A systematic review of the study design and statistical practices in randomized controlled trials on myopia treatment (2019-2023).

BACKGROUND: Myopia is a global public health concern affecting billions worldwide, with its prevalence rising at an alarming rate. High-quality RCTs for evaluating new interventions to slow myopia progression are needed. This systematic review aims to analyze the study designs and statistical practices of randomized controlled trials (RCTs) on myopia treatment published between 2019 and 2023. METHODS: A systematic search of PubMed and Embase was conducted in February 2024 and May 2025 using keyword-based and MeSH term-based queries, filtered for RCTs published between January 1, 2019, and December 31, 2023. All retrieved studies were manually screened to ensure they reported primary results from RCTs investigating myopia treatment or progression control, assessing at least one intervention for efficacy and/or safety. RESULTS: 73 trials performed in more than 15 countries across 36 journals were identified. The most common interventions involved atropine (n = 19, 26.0%), lens (n = 22, 30.1%), surgery (n = 9, 12.3%), and light therapy (n = 8, 11.0%). These trials used one-eye design (n = 31, 42.5%), two-eye design (n = 27, 37.0%), paired-eye design (n = 9, 12.3%), mixed-eye design (n = 3, 4.1%). The most common primary outcomes were spherical equivalent refraction (n = 48, 65.8%), axial length (n = 45, 61.6%), and uncorrected distance visual acuity (n = 10, 13.7%). 66 trials (91.7%) utilized cycloplegia for measuring refractive error. Among 39 trials involving two eyes, inter-eye correlation was accounted for in the analysis of 23 trials (59.0%), with 8 (34.8%) using a paired t-test, 12 (52.2%) using a mixed effects model, and 3 (13.0%) using generalized estimating equations. Among 60 trials with missing data, 24 trials (40.0%) did not mention any statistical methods for missing data. CONCLUSION: The methodological rigor of myopia RCTs can be improved by carefully considering eye study designs, employing robust statistical techniques, ensuring cycloplegia for refractive error measurements, and properly accounting for inter-eye correlation and missing data when applicable. Standardizing these practices will improve the quality of future myopia trials.