META-ANALYSIS TO ESTIMATE EXPECTABLE DROPOUT RATES IN RANDOMIZED CONTROLLED CLINICAL TRIALS ONAGE-RELATED MACULAR DEGENERATION - SUPPORTING FOR OPTIMIZED SAMPLE SIZE CALCULATION

Topic:Over- or underestimation of participants in clinical trials may lead to serious problems regarding ethics, economics or significance of results. A realistic net sample size calculation is essential for planning; which also includes an expectable dropout profile to ensure sufficient number of subjects for statistical analysis. This meta-analysis determine dropout rates in randomized and controlled clinical trials (RCTs) on age-related macular degeneration (AMD), to optimize sample size calculation according to the requirements of the current guidelines(BGBI (2004), Moher et al., 2010). Methods: A meta-analysis was conducted to estimate the dropout rate in RCTs of AMD. Data collection was performed by a full text handsearch in six subject-specific journals of the Anglo-American language area with peer review system and high impact factors(publication period01/2004- 12/2013).Raw data extraction on reported dropout rates was performed by two independent parallel readers.Meta-estimation of the reported 12 and 24-month dropout rates was based on the random effects model. Results:45 RCTs (16,555 patients) out of 1,062 publications were included in the analysis after 12-month follow-up, and 22 RCTs (7,845 patients) met the inclusion criteriafor the24-month follow-up period. The meta dropout rate for the primary endpoint was estimated as 9.2% (95% CI 7.1% - 11.9%) and increased up to 19.0% (95% CI 14.2% - 25.0%) after 24 months. Conclusion:Sample size calculation in clinical trials for AMD should account for dropouts of at least 10% during a 12-month follow-up and of at least 20% during a 24-month follow-up period.