Placebo and nocebo responses in restless legs syndrome: A systematic review and meta-analysis

To estimate the placebo and nocebo responses in restless legs syndrome (RLS) and explore their determinants.

Databases were searched up to October 2015. Randomized, double-blind, placebo-controlled trials of patients with RLS were included if quantitative data were extractable in the placebo arm. Placebo response was defined as the within-group change from baseline, using any scale measuring RLS severity or disability. Nocebo response was defined as the proportion of patients experiencing adverse events in the placebo arm. Random-effects meta-analysis was used to pool data. Statistical heterogeneity was assessed with I² statistic. Several predetermined subgroup and sensitivity analysis were performed. PROSPERO registration number is CRD42015027992.

We included 85 randomized controlled trials (5,046 participants). Pooled placebo response effect size was -1.41 (95% confidence interval [CI] -1.56 to -1.25, 64 trials, I² = 88.1%), corresponding to -6.58 points in the International RLS Study Group Scale (IRLS). Pooled nocebo response was 45.36% (95% CI 40.47%-50.29%, 72 trials; I² = 89.8%). The placebo and nocebo responses were greater in trials with longer duration, evaluating pharmacologic interventions and idiopathic RLS, and in industry-funded and unpublished studies. The placebo response was considerably smaller in objective as compared to subjective outcomes. In addition, the nocebo response increases proportionally with the placebo response, and has the same predictors.

The magnitude of the placebo response in RLS is above the threshold of minimal clinical important difference, and the frequency of adverse events is also considerable. These results are relevant to inform the design and interpretation of future clinical trials.