Mental health interventions targeting children and young people: A mapping review of interventions, follow-up, and evidence gaps

The evidence: Inclusion criteria and search strategy

There is an abundance of systematic reviews summarising the effects of interventions for child and young people’s mental health. We have in the above-mentioned previous work completed eight OoOs including systematic reviews evaluating all interventions targeting children and adolescents (0-18 years) with; ADHD, anxiety, depression, bipolar disorder, psychosis (including schizophrenia), OCD, self-harm and trauma (including PTSD).^11–18 The analysis reported on in this paper is in its entirety based on these eight OoOs and represents a secondary analysis of what we consider to be complete summaries of the effectiveness of interventions reported on in high quality systematic reviews.

Our analysis is restricted to the inclusion and exclusion criteria of the eight OoOs. The protocols for these overviews have been published in PROSPERO (ADHD, CRD42020159885; Anxiety, CRD42020159884; Bipolar, CRD42020176356; Depression, CRD42020159883; Psychosis, CRD42020212244; Self-harm, CRD42019117942; Trauma CRD42019120078; OCD, CRD42020221081), however we briefly summarise the approach we applied here.

All OoOs were conducted adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist and according to “best practice”-principles of review methods.¹⁰ The search strategies for the OoOs were largely based on the database for systematic reviews IN SUM (indexing publications from all major databases. The search strategy applied by the IN SUM database can be accessed here). All systematic reviews indexed in IN SUM were considered for inclusion. We also hand searched national databases of evidence-based guidelines in Sweden, Denmark and the UK (NICE). The flow chart describing each step of the inclusion and exclusion of publications can be seen in Figure 1.

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart.

PICO, patient intervention comparison outcome; OCD, obsessive compulsive disorder; ADHD, attention deficit/hyperactivity disorder; OoO, overviews of systematic reviews.

To be included in one of the OoOs, a review had to meet a set of minimum criteria (4 out of 5) for being considered “systematic”, set by DARE.²²

Furthermore, all interventions intended to improve an aspect of mental health for children and adolescents under the age of 18 were included.

Some of the OoOs included both preventive and treatment interventions. For this mapping review, we only considered interventions targeting children and adolescents with a clinical diagnosis of mental illness or otherwise judged by the systematic review authors to have moderate to serious symptomatology. The reason for this is that not all the OoOs included preventive studies and would thus render this secondary analysis incomplete.

The OoOs excluded studies where pharmacological interventions were not compared to active interventions, thus placebo-controlled studies were excluded. Furthermore, studies evaluating mental health interventions targeting people where the primary health concern was somatic (e.g., patients with asthma and co-occurring anxiety) were excluded.

The overviews are “living reviews”, and so for the sake of this mapping review we based our data collection on the latest updated version of each overview respectively, with the most recent of these updates including evidence published in April 2021 or later. The data collection for this mapping review took place between September 2021 and December 2022.

Data collection

We mapped all interventions evaluated. In the results section, we present these findings in descriptive charts (Sunburst-diagrams) for each patient group respectively. We also mapped the number of treatment comparisons evaluated and the longest time of follow-up included in these assessments (for any outcome).

To create a mega-map of intervention effects across patient groups we also extracted data on the primary outcome for each intervention, and the judgement of certainty for this outcome.^23,24 All data were entered into Excel. The primary outcome was conceptualised as overall symptomatology associated with the relevant condition, such as anxiety symptoms for those with anxiety and depression symptoms for those with depression. In cases where the same outcome was assessed using different outcome measures, we extracted the outcome with the longest follow up and highest certainty.²³ If a review did not report findings on overall symptoms, we reported other symptoms as proxy, for example inattention in treatment of ADHD when total symptoms were not available.

Since this is a mapping review, we based our analysis on the effect-sizes and judgements of certainty as they were reported in the OoO. Judgements about certainty included in the OoO were made using the GRADE-criteria.^23,24 GRADE (Grading of Recommendations, Assessment, Development and Evaluations) is a transparent and widely adapted tool for developing and presenting summaries of evidence.²³ Using this framework, the evidence is judged to have high, moderate, low or very low certainty by considering the following criteria: risk of bias, imprecision, inconsistency, indirectness, publication bias, magnitude of effect, dose-response gradient and residual confounding.

