Analyzing Radiographic Outcomes from Longitudinal Rheumatoid Arthritis Trials

Jingyuan Yang1, Lifen Zhou2, Grace Park2, Huei Wang2 and Yun Chon2
1Ohio State University, OH
Amgen Global Biostatistics and Epidermiology, Thousand Oaks, CA

Abstract

Structural joint damage in rheumatoid arthritis (RA) can be assessed by radiographic imaging. Radiographic scores quantify the severity of the disease, and thus become popular endpoints in longitudinal trials for efficacy of new care of RA. In this report, statistical characteristics of radiographic outcomes are discussed, and conventional non-parametric and parametric methods for analysis of radiographic outcomes are reviewed. We propose new parametric methods to account for the special distribution of the radiographic outcomes. Conventional and proposed methods are applied to simulated data generated in two different ways and to data from two clinical studies. Based on our assessment, we recommend using cross-sectional non-parametric tests to compare treatment groups in longitudinal trials of rheumatoid arthritis.