Test-retest variability of Randot stereoacuity measures gathered in an unselected sample of UK primary school children

Abstract

To determine the test-retest reliability of the Randot stereoacuity test when used as part of vision screening in schools. METHODS: Randot stereoacuity (graded-circles) and logMAR visual acuity measures were gathered in an unselected sample of 139 children (aged 4-12, mean 8.1+/-2.1 years) in two schools. Randot testing was repeated on two occasions (average interval between successive tests 8 days, range: 1-21 days). Three Randot scores were obtained in 97.8% of children. RESULTS: Randot stereoacuity improved by an average of one plate (ie, one test level) on repeat testing but was little changed when tested on the third occasion. Within-subject variability was up to three test levels on repeat testing. When stereoacuity was categorised as 'fine', 'intermediate' or 'coarse', the greatest variability was found among younger children who exhibited 'intermediate' or 'coarse'/nil stereopsis on initial testing. Whereas 90.8% of children with 'fine' stereopsis (</=50 arc-seconds) on the first test exhibited 'fine' stereopsis on both subsequent tests, only approximately 16% of children with 'intermediate' (>50 but </=140 arc-seconds) or 'coarse'/nil (>/=200 arc-seconds) stereoacuity on initial testing exhibited stable test results on repeat testing. CONCLUSIONS: Children exhibiting abnormal stereoacuity on initial testing are very likely to exhibit a normal result when retested. The value of a single, abnormal Randot graded-circles stereoacuity measure from school screening is therefore questionable.