Efficient referral thresholds in autorefraction-based preschool screening.

Abstract

To determine economically efficient referral criteria for a preschool vision-screening program using autorefraction. Retrospective economic evaluation of a screening protocol. Preschoolers in 2 cities received preschool-based Retinomax screening with a standard referral protocol and as-needed comprehensive eye examinations in 2012-2013. Positive predictive values and referral criteria that minimized cost per case detected were derived using data from San Francisco. These modeled referral criteria were then retrospectively tested for cost-effectiveness against other common criteria in Oakland with sensitivity analysis. Cases were defined by American Association for Pediatric Ophthalmology and Strabismus (AAPOS) criteria for amblyopia risk factors. The perspective was a third-party payer. Costs were obtained from a regional insurance provider. In San Francisco, 3974 children were screened, 631 referred, and 412 examined. Forty-eight percent of referrals, 301 children, met more than 1 of the referral criteria. Positive predictive values ranged from 7% for myopia to 56% for astigmatism. In Oakland, 2359 children were screened and 269 were examined. When applying the modeled referral criteria derived from San Francisco to the population of Oakland, the cost per case detected was $258. When compared in Oakland, the original referral criteria and criteria based on Vision in Preschoolers study cost, respectively, $424 and $371 per additional case detected. The modeled referral criteria had a lower cost per case detected across sensitivity analysis. More stringent referral criteria may reduce the cost per case detected in vision screening and allow more at-risk children to be detected with the same financial resources.