Autorefractors

Automated refraction is another method of vision screening preschool children.  The most widely available autorefractors include the Welch Allyn SureSight and the Nikon Retinomax both which were formally validated by the VIP study. The  autorefractors utilize ultrasonic measurements of the wavefront to estimate refractive error, which can then be used to predict the actual refractive error in screened children.  However, these instruments cannot control accommodation and therefore their estimates do not correlate well with refractive error determined under cycloplegic conditions.

Condonnier and Kallay have used the handheld Retinomax autorefractor to detect refractive errors in 12,012 children in Brussels, Belgium.  They found this device to have a sensitivity ranging from 37-87% and a positive predictive value from 19-69% depending on the type of pathology observed.  The VIP study found more definitive results reporting a positive predictive value ranging from 50-60%, a 85% sensitivity to detect amblyopia and a 63% overall sensitivity to detect all targeted conditions (amblyopia, strabismus, significant refractive error, and unexplained reduced VA)all at a specificity of 90%.  The VIP study found the Retinomax and SureSight to have similar high sensitivities for all targeted conditions which exceeded those of all other tested methodologies.

The handheld SureSight measures and indicates abnormal readings without provider interpretation.  The SureSight has been much more extensively studied than the Retinomax, likely due to its lower cost, high marketability, and high visibility.  Multiple referral criteria have been proposed for the SureSight device.  The criteria proposed by the manufacturer have the highest sensitivity but the highest referral rate and the lowest positive predictive value.  The VIP study demonstrated the SureSight to have a sensitivity of over 90% to detect high magnitude refractive error when the manufacturer’s criteria were used, but experience from the Vanderbilt group in field studies is that this set of criteria yields substantial over-referrals with a positive predictive value under 10%.  When specificity was set to 90%, the results of the VIP study indicated that in the hands of eye care professionals, the SureSight is as accurate as the LEA Symbols test, the Retinomax Autorefractor, and noncycloplegic retinoscopy in detecting children who had one or more of the targeted conditions (amblyopia, strabismus, significant refractive error, and unexplained reduced VA).  

The VIP study proposed a second set of referral criteria in an attempt to increase specificity for the SureSight.  These criteria had a post-hoc specificity of 94% in increasing the predictive value to 50% but decreased the referral rate substantially.  Unfortunately, more specific referral criteria are not currently commercially available and it is unclear whether further modification of criteria can produce corresponding increases in referral rate and predictive value without substantially jeopardizing sensitivity.