Our systematic review identified scientific evidence concerning the benefits and harms of school-based interventions, which were directed at students 18 years of age or younger to influence MH discrimination. However, five limitations prevent us drawing conclusive inferences regarding interventions' risk-minimizing ability to eliminate or prevent it.
Consequently, we cannot determine which interventions or intervention types "work," which "work best" or "better," for whom, in what terms (i.e., outcomes) or under what conditions (e.g., setting). Likewise, we cannot identify the intervention types, activities (e.g., contact), materials (e.g., video) or contents (e.g., MHDs portrayed) that are necessary or sufficient to produce population-specific or population-independent benefits while also avoiding harm. Few studies reported having investigated possible harms. As a result, we cannot recommend any single school-based intervention or intervention type.
Yet, the evidence also does not permit us to identify those intervention types, interventions or characteristics that lack the potential to reliably produce benefits. Virtually none of the efforts to date have entailed appropriate appraisals of this potential. New research should likely begin by correcting this state of affairs. There is likely little sense in trying to "reinvent the wheel" when there are many approaches, whose value is largely unknown. Moreover, there are numerous school-based interventions, whose impacts have not yet been evaluated [11, 69–77]. Yet, going beyond existing philosophies and practices should not be ruled out.
Additional recommendations for future research are informed by largely consistent viewpoints obtained from three sources (see below): a) investigators' interpretations of their and others' suggestive yet inconclusive results from studies that we reviewed; b) researchers whose efforts to influence MH stigma focused on individuals outside our population of interest; and c) those youth, service users, advocates, volunteers, researchers, educators, clinicians, and policy-makers who attended a recent international workshop . The workshop was organized in response to the findings of our review, and was intended to derive a meaningful research agenda by further examining the state of the science from the perspective of "real world" experience and expertise. Foremost among the shared views is that "contact" is likely necessary but alone may not be sufficient to produce maximally beneficial outcomes (see below).
What we propose cannot be drawn solely from those interventions, which our review highlighted, since virtually none entailed ongoing (e.g., multiple exposures), curriculum-based efforts. Rather, they tended to be one-off, brief events whose typically cognition-focused outcomes (e.g., knowledge, attitudes, stereotypes) were evaluated over the short-term in mostly older children; and, many did not aim to facilitate an experiential engagement with individuals experiencing MHDs.
It is our view that interventions should be employed to develop a sustainable, self-regulating "compass," which by its very nature makes highly unlikely, if not impossible, any inclination to perceive or behave in ways that discriminate against those experiencing MHDs. But, to achieve this aim, school-based interventions should likely exhibit certain characteristics. Otherwise, we expect that any changes brought about by strategies that do not satisfy these conditions will not be substantive enough to assure their generalization much beyond the specific context or time period in which they were implemented, or the developmental stage of those students who were exposed to them.
Interventions should likely involve experiential activities, which in facilitating students' interaction with other human beings, engage students' feelings and behaviour, not just cognition-based points of view [33, 45, 50, 79, 80]. As well, given how early in life discriminatory viewpoints and behaviours can appear, early intervention is a reasonable aim [10, 32, 66]. While for some students this might constitute primary prevention, for others it would represent secondary prevention.
The intervention should likely be implemented multiple times within and across the school years (e.g., "critical periods") [40, 43, 50, 51, 59, 65]; and, starting early could maximize the number of exposures to activities, materials and contents that are repeated both within and across successive stages of cognitive and affective development. Yet, to be able to foster a deepening integration of the benefits afforded by these exposures, the activities, materials and contents should be modified in ways, which over successive implementations, are incrementally challenging yet always developmentally-appropriate. In short, we propose a curriculum, whose implementations reinforce and build upon prior ones [32, 45, 67].
