Maarten Cozijnsen

63 Chapter 3 TISKids: an international multicenter RCT Data collection and monitoring Data are collected in Castor EDC (26), a web-based CRF database enabling the central study coordinators to follow and check the CRF input of each of the collaborating centers online. Additionally, a certified monitor will visit each site every year. A Data Safety Monitoring Board was not appointed, as the risks of adverse events associated with this study are considered low, because only approved therapies are used and treatment is not blinded. Statistic methods Subject baseline and demographic data as well as baseline disease characteristics data will be summarized by treatment group. Parametric variables will be described by their mean and standard deviation, and compared with use of the T-test, and non-parametric variables will be described by their median and interquartile range and compared using the Mann- Whitney U test. Categorical variables will be summarized using counts and percentages, and compared using the Chi-squared test, or the Cox proportional hazard test in case of time-dependent categorical variables. Correlations will be assessed using either the Pearson correlation coefficient (parametric) or the Spearman’s rank correlation coefficient (non-parametric). Analyses will be performed on an intention-to-treat basis. All statistical testing will be 2-sided and significant at the 0.05 level. Missing data will be reported and left out of the analyses.