Main findings
The WISE intervention did not appear to have an effect on the primary outcome of teacher well-being. There was also no evidence of an effect on teacher depression, student well-being or psychological difficulty, or on most other individual- or school-level outcomes. There was one small effect for teacher absence as a continuous variable in the fully adjusted models, but in favour of the control arm. This may indicate that the intervention actually led to poorer teacher well-being and resultant absence, but it could also mean that teachers in the intervention schools had become better at identifying when they need to take time off because of increased awareness of the importance of self-care. Given the number of analyses that were run, a third explanation could be that this finding was due to chance. There was weak evidence of a change in the direction that favoured the control schools or lower implementing schools in some of the analyses, which may indicate harm resulting from the intervention, something explored further below. However, it must be emphasised that these observed changes were not large and at least some may have been the result of chance findings due to the number of analyses undertaken. There was also the suggestion of an interaction between perceptions of school attitude towards student well-being and teacher well-being, with teachers who were less likely to believe their schools cared about student mental health at baseline having higher well-being by the end. Other studies have found universal interventions to have heterogeneous effects in different subpopulations,89 although it must be emphasised that this analysis was underpowered and so was exploratory. Furthermore, it could be a chance finding, given that we examined a number of different secondary outcomes.
We found insufficient evidence that the costs of the WISE intervention were justified by any improvements in staff-, student- or school-level outcomes. The total mean cost per school was relatively small (£9103). The model of training HSCs to deliver the training used in Wales was slightly more expensive within the context of this trial. However, it has the potential to be cheaper in the long-term, as it would mean the MHFA training could be extended to a larger number of schools. It is clear from the process evaluation that the intervention also has opportunity costs for peer supporters. Although the proportion of teachers who reported accessing the support was relatively low, the typical duration of contacts (i.e. 5–30 minutes) and the timing of those contacts (i.e. lunchtime or before and after school) will have an impact on peer supporters’ time for other activities. Furthermore, there may be other potential spillover costs, for example increased mental health awareness leading to staff taking time off for self-care or to seek further help. The intervention also has potential effects on students in other years, family members and associates of staff that were not captured in the trial. It is also possible that the full effect of the intervention might become evident over a longer follow-up period. However, we found no evidence to suggest that either of these factors was likely to have an important impact on our conclusions. Although there are no direct comparisons with economic evaluations of interventions aiming to improve the mental health and well-being of teaching staff, several recent trials have evaluated interventions aimed at improving student social, emotional and mental health.33,77,90 The interventions evaluated in these trials are diverse and target different aspects of mental health in different age groups. Nevertheless, some of the economic findings may be generalisable to other school-based interventions to improve mental health. The choice of valuation method for staff time in intervention training and delivery is likely to have a strong influence on cost-effectiveness estimates. Interventions that require teachers to be absent for lengthy training periods will incur costs for the school in obtaining supply teacher cover. Intervention delivery may have no financial costs for the school, but will have an opportunity cost if it displaces other aspects of the curriculum. Effect sizes, where present, tend to be small in public health interventions aimed at the general population of teachers and/or students. Therefore, cost-effectiveness may be hard to demonstrate within a trial. In this context, it is particularly important for policy-makers to consider the impact on subgroups of the population (e.g. those with poorer mental health), the sustainability of the intervention (e.g. because of high staff and student turnover) and the potential spillover effects across the school year groups and beyond the school.
It is important to draw on the findings from the mixed-methods process evaluation to understand why the intervention did not have the hypothesised impact on teacher or student outcomes. Regarding how well implementation was achieved, fidelity and quality of the 1- and 2-day MHFA training were rated as good by research team observers, course participants and trainers alike. As has been found in other studies evaluating MHFA,49 the training improved teachers’ knowledge and confidence to help others. There were challenges in terms of delivery in the school context, with the risk of interruptions and timetabling issues. There was no evidence that this had an impact on fidelity or quality because of the skills of the trainers at working flexibly, but it may have meant that for a few participants they were distracted or even had to miss parts of the training. The prespecified dosage of the MHFA training (in total, 16% of all staff in each school) was not achieved in one-third to one-quarter of schools. A previous RCT found an effect when 15% of the student body was trained to be a peer educator.90 The failure to achieve the target dose in this study may have meant that fewer students or staff received help from a trained individual, thereby affecting any population-level improvement. Fidelity of the 1-hour awareness session delivered to all teachers was also good and most participants reported learning gains. However, this session was more variable in terms of acceptability, with some participants suggesting improvements relating to the focus and delivery approach.
