Monte Carlo for Clinical Trial Biostatisticians

Six months before the analysis. The FDA reviewer asks why the protocol’s expected power of 0.90 doesn’t match the simulator’s 0.78 under realistic non-proportional hazards. The room goes quiet. The biostatistician opens a textbook and finds the Schoenfeld formula. The reviewer waits.

Across fourteen chapters built around four worked-example trials — a Phase 3 immune-checkpoint comparison, a Phase 1 dose-finding study, a Phase 2 basket trial with hierarchical borrowing, and a rare-disease hybrid design with a propensity-weighted external control — Dr. Ingrid Voss develops Monte Carlo simulation as the working biostatistician’s primary tool for the regulatorily-relevant questions of the late 2020s. You will learn to size a trial against the data-generating process it actually produces, to choose a primary analysis the regulator will accept, to defend every cell of every operating-characteristics table at the Type C meeting, and to write simulators a third party can run on a clean machine five years later and reproduce your reported numbers to the digit.

The R code is in the appendix. The opinions are explicit. The discipline is what thirty-one years of regulatory negotiations have actually produced.