Efficient Study Design and Analysis of Longitudinal Dose–Response Data Using Fractional Polynomials

• DOI: 10.1002/pst.2425
• Publication Date: 2024-07-29T00:34:47Z
• Authors: Benjamin F. Hartley; Dave Lunn; Adrian P. Mander
• Keywords: 03 medical and health sciences; 0302 clinical medicine

ABSTRACT Correctly characterising the dose–response relationship and taking correct dose forward for further study is a critical part of drug development process. We use optimal design theory to compare different designs show that using longitudinal data from all available timepoints in continuous‐time model can substantially increase efficiency estimation compared single timepoint model. give theoretical results calculate gains large class these models. For example, linearly growing Emax population with between/within‐patient variance ratio ranging 0.1 1 measured at six visits be estimated between 1.43 2.22 times relative gain, or equivalently, 30% 55% reduced sample size, final timepoint. Fractional polynomials are flexible way incorporate repeated measurements, increasing precision without imposing strong constraints. Longitudinal models two fractional polynomial terms robust mis‐specification true process while maintaining, often large, gains. These have applications interim analyses.