Implementation of the peer support service was more variable. A service was set up in all 12 intervention schools, and there was evidence that each service was used by some staff. However, peer supporter dropout created further dosage reductions in some schools, which may have contributed to the lack of impact. In addition, key aspects of the guidance were not always followed, specifically senior leaders visibly championing the service and supporting its delivery in practical terms, the establishment of a confidentiality policy and advertising the service at least once each academic year. These issues were likely to have had an impact on the key barriers to service use that were highlighted by participants, namely concerns about confidentiality, concerns that peer supporters did not have time or a safe space to provide support and not knowing about the service. As a result, although all teachers did have increased access to support, which was the first step in the logic model, they did not always consider this an acceptable form of support, which contributed to low service use. This may explain the lack of a population-level impact on staff outcomes. The importance of having an intervention champion and support from senior management have been identified as key for success in the child and adolescent health literature.91
Next, we turn to whether or not the mechanisms of change were successfully deployed, as theorised in the logic model. As already noted, there was evidence that teachers did have improved access to support if they chose to make use of this, and they also had increased awareness regarding their own and their students’ mental health and knowledge of how to provide support. According to course attendees, the MHFA training also improved the quality of support that they were able to provide to students. However, there was no evidence that this had an impact on the number of students who reported receiving support from teachers or on their perceptions of how supportive teachers were. This may be because the number of teachers completing the 1- or 2-day training was too low to see this effect and there was little evidence of other staff making use of their trained colleagues or learning from them. However, over half of all teachers received at least 1-hour of awareness raising. An alternative explanation is that the barriers to seeking help from the student perspective were not addressed. Specifically, students discussed concerns about confidentiality and a perception that teachers are interested in their academic performance only.
There was no evidence that the intervention had an impact on the subsequent steps in the logic model, namely improved teacher–student relationships in the schools and a reduction in stress related to supporting students. Qualitative findings from the student focus groups in both study arms included some very negative comments about their relationships with some teachers, although others acknowledged that students were also responsible for poor-quality relationships, and a majority in the student survey agreed that their school and their teachers cared about them. Counterintuitively, some of the measures in the student surveys showed that students in the control schools had a higher opinion of teachers’ attitudes towards them and their well-being by T2. This may indicate an unintended negative effect of the intervention, although the ‘dark logic’ by which this may have come about is not clear.82 There was some suggestion in the qualitative data of teachers in the intervention schools becoming disillusioned, which may have affected their commitment to supporting the students, although there is no evidence from the process evaluation to support this theory. This sense of disappointment that the intervention did not lead to greater change may also explain the weak tendency towards lower well-being in the intervention schools and among those trained at the final follow-up than among those in control schools. It would be important in any future study focusing on teacher well-being to explicitly monitor this potential for disillusionment and any subsequent implications for teacher or student outcomes. There was also no evidence that teacher stress had been alleviated or satisfaction at work improved by the intervention. This is not surprising, as, if quality of relationships had not improved and large numbers of teachers were not accessing more support, then there would be no reason for stress to have reduced. Given that many of the mechanisms of change were not activated, at least for large numbers of the teacher and student populations, the null effect for the main outcomes is perhaps not surprising.
The final question, which may explain the null finding, concerns the extent to which the WISE intervention differed from usual practice, and whether there was any indication of contamination, or of control schools taking up other, equally effective interventions. There was evidence of some new activity in the control schools to support students and staff over the course of the study. These activities were not the same as those within the WISE intervention, but this could have diluted any effects if these activities were equally effective. However, there was also evidence of similar new activity in the intervention schools, so if the WISE intervention was effective we might have expected to see this over and above the impact of these other activities. The fact that schools across arms appeared to remain broadly similar in terms of delivering other activities shows that the WISE intervention did not have a harmful effect through causing a reduction in other activities supportive of mental health.
