Reliability of Stereoacuity Tests for the Detection of Ocular Conditions in Pre-School Children: the Sydney Pediatric Eye Disease Study

Abstract

Purpose: : To assess stereoacuity testing using Randot and Lang II in screening for unilateral visual impairment, anisometropia and strabismus in pre-school children

Methods: : The Sydney Pediatric Eye Disease Study examined 2461 children during 2007-9. Stereoacuity testing was attempted on all participants aged ≥ 30-72months, using the Lang II stereo card (Forch) and the Randot Preschool Stereoacuity(Stereo Optical Company) held at 40cm.Monocular visual acuity (VA) was measured using the Electronic Visual Acuity system,and in a subset LogMAR chart. Cover and prism tests were used to detect strabismus at fixation distances of 33cm and 6m. Cycloplegic auto-refraction was performed using Canon RK-F1 or Retinomax K-Plus 2.Unilateral visual deficit (VD) was defined as presenting VA ≥ 2 line difference between two eyes, anisometropia as SER difference ≥1.00D between eyes, and strabismus as a heterotropia at distance and near.For each disparity level of the Randot and Lang II tests, sensitivity and specificity, receiver operator characteristics (ROC) curves and area under curve (AUC) were calculated for the detection of unilateral visual impairment, anisometropia and strabismus.

Results: : Randot with disparities 40,60,100,200,400,800secs/arc produced an AUC=0.79 for unilateral VD. ROC curves indicated the Randot 60secs/arc was the most discriminative level, with a sensitivity and specificity of 85% and 61% respectively. This was also found for anisometropia, where 60secs/arc had highest selectivity,sensitivity and specificity 71% and 59% respectively AUC=0.72. Selectivity of strabismus with Randot was highest at 100secs/arc disparity sensitivity of 73% and specificity of 85%, AUC=0.85. Lang II test with disparities of 200,400,600 secs/arc had lower AUC of 0.60 for strabismus and 0.57 for both unilateral visual impairment and anisometropia. This test had the highest selectivity at disparity of 200secs/arc using ROC curves, sensitivity and specificity respectively for unilateral visual impairment was 19% and 91%, for anisometropia 29% and 88% and for strabismus 37% and 88%.

Conclusions: : In a pre-school setting, the Randot test was found to be more reliable than Lang II, with results suggesting disparities of 60 to 100 secs/arc to have more diagnostic value in detecting binocular dysfunction.

Keywords: binocular vision/stereopsis • screening for ambylopia and strabismus • infant vision