Mental Health Promotion for the ‘In-Betweeners’: The Rationale and Effectiveness of Community-Based Mentoring and Coaching Schemes for Primary School-Aged Children.

O’Connell et al9 identified four key features of normal development that need to be considered when designing promotion and prevention interventions, these are:

related patterns of competence and disorder- it is recognized that the development of specific ompetencies occurs throughout life but that competencies acquired in early childhood (also referred to as developmental assets) establish a foundation for developing other competencies in the future. Failure to develop certain competencies at an early age may affect a broad range of functional domains (e.g. physical, cognitive), and/or behavioural decision making at a later age (e.g. risk taking behaviour). It is suggested that the more competencies developed, the greater the individuals’ ability to tolerate adversity in the future^{10, 11}.

this refers to expectations regarding an individuals’ behaviour in given social contexts; these may vary with age, gender, culture and across generations or over time (decade by decade). Examples include developing secure attachments, appropriate conduct, etc. The young person and/or other people judge the level of success in completion of these tasks and perceived failures affect subjective or observer views of competence and self-confidence¹⁰. For example, children who failed to learn behavioural self-regulation and manifested conduct problems in early schooling were more likely to experience academic failure in adolescence and this group had a higher risk of depression in early adulthood¹².

individual development at different ages occurs in multiple contexts: family/home, school, neighbourhood, community, and culture etc. These micro-, meso- and macro-systems, influence developmental processes and the acquisition of competence requires individuals to adapt to the demands of different contexts and to negotiate transitions between elements of the system (e.g. between family and school, etc.).

Interactions between Biological, Psychological & Social Factors- many complex interactions occur between different factors, (e.g. genes and environment can influence developmental trajectories, offering potential opportunities to understand and/or intervene in selected processes or pathways). Also, it is known that temperament and personality characteristics can influence the events to which a person is exposed and/or how they react to these.

In summary, mental health promotion attempts to establish specific competencies, increase the completion of developmentally-appropriate tasks, enhance the chances of positive development, and strengthen an individuals’ adaptability and tolerance of adversity⁹. Theoretically, several different ‘promotion’ models can address the elements highlighted in the developmental framework. In practice, the most commonly employed interventions all include some form of mentoring or coaching.

The terms developmental mentoring and coaching are used in diverse ways, but in this review, they are used to describe a stable, supportive, created relationship with an ‘attuned’ unrelated adult^{13, 14}, where the main goal is to use this relationship as a vehicle to facilitate social, emotional and academic skills development, etc^{15, 16}. As compared to other models or prevention strategies, developmental mentoring or coaching is usually used for children without pre-existing evidence of mental disorders or problems (and is the most relevant universal mental health promotion approach for this review).

Developmental mentoring and coaching can be delivered as a single intervention or as part of a multi-component ‘youth development’ strategy⁶. According to the National Research Council and Institute of Medicine¹⁷, the approach should be undertaken in a positive development setting, which are characterized by: physical and psychological safety; appropriate structure; supportive relationships; opportunities to belong; positive social norms; support for efficacy and mattering; opportunities for skill building; and integration of family, school and community efforts. Lerner et al¹⁸ identify that such settings support developments in five ‘C’ domains: confidence, competence, character, connectedness and caring.

As summarized in Box 1, Rhodes and colleagues have proposed that once a strong emotional bond exists between the mentor and mentee, the relationship may contribute to positive outcomes via three interacting developmental processes: social-emotional, cognitive and identity^{15, 18, 19}. For example, the mentor may enhance social skills and emotional well-being by modelling effective communication, and helping the child understand, express, and regulate their emotional responses. The mentor may help to improve cognitive skills through dialogue and listening (and by modelling problem-solving skills in situations involving alternative choices). In addition, the mentor can serve as both a role model and advocate by demonstrating appropriate actions and behaviours¹⁵. Importantly, the model assumes that the positive social-emotional learning experiences will generalize to interactions with other people and other environments¹⁹. For example, studies have shown that mentees who developed a close relationship with a mentor then developed improved relationships with their parents and the latter mediated an improvement in academic performance¹⁸.