Implementers, who include those experiencing MHDs, should likely be those with whom the children or youth are most likely to identify (e.g., those most similar to themselves) [50, 67]. Yet, actively involving their teachers, other school staff, the school administration and parents could maximize the likelihood of making a sustainable difference. Moreover, child and adolescent involvement in creating, refining and pilot-testing the curriculum is likely essential to maximize the relevance and developmental appropriateness of its components and the timing of their implementation.
Our review team echoes what other sources have opined about the need for contact-based interventions, which reflect an experiential approach, to produce substantive, especially behavioural, change [39, 40, 43, 51, 52, 62, 65, 68, 78, 79, 81–83]. For example, a recent, large and well-controlled meta-analysis of study data, which were identified without specific reference to our topic, found that intergroup contact typically reduces intergroup prejudice across a broad range of groups and contexts . Yet, while the creation of carefully structured contact situations enhanced the magnitude of benefits, it was not required to produce them.
As well, some contact-based interventions, whose evaluations we appraised, did produce some suggestive evidence of benefit, which might be confirmed by high quality RCTs [43, 48, 49, 67]. Perhaps most notable, however, are the conclusions offered by investigators, whose education-only interventions we reviewed. Several claimed that contact is likely necessary to produce substantive change [40, 65]. Unsurprisingly, contact's candidacy as the most promising way to do so also comes from the observation that research on a frequently employed alternative has not engendered confidence in the latter's capacity to reliably produce such benefits. Education-only strategies sometimes produce shifts in knowledge, attitudes or stereotypes , yet these do not reliably predict behaviour (see below) [33, 40, 46–48, 50, 60, 67].
It is our view that the greater promise of contact interventions to produce substantive change stems from the assumption that, compared with approaches that typically aim to influence responses to short-term, post-intervention queries concerning attitudes, stereotypes or knowledge, they are more likely to kindle the development of empathy. Our review did not identify a single contact strategy that was explicitly evaluated for its impact on empathy as an intermediate outcome, yet we hypothesize that empathy is the mechanism by which contact can produce substantive, behavioural change [33, 50, 86–89]. We also posit that the development of empathy, which might underlie a self-regulating "compass," likely requires the stimulation of affect and affect-based understanding within contact scenarios.
A recent controlled study conducted outside a school context, which involved naturalistic contact, found that stimulating affective responses can yield prominent change . Pairing volunteers with individuals, who were experiencing severe MHDs, reduced negative affective responses in the former with reference to the latter; and, changes in affective response were directly related to the quality of the contact. Furthermore, Angermeyer and Matschinger found that the tendency for pro-social action toward those individuals experiencing MHDs depended on emotional reactions ; and, personal stories shared by those experiencing MHDs can produce an affective response in children and youth [33, 50].
Authors of a study we evaluated observed that, when asked to select less stigmatizing descriptions, study participants tended to focus on others' feelings . Evolutionary theory notes that the perception of the emotional state of another living being appears to automatically activate the creation of a matching emotional state in the observer . Then, with increasing cognitive capacity, "state-matching" – which underlies the fundamental, nonverbal experience of empathic identification – appears to evolve into more complex forms, which include concern for the other and perspective-taking. Whether, and how, "mirror neurons" play a mediating role in state-matching for different species remains to be confirmed . Nonetheless, at least for human beings, the nature of the changes brought about by contact can also involve a notable perceptual-cognitive shift in how we typically experience reality and self.
The distinctions and dualities produced by our natural capacity to "difference-make" are the hallmark of the lifelong project by which our sense of personal (or collective) identity (i.e., self-sameness) develops and is (pre)served, and they allow us to make sense of reality and self [92–98]. However, especially unself-reflective use of this capability can culminate in the typically fear-imbued identification of "others" from whom separation by way of acts of exclusion (i.e., boundary-making) comes naturally, and against whom self-(pre)serving, discriminatory attention can be directed with little conscious effort [92–94, 98].
