Influence of active versus placebo control on treatment responses in randomised controlled trials in rheumatoid arthritis

Abstract

Objectives To investigate whether treatment effects of pharmaceutical compounds compared with placebo controls are systematically different to the effects of the same compounds compared with active treatment controls in rheumatoid arthritis (RA) clinical trials. Methods We systematically identified randomised controlled trials (RCTs) in RA, and matched active treatment arms with comparable regimens, populations, background therapy, and outcome reporting, by the nature of their control group (active comparator or placebo). Medline, EMBASE and CENTRAL were used to identify RCTs investigating disease modifying anti-rheumatic drug therapies until December 2021. Using mixed-model logistic regression we estimated OddsRatios (OR) for achieving an American College of Rheumatology (ACR) 20/50/70% response at weeks 12 and 24. Risk of bias was assessed using the Cochrane Tool. Results We screened 8328 studies and included 40 for analysis after detailed review of 590 manuscripts; unique compounds had significantly higher responses in active comparator trials compared with their effects observed in placebo controlled trials, with ORs of 1.67 (95% CI 1.46 to 1.91; p<0.001) for ACR20, 1.50 (95% CI 1.29 to 1.75; p<0.001) for ACR50 and 1.65 (95% CI 1.30 to 2.10; p<0.001) for ACR70 (week 12); corresponding ORs for ACR 20, 50, and 70 (week 24) were 1.93 (95% CI 1.50 to 2.48; p<0.001), 1.75 (95% CI 1.32 to 2.33; p<0.001) and 1.68 (95% CI 1.21 to 2.34; p<0.001), respectively. Sensitivity analyses showed consistent results. Conclusion Placebo controlled trials lead to smaller effect sizes of active compounds in RCTs compared with the same compound in head-to-head trials. This difference may be explained by potential nocebo effects in placebo-controlled settings and needs to be considered when interpreting head-to-head and placebo-controlled trials, by patients, investigators, sponsors and regulatory agencies.