Dolan and Brady²⁰ argue that effective models of mentoring or coaching involve both a positive, high quality relationship and engagement in regular shared activities over an extended period of time. Further, they comment that the rationale for using formal programmes stems from several recent observations, not least of which is that ‘natural’ (or informal) mentors are less available than in the past. As such the opportunities for informal interactions between younger and older people have diminished (probably because of changes in social structures such as family make-up, parental employment patterns, family mobility, and contact with members of the extended family or others in the neighbourhood, etc.). Also, it has been suggested that primary school age children may be more receptive to an adult mentor, develop a closer relationship, and be less likely to terminate the interaction prematurely than adolescents (who may be less trusting or committed to the relationship, or more vulnerable to peer pressure)^{5, 21}.

In summary, all these approaches assume that addressing basic developmental needs across multiple contexts through personal relationships and constructive activities can promote healthy outcomes and protect against engagement in risky behaviours. Having noted the theoretical models and rationale, we then examined the evidence for community-based mentoring or coaching for children aged 5-11 years. In addition, we highlight any reports that examine some of the proposed mediators or moderators of benefit.

Evidence-Mapping: Is developmental mentoring or coaching effective in children?

For the purposes of the evidence mapping exercise, we used a published definition of mentoring or coaching¹⁵, which described as ‘a relationship between an older, more experienced adult and an unrelated, younger mentee (protégé)…in which the adult provides ongoing guidance, instruction, and encouragement aimed at developing the competence and character of the mentee’.

Whilst there are an enormous number of individual studies and large-scale reviews of mentoring, very few reviews directly address OST programmes for 5-11 year olds, or report on any differences in effectiveness due to the age of mentees and/or the context/structure or primary goals of the programme. Furthermore, many studies that address some of these issues are published as book chapters rather than peer reviewed papers, whilst other publications are not of the quality required for inclusion in a systematic review. Interpretation of the data is also hampered by the wide range and varying reliability and validity of the subjective or observer rated measures employed, and many studies do not used objective data (such as physical health measures, national school records, etc.). We therefore undertook an umbrella review (or a ‘review of reviews’) to synthesize the ‘state of the art’. The review aimed to documented contextual information on how mentoring for young children has been organised, implemented and delivered. We used an ‘umbrella review’ methodology as this has increasingly been shown to be useful in public health and has been found to provide important insights into the social determinants of health and medical illnesses. The approach uses a typical systematic review methodology but focuses on locating, evaluating and synthesizing published ‘review-level’ evidence (narrative reviews, systematic reviews and meta-analyses) and then extracts information on specific studies that may be pertinent to answering the key questions being addressed.

This approach enabled us to build an evidence-map of the findings from previously published, peer reviewed meta-analyses of outcome data in which some or all of the eligible studies included-

(a) primary school age children engaged in developmental mentoring programmes delivered in the community/OST,

(b) mentoring interventions that were undertaken on a one-to-one basis or as one component of a youth (child and adolescent) development programme,

(c) similar interventions (not necessarily referred to as mentoring) that were explicitly based on the developmental framework and conceptual model outlined (i.e. social learning; positive youth development ^PYD; self-esteem/self-concept; physical health promotion),

(d) outcome data for more than one domain associated with mentoring processes was reported (i.e. social-emotional, cognitive, identity, developmental competencies, etc.), or an intervention that addressed one domain was assessed by outcomes in another domain (e.g. the influence of exercise on self-esteem or self-concept).

Thirty-three systematic reviews were screened, of which nine were meta-analyses that met criteria for inclusion in the evidence map. As shown in Table 1, blank boxes indicate ES unavailable for that category eight publications^{13, 22, 21, 22, 23, 24, 25, 26, 27, 28} focused on outcomes of individuals participating in programmes: two reviews specifically examined mentoring programmes, two examined OST programmes and four examined health promotion, social competence and/or self-esteem (or self-concept ). A further review²⁹ examined the effects of development programmes from the perspective of the social systems involved (family, school, community).