That said, our equally natural capacity for empathy affords an experience of experiential identification and understanding ("resonance"), which allows the observer to transcend the presumed boundary between self and "other" and thereby makes it more difficult to (exclusively) relate to this "other" in ways characterized by exclusion or exclusivity . Empathy may therefore be thought of as promoting an experience of inclusion and inclusiveness, which typically entails noticing, and acting upon the recognition of similarity or sameness. This capacity can balance our inclination to notice, and act based upon, difference at the same time that it can cultivate a sense of community. Finally, the development of empathy is likely integral to the realization of one facet of our transpersonal human potential, which is the compassion that knows neither boundaries nor "others" .
So, how might an empathy-facilitating, contact-centered curriculum be described? Each implementation could involve students being brought into direct contact with those experiencing MHDs. However, based on work which has not had our population of interest as its explicit or sole focus, it has been suggested that certain criteria should be met in order for contact to stimulate meaningful change [81, 82].
Interventions should promote interactions between those experiencing MHDs and their "audience," which demonstrate their equal status, provide an opportunity for them to get to know each other, and foster their active co-operation in the pursuit of a mutual goal such as sharing information that challenges negative stereotypes . Yet, a narrative review of that literature found that many of the studies from which these criteria were derived were fraught with the same methods-related problems that we observed in our review . As an aside, failing to satisfy some of these criteria may explain why the behaviour of some MH professionals (e.g., "playing the elevated, expert interpreter and labeler of others' experiences and reality") [99, 100] can be seen as stigmatizing by those who experience MHDs .
The typical contact approach, which invites into the classroom those who reflect upon their experiences with MHDs, is likely too large a cognitive challenge for the youngest students (e.g., those in kindergarten or the first few grades). Instead, a generic form of contact could be employed, which aims to stimulate and develop empathy. Only as these children develop would the typical types of contact intervention be employed.
An exceptional candidate for inclusion in an empathy-centered curriculum, which has been successfully implemented as early as kindergarten and as late as early high school, is the Roots of Empathy program [86, 89]. It does not assert as one of its objectives the elimination or prevention of MH discrimination. Rather, it aims to cultivate the development of empathy and emotional literacy, to reduce levels of bullying, aggression and violence, and to promote pro-social behaviour, among various civility-related aims.
The program brings a neighborhood infant and parent into a classroom every three weeks for one school year. Using a structured, manual-based curriculum, which describes themes that are broken down into four age ranges, a trained instructor coaches students to observe the baby's development and label the child's feelings. Controlled, prospective studies have shown a significant decrease in aggression and bullying, along with an increase in pro-social behaviour [86, 89]. The program appears to promote certain "positive" facets of MH. These benefits have been observed immediately following completion of the program, and some were maintained after three years.
But, its usefulness in developing social and emotional learning could contribute either directly or indirectly, through its impact on intermediate conditions such as social inclusion and inclusiveness, to the prevention or elimination of various forms of (e.g., MH) discrimination. Pilot-testing should reveal how, when, and if this program could become a part of a curriculum that strives to achieve these aims. But, even if it were found to be an essential component, any plan to directly address MH discrimination would likely need, at some point in the curriculum, to employ contact elements that incrementally and appropriately expose students to individuals and issues with a specific focus on MH, MHDs and related discrimination. Contact opportunities might benefit from establishing certain favorable conditions (e.g., equal status).
Activities, materials and contents could be modified in ways that deepen and extend the development of empathy. For example, exercises (e.g., perspective-taking) could make increasingly explicit the relevance of empathy for dealing with issues of MH, MHDs and MH discrimination. Modeling and role-playing [33, 39, 51, 60] might be used to explore and practice appropriate ways to express empathy-guided behaviour.
