Improving RCT Power Using Prognostic Score Adjustment

This tutorial presents a guide on enhancing the power of randomized clinical trials (RCTs) through the use of prognostic score adjustment for linear models. The document outlines the theoretical framework that supports the use of historical data to improve statistical power without increasing the risk of type I errors. It discusses various methodologies for estimating average treatment effects (ATE) and provides guidelines for implementing linear adjustment techniques. The tutorial includes a simulation study that compares linear prognostic score adjustment with other methods, such as propensity score matching and ANCOVA. Additionally, it details a case study involving a phase IIIb clinical trial for type 2 diabetes, showcasing how linear adjustment can lead to a reduction in participant numbers while maintaining the integrity of the trial's results. The paper concludes with a discussion of the limitations and considerations when applying this method in clinical research.