The evidence is reported from three perspectives:

Do the interventions work?

As the studies included in each meta-analysis used many different pre-post intervention measures, outcome data were grouped into broad categories. These included: academic competence, self-esteem/self-confidence, connectedness/relationships, or global outcome (which represents a composite measure that the researchers derived from all the assessments), etc.

What is the magnitude of any effect (reported as effect sizes: ES)?

The ES are reported for as many post-intervention outcome categories as feasible as well as ES for the limited number of follow-up evaluations (see Table 1). The recommendations for educational research suggest that the thresholds for significance are lower for promotion/prevention interventions compared to those applied in clinical trials³⁰. As such, ES between .2 and .4 represent a small but significant impact, with ES between .4 and .6 representing a medium level of impact and ES >.6 regarded as a large impact.

What factors influence the effectiveness of mentoring?

Analyses of moderators of mentoring effects were identified and ES or correlations were noted.

As shown in Table 1, two reviews by Dubois and colleagues^{13, 22} offer important insights into the potential benefits of mentoring. These meta-analyses of mentoring studies in children and adolescents examined several key domains such as emotional/psychological functioning, social competence, academic performance, problem/high risk behaviours, etc. The first meta-analysis was of 55 studies published before 1999, whilst the second reviewed 73 studies published from 2000 onwards.

Dubois et al¹³ demonstrated that there were small but definite benefits from participation in mentoring (ES for global outcome= .14) and a limited number of follow-up studies were identified that showed that some of these benefits were maintained post-intervention for up to two years (ES .10). The second meta-analysis22 showed marginally higher ES for global outcome (.22), and the available follow-up data again suggested that there were small but enduring positive benefits (.17). The later paper also reported a small non-significant effect on physical health (.06), although the data are difficult to interpret as only a small proportion of studies included any measurement of this domain. These two reviews of mentoring interventions provide the most detailed analyses of moderators of mentoring effects. Significant influences on the ES included the presence or absence of parental support/involvement with the intervention (no parental support: .13; support: .27) and the ‘risk’ status of the mentee (no risk: .15; high risk: .26). When risk was broadly divided according to environmental (e.g. lower socio-economic status) or individual risk (e.g. the presence of emotional or behavioural problems), it was found that the benefits of mentoring were significant for the environmental group (.17) but not for the individual risk group (.03). Interestingly, further analysis suggested that the individual risk group might benefit if the mentor had a ‘helping’ profession background (e.g. teaching, health professional, etc.) rather than being a ‘community volunteer’.

Dubois et al¹³ also found that, after controlling for other variables that may explain the outcomes, the age of the mentee (child: .17 versus mid-late adolescent: .13) and the context (community-based: .15 versus school-based: .11) moderated the effects of mentoring. However, these factors had less impact on mentoring outcomes than the structure and quality of the intervention programme. These had a highly significant influence on the outcome, with key elements being: the use of structured activities; expectations (regarding the contact with the mentor); the intensity and ‘quantity’ of the programme (i.e. emotional closeness, frequency and regularity of weekly contacts, duration of the programme); the ongoing support and training for mentors; and the monitoring of the overall implementation of the programme. They concluded that theory-based, carefully constructed interventions (i.e. those that employ ‘best practices’) were significantly superior to ‘atheoretical’ models undertaken by organizations that lacked the infrastructure to assure the quality of the intervention and/or that paid less attention to evidence-based approaches (ES .22 versus .07).

