Placebo response rates and potential modifiers in double-blind randomized controlled trials of second and newer generation antidepressants for major depressive disorder in children and adolescents: a systematic review and meta-regression analysis

Abstract

Children and adolescents with major depressive disorder (MDD) appear to be more responsive to placebo than adults in randomized placebo-controlled trials (RCTs) of second and newer generation antidepressants (SNG-AD). Previous meta-analyses obtained conflicting results regarding modifiers. We aimed to conduct a meta-analytical evaluation of placebo response rates based on both clinician-rating and self-rating scales. Based on the most recent and comprehensive study on adult data, we tested whether the placebo response rates in children and adolescents with MDD also increase with study duration and number of study sites. We searched systematically for published RCTs of SNG-AD in children and/or adolescents (last update: September 2017) in public domain electronic databases and additionally for documented studies in clinical trial databases. The log-transformed odds of placebo response were meta-analytically analyzed. The primary and secondary outcomes were placebo response rates at the end of treatment based on clinician-rating and self-rating scales, respectively. To examine the impact of study duration and number of study sites on placebo response rates, we performed simple meta-regression analyses. We selected other potential modifiers of placebo response based on significance in at least one previous pediatric meta-analysis and on theoretical considerations to perform explorative analyses. We applied sensitivity analyses with placebo response rates closest to week 8 to compare our data with those reported for adults. We identified 24 placebo-controlled trials (2229 patients in the placebo arms). The clinician-rated placebo response rates ranged from 22 to 62% with a pooled response rate of 45% (95% CI 41–50%). The number of study sites was a significant modifier in the simple meta-regression analysis [odds ratio (OR) 1.01, 95% CI 1.01–1.02, p = 0.0003, k = 24) with more study sites linked to a higher placebo response. Study duration was not significantly associated with the placebo response rate. The explorative simple analyses revealed that publication year may be an additional modifier. However, in the explorative multivariable analysis including the number of study sites and the publication year only the number of study sites reached a p value ≤ 0.05. The self-rated placebo response rates ranged from 1 to 68% with a pooled response rate of 26% (95% CI 10–54%) ( k = 6; n = 396). This meta-analysis confirms a high pooled placebo response rate in children and adolescents based on clinician ratings, which exceeds that observed in the most recent meta-analysis of placebo effects in adults (36%; 95% CI 35–37%) published in 2016. However, and similar to findings in adults, the pooled response rates based on self-ratings were substantially lower. In accordance with previous meta-analyses, we corroborated the number of study sites as significant modifier. In comparison to the recent adult meta-analysis, the substantially lower number of pediatric studies entails a reduced power to detect modifiers. Future studies should provide more precise and homogenous information to support discovery of potential modifiers and consider no-treatment—if ethically permissible—to allow differentiation between placebo and spontaneous remission rates. If these differ, practicing clinicians should facilitate placebo effects as an addition to the verum effect to maximize benefits. Further research is required to explain the discrepant response rates between clinician and self-ratings.

Electronic supplementary material

The online version of this article (10.1007/s00787-018-1244-7) contains supplementary material, which is available to authorized users.