Exploring the effect on primary endpoints in trials testing targeted therapy interventions for rheumatoid arthritis: a meta-epidemiological study on the appropriate use of a core outcome set.

Objectives: To explore which core domain is best associated with the American College of Rheumatology (ACR) 20% response in trials assessing the effect of targeted interventions in rheumatoid arthritis (RA).

Methods: A meta-epidemiological study was performed on randomised trials investigating biologics and targeted agents compared with placebo or conventional disease-modifying antirheumatic drugs in patients with RA. The main outcome measures were ORs for the ACR 20% response and at least one of the eight core domains according to the existing RA core outcome set (COS) analysed based on standardised mean differences.

Results: 115 trials involving 55 422 patients with RA were eligible. The OR for achieving ACR 20% response was 3.19 (95% CI 2.96 to 3.44) for the experimental interventions relative to the comparators. The median number of COS domains reported was 6; 18 trials reported only 1 domain, 17 all 8. Univariable meta-regression analyses indicated that each of the eight core domains was significantly associated with ACR 20% response, yet improvements in physical disability explain a successful ACR 20% response the most. Including only trials reporting on all eight core domains, univariable meta-regression analyses proved improvement in fatigue to explain a successful ACR 20% response the most.

Conclusions: Within this dataset, it is evident that the conclusions concerning our primary objective were significantly influenced by both the amount and characteristics of missing data. Our data suggest that fatigue could be more important for the primary endpoint than previously assumed, but this is based on limited data.