Strengths and limitations
Strengths
To our knowledge, this was the first RCT to evaluate the clinical effectiveness and cost-effectiveness of an intervention aiming to improve teacher mental health. The prospective collection of cost data allowed us to evaluate the trade-off between the investment (i.e. of time and money) required of schools against the potential for improved staff and student health and well-being.
The high response rates from both teachers and students, and the retention of all recruited schools ensured that the study had sufficient statistical power to identify important differences in mental health outcomes between study arms. In addition, high response rates and the use of multiple imputation in our analysis helped limit the risk of reporting bias due to missing outcome data. Collecting baseline data before randomisation helped to further mitigate risk of bias. The stratified approach to recruitment and sampling helped ensure that samples were balanced across study arms at baseline, with the exception of student attainment and teachers leaving for reasons other than retirement, both of which were higher in the intervention schools. An extensive process evaluation enabled us to answer important questions about how the intervention was delivered, participants’ experiences of the intervention and the context in which the intervention was delivered, which helped to shed light on why it did not lead to improvements in teacher or student outcomes. We successfully achieved delivery of all intervention components in all intervention schools and, to the best of our knowledge, no control schools took up the intervention during the study.
Limitations
The lack of blinding of participants was a limitation, as we cannot be sure that knowing they were in an intervention school did not affect their responses. Indeed, there were signs that some teachers in the intervention schools became frustrated by the lack of whole-school change, which may have led to reporting bias, particularly given that a self-report questionnaire was the main source of outcome data. A second problem with using self-report measures of well-being and mental health is the risk that gaining knowledge about mental health may affect how individuals then report their own mental health. This problem has been discussed in the mindfulness literature, in which those who are more experienced at mindfulness may paradoxically report themselves as less mindful because of this greater insight.92
An important potential limitation for our economic analysis is that we did not directly collect information on school staff or student use of mental health services during the follow period. Direct measurement at 12 months (teachers) and 24 months (teachers and students) would probably have resulted in recall bias,93 and potentially differential bias between study arms in an unblinded study. An effective peer support intervention might either increase (because of signposting) or reduce (because of addressing distress early) health-care use. We know from the log reports that peer supporters discussed available services in approximately 23% of cases (see Table 31). However, we do not know how many of these cases were acted on or which help sources were involved. If the trial had demonstrated that the intervention was effective and efficient from the school’s perspective, it would be less efficient from an NHS perspective if the peer supporters signposted many more teachers on to general practitioner and other NHS services. Given that we found no improvement in staff well-being, schools are unlikely to invest in or sustain the intervention regardless of any possible impact on health services.
Regarding limitations of the process evaluation, it was difficult to confidently measure use of the peer support service because of the informal way in which it operated, in the context of pre-existing relationships. It is, therefore, hard to be sure of how extensively the service was used. Furthermore, because we did not manage to identify enough users of the peer support service willing to take part in an interview, we were unable to evaluate the quality of the support provided beyond a closed question in the teacher questionnaire. We did not manage to collect all intended data from all schools; specifically, we did not collect peer supporter feedback from all schools 1 year after service set up and we did not manage to complete a T2 audit for all schools.
There were aspects of the intervention that did not go to plan. We did not manage to train the specified number of teachers in all schools, and numbers reduced further in some cases because of staff leaving. We were unable to build in more formal external support for the peer supporters from local CAMHS link workers or HSCs, which may have led to the peer supporters finding the role more challenging and may also have contributed to the sense that this new role had not been properly acknowledged. Finally, as discussed above, we were also unable to ensure that the peer support service was delivered as intended, including, crucially, with support from senior leadership, and with a clear confidentiality policy.
Links to the wider literature
Despite the lack of impact on student mental health, MHFA was viewed as valuable by participants in terms of improving their knowledge, confidence and skills in responding to the needs of students that they experienced in their day-to-day roles. It is therefore important to continue to explore effective ways in which mental health training might be able to be delivered. Building mental health training into initial teacher training might be one way to ensure that all teachers are equipped to respond appropriately to vulnerable students in their care, which may lead to greater and more sustainable changes relating to teacher–student relationships and mental health outcomes. Most MHFA evaluations have not measured impact beyond 6 months.49 One evaluation of the impact of training parents of teenagers found improvements in knowledge and confidence were still visible at 2 years, but had decreased in effect size.93 Therefore, examining the possibility of delivering regular refresher training would be important as part of any training delivered to all teachers.
