Article type
Year
Abstract
Background:
Lessebo effects describe decreased effects of active treatment and placebo in randomized controlled trials (RCTs) with objectively lower chances of receiving active treatment. These effects may bias estimations of effect sizes in RCTs and systematic reviews.
Objectives:
To investigate the presence of lessebo effects in RCTs of the treatment of panic disorder with selective serotonin reuptake inhibitors (SSRIs) in adults.
Methods:
We searched PubMed to find relevant studies up to February 27, 2018. We also handsearched reference lists of relevant studies, systematic reviews and meta-analyses.
SSRI and placebo arms from randomized, double-blind, placebo- or active-controlled trials of the treatment of panic disorder with SSRIs in adults were eligible for inclusion.
We extracted the proportion of patients free from panic attacks and changes in panic disorder symptom severity. We pooled outcomes by trial design: outcomes in SSRI arms were pooled for active-controlled, and for placebo-controlled trials; outcomes in placebo arms were pooled for trials with multiple active treatment arms and for trials with only one active treatment arm.
Results:
773 records were identified, 68 abstracts were screened. 49 full-text articles were assessed for eligibility. 53 trial arms consisting of 36 SSRI arms and 17 placebo arms from 27 studies were included in this study. Random-effects meta-regressions demonstrated that the proportion of patients free from panic attacks in SSRI arms was lower in placebo-controlled trials compared to active-controlled trials (48.3% vs 63.0%, P = .007) in the ITT-sample. However, this lessebo effect was accounted for by systematically lower completion rates in placebo-controlled compared to active-controlled trials. No further lessebo effects were identified.
Conclusions:
Our study indicates that lessebo effects are quite small, if they exist in RCTs of the treatment of panic disorder with SSRIs in adults. Our data also suggest that lessebo effects found in ITT-analyses may be, at least partially, due to systematic differences in completion rates across trial designs. Analyses of higher power are needed to clarify the size and the robustness of lessebo effects. We emphasize that trial design should be considered whenever treatment effects in RCTs are interpreted.
Patient or healthcare consumer involvement:
Even though no patients or healthcare consumers were directly involved in this investigation, this study's analyses were based on previously collected data from patients participating in RCTs.
Since lessebo effects may bias estimations of effect sizes derived from RCTs and systematic reviews, patients and health care consumers referring to such studies may over- or underestimate the beneficial effect of a treatment, just like investigators and clinicians. Furthermore, lessebo effects may contribute to understanding subjective experiences made by patients while participating in a RCT.
Lessebo effects describe decreased effects of active treatment and placebo in randomized controlled trials (RCTs) with objectively lower chances of receiving active treatment. These effects may bias estimations of effect sizes in RCTs and systematic reviews.
Objectives:
To investigate the presence of lessebo effects in RCTs of the treatment of panic disorder with selective serotonin reuptake inhibitors (SSRIs) in adults.
Methods:
We searched PubMed to find relevant studies up to February 27, 2018. We also handsearched reference lists of relevant studies, systematic reviews and meta-analyses.
SSRI and placebo arms from randomized, double-blind, placebo- or active-controlled trials of the treatment of panic disorder with SSRIs in adults were eligible for inclusion.
We extracted the proportion of patients free from panic attacks and changes in panic disorder symptom severity. We pooled outcomes by trial design: outcomes in SSRI arms were pooled for active-controlled, and for placebo-controlled trials; outcomes in placebo arms were pooled for trials with multiple active treatment arms and for trials with only one active treatment arm.
Results:
773 records were identified, 68 abstracts were screened. 49 full-text articles were assessed for eligibility. 53 trial arms consisting of 36 SSRI arms and 17 placebo arms from 27 studies were included in this study. Random-effects meta-regressions demonstrated that the proportion of patients free from panic attacks in SSRI arms was lower in placebo-controlled trials compared to active-controlled trials (48.3% vs 63.0%, P = .007) in the ITT-sample. However, this lessebo effect was accounted for by systematically lower completion rates in placebo-controlled compared to active-controlled trials. No further lessebo effects were identified.
Conclusions:
Our study indicates that lessebo effects are quite small, if they exist in RCTs of the treatment of panic disorder with SSRIs in adults. Our data also suggest that lessebo effects found in ITT-analyses may be, at least partially, due to systematic differences in completion rates across trial designs. Analyses of higher power are needed to clarify the size and the robustness of lessebo effects. We emphasize that trial design should be considered whenever treatment effects in RCTs are interpreted.
Patient or healthcare consumer involvement:
Even though no patients or healthcare consumers were directly involved in this investigation, this study's analyses were based on previously collected data from patients participating in RCTs.
Since lessebo effects may bias estimations of effect sizes derived from RCTs and systematic reviews, patients and health care consumers referring to such studies may over- or underestimate the beneficial effect of a treatment, just like investigators and clinicians. Furthermore, lessebo effects may contribute to understanding subjective experiences made by patients while participating in a RCT.