In response to a growing concern about the wellbeing of young people, the government published a green paper last December outlining the role schools can play in providing mental health services and education. Schools are an obvious place to access young people and to intervene both proactively, to encourage positive mental health, resilience; and reactively, to target those who are struggling. The role of schools in mental health provision is eloquently discussed in Pooky Knightsmith’s Mental Elf blog (School based mental health services: a good idea?) earlier this year.
Resilience is a buzzword in schools at the moment. With funding tighter than ever and pressure to have every child performing academically above national average (yes, that is the aim of every school: clearly someone missed a maths lesson!), mental health is not at the top of the agenda, but it is unavoidable and teachers are on the front line. Schools are crying out for low cost, easy to implement strategies to improve mental health and wellbeing. In her Mental Elf Blog (Preventing mental health problems: what can we do?), Josefine Breedvelt discusses prevention as one of the key ideas for a cost effective solution is to encourage school based interventions. Buying in services or training teachers to deliver these sorts of universal, preventative school-based intervention programmes seems an ideal solution.
However, many questions arise from these sorts of interventions:
- Can they be effectively delivered by teachers?
- Are they effective in the long term?
- Do they have the same effects across all populations and subgroups?
This systematic review by Dray et al (2017) sets out to review universal, school-based, resilience-focused interventions on the mental health of children and adolescents and seeks to answer some of the questions above.
Reading this paper, I had just finished Martin Seligman’s book ‘The Optimistic Child’, his Penn Resilience Programme (PRP) (mentioned in the review and run by several of the included trials) was the basis for the book. Seligman talks about his research, how effective it is, even two years after the end of the programme. As I read the review I had high hopes that such interventions would show overwhelming evidence that, by building resilience in young people, schools could play a fundamental role in staving off mental ill health in young people.
The studies included in the review were randomised controlled trials that considered at least 1 of 7 mental health problems (e.g. depressive symptoms, anxiety and conduct problems) in their outcome measures. The interventions needed to address at least 3 internal resilience protective factors such as cognitive competence, problem solving and coping skills. Six databases were searched, also the Cochrane Central Register of Controlled trials, Google Scholar, and a variety of key journals and references.
In total, 57 studies were included, 49 of which contributed to the meta-analyses. The comparative meta-analyses conducted were on:
- Age (5-10 years and 11-18 years)
- Length of follow-up
- short: post-≤12 months
- long: post-≥12 months
- Therapeutic basis for intervention: CBT vs. Non-CBT
However, there were too few studies providing data about gender for them to be amenable to meta-analysis.
The majority of trials measured the effect of the intervention on depressive symptoms (n=41) with most (n=39) having a follow-up of immediate-post to 12 months, 18 trials reported on a longer follow-up. The largest number of trials based their intervention on Cognitive Behavioural Therapy (CBT) (n=31), with others focusing on things such as life-skills, coping skills or psychological well-being therapy. The largest number of trials were delivered by teachers (n=24), with 20 being delivered by clinicians/trained facilitators and 13 using a combination of the two.
Risk of bias
The review reports a high risk of bias in the included studies, predominantly because of the lack of blinding of both those delivering and those participating in the intervention and the use of self-report measures. The reviewers also report a high risk of publication bias, downgrading the quality of evidence for depressive symptoms to ‘low’ and all the other mental health outcomes to ‘moderate’ due to methodological problems.
Overall it appears that, compared to controls, there was a significant effect of the intervention on 4 of the 7 outcomes:
- Depressive symptoms
- Internalising problems
- Externalising problems
- General psychological distress.
There was no significant effect on:
- Conduct problems.
There appears to be different effects on the two age groups:
- For the younger children (5-10 years old) there is a significant overall effect on anxiety and general psychological distress
- For older children (11-18 years old) an effect was only seen on internalising problems.
Differences in gender were reported using a narrative synthesis:
- For depressive symptoms, 15 studies reporting gender data found no difference, two found an improvement for the intervention: one favoured males, the other females; and a final study found an improvement for females but a increase in depressive symptoms for males
- In terms of anxiety most trials found no gender difference (n=10) but two did find a reduction in symptoms for females but not males
- Single trials considered the gender differences for hyperactivity, internalising problems and total difficulties, none found any significant differences.
For all 7 outcomes, an analysis of short term follow up was possible and indicated a significant effect for depressive symptoms and anxiety symptoms. For longer term follow-up only, comparison of 5 out of the 7 outcomes was possible and the only significant effect was on internalising problems. No trial reported on the long term effect of conduct problems and only one trial reported on generalised psychological distress (reporting a significant effect of the intervention at 18 months).
Finally when comparing CBT to Non-CBT based interventions a comparison of 3 outcomes was possible: depressive symptoms, anxiety symptoms and general psychological distress. For the CBT based interventions a significant effect was found on all three outcomes, where there was no effect for the non-CBT based interventions.
The review concludes that CBT-based programmes are the most effective universal resilience-focused programmes for children and adolescents, at least for short-term reductions in depressive and anxiety symptoms. There was too little high quality data to fully evaluate the effect of age, length of follow and gender on the efficacy of resilience-focused interventions. Thus understanding how the different aspects of the interventions achieve the best results in different school populations is an important area for future research.
