A Comparison of Two Automated Tests of Refractive Error

Abstract

Purpose: Although early detection of visual disorders is critical, this is difficult as current screening tests often require tester expertise and/or place excessive cognitive demands on the child. Objective, automated tests of refractive error offer an alternative over traditional vision screening tests. Here we provide the first comparison of the validity and screening effectiveness of two devices that measure refractive error; the Welch-Allyn SureSight Autorefractor (WASS) and the PlusoptiX S09 vision screener (PS09).

Methods: Participants included 61 children between the ages of 6 months and 12 years (Mean=5.8 yr±3.2 yr). Thirty-seven participants were attending eye exams with one of two local ophthalmologists. The remaining 24 participants were tested as part of a local preschool vision screening program and then received eye exams with one of two local optometrists. All participants were tested with the WASS and the PS09, and all subsequent eye exams included cycloplegic retinoscopy (CR). To determine the validity of each device, refractive error measures were compared to those obtained using CR by Coefficient of Repeatibility (COR) analyses. To determine screening effectiveness, the sensitivity, specificity, and accuracy of each device was calculated based on refractive error diagnoses from CR.

Results: Compared to CR, both devices underestimated hyperopia (CR=1.41D, WASS=0.64D, PS09=0.39D, all p<0.01). Estimates of cylindrical refractive error obtained using the PS09 and CR did not differ (0.83D and 0.74D, respectively, p=0.21), while the WASS provided higher estimates of cylindrical refractive error than CR (1.07D v. 0.74D, respectively, p<0.001). COR analyses indicated that compared to the WASS, the PSO9 showed better agreement with CR on both spherical (CORs = 2.74D v. 4.03D) and cylindrical refractive error (CORs = 1.05D v. 1.27D). Despite these findings, the WASS yielded higher sensitivity and accuracy scores than the PS09 (sensitivity = 81% v. 63%; accuracy = 80 v. 72%). The two devices did not differ on specificity (79%).

Conclusions: Both devices underestimated hyperopia and the WASS overestimated astigmatism. Although the PS09 showed better agreement with CR on both spherical and cylindrical refractive error, the WASS yielded superior measures of accuracy and sensitivity. However, screening effectiveness data must be interpreted cautiously as they depend on the chosen referral criteria.

Keywords: 709 screening for ambylopia and strabismus • 676 refraction