Purchase this article with an account.
G.–S. Ying, Vision in Preschoolers Study Group; Impact of Confidence Number on the Accuracy of the Retinomax and SureSight Autorefractors as Screening Tools . Invest. Ophthalmol. Vis. Sci. 2006;47(13):693.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To assess whether repeated testing to achieve a reading with higher confidence number improves the accuracy of screening by the Retinomax and SureSight Autorefractors.
Nurse and lay screeners trained in the use of Retinomax and SureSight Autorefractors screened 1452 children enrolled in the VIP Phase II study. All of these children also received a comprehensive eye examination. Using the data from children (∼30%) who had repeat readings because of a low confidence number for one or both eyes on the initial reading, we determined the screening failure criteria that maximized the overall sensitivity with specificity set to 90% and 94%. We compared the sensitivities (at both 90% and 94% specificity) from using the lowest confidence number reading vs. using the highest confidence number reading for the same children. The comparison includes the sensitivity for detecting any conditions, conditions grouped by severity (Group 1, 2, 3) and by the 4 VIP–targeted conditions (amblyopia, strabismus, significant refractive error, and unexplained decreased visual acuity (VA)). Statistical comparison of correlated proportions was performed.
For the Retinomax Autorefractor (N=970 pairs of repeated readings), the confidence number of the repeated test was on average 3 units better, with 88% readings above the recommended minimum value (8). Sensitivity improved by 0% to 9% at 90% specificity, and by 2% to 19% at 94% specificity with the higher confidence number. The improvements in sensitivity are significant for detecting any conditions (3% at 90% specificity, 13% at 94% specificity), Group 1 condition (15% at 94% specificity), Group 2 condition (16% at 94% specificity), significant refractive error (14% at 94% specificity) and reduced VA (7% at 90% specificity, 19% at 94% specificity).
For the SureSight Vision Screener (N=801 pairs of repeated readings), the confidence number of the repeated test was on average 2 units better, with 82% of readings above the recommended minimum value (6). The improved confidence numbers only minimally changed the sensitivity at 90% specificity (range of change in sensitivity: –2% to 6%), but significantly improved the sensitivity at 94% specificity for detecting any conditions (10%), Group 1 conditions (12%), Group 2 conditions (12%), significant refractive error (12%) and amblyopia (18%).
Higher confidence number of a reading from repeated testing improved the accuracy of Retinomax and SureSight Autorefractors, suggesting repeated testing to reach the manufacturers’ recommended minimum value is worthwhile in screening.
This PDF is available to Subscribers Only