Translating Treatment Effects Between Rheumatoid Arthritis Activity Measures and American College of Rheumatology Responses in Direct Comparison Trials.

Abstract

Direct comparison trials in rheumatoid arthritis (RA) increasingly use changes in continuous disease activity measures as endpoints. However, the between-arm differences in these scores that are clinically meaningful are uncertain. To aid interpretation of clinical trials that use the Disease Activity Score in 28 joints (DAS28) or Simplified Disease Activity Index (SDAI) as endpoints, we developed statistical equivalences between changes in these measures and American College of Rheumatology (ACR) responses. For superiority trials, we computed the minimal detectable difference in DAS28 changes and SDAI changes that correspond to the ACR criteria for 20% improvement (ACR20) and 50% improvement (ACR50) responses at the same type I and type II errors and same sample size. For noninferiority trials, we computed noninferiority margins that were statistically equivalent across measures. Standard deviations of the changes in the DAS28 and SDAI from a recent observational study were used as the basis of calculations in our examples. In the base scenario with type 1 error 0.05 and power 0.80, a trial with 300 subjects per arm would detect a 0.31-point difference in mean DAS28 change scores and 3.71-point difference in mean SDAI change scores as statistically equivalent to an absolute difference of 11% in ACR20 between treatment arms. We developed a web-based utility that provides equivalent differences among these measures for customized sample sizes, error rates, and standard deviations of the DAS28 and SDAI between-arm differences. The DAS28 and SDAI responses can be related to statistically equivalent changes in ACR responses, which can aid the interpretation of trials that use these measures.