Factors Influencing Placebo Responses in Rheumatoid Arthritis Clinical Trials: A Meta-Analysis of Randomized, Double-Blind, Placebo-Controlled Studies.
Kota Nagai, Keisuke Matsubayashi, Kazuki Ide, Kahori Seto, Yohei Kawasaki, Koji Kawakami
Author Information
Kota Nagai: Department of Pharmacoepidemiology, Graduate School of Medicine and Public Health, Kyoto University, Yoshida-Konoe-cho, Sakyo-ku, Kyoto, 606-8501, Japan.
Keisuke Matsubayashi: Department of Pharmacoepidemiology, Graduate School of Medicine and Public Health, Kyoto University, Yoshida-Konoe-cho, Sakyo-ku, Kyoto, 606-8501, Japan.
Kazuki Ide: Department of Pharmacoepidemiology, Graduate School of Medicine and Public Health, Kyoto University, Yoshida-Konoe-cho, Sakyo-ku, Kyoto, 606-8501, Japan.
Kahori Seto: Department of Pharmacoepidemiology, Graduate School of Medicine and Public Health, Kyoto University, Yoshida-Konoe-cho, Sakyo-ku, Kyoto, 606-8501, Japan.
Yohei Kawasaki: Biostatistics Section, Clinical Research Center, Chiba University Hospital, 1-8-1 Inohana, Chuo-ku, Chiba-shi, Chiba, 260-8677, Japan.
Koji Kawakami: Department of Pharmacoepidemiology, Graduate School of Medicine and Public Health, Kyoto University, Yoshida-Konoe-cho, Sakyo-ku, Kyoto, 606-8501, Japan. kawakami.koji.4e@kyoto-u.ac.jp. ORCID
BACKGROUND AND OBJECTIVE: A better understanding of placebo responses and the specific factors influencing these outcomes is important for clinical trial design. We investigated the magnitude of placebo responses at 3 months and the potential factors influencing these outcomes in rheumatoid arthritis (RA) clinical trials. METHODS: We conducted a systematic review of randomized placebo-controlled trials of pharmacological agents for RA identified from PubMed, Embase, and Cochrane Central Register of Controlled Trials databases. The primary placebo outcome was American College of Rheumatology 20% response rate (ACR20). Data were pooled with a random-effects model. Factors influencing placebo response were assessed by meta-regression analyses. Subgroup analyses were performed for studies conducted in non-Western countries only versus in Western countries (North America and/or Europe) only or both. RESULTS: The meta-analysis included 88 studies comprising 8406 patients receiving a placebo. The pooled estimate of placebo ACR20 was 29.0% (range 10.0-46.2; 95% confidence interval 27.2-30.9). Placebo ACR20 was negatively associated with trials in non-Western (Asian) countries and patient populations showing an inadequate response to biological disease-modifying antirheumatic drugs (DMARDs) in the multivariable analysis, whereas it was positively associated with the year of publication. No background DMARD treatment was also a negative predictor (albeit statistically non-significant). In subgroup analyses of Western and multiregional studies, study population and publication year were significant factors. CONCLUSIONS: Our meta-analysis suggests that study location, patient population, and a background DMARD treatment influence placebo ACR20. These along with placebo response temporal profiles have important implications for designing and interpreting RA clinical trials.
