The Cochrane Methodology Review Group produces reviews that examine the methods used in research and, in April 2022, we published a new review of the effects of having different types of control group in trials among people with mental health disorders. Here's lead author, Erlend Faltinsen from Cochrane Denmark, to tell us about the review and its findings
Mike: Hello, I'm Mike Clarke, podcast editor for the Cochrane Library and co-ordinating editor of the Cochrane Methodology Review Group. The Group produces reviews that examine the methods used in research and, in April 2022, we published a new review of the effects of having different types of control group in trials among people with mental health disorders. Here's lead author, Erlend Faltinsen from Cochrane Denmark, to tell us about the review and its findings.
Erlend: In randomised trials of new treatments, we include a control group to obtain a reliable estimate of the effects of the new treatment, and the design and content of these control groups may have an important influence on the trial's results. In trials that include patients with a mental health disorder, a wide variety of control interventions are used, with the most common being different types of placebo interventions, treatment as usual, waitlist and no-treatment. A consensus on how to properly select, design and report on such control interventions is lacking in mental health intervention research, and we conducted this review to see how their effects might differ.
We wanted to investigate the benefits and harms of the common control interventions when they were directly compared in trials. We searched 11 databases and six trial registries in March 2018 looking for randomised trials with two control interventions that were a placebo or treatment as usual and no-treatment or waitlist.
After checking nearly 37,000 records, we found 96 randomised trials with 4200 participants in total, but only 83 trials provided usable data. The trials involved participants with fifteen different mental disorders, and the most common were anxiety disorders (included in 25 trials), major depressive disorder (in 16 trials) and sleep-wake disorders (in 11 trials).
When we combined the evidence from all the included mental health disorders, we did not find a significant effect in favour of usual care when it was compared with waitlist or no-treatment. The differences between placebo interventions and waitlist and no-treatment were small to moderate.
In trials of psychological treatment, when a placebo was compared with waitlist and no-treatment, there was a moderate effect in favour of the psychological placebo. But in trials of drug treatments, we were not able to detect any significant differences for pharmacological placebos and there was only a small effect for placebos in trials of physical interventions.
Looking at the specific mental disorders, we found evidence of significant differences in favour of all placebos versus no-treatment or wait-list for sleep-wake disorders, major depressive disorder and anxiety disorders, but the results were imprecise.
All in all, we found large variations in effects between different control interventions, but with significant methodological and clinical variability across the trials. Our key message is that the choice of a control intervention has a large impact on the effect estimate for mental health treatments in randomised trials, which raises the need for guidelines to reach a consensus on how to properly select, report and design control interventions in randomised trials with patients with mental disorders.
Mike: Information on all the studies in Erlend's review and finer detail on the control interventions is available in the full review, which is available at Cochrane Library dot com with a simple search for 'control interventions in mental health trials'.