New Cochrane Review: Control interventions in randomised trials for people with a mental health disorder
- Researchers use many different control interventions in randomized trials on treatments for patients with mental health disorders, but there is little consensus on how to report and adequately design these controls. This practice has widespread consequences for the evidence base underpinning psychiatric treatments
- The choice, design and reporting of a control intervention is just as important as the experimental treatment in a randomized trial with psychiatric patients. This is not reflected in most randomized trials with mental health patients, as control interventions are often poorly reported upon and lack methodological rigor
- Some psychiatric treatments may be recommended based on just having compared the treatment with a waitlist or no-treatment control in a randomized trial, which may give a misleading picture of how effective the treatment is
Erlend Faltinsen, lead author, commented, "There is a need to develop methodological guidelines on how to design and report upon control interventions in randomized trials on psychiatric treatments, as trialists working in the field of mental health do not have a solid evidence-based framework to draw from on this issue."
Why was this review conducted?
The review investigates the beneficial and harmful effects between different control interventions in randomized trials with mental health patients. We wanted to investigate how control interventions differ from each other and to lay the empirical groundwork to develop methodological guidelines on reporting, and the design of control interventions in psychiatric randomized trials.
What did the authors do?
The authors conducted a Cochrane systematic review and meta-analysis to assess the benefits and harms between placebo, usual care (or treatment as usual) and waitlist controls versus receiving no treatment. In that way they assessed how effective and harmful different control interventions in psychiatric randomized trials are.
What did they find?
- 96 randomized trials were included and the trials involved 15 different types of mental health disorders
- When combining three different types of placebos, the beneficial effects compared with no-treatment or wait-list controls was small to moderate
- There was no significant difference between usual care controls and wait-list or no-treatment on benefits. The same was true for waitlist versus no-treatment controls
- Psychological placebos (non-active controls used mostly in psychotherapy research) showed moderate effects compared with no treatment whereas placebos used for physical treatments like surgery showed a small effect. We found no significant effects of pharmacological placebos versus no treatment
- There was little data on harms between the control interventions and the findings on harms were insignificant
- The control interventions were mostly poorly reported upon and there was little rationale for why a given control was used in most reports
What are the limitations of the evidence?
The certainty of the evidence was rated low to very low, and the risk of bias was rated high in all studies. The authors mostly included randomized trials with three intervention arms to compare two controls, which causes issues with blinding of participants and trial personnel. This limitation was due to the methodological objective of the review and may be viewed as part of the review itself rather than a flaw in the evidence-base. Many of the studies were small, however, leading to risk of small-study effects, which limits the evidence.
What gaps did the authors identify?
There is a need to develop methodological guidelines on control interventions in psychiatric randomized trials, as trialists working in the field of mental health do not have a solid evidence-based framework to work from when choosing and reporting upon controls
What important related questions were not addressed in this review?
The review did not compare usual care with placebo interventions, which would have been relevant. It did not quantify the level of reporting issues and rationale for choosing a control in a psychiatric randomized trial, which also would have been relevant.
Who will find this review most relevant?
Psychiatric researchers and especially those who conduct randomized trials will find the review relevant for their research work. It may also appeal to trialists in general and methodologists.
How up to date is this review?
The search was conducted in March 2018. The researchers were not able to screen records after this date, since the search process was very large and extracting data was exceptionally time-consuming.