Interobserver agreement for measurement of grating acuity and interocular acuity differences with the Teller acuity card procedure

Interobserver reliability of the Teller acuity card (TAC) procedure for estimating acuity and interocular acuity differences (IADs) was assessed with 342 infants and children who had been treated in a neonatal intensive care unit for preterm birth and/or perinatal complications. Subjects were tested binocularly at term and monocularly at 4, 8, 11, 17, 24, 30, 36, and 48 months corrected age with TACs. Testers were masked to the location and spatial frequency of the grating on each card. Of the interobserver test-retest scores, 67% differed by no more than 0.5 octave, and 87% of the test pairs differed by no more than 1 octave. Of the test-retest comparisons of a subject's IAD, 54% showed agreement of 0.5 octave or better, and 76% differed by no more than 1 octave. Interobserver agreement for binocular and monocular tests was similar to that reported previously for visually and neurologically at-risk infants and children tested with the forced-choice preferential-looking procedure or with prototype acuity cards. Interobserver agreement for IAD estimates was somewhat less than that reported for a sample of infants with ocular disorders. There were no systematic differences in interobserver agreement between eyes tested first and eyes tested second, nor was interobserver agreement related to subject's medical diagnosis. Interobserver agreement was influenced, however, by the spatial frequencies of the particular gratings used during testing and, to a limited extent, by the age of the child. The duration of individual tests and observers' ratings of confidence in their acuity estimate were not reliable indicators of test-retest pairs that were not in agreement. The results demonstrate the reliability of the TAC procedure, but suggest that acuity estimates critical to a patient's diagnosis or treatment should be confirmed by repeat testing.