Impact of active placebo controls on estimated drug effects in randomised trials
2Department of Psychology, Faculty of Health Sciences, UiT–The Arctic University of Norway, Norway
3Novo Nordisk; Department of Clinical Biochemistry and Pharmacology, Odense University Hospital; Clinical Pharmacology, Pharmacy and Environmental Medicine, Department of Public Health, University of Southern Denmark, Denmark
4Department of Internal Medicine 2, University Hospital St. Poelten, Austria
5School of Psychology, Faculty of Science, The University of Sydney, Australia
Background: Active placebo controls are designed to mimic the nontherapeutic adverse effects of drugs in randomised trials. Active placebos are rarely used but could reduce the risk of bias due to unblinding.
Objectives: We aimed to estimate the difference in drug effects when an experimental drug is compared with an active placebo versus a standard placebo control intervention, and to explore causes for heterogeneity. In the context of a randomised trial, this difference in drug effects can be estimated by directly comparing the effect difference between active placebo and standard placebo intervention.
Design: A systematic review.
Methods: We searched PubMed, CENTRAL, Embase, and other sources up to October 2020 and included randomised trials directly comparing active placebo versus standard placebo. Our primary inverse-variance, random-effects meta-analysis used standardised mean differences (SMDs) of participant-reported outcomes at earliest post-treatment assessment. An SMD <0 favoured the active placebo. We conducted seven sensitivity analyses, four subgroup analyses and a meta-regression. In secondary analyses, we investigated observer-reported outcomes, harms, attrition, and cointervention outcomes.
Results: We included 21 trials (1,462 participants). The pooled SMD in our primary analysis was -0.08 (95% CI -0.20 to 0.04; Figure 1). The difference was more pronounced and statistically significant in the sensitivity analyses restricting to trials with overall low risk of bias (pooled SMD -0.24; 95% CI -0.34 to -0.13) and using a fixed-effect model (pooled SMD -0.15; 95% CI -0.23 to -0.07). The pooled SMD of observer-reported outcomes was similar to the primary analysis. The pooled odds ratio for harms was 3.08 (95% CI 1.56 to 6.07), and for attrition, 1.22 (95% CI 0.74 to 2.03). Cointervention data were limited. Meta-regression found no statistically significant association with adequacy of the active placebo or risk of unintended therapeutic effect.
Conclusions: We did not find a statistically significant difference between active and standard placebo interventions in our primary analysis, but the result was imprecise and compatible with a difference ranging from important to irrelevant. Furthermore, the result was not robust, for example, when restricting to trials with overall low risk of bias.
Patient, public and/or healthcare consumer involvement: None.