The moderator analyses in the later publication were more circumscribed²². However, it revealed that the association between risk status and mentoring outcomes showed a curvilinear relationship: mentees with a medium level of risk (ES .32) showed greater ES for mentoring than mentees with either low (ES .25) or high risk (ES .22). Furthermore, the analyses demonstrated that differences between mentored and non-mentored individuals could be explained by a relative slowing in the emergence of negative behaviours in the mentored group, compared with the more overt expression of behaviour problems (such as conduct issues or risk taking) in the non-mentored groups. Also, there was a trend for the mentored group to show improved developmental trajectories for positive assets.

Several of the findings from the reviews by Dubois’ group were confirmed in other meta-analyses. For example, it was shown that the theoretical underpinnings and infrastructure of the programme are important moderators of mentoring outcome²⁸; that OST programmes are effective, and that, whilst simple (single target approaches) and multifaceted programmes are both beneficial, greater benefits may accrue from multifaceted programmes in those aged older than nine years²⁵. The data on age sub-groups is not straightforward as most programmes identify some gains for most age groups. Although Durlak and Wells²⁶ suggest the gains in 2-6 year olds are much greater than 7-11 year olds, this finding should be interpreted with caution as some studies lacked appropriate comparison groups. As such, whilst greater ES may be uniquely related to the intervention, the magnitude of the ES could also reflect the normative developmental gains that occur in each age period (which are usually largest in the early age group).

All eight meta-analyses demonstrate that the benefits of interventions are not just ‘domain specific’ and, for example, non-cognitive social interventions may promote improvements in cognitive skills and academic performance. Also, Ekeland et al²⁷ demonstrate the significant impacts on self-concept and self-esteem of exercise interventions. However, these findings should be regarded as offering preliminary insights on this topic rather than definitive answers as there were few other meta-analyses for comparison.

Only one publication to date that synthesizes data on the effects of ‘developmental’ programmes on school, family and community systems²⁹. The review included over 500 studies, but the paucity of data assessing certain domains (e.g. community bonds) made some ES difficult to interpret. Despite its shortcomings, the publication is important given the potential bi-directional influences of programmes and ‘system’ elements. These influences included the social capital of a community (e.g. networks and cohesion), and the socio-demographic make-up of the location where the programme is undertaken, etc³¹. Overall, Durlak and colleagues²⁹ demonstrated significant ES for the effects of the intervention with regard to the family (family environment: .34; parenting practices: .41), family-school relationships (.49) and various aspects of the school environment (ES ranged from .47 to .78).

What are the costs of developmental interventions?

It was not feasible to identify the cost of developmental mentoring programmes for children separately from youth. Furthermore, the nature and quality of individual studies of developmental mentoring (or other similar interventions) make a systematic evaluation of cost-effectiveness impossible. Indeed, the only detailed health economics review of programmes (undertaken in the USA)³² showed that the cost of delivery of approaches labelled as either mentoring or as youth development programmes ranged from $8.00 to $26,000 per participant, with gains per dollar invested ranging from $0.42 to $28.42. Asos et al³² also undertook ‘costs versus savings’ calculations based on a broad range of potential adverse outcomes that could occur within a short time frame (such as rates of juvenile delinquency; school/college dropout) or represent longer-term ‘down-stream’ consequences (lower earning capacity; development of mental disorders requiring clinical treatment, etc.). Using this approach the overall ‘benefit minus cost’ estimates ranged from +$9,800 to -$15,000 per person offered the intervention.

Most importantly, Asos and colleagues³² were unable to determine whether programmes with more funding were associated with better outcomes. The researchers stated that rigorous cost estimates will not be possible until organizations initiating mentoring or other schemes place a greater focus on implementation strategies and ongoing quality control of the programmes (including outcome and cost-effectiveness evaluations).

The current best estimate of the cost of delivering a community-based mentoring intervention with OST activities is about $1500. Although this appears to be higher than the average cost reported a school-based programmes (range ~$500-1000 depending on the structure and goals of the programme), it is notable that mentees spend twice as much time per month with their mentor in community-based programmes (up to 12 hours) `compared to school-based interventions³³.