2.1 Design

The methods and results are presented in line with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.¹⁶ The review protocol was developed and revised by the authors. The protocol is registered in PROSPERO (CRD42019122057 available from https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42019122057).

2.2 Search strategy and study selection

Ten databases (PsycInfo, EMBASE, Medline [OVID], Scopus, ASSIA [ProQuest], SSCI, SCI [Web of Science] CINAHL PLUS, Social Sciences full text [EBSCO]) were systematically searched from inception to July 2020. Initial searches were completed in January 2019 and again in July 2020. The search strategy was originally developed in two databases and customized to each of the other databases searched. Controlled vocabulary was included as medical subject heading and relevant articles were assessed for keywords to optimize the identification of articles of interest for the final searches. Four key components were used to structure the search strategy using the population, intervention, comparison, outcome (PICO) framework: (1) mental health and wellbeing, (2) LMICs, (3) perceptions and (4) child and adolescent populations. In each component, synonyms were combined using the Boolean operator ‘OR’ and across components using ‘AND’. Searches were adapted as necessary for individual databases. An example search strategy is available from the author on request. Forward citation tracking was undertaken for included studies up to April 2020.

2.3 Eligibility criteria

This review included original research that utilized primary data to examine the conceptualisations, views and perceptions of children and adolescents regarding positive mental health and wellbeing. Studies that had used qualitative, quantitative or mixed-method designs were included. Also, studies that reported on samples of children and adolescents in LMICs with a mean age under 18 were included. LMIC countries were defined by the Organization for Economic Cooperation and Development (OECD) Development Assistance Committee 2018–2020. Peer-reviewed journal articles and dissertations were included. Conference paper authors were contacted to include peer-reviewed, full-text articles of studies where available. Non-English studies were included, and data were extracted by bilingual researchers affiliated with the study team. No date restrictions were used, and studies were not excluded based on the results of the quality assessment. Full inclusion/exclusion criteria can be found in Table 1.

2.4 Screening

Returned records from database searches were combined, duplicates removed using Endnote software, and remaining references imported to the Covidence tool (https://www.covidence.org/). Two reviewers independently screened the title and abstracts of each study for relevance during the first stage of screening. Full texts of potentially relevant articles and those which did not contain sufficient information at the level of title and abstract were obtained and double-screened by two reviewers. Disagreements regarding inclusion or exclusion that did not achieve consensus were resolved by a third reviewer not involved in the original decision. The authors met regularly throughout the review process to discuss the process of screening and resolve any difficulties or challenges in the process. Reasons for exclusion at the level of the full text are documented in the PRISMA diagram¹⁶ (please see Figure 1).

2.5 Data extraction

We extracted data using electronic forms created in Microsoft Excel expressly for the purpose of organizing data from selected studies. We piloted extraction using 10 studies across the review team to ensure field titles and descriptors were interpreted and extracted consistently allowing further iterations before extracting all data. Data extraction and quality assessments were conducted simultaneously by the study team (L. R., H. B. and R. P.).

Primary data were extracted from quantitative, qualitative and mixed methods studies relating to the review questions simultaneously. Data relating to study conduct were also extracted including publication, country, and setting (community, school-based, clinical), study design, primary aim, metholodological approaches employed and other relevant contextual information.

2.6 Quality assessment

We anticipated a wide range of methodologies in this review signalling the need for a mixed methods appraisal approach and quality assessment of included studies was undertaken using the Mixed Methods Appraisal Tool (MMAT).¹⁷ We used the criteria corresponding to qualitative, quantitative descriptive and mixed methods designs based on the studies selected for review. Scores were expressed as a percentage of possible items divided by affirmative items. Each study was then classified as weak (≤50%), moderate–weak (51%–65%), moderate–strong (66%–79%) or strong (≥80%) based on a methodological scoring system.¹⁸ Quality was individually assessed by reviewers with 10% checