Understanding side effects of psychotherapies: implications for clinical practice and research trials

Based on efficacy and effectiveness data from randomized controlled trials (RCTs) spanning several decades, psychotherapies are considered first-line treatments for numerous psychiatric conditions1. There is, however, a remarkable lack of evidence regarding the risks associated with these treatments. This is striking, given the focus on risk assessment and harm mitigation that underpins clinical practice across the fields of psychiatry and psychology. There have been, therefore, increasing calls to improve the reporting of side effects in psychotherapy trials2, 3. Here we outline the prevalence and types of side effects of psychotherapies (in both real-world settings and clinical trials), summarize tools and resources used for detecting and reporting these side effects, and provide recommendations aimed to promote consistency in trial practices to better inform clinicians and patients of the risk-benefit profile of these treatments. When looking into the literature from real-world settings, the side effects of psychotherapies appear to be common. A survey of 224 people who received a psychotherapy revealed the most common side effects to be unpleasant memories (57.8%) or feelings (30.3%), and a perceived lack of understanding by the therapist (18.4%)4. When looking into the clinical trial literature, however, the prevalence of side effects appears much lower. One review reported that 5% of participants experienced an (unspecified) adverse event, 2.5% a serious adverse event (e.g., unplanned hospitalization), and 4% attempted suicide5. It is difficult, however, to estimate and compare the side effects of psychotherapies, because they are seldom and inconsistently reported and ill-defined in clinical trials. A recent review of 115 psychotherapy trial protocols found considerable variation in what and how side effects and harms were recorded. Most only monitored suicidality, completed suicides or psychiatric admissions6, which provides an incomplete picture of the potential harms associated with these treatments. This dearth of evidence is driven by a lack of consensus around what constitutes a side effect of treatment. Side effects can be expressed as adverse events (defined as any untoward medical occurrence), that are commonly self-reported, clinician-reported or researcher-reported, or derived from or validated against medical records in clinical trials in other areas of medicine. What makes psychotherapy trials somewhat unique, however, is that side effects of treatment are more difficult to identify, manifesting for instance as treatment dissatisfaction, lack of therapeutic alliance, anxiety or rumination about therapy, clinical deterioration (estimated to occur in 5% of psychotherapy trial participants7) or perceived negative effects of treatment on family/friends. Moreover, contrary to trials of pharmacotherapies – for which adverse events are required to be monitored, adjudicated or reported as part of regulatory processes required by bodies such as the US Food and Drug Administration – psychotherapy trials are not subject to these regulations, which encourages under-reporting. There have been some attempts to promote consistency in monitoring side effects in clinical trials evaluating psychotherapies. One example was a 2019 systematic review of tools that can be used to monitor these side effects8. Nine instruments, covering seventeen conceptual domains of negative effects of psychotherapy, were identified. They included the Vanderbilt Negative Indicators Scale (VNIS), the Unwanted Event to Adverse Treatment Reaction checklist (UE-ATR), the Inventory for the Assessment of Negative Effects of Psychotherapy (INEP), the Experiences of Therapy Questionnaire (ETQ), the Exploitation Index (EI), the Negative Effects Questionnaire (NEQ), the Unwanted Events and Adverse Treatment Reactions in the context of group psychotherapy (UE-G), the Side Effects of Psychotherapy Scale (SEPS), and the Positive and Negative Effects of Psychotherapy Scale (PANEPS). Among the most common conceptual domains across these nine instruments were stigma, symptom deterioration, therapeutic relationship (e.g., dependency), and treatment response. It is notable that issues such as therapeutic misconduct and quality of therapy were mixed up with the issue of side effects of treatment. The UE-ATR covered the highest number of side effect domains. While there were limited data on the psychometric properties of the examined instruments, the test-retest reliability was reported to be 0.76-0.96 for the ETQ, and the inter-rater reliability to be 0.89-0.96 for the VNIS8. Despite the availability of these tools, covering a wide range of harm concepts, a review found that only 11% of authors of psychotherapy trials used harm assessment instruments5. It is not uncommon for mental health clinical trials to adopt more generic adverse event reporting questionnaires used in other areas of medicine. While this is helpful for comparing the safety profile of a mental health therapy to drugs or devices used in other disciplines, it risks underreporting harms related to psychotherapies, because these questionnaires often neglect the key theoretical domains relevant to psychotherapy evaluation. For clinical trials of psychotherapies to be clinically salient to therapists and consumers, we recommend as a minimum to embed the following within clinical trial protocols: multiple avenues and opportunities by which to collect data on side effects (e.g., participant self-report aided by validated instruments which include questions about the impact of therapy on the above-mentioned domains; therapist observations/notes/recordings; trial staff interaction with participants; carer- or family-reported effects); the establishment of an independent data safety and monitoring board or equivalent (responsible for assessing participant safety, including the reporting of adverse events, and implementing stopping rules); adequately trained and qualified trial personnel who are responsible for assessing the relatedness of a side effect to the therapy being evaluated, and its seriousness; and reporting the percentages of side effects related to the therapy and the comparator in the manuscript detailing the results of the trial. Such activities are not without budgetary implications. Prospectively embedding adverse event recording, monitoring, adjudication and reporting activities into clinical trial budgets is critical to allow for accurate detection of these events. Preparing protocols to align with the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guidelines and with the reporting requirements of CONSORT-Harms (with relevant extensions such as CONSORT-Social & Psychological Interventions9) can be considered. The current practice of under-reporting side effects contributes to the assumption that psychotherapies are without risk. However, the absence of reporting of adverse events does not mean an absence of such events. Identifying adverse events is not a sign of poor, but of good clinical practice5. Providing clinicians with evaluations of a therapy which comprise both harms and benefits data will generate a more complete picture, better informing clinical decision-making.