Developmentally-appropriate discussions could be scheduled strategically over the years, which successively focus attention on issues concerning: a) difference-making, which involves the identification of differences (e.g., "bad" versus "good") integral to acts of (social) exclusion and discrimination; b) the perception of sameness, which can be associated with acts of (social) inclusion; c) MH and wellness (e.g., resilience); d) MHDs and help-seeking; and, e) starting with older children, the power dynamics of MH discrimination. Engaging their conceptual frameworks would depend upon children's cognitive and affective readiness. Educational materials, whose nature would need to be determined, could also be used to reinforce and extend issues raised through contact (e.g., focusing on similarities) . Both "heart" and "mind" could be educated concurrently within the curriculum .
Any resistance by teachers or schools to the inclusion of such a curriculum could be resolved by recognition of the following. Since what it aims to achieve with reference to MH discrimination employs at least one element (e.g., developing meaningful consideration for others) that is represented somehow within some, extant school (e.g., "anti-bullying") programs, it might be possible to unify these efforts. A single curriculum could be developed, which fosters the development of empathy, and which in promoting social inclusion and inclusiveness, serves to prevent or eliminate various forms of negative attention that can be directed at "others" (e.g., racism, ageism, sexism, MH discrimination, despoiling the environment). Lastly, getting different sectors (e.g., MH, health, education, youth justice) to recognize the potential of an empathy-based curriculum, which concurrently or sequentially influences unique and overlapping outcomes of pertinence to their respective mandates, might be enough to get these stakeholders to meet and collectively support its use.
This discussion, when seen against the backdrop of our review, suggests several questions that could be investigated in future research. Given our hypothesis that school-based interventions require empathy-inducing contact in order to produce substantive change, and the recognition that education-centered components at least in principle could contribute to eliminating or preventing MH discrimination, we might ask whether education meaningfully enhances any of the benefits that might be produced by contact alone. It may be the case that contact is necessary to produce substantive change yet alone it may be insufficient to do so. To bring about this change, education may be needed as an add-on. Yet, this view suggests that we already know that contact alone reliably produces meaningful benefits. We do not know this.
Therefore, several questions require investigation: Does contact (C) produce substantive changes, whose nature and magnitude are pre-established? Does education (E), when added to contact (C), produce substantive changes, whose nature and magnitude are pre-established? Does adding education to contact (C+E) produce significantly greater change than that produced by contact (C) alone? These questions may be investigated concurrently within a single research design. But, before we describe it, we turn our attention to the issue of appropriate outcomes.
Changes in knowledge, attitudes and stereotypes do not reliably predict behaviour [33, 40, 46–48, 50, 60, 67]; and, when evaluated especially in the short-term, these outcomes appear to be vulnerable to socially desirable response sets shaped by expectations that can be readily communicated via the nature of the intervention or the contents of pre-intervention assessments [33, 39, 40, 60, 64]. Moreover, it is discriminatory action or inaction that is particularly noxious. Consequently, the primary outcome should be behaviour, which can be observed especially under naturalistic conditions. Virtually no studies that we identified utilized such outcomes .
The outcomes also need to reflect the kinds of activities, materials and contents that hitherto have been provided. For younger children, schoolyard play, which focuses on the inclusion/exclusion of peers who are perceived to be different on some or any basis, could be assessed. Starting with older children, who are exposed to individuals and issues that focus on MH, MHDs or related discrimination, acts of inclusion/exclusion could be assessed with reference to others whose differences are defined by their MHDs.
Secondary outcomes might elucidate the perception of social distance, which as a possible measure of inclusiveness, could capture students' expressed inclinations to interact with those who are different in some way (i.e., for the youngest children) or specifically because they experience MHDs (i.e., for the oldest children). One subtle way to evaluate respondents' intended behaviour might be to ask them to describe their peers' intended behaviour toward someone experiencing MHDs. These observations, which in one study pertained to a child with autism, better predicted students' behaviour than did descriptions of their own intended behaviour . The latter may be more susceptible to socially desirable responding. Finally, the judicious inclusion of a few tertiary measures such as knowledge, attitudes and stereotypes could allow us to begin to identify the possible causes or correlates of (failed) behaviour change.
