Effect of a Randomized Interventional School-Based Vision Program on Academic Performance of Students in Grades 3 to 7 - A Cluster Randomized Clinical Trial (Vision for Baltimore)

2,304 students who received eye examinations and eyeglasses; 120 schools across 3 randomized cohorts (964 / 775 / 565 students in 41 / 41 / 38 schools)

Grades 3-7 in Baltimore City Public Schools, 2016-2019; 77.6% Black, 16.8% Latinx, 17.6% in special education; mean age 9.4

Stepped cluster RCT - schools randomized 1:1:1 to receive eye exams and eyeglasses in one of three successive school years, so later cohorts serve as controls for earlier ones. ITT with hierarchical linear modelling. Two independent standardized outcomes were used and they disagree, which is the single most informative feature of this trial: i-Ready (commercial adaptive progress-monitoring) showed an effect; PARCC (the state accountability assessment) did not.

Overall 1-year impact on i-Ready reading was ES 0.09 (P = .02) in school year 2016-17. Larger subgroup effects: female students ES 0.15, students in special education ES 0.25, and students in the lowest baseline quartile ES 0.28 (all P < .001); elementary-grade i-Ready mathematics ES 0.03. Critically, the intervention did NOT show sustained impact at 2 years, and showed NO impact on PARCC testing at all. The pattern is a small first-year gain on one standardized measure that neither persists to year 2 nor replicates on the higher-stakes state test.

Low for the treatment contrast: schools were randomized. Worth noting for the hereditarian lens that the largest effects were in the lowest-achieving baseline quartile and in special education - exactly where an uncorrected sensory barrier is most likely to have been misattributed to low ability.