Background The effectiveness of rheumatoid arthritis (RA) treatment in improving physical functioning is most often measured using the HAQ-DI and SF-36v2 physical functioning (PF) scales. Items of the HAQ-DI assess lower level of physical functioning, such as eating, dressing, and reaching, while items of the SF-36 PF scale assess relatively higher levels of physical functioning, such as lifting and moving heavy objects or climbing flights of stairs. Combining the items of both scales may improve the ability to detect treatment effects by extending the range of physical functioning measured.
Objectives To cross-calibrate items of the HAQ-DI and SF-36 PF scales to score a single composite index of physical functioning and evaluate the ability of such index to detect secukinumab treatment effects.
Methods Adult RA patients (n=237) on methotrexate were randomized equally to receive monthly s.c. injections of secukinumab 25mg, 75mg, 150mg, 300mg or placebo. The SF-36v2 and HAQ-DI were administered at baseline and weeks 2, 4, 8, 12, and 16. A generalized partial credit interactive response technology (IRT) model was used to cross-calibrate the items of the SF-36 PF and HAQ-DI scale and weighted maximum likelihood estimation was used to score a composite PF index. Change scores from baseline to week 16 were calculated for the composite PF index and the SF-36 PF and HAQ-DI scales. Analysis of variance and effect sizes (mean change score divided by SD) were used to evaluate and compare the sensitivity of each physical functioning measure in responding to secukinumab treatment. A ratio of F-statistics was calculated to determine the relative validity (RV) of each physical functioning measure in responding to treatment in both a within groups (difference from 0) and between groups (treatment vs. placebo) analysis.
Results From the within groups analysis the composite PF index was found to be the most responsive to secukinumab treatment for the 25mg, 75mg, and 150mg groups. In each of these 3 dose groups the effect size for the composite PF index was 10%>40% larger than the effect sizes observed for the SF-36 PF and HAQ-DI scales alone. In addition, the F-statistics testing the difference in change score from 0 within each of these 3 dose groups was largest for the composite PF index, indicating a greater response to treatment. In the between groups analyses only the 150mg dose group showed significant improvement in physical functioning compared to placebo. The composite index PF index was most sensitive with an effect size of 0.49 compared to an effect size of 0.43 for the HAQ-DI and 0.32 for the SF-36 PF scale. The F-statistic was largest for the composite PF Index (F=6.6, p<0.01, RV=1.0) compared to the HAQ-DI (F=4.4, p<0.05, RV=0.67) and SF-36 PF (non-significant).
Conclusions IRT methods offer a promising approach to improve the treatment benefit assessment when measuring physical functioning in treatment studies of RA.
Disclosure of Interest A. Gnanasakthy Shareholder of: Novartis Pharmaceuticals Corporation, Employee of: Novartis Pharmaceuticals Corporation, M. Kosinski Consultant for: Novartis Pharmaceuticals Corporation, H. Kellner: None Declared, U. Mallya Shareholder of: Novartis Pharmaceuticals Corporation, Employee of: Novartis Pharmaceuticals Corporation, S. Mpofu Shareholder of: Novartis Pharma AG, Employee of: Novartis Pharma AG