Abstract
PURPOSE: To determine the proportion of preschoolers testable with the Random Dot E (RDE) stereotest and the between-tester reliability. METHODS: Subjects were 1257 3- to 5-year-olds who were participants in Head Start programs in five communities. The sample was over-weighted with children who failed the routine Head Start vision screening (58% failures; 42% nonfailures). Each child had stereotesting attempted by two licensed eye care professionals using the RDE test, first in a screening setting and later in a comprehensive eye examination. Pretesting was performed at 50 cm. Children who discriminated between the nonstereo E and blank cards on 4 of 4 or 4 of 5 presentations were considered testable. Stereotesting was conducted at 50, 100, and 150 cm corresponding to 504, 252, and 168 arcsec disparity). Extent of agreement beyond chance between the screening and examination test results was assessed using the kappa (kappa) and weighted kappa (kappa(w)) statistics. RESULTS: Testability increased with age at both the first (86% of 3-, 89% of 4-, and 93% of 5-year-olds; p = 0.02) and the second (90% of 3-, 94% of 4-, and 98% of 5-year-olds; p = 0.0001) sessions. Overall, stereopsis was better at the second session. Agreement between sessions was moderate (kappa(w) = 0.43; 59% identical scores). While identical scores increased with age (53% of 3-, 59% of 4-, and 63% of 5-year-olds), kappa(w) statistics did not differ with age (p = 0.49). CONCLUSIONS: Child testability on the RDE stereotest was significantly better for older than younger preschool-age children. Overall, test-retest agreement was moderate and did not differ according to age. Testability rates of 98% and higher have been achieved in the same population with other screening tests including another random dot stereotest.