Statistical Analysis

We calculated the sample size for primary outcome assuming a power of 90% with 105 children required. All analyses were performed with Bayesian method. For the initial evaluation, we realized a concordance study to primary outcome analysis to test equivalence of both VA tests (standards vs app). We used Bland-Altman method and we calculated intraclass correlation coefficient for continue quantitative variables with their 95% credible intervals. VA, painfulness, attention and comprehension between both VA tests were compared with mixed beta regression model. Duration between two tests was compared with mixed gamma regression model. The probability that the coefficient associated with the tests was greater than 0 was calculated. It was considered as no difference a probability comprised between 0.025 and 0.975. Finally, a non-inferiority study was realized by calculating the mean and its credible interval regarding the posterior distribution of the difference between both tests. The non-inferiority margin was fixed at 1 for decimal scale and 0.10 for logarithmic scale. Statistical analyses were performed using the R software 2018 version 3.5.0.