By comparison, the peer support service, although viewed positively as an idea, was not implemented with the same consistently high fidelity and, partly as a consequence of this, was not well used. Other evaluations of peer support work have uncovered similar challenges in terms of implementation and identified the need for such services to have clear expectations around the role and how it fits within the wider organisation, something that was not always achieved in the WISE intervention.94,95 For such a service to work well in the school context, it is clear that more work needs to be carried out with schools and senior leaders to explore where they see this service fitting into the broader context, and how the role can be more formally recognised as part of the wider work of staff within the school. In addition, if such an intervention were to be pursued in schools, evidence from the process evaluation indicates that peer supporters would benefit from more formal supervision, similar to that received by counsellors. This would of course increase the costs of the intervention.
The failure of the intervention to address barriers to help-seeking expressed by both teachers and students ultimately meant that there had been no significant change to whole-school culture in relation to the stigma of seeking help, something reported elsewhere by teachers.24 A theme running throughout the process data was the difficult job that teachers do and how the situation is worsening. Teachers in our sample had poor well-being and a high level of depressive symptoms compared with the general population, and 85% felt that their job was moderately to extremely stressful at baseline. A recent survey published by the Education Support partnership of educational professionals painted a similar picture, with 72% of respondents describing themselves as stressed and the sample having a mean well-being score on the WEMWBS of 44.7.95 Furthermore, resonating again with themes from this study, 60% of respondents said that they would not feel confident disclosing stress or mental health issues to their employer and 43% felt that their employer did not support staff mental health adequately. Evaluations of public health interventions need to take into account the ways in which context may constrain or enable delivery.96 This wider picture, although confirming the need for interventions to improve support for teachers, also created a challenging context for the WISE intervention to be implemented in, and to have a significant and sustainable impact.
Against this backdrop of challenges within the teaching profession, the WISE intervention had initial high acceptability, as it fitted a wider developing agenda around staff mental health. However, this acceptability waned over time, as participants realised that it had been minimally disruptive to the wider, deeply entrenched culture in which they worked,97 characterised by demanding workloads, high levels of accountability and a fear of asking for support. Hawe98 suggests that the success of an intervention delivered into a complex system, such as a school, depends on the extent to which that intervention harnesses and ultimately reconfigures the dynamic properties of that system, including the social networks therein.98 Schools are complex systems in the sense that they are made up of a series of interacting subsystems, including the teacher system, the senior leader system, the non-teaching staff system and the student system. In the case of the WISE intervention, those interactions were not sufficiently changed to create new customs, norms and expectations within the whole system. Previous studies have reported that school culture (namely the extent of accountability, how far senior leaders are seen to care about teacher welfare and connectedness among teachers) can be more influential on teacher stress and satisfaction than discreet interventions aiming to support them and, indeed, that these school-level factors can interfere with the successful implementation of such interventions.99–101 In the present trial, the key initial gains of the WISE intervention (i.e. learning from the MHFA training and provision of a peer support service) failed to become embedded into ongoing practice in any meaningful way. Therefore, they did not lead to changes at the system level in terms of the norms regarding expectations made of teachers and how far mental health is prioritised. The evidence suggested that this led to a certain amount of disillusionment from teachers who had initially invested willingly and enthusiastically in the intervention, leaving them with the view that the WISE intervention had been treated as a tick-box exercise. The difficulties experienced by the peer supporters, particularly regarding role credibility and feeling what they were doing was a core part of the organisation, have been reported elsewhere.102 The perceived failure of senior leadership to prioritise and embed the intervention was raised as a particular sticking point by participating teachers, and, therefore, future work exploring how to better engage educational leaders in the task of supporting teachers’ and students’ mental health would be valuable. One recent workplace RCT delivered mental health training to senior managers within the fire and rescue service, and reported a positive effect in terms of reduced levels of sickness absence among the staff that they managed.54 Another important avenue to explore is support for senior leaders themselves. The Education Support survey found that 74% of all education professionals described themselves as stressed and this rose to 84% for senior leaders.95
It was clear from the audit data that a large number of potentially distressing events and changes can happen in schools over the course of 2 years. However, negative inspections were reported in intervention schools only, in our sample at T2. Despite regular reports of the stressful impact of an Ofsted report in the media, there is very little research examining this as a determinant of stress or poor mental health. Two small-scale studies103,104 have shown associations with increased levels of stress and sickness absence among teachers during the period just before and just after an inspection. In addition, one of the studies104 showed that staff morale is generally affected after any inspection, regardless of the outcome. It is, therefore, possible that this stressor contributed to a particularly difficult culture among those intervention schools affected, which may have over-ridden any potential impact of the intervention.
