Multi-Arm Non-Inferiority Tests for Survival Curves

This technical report details the methodology for calculating power and sample size for testing the non-inferiority of multiple treatment hazard rates against a common control hazard rate using Cox's proportional hazards regression in a cluster-randomized design. It explains that survival times are often not normally distributed and may be censored, making Cox's regression a suitable analytical method. The report outlines the significance testing for Cox regression coefficients and discusses the implications of cluster-randomized designs, including the potential loss of efficiency and the necessity for adjusted sample size calculations. It also covers the Bonferroni adjustment for type I error rates when multiple comparisons are made. Additionally, the report presents the statistical hypotheses for non-inferiority tests and provides examples of sample size calculations necessary for achieving desired power levels in clinical trials. The document serves as a comprehensive guide for researchers conducting non-inferiority tests in survival analysis.