Strengths and limitations
The review was registered with PROSPERO and followed PRISMA guidelines. All studies were independently assessed for bias by two raters and there was a high degree of agreement between the raters. There are several sources of possible bias: of note is the publication bias (the searches conducted were all of published work and studies which were not in English were excluded); the lack of blinding in many of the trials that then used self-report outcome data produced a high degree of performance and detection bias. These are acknowledged by the authors. The screening process for potential studies was thorough, with authors of included studies being contacted.
One of the greatest issues for the reviewers is the heterogeneity across the included studies. As such, in several cases meta-analysis was not possible and so understanding the interaction between the different sub-groups of the study was limited. For example conduct problems as an outcome measure was only measured in one adolescent trial.
Measures of all outcomes were based on reliable and validated scales such as RCADS, KSQ, SDQ etc. If a school were to implement such an intervention and wished to monitor the outcomes, such scales may be impossible to use (due to time constraints, access and knowledge) therefore it would be more practical to have reported on other measures of school performance such as attendance data, behavioural data and academic performance data all of which correlate with mental health (Faulconbridge et al, 2017).
The authors provide a supplementary document detailing much of the included trials’ information. However it does not include information about the number of participants in the trials or the type of school the trial was run in. This affects generalisability: delivering an intervention to a small group vs a large group; socio economic status may affect which outcomes are measured and how effective the intervention is; school culture may also have a bearing on the outcomes. Without this information it is difficult to know whether the interventions are truly ‘universal’.
The study briefly mentions that the interventions were delivered by either teachers, trained clinicians/facilitators or combination of both. However the authors do not consider whether this variable has an effect on the outcomes. In another Mental Elf blog (School-based CBT for anxiety and low mood), Alan Underwood discusses the implications of this, as the teacher delivery seems to differ significantly to the trained professionals and this could have implications for the effectiveness of the trials.
Whilst we hope that universal interventions will be positive for the majority, there is the acceptance that there may be no benefit for some. When we lump together individual data we may overlook the fact that for a minority the intervention may in fact do harm. In one case we see that, for the boys, taking part the intervention increases their depressive symptoms. Even if it is not a statistically significant number perhaps greater awareness should be paid to those sub-groups or single data sets that show a negative effect, because we should be confident that at the very least any universal intervention will do no harm.
Overall I feel frustrated by the review’s inability to really answer the question it set out to answer. There appears to be some benefit to using these types of universal resilience focused interventions, but it is limited. However, I am unsure whether the limits are because the interventions themselves are not appropriate, the trials are poorly conducted or because the number of studies for the different sub-groups do not include enough of the measures to show a difference.
Implications for practice
This review does not fill me with hope. As a teacher I would worry that the impact of these sorts of programmes are short lived and only partly effective. However, the review gives some guidance as to how to tailor such programmes and make them as effective as possible. One clear finding is that CBT based approaches do seem to be more effective than non-CBT based approaches, although it is not possible to say exactly which parts of the CBT interventions are more or less effective it gives a basis for fine tuning such resilience programmes.
Tailoring programmes to age; targeting protective factors that are developmentally age appropriate also seems to an appropriate next step. In addition, gender differences in mental health issues (Patalay & Fitzsimons, 2017) would suggest that resilience-focused programmes may benefit from being tailored for girls and boys. Again further research into the interactions between gender, resilience and mental health would be a good next step.
Finally, and possibly for me the most important, is that the long term effects of these programmes seems to be limited. Therefore further consideration of how schools can foster a wider culture of resilience would perhaps be beneficial to encourage longer and deeper embedding of the resilience skills in children and young people. In my experience schools are looking for a quick fix, parachuting in a resilience programme that is quickly forgotten means that the long term impact of it is minimal. All too often the results are not immediate and the school moves on to the next shiny package. Understanding the culture and ethos of a school and how resilience can be embedded into everything a school does may be an important part of any intervention programme, but this is often overlooked by researchers.
Having spent the last year in a variety of schools, talking to teachers and professionals about their school mental health and wellbeing provision, I have come to understand the complexity of the landscape. Whilst I would love to be able to say that a single universal intervention would solve all the problems, this review has consolidated my realisation that there are too many interacting variables within a school context for anyone to confidently say that an intervention that works in one school will generalise to other schools and contexts. Gender, age, delivery quality, parental support and understanding, school ethos and culture (including faith), teacher support, socio-economic grouping, Ofsted ratings, school behavioural policy are possibly a fraction of the considerations that may affect whether or not such interventions work. I would like to suggest that the key to enabling universal preventive interventions that work, is to give schools the skills to understand what is best for their context, the tools to measure the impact of any intervention, and the ability to embed good practice throughout the school.
Conflicts of interest
Dray, J et al. (2017) Systematic Review of Universal Resilience-Focused Interventions Targeting Child and Adolescent Mental Health in the School Setting. Journal of the American Academy of Child & Adolescent Psychiatry, 56 (10), 813-824.
Faulconbridge, J., Hickey, J., Jeffs, G., McConnellogue, D., Patel, W., Picciotto, A. & Pote, H. (2017). What good looks like in psychological services for schools and colleges: Primary prevention, early intervention and mental health provision. Child & Family Clinical Psychology Review, 5.
Seligman, M. (2007) The Optimistic Child. New York: Houghton Mifflin.
Patalay P & Fitzsimons E. Mental ill-health among children of the new century: trends across childhood with a focus on age 14. September 2017. Centre for Longitudinal Studies: London.