Abstract
Abstract: :
Purpose: The Welch Allyn SureSight autorefractor has been calibrated for preschool children. We compared the SureSight against a) the Retinomax K-Plus and b) traditional subjective methods in the screening of preschool children. Methods: A subset of children (N=147; Aged 51 ± 9 months) was selected from a preschool vision screening programme (N=1907), which used tests of single letter acuity (Cambridge Crowding cards, pass at 6/6) and stereoacuity (StereoFly, pass at 100 sec). Seven strabismic children were removed from the analysis. The SureSight Prototype (WA SS, Welch Allyn Inc., S/W 3.01) was tested without cycloplegia and compared to cycloplegic retinoscopy (CR) and non-cycloplegic and cycloplegic Retinomax-K Plus (RKP, Nikon, Inc.). CR taken in a blind protocol served to define vision problems using AAO criteria (Hyperopia 2D, Myopia 1D, Astigmatism 1D, Anisometropia 1D). Older children (N=90, Aged 6-13 years) were added to the sample to determine if the WASS calibration was robust beyond preschool age. Results: A) Autorefractor bias: 1) Bias was defined as mean CR equivalent sphere (ES) subtracted from the mean Autorefractor ES. Bias for non-cycloplegic WA SS showed a small hyperopic trend (0.22D, p= .006) similar to that found for cycloplegic RKP (0.24D, p< .01). However this bias declined and changed direction with increased hyperopia in a linear fashion (p< .01). Non- cycloplegic RKP showed a strong "myopic" bias (-1.54, p<0.01), which increased in proportion to hyperopia (p< .01). B) WASS and age: The WASS showed a more hyperopic bias in older children compared to pre-schoolers, however multiple regression analysis showed this to be driven more (64%) by the difference in refractive error (ES=0.67D vs. 1.49D) than age itself (17%). C) Validity tests: The highest sensitivity (89.39%) and specificity (67.90%) was found for cyclopleged RKP. Of all the screening options the WA SS showed the highest sensitivity (72.73%) as compared to VA & SA (42.42%) and the NRK (39.39%), and the RKP showed better specificity (69.14%) than the WA SS (38.27%) and VA & SA measures (67.9%). Conclusions: 1) Cycloplegic RKP agrees closest with CR. The RKP without cycloplegia leads to a significant underestimation of hyperopia, contributing in part to its weak screening sensitivity. 2) The small hyperopic bias of the WA SS largely accounts for its higher sensitivity but lower specificity. 3) The calibration of the WA SS appears robust across school age children since group differences are explained by differences in refractive error. Removing WASS dependence on refractive error magnitude should improve its screening performance.
Keywords: refraction • hyperopia • screening for ambylopia and strabismus