This book is meant to provide practitioners the background theory and practical knowledge to apply covariate adjustment in randomized trials. The main focus of the material is on marginal estimates of treatment effects for various quantities of interest (estimands). The material should be accessible to individuals with a background in probability, statistics, regression modelling, time-to-event/survival analysis. Familiarity with causal inference methods and clinical trial methodology may be beneficial, but are not necessary.
Example data are provided which are simulated to mimic key features of actual randomized trials, including the distribution of covariates and patterns of missingness. Code is provided to help users assess covariate balance, perform unadjusted and covariate-adjusted analyses, interpret results, and assess the potential gains in efficiency. Worked examples are provided for continuous, binary, ordinal, and time-to-event outcomes.
After addressing fixed sample size designs, group-sequential and information adaptive designs are addressed.