Whatever the outcomes, they should be validated as well as developmentally-appropriate. At what time intervals or developmental levels they should be assessed within the curriculum would need to be determined via pilot work. Finally, possible harms require a direct yet potentially subtle and qualitative evaluation [51, 66, 102] since efforts to prevent harmful events can inadvertently produce them [33, 63]. Harms data were under-evaluated within our evidence base, although unidentified harms could account for some studies' notable numbers of withdrawals or dropouts [40, 57, 65].
Returning to the proposed research, pilot work could justify conducting a two-group RCT (i.e., C+E versus C). Independent, uncontrolled before-after designs would have to demonstrate that each intervention can produce predefined types and magnitudes of benefit as well as minimize harm. As stated earlier, our identification of five RCTs suggests the absence of ethical or practical barriers to conducting these designs to investigate our topic [33, 39, 44, 46, 50, 57, 58, 64, 66].
The most appropriate design is the cluster RCT, whereby schools that exhibit similar profiles of relevance to the study (e.g., same basic curriculum; presence of children with MHDs) would be randomized to study arms to minimize the contamination that can arise when students within the same school are allocated to different study groups and inevitably discuss their respective exposures . Contamination can wash out real effects [40, 58, 59].
Yet, other criteria require satisfaction in order to allow us to meaningfully attribute any (lack of) observed benefits exclusively to an intervention. While educational activities, materials or contents would be used to complement contact (i.e., C+E), those participants in the condition (i.e., C) that does not receive them would need to have something provided in order to control for the "generic" elements (e.g., attention paid, time spent, a novel event) that are given to the C+E group. Attention could be directed to general health issues.
Outcomes would be assessed at baseline as well as following interventions in order to determine whether, in addition to between-group differences, meaningful changes occur within each study group. One possible result is that both interventions produce the same magnitude of meaningful benefit, which would allow schools to select between two types of curriculum (i.e., C+E or C).
Since pre-intervention exposure to those with MHDs, a personal history of MHDs and help-seeking, prior experiences of MH discrimination, pre-study levels of empathy, baseline cognitive-affective capacity, and pre-intervention types and intensities of discriminatory knowledge, attitudes or stereotypes about MH can each influence how students respond to a review-relevant intervention, these variables require pre-study evaluation (and possible experimental control) [11, 33, 39, 40, 43, 44, 46, 48, 50, 53, 58–60, 62, 64, 66]. The presence and absence of notable pre-study biases would identify those students for whom the intervention is secondary and primary prevention, respectively [33, 39, 44, 46, 60, 67].
However, since evaluating pre-study characteristics can serve as cues that sway responding in expected or desired directions as well as establish a ceiling on the magnitude of possible changes from baseline, it might be best to conduct pre-study assessments for only a randomly selected half of the schools that are allocated to each study condition [50, 67]. Participants at the other schools would be asked about contents outside the study's focus (e.g., general health). This strategy would permit an evaluation of the impact of conducting pre-study assessments on results.
Primary analyses would be completed with data obtained from the intention-to-treat population, while the impact of potential, confounding factors could be investigated within secondary analyses. Additional, candidate confounders, which demonstrated some potential to influence outcomes within our evidence base, include: socio-demographic factors; religious beliefs; and, the presence at school of students who are experiencing MHDs [11, 44–47, 49, 50, 59, 60, 66, 71–73].
Finally, it could be argued that, to be most salient for older children and youth, the MHDs portrayed via contact and education exposures should be ones that are the most prevalent within this population. Instead of focusing on schizophrenia, which is the case within the WPA program, interventions could highlight MHDs that students are more likely to encounter amongst peers, for example (e.g., anxiety disorders).
Limitations of our review include being unable to contact all authors to clarify poorly reported study details, although it is unlikely that successfully obtaining these particulars would have changed our observation that almost no studies exhibited even adequate methods-related quality. Our focus on students 18 years of age or younger means that we did not review studies that exclusively enrolled older students. A list of these studies is available upon request.