, TVEP amplitudes at both 4/8 months and for two pattern sizes were significant. Splitting the tests above/below the mean produced significant Cohen’s Kappa reliability values for 4 month TVEP amplitudes from 30’ checks. Conclusions: In this analysis of predictive validity and reliability in 127 infants retested at 5 YO there were no consistent trends in significance suggesting that only broad categorical statements can be made regarding an infant’s visual potential.

Keywords: visual development: infancy and childhood • infant vision • visual development