Research question: Will providing free glasses to myopic rural Chinese students, with a teacher incentive to promote use, increase academic high school attendance?
Design: Cluster-randomised controlled trial
Rationale: Rural Chinese children have high myopia prevalence, but poor access to glasses. Our previous trials show giving free glasses significantly improves academic performance, with greater effect size than parental education or family income, equaling or exceeding other classroom-based medical interventions. Non-vocational (academic) high school attendance is only 30% in rural western China. Strong evidence of educational benefit from glasses is needed to spur adoption of national distribution programs.
Methods: Children in Year 1 at 111 randomly-selected middle schools in Liaoning, Northern China, will be randomized by school to receive free glasses and a trial-proven teacher-based incentive to promote wear (Intervention) or prescriptions only (Control). The main outcome 3 years later will be high school attendance (powered to detect 10% difference between study groups); secondary outcomes of compliance and use of near versus distance classroom learning aids will elucidate biological plausibility of a causal pathway between myopia correction and learning. Local knowledge and attitudes about myopia and spectacle use and intervention cost-effectiveness will be studied.
Statistical methods to be employed in the analysis and justification for the choice of sample size:
Adjusted- and unadjusted-comparison of the difference between study groups
Principal analyses for the main trial will include:
* Of attendance at academic high school after Year 3 of middle school (as opposed to following a vocational pathway or leaving school).
* of observed spectacle wear at un-announced examinations at school and for self-efficacy scores.
* of mental health outcomes after Year 1 year of intervention
Sample size: Assuming 120 students per school undergoing vision screening, 50% of children failing vision screening, 30% of these needing glasses, α=0.05, intra-class correlation=0.10, explained variation by covariates=0.40, difference between the groups in the main study outcome=10%, high-school attendance rate in the Control group=50%, a sample size of 111 middle-schools (55 or 56 in each group) will provide power=80%.
Loss to follow-up in our previous school-based trials in China, SWLW I and II, was approximately 4% over the course of one school year, and so our calculations here assume a loss to follow-up of 10% over 2-3 years. However, we have not adjusted for non-compliance because we will use a strict intention-to-treat analytic approach where children in schools allocated to the Intervention group will be analysed as such irrespective of compliance and because our intention is to test the policy of making free glasses available to those students who need them, rather than the effect of those students actually wearing the glasses. Further, our expected effect size is based conservatively on previous trials with compliance rates of approximately 40%, which we feel can be improved on in the current trial with teacher incentives as above.