Summary of research recommendations
Future school-based research to improve mental health should focus on the following.
How to fully engage senior leaders in mental health interventions
This may include focusing on training for the leaders themselves, addressing the pressures that they experience and ways in which these pressures can be alleviated, and identifying facilitators that would enable them to prioritise mental health. The inclusion of assessment of mental health support for staff and students in Ofsted inspections may be one way to incentivise senior leaders to give such interventions the status and investment they require.
How to address the structural determinants of poor teacher mental health, such as high workload and pressure to perform
The application of systems thinking to public health improvement has become increasingly popular, in which stakeholders map aspects of the system that influence the outcome of interest and where change is required. Such an approach could be applied to the school system, with subsequent evaluation of system-level changes. One characteristic of the system identified in this study is the culture of silence and stigma around mental health that still pervades much of teachers’ working lives. Research to understand how this culture is perpetuated and how it can be challenged would be an important aspect of such system-level work.
How to strengthen relationships within schools (between staff, staff and students, and staff and senior leaders)
Introducing mental health training to a small group of staff was not enough to see a change in these relationships. It may be that whole-school changes involving policy and practice are needed to create more positive and supportive interactions as the norm.
How to build support for schools within the wider mental health services system
In this study, we explicitly planned to set up support for the peer supporters via CAMHS and HSC teams that were already working in schools, but this proved too challenging in practice. The introduction of education mental health practitioners to certain local authorities as a consequence of the government’s Transforming Children and Young People’s Mental Health Provision: A Green Paper16 may provide a model in which this can be achieved, although the focus here is on student mental health and not staff mental health. In the absence of sufficient specialist provision for children and young people’s mental health, expectations are likely to continue to fall on schools to support ‘low-level’ mental health needs, creating further stress for busy teachers. One important avenue for future research could be to examine the feasibility of external services providing formal supervision for school staff to support them in this role.
Whether mental health first aid or other mental health training does have a place within schools alongside system-level change
The training delivered as part of the WISE intervention was well received by teachers, but there was evidence that not all knowledge was retained and dosage may not have been sufficient. Future studies could investigate whether or not the impact increases with a higher dosage and regular top-up sessions. Consideration could also be given to whether or not particular teachers would benefit most, for example those who are still in training or newly qualified, and whether or not learning should be more formally cascaded to colleagues. In addition, ways to more formally recognise the contribution of those who have received the training to supporting mental health should be explored, for example releasing staff for a set amount of time each week to engage in support work, although this would add to the costs of the intervention.
Conclusions
Schools are at a ‘tipping point’. Teacher mental health is becoming increasingly acknowledged as an important focus for interventions because of high levels of stress and distress, and poor well-being in this population. The WISE intervention was acceptable, those who engaged with it found it useful and there was some evidence of a general raising of awareness of the importance of mental health and its impact in schools. However, it did not have an impact on teacher or student mental health. This is likely to be because, following the intervention, there was not enough of an increase in support accessed by staff or students, and there was no evidence of better-quality relationships, reduction in work-related stress or changes to a school culture characterised by high workload, accountability and fear of asking for help. Ultimately, the WISE intervention did not create a system-level change.105 Future studies aiming to support teacher or student mental health need to take a more radical approach that seeks to understand how to be more disruptive of these systemic, deeply embedded cultural norms. This needs to include further exploration of how to reduce the stresses faced by senior leaders themselves, and how to better engage them in supporting sustained and meaningful work towards improved teacher and student mental health.