General methods

All main analyses will be performed on the mITT and PP populations and complementary analyses will be performed on the PP population. The primary analysis and all secondary analyses of time-to-event variables between the two randomised groups will be analysed unadjusted with the log-rank test, and the results will be displayed with Kaplan-Meier graphs. Event incidence will be portrayed through event rates, calculated as the total a number of events divided by the total number of follow-up time. To estimate the precision, 95% CIs will be computed using the exact Poisson limits. The magnitude of the effect size will be described by HRs derived from Cox proportional hazards models. In time-to-event models, where not all causes of death are included, competing risk will be considered.

The study used optimal allocation based on donor age, gender of the donor, recipient age, gender of the recipient and recipient preoperative ventricular assist device (yes/no). For adjusted analysis between the two randomised groups regarding time-to-event variables, multivariable Cox proportional hazards models will be used adjusted for the stratification variables. If the proportional hazard assumption is not fulfilled for some covariates, then these variables will be stratified for in the multivariable Cox model.

If any confounding variables are identified, those that exhibit significant associations with both the outcome and the treatment group, complementary analyses will be conducted adjusting for these statistically identified confounders. For unadjusted comparison between the two randomised groups, Fisher’s exact test will be used for dichotomous variables, Fisher’s non-parametric permutation test for continuous variables, Mantel-Haenszel χ² trend test for ordered categorical variables and χ² test for non-ordered categorical variables. For analyses of number of events, Poisson regression models will be considered.

For all unadjusted and adjusted analyses between the two groups, mean differences and ratios with 95% CI will be calculated whenever possible. All significance analyses will be two sided and conducted at the 5% significance level if not otherwise stated.

Standard methods for descriptive analysis will be used. Continuous variables will be described with mean, SD, median, first quartile, third quartile and categorical variables with numbers and percentages.