ABSTRACT
BACKGROUND: Substantial placebo response has been observed in trials assessing treatments in Raynaud's Phenomenon (RP), which makes any treatment effect difficult to detect. However, whether this response is due to a real placebo effect or to other nonspecific effects, such as regression towards the mean (RTM), has not been explored. Our objectives were to explore and quantify placebo response in RP, and to evaluate the magnitude of RTM contribution. METHODS: We combined trial-level and individual-level data from a series of n-of-1 trials and a network meta-analysis, respectively. Main outcomes were the daily frequency and the mean duration of RP attacks, as well as the Raynaud's Condition Score (RCS). We estimated the placebo response by the mean difference between the placebo period (or arm) and the baseline. RTM was estimated by the relationship between placebo response and baseline, and with Galton squeeze plots. Finally, we simulated the effect of the threshold used for inclusion in clinical trials on RTM. FINDINGS: We observed a large and significant placebo response from both individual and trial data for RCS [-1.20 (-1.63, -0.77) and -0.65 (-0.89, -0.41)] and the daily frequency of RP [-0.61 (-0.85, -0.37) and -0.75 (-0.95, -0.54)]. Outcome at baseline was significantly associated with placebo response, suggesting the presence of RTM. The latter was confirmed on individual data, through Galton squeeze plots. INTERPRETATION: Placebo response is large in RP trials, and likely due to regression towards the mean rather than 'true' placebo effect. This should be carefully considered when designing future trials. FUNDING: This work has been partially supported by MIAI @ Grenoble Alpes (ANR-19-P3IA-0003).
