Control interventions in randomised trials for people with a mental health disorder


This systematic review assesses the effects of different control interventions in randomised trials including patients with a mental health disorder. In randomised trials, patients are assigned by chance to one of two or more groups – usually an experimental intervention and a control intervention. There are many types of control interventions in mental health intervention research. Some of the most common are different types of placebos that lack what is assumed to be the active component in the experimental intervention, and usual care, where patients receive the standard treatment for their mental health disorder in the area where they live. Two other types of control interventions are wait-list or no-treatment where patients receive no trial-related care during the study (although some patients may receive care outside the studies). Wait-list patients are often offered the experimental intervention after the trial has been finalised if it is likely to provide more benefits than harms, while no-treatment participants are not offered the experimental intervention by the researchers.

We searched for randomised trials with patients with mental health disorders where wait-list, usual care, or placebo interventions were compared with either wait-list or no-treatment. We looked at differences between all the types of control interventions on beneficial effects and whether they caused any adverse effects. We included 96 trials with a total of 4200 participants. Only 83 trials (3614 participants) provided usable data. Fifteen different mental health disorders were included. We found that all the trials were at high risk of bias in how they had been conducted, which reduced the interpretability of our findings. However, the risk of bias was mostly due to lack of blinding in the placebo studies, which may be seen as an aspect of the review's methodological question rather than a flaw with the review itself. We found no clinically important differences for usual care or wait-list control interventions in the main analyses, however in our secondary analyses we found a clinically important favourable difference for usual care. In general, placebo control interventions tended to be favourable over no-treatment or wait-list control interventions across mental health disorders. We found no clinically important differences on adverse events.

This review suggests that different control interventions have a tendency to yield very different estimates for the effects of the experimental intervention and that the choice of control intervention has a large impact on how effective a mental health treatment appears to be. Control interventions in trials with patients with mental health disorders are often poorly reported upon, and guidelines are needed to inform researchers on how to properly design, report, and interpret these trials.