We extracted the effect sizes and categorized these “small”, “moderate” or “large” based on statistical rule-of-thumb judgements (see Table 1).^10,24

What is considered “a meaningful effect” depends on the individual patient and context. Thus, we made no attempt to consider the clinical importance of these results as this is best judged by those delivering and receiving treatments. In some cases, it was difficult to judge the size of the effect as the results were not reported using standardized or relative effect estimates. In these circumstances we used the systematic review authors’ own judgements and have annotated these in the results table. If no judgements were made by the review authors, we marked this as “effective” without making judgements about the size of effect. Trivial or small not statistically significant differences in effect was coded as “little or no difference” according to the GRADE-recommendations.²⁵

All data extraction was done by one author and double checked by another co-author. Any difference in opinion between these were discussed with a third co-author.

Treatment comparisons and follow-up

We reviewed 116 systematic reviews included in the eight OoOs (see Figure 1). As the OoOs are living documents and regularly updated, we have included a list of the systematic reviews included in the summaries at the time data was extracted for this study – see Underlying data.³⁰

The evidence underlying treatment options of mental illness for children and young people includes 200 treatment comparisons evaluating the primary outcome for each specific diagnostic group. See Underlying data²⁹ for more information.

Overall, 49.5% of the treatment comparisons were non-pharmacological interventions, 36% included pharmacological or nutrition/supplemental interventions, and 14.5% were combination treatments including both pharmacological and non-pharmacological treatments.

The number of treatment comparisons varied greatly across the patient groups; from 69 comparisons for ADHD to 7 for OCD (see Table 2).

Interventions evaluated by diagnostic group

Interventions by treatment categories are displayed in Sunburst-diagrams by patient groups (see Figures 2–10). The evidence informing ADHD-treatment includes a range of psychological, psychosocial, dietary, systemic, physical activity, skills training and pharmaceutical interventions. About half of the ADHD-interventions are pharmacological or combination therapies including medication.

Figure 2. Unique intervention categories evaluated for attention deficit/hyperactivity disorder (ADHD).

CBT, cognitive behavioural therapy; DBT, dialectic behavioural therapy; NRI, norepinephrine reuptake inhibitors.

Figure 3. Unique intervention categories evaluated for anxiety.

CBT, cognitive behavioural therapy; MBSR, mindfulness based stress reduction.

Figure 4. Unique intervention categories evaluated for bipolar disorder.

CBT, cognitive behavioural therapy; DBT, dialectic behavioural therapy.

Figure 5. Unique intervention categories evaluated for depression.

CBT, cognitive behavioural therapy; DBT, dialectic behavioural therapy; ECT, electroconvulsive therapy; rTMS, repetitive transcranial magnetic stimulation; MBSR, mindfulness based stress reduction.

Figure 6. Unique intervention categories evaluated for obsessive compulsive disorder (OCD).

CBT, cognitive behavioural therapy; iCBT, internet-based cognitive behavioral therapy; ERP, exposure and response prevention; SSRI, selective serotonin reuptake inhibitors.

Figure 7. Unique intervention categories evaluated for psychosis.

CBT, cognitive behavioural therapy.

Figure 8. Unique intervention categories evaluated for self-harm.

CAT, cognitive analytic therapy; CBT, cognitive behavioral therapy; DBT, dialectic behavioural therapy; MBT, mentalization-based therapy.

Figure 9. Unique intervention categories evaluated for post-traumatic stress disorder (PTSD).

CBT, cognitive behavioural therapy; EFT, emotion-focused therapy; EMDR, eye movement desensitisation and reprocessing; TF-CBT, trauma focused cognitive behavioral therapy.

Figure 10. Unique intervention categories evaluated for trauma.

EMDR, eye movement desensitisation and reprocessing; CBT, cognitive behavioural therapy; TF-CBT, trauma focused cognitive behavioral therapy.

Most treatment evaluations for depression are non-pharmacological and include different types of psychological therapies, physical activity, art therapy and medical treatments such as electroconvulsive therapy (ECT) and repetitive transcranial magnetic stimulation (rTMS). The evidence also includes pharmacological interventions including use of serotonin-norepinephrine reuptake inhibitors and selective serotonin reuptake inhibitors (SNRIs/SSRIs) in combination with cognitive behavioural therapy (CBT) or SSRI used alone.

The evidence for treating OCD represents less diversity, and most interventions are CBT-based, either used alone or in combination with medications.

Treatment evaluations for bipolar disorder and psychosis include mainly pharmacological treatments, either as direct comparisons or in combination with psychological treatments or other medications (bipolar disorder). Psychological treatments used for psychosis include psychoeducation, computer-based cognitive training, CBT and group- and family therapy. For bipolar disorder, evaluations also include psychoeducation, CBT, interpersonal therapy, Dialectic behavioural therapy (DBT) and family therapy.

Self-harm and trauma/PTSD-interventions are all psychological, psychosocial, or other therapeutic interventions such as psychoeducation. Several organisational or systemic treatments used for preventing suicide or reoccurrence of self-harm have also been evaluated, including interventions such as the use of “emergency green card” (self-admittance to specialist health care).

Time of follow-up

The longest follow-up for any outcome could be found for anxiety (6 years), followed by ADHD and depression (3 years), bipolar, self-harm and PTSD (2 years), trauma (1 year) and OCD (6 months).

Overall, approximately 40% of the treatment comparisons included only short-term follow-up (< 3 months). A total of 10% of the treatment comparisons included follow-up beyond 12 months (see Table 3).

Follow-up times by diagnosis and category can be seen in Table 4. Overall, it can be observed that for most patient groups, treatment comparisons of pharmacological treatment included very short follow-up.

Treatment uncertainty and comparative effectiveness

There was a large diversity of treatment comparisons evaluated. We identified 24 unique groups of treatment comparisons evaluated for more than one patient group. These can be seen in Table 5.

Most treatment evaluations were associated with low or very low certainty. The certainty was higher for treatment evaluations including CBT, which also demonstrated moderate to large effects compared to treatment as usual across different patient groups. DBT compared to treatment as usual has been evaluated for depression and self-harm with low to moderate effect sizes on primary symptoms (low certainty). Physical activity is found to have a moderate effect on primary symptoms of ADHD and depression (low certainty). Mindfulness interventions also result in small but beneficial effects compared to treatment as usual (TAU) for anxiety and depression (moderate to low certainty).

Little or no important difference in effect was found when comparing psychological treatments with pharmacological treatments for depression and OCD, however a small but beneficial effect in favour of psychological treatment was found for anxiety (all low certainty). Combination therapy of CBT and pharmaceutical treatments was found to have a moderate effect on anxiety, but there was little or no difference for ADHD when compared to psychological treatment alone. The certainty of the evidence for combination therapy compared to pharmacological therapy alone, was very low for three out of four patient groups evaluating this treatment comparison (anxiety, bipolar and psychosis). For ADHD, combination treatment was found to produce large effects on the primary outcome, although with low certainty.

Treatment comparisons including psychoeducation (vs TAU), family therapy (vs TAU), psychodynamic (vs TAU), Eye Movement Desensitization and Reprocessing (EMDR) (vs TAU), massage therapy (vs TAU) is very uncertain across all patient groups evaluated.

Limitations and strengths

Our analysis is based on eight recent overviews of systematic reviews summarising the evidence reported in 116 systematic reviews.^11–18 This analysis includes only direct comparisons of treatments reported on in high quality systematic reviews. To our knowledge this is the first mapping review of its kind supporting treatment of child and youth mental illness, and our findings is of great relevance to clinicians, funders of new research initiatives and researchers with a blueprint for planning future research activities. Furthermore, acknowledging research uncertainties is an important step-stone in good patient care.

One limitation to our work is that research is accumulating rapidly in this field, and that while this report is being written, new evidence may have emerged. However, considering the substantial evidence gaps we identified we are confident that our conclusions may continue to be valid some time to come. Another limitation may be that our analysis is based on reviews, and that the results are impacted by the methodological choices and reporting by the authors of the individual reviews included in each OoO, but also by the inclusion criteria by the OoO. For example, this analysis does not include placebo-controlled pharmaceutical studies or comparisons of preventive treatments. As such, we would like to emphasise that there is a large body of evidence evaluating these treatment comparisons and that the results of these and the corresponding evidence gaps should be considered together with our analysis. Network analyses of indirect comparisons are an important contribution to this evidence base.²⁶ Furthermore, our analysis is based on treatments evaluated in specific patient groups. Other reviews have summarised the effects of interventions in diverse patient groups (not restricted to diagnostic criteria). Thus, implementation of findings coming out of our analysis should take into consideration the findings of such transdiagnostic reviews.

Summary of results

We identified 200 treatment comparisons. The most striking finding of this review is the short follow-up times for most treatment comparisons. This is the case for pharmaceutical treatments in particular. Approximately 40% of all treatment comparisons included only short-term follow-up (<3 months), with only 10% of the treatment comparisons including a follow-up beyond 12 months. Thus, there is great uncertainty associated with the long-term effects of mental health treatments for children and young people.

We found large diversity in treatments evaluated. Our descriptive analysis shows that the largest diversity can be found for ADHD and less diversity was found for OCD. Furthermore, from the research that has been conducted so far it can be deducted that the etiology and assumed relevant interventions are understood differently depending on the diagnosis. Some conditions are, for instance, addressed mostly by combination treatments or pharmaceutical treatments alone whereas others are treated with psychological or psychosocial interventions. For example, most treatments for bipolar disorder were pharmacological, whereas treatments for children with PTSD/trauma were all psychological or psychosocial therapies. Treatments for ADHD and psychosis include a combination of both pharmaceutical and psychological or psychosocial treatments. This is an important contribution to the epistemology of mental health, and the understandings of what may reduce or increase mental health depending on the symptoms we identify (diagnosis).

The certainty of most treatment outcomes is low or very low. However, there are treatments with promising effects, which also seem to be effective across patient groups. CBT demonstrates moderate to large effects (compared to treatment as usual) across four patient groups. DBT therapy, physical activity and mindfulness interventions also result in small to moderate beneficial effects compared to TAU (moderate to low certainty). Combination therapies including CBT and pharmacological treatments was either found to have similar effects on primary symptoms compared to either CBT or pharmacological treatment alone, or to be more effective (low certainty).

Implications for research and practice

Practitioners and decision-makers should be aware of important research uncertainties and make these explicit in communication with patients and when developing guidelines. Our report may be an important platform for doing so.

For many treatment evaluations the evidence is limited. Nevertheless, some treatments have shown convincing treatment effects and should be considered as first choice when treating children and young people with mental illness. Some treatments identified in our review may be well documented for use in adults and may therefore also be applied to younger people. Although this may be theoretically sound, the transferability to and the effectiveness for younger people is unclear and should be addressed. There is therefore a need for more high-quality research on the effectiveness of treatments with uncertain effects on children and young people. Future research efforts should also include long-term follow up assessments, as this is fundamental to patient safety.

We have made no judgements about the clinical importance of the effect-sizes we mapped in this review, as this is a decision which is best left up to health professionals and their patients. Furthermore, we did not review or make any judgements about which outcomes were evaluated in the treatment comparisons we mapped. For pragmatic reasons, our analysis depended on the primary outcome (change in symptoms) for each diagnostic group using statistical rule of thumb thresholds. It is important to emphasise that any changes in symptomatology should be supplemented with measuring other relevant outcomes. The opinions of patients and health professionals about the relevance of outcomes may differ from those of researchers. Thus, future studies should include the opinions of patients and should consider any established core outcome sets.^27,28

Underlying data

Zenodo: Included comparisons_mapping review_2022. https://doi.org/10.5281/zenodo.6948764.²⁹

This project contains the following underlying data:

Zenodo: Supporting materials_mapping review_2022. https://doi.org/10.5281/zenodo.6815213.³⁰

This project contains the following underlying data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).