Purchase this article with an account.
Marjean T. Kulp, Gui-Shuang Ying, Jiayan Huang, Maureen G. Maguire, Graham E. Quinn, Elise Ciner, Lynn A. Cyert, Bruce Moore, Deborah A. Orel-Bixler, Vision In Preschoolers Study Group; Validity of Noncycloplegic Retinoscopy, Retinomax Autorefractor and SureSight Vision Screener for Detecting Significant Refractive Errors. Invest. Ophthalmol. Vis. Sci. 2011;52(14):1601.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate, by receiver operating characteristic (ROC) analysis, the ability of non-cycloplegic retinoscopy (NCR), Retinomax Autorefractor (Rmax) and SureSight Vision Screener (SSight) to detect significant refractive errors (RE) among 3- to 5-year-old Head Start children.
In the 2-year VIP Phase I, 2588 preschoolers were screened with NCR, Rmax and SSight by pediatric eye care professionals. In VIP Phase II, 1452 preschoolers were screened with Rmax and SSight by nurse and lay screeners. All children underwent cycloplegic retinoscopy by masked examiners, and significant RE was defined as: hyperopia >+3.25D; myopia >-2.00D; astigmatism >1.50D; and anisometropia >1.00D interocular difference in hyperopia, or >3.00D interocular difference in myopia, or >1.50D interocular difference in astigmatism. The ability of each screening test to identify any significant RE and by RE type and by severity was summarized by the area under the ROC curve (AUC), calculated from weighted logistic regression models. Detection of RE type was based upon the child’s worse eye and using the following SSight, Rmax, and NCR
most positive meridian for hyperopia, most negative meridian for myopia, cylinder power for astigmatism, and maximum intereye difference in sphere and cylinder for anisometropia.Results: AUC was better for detecting the most severe levels of RE only than for the most and moderately severe levels combined (AUC .97 to 1.00 vs. .92 to .96). AUC of each screening test was high for myopia (AUC .97 to .99). NCR and Rmax performed better than SSight for hyperopia (AUC .92 to .99 and .90 to .98 vs. .85 to .94, p≤0.02), Rmax performed better than NCR for astigmatism (AUC .95 to .97 vs. .90 to .96, p=0.01), and SSight performed better than Rmax for anisometropia (AUC .85 to 1.00 vs. .76 to .96, p≤0.01).
No test consistently outperformed the others. AUC was excellent for the most severe levels of RE and very good for detection of significant RE overall.
This PDF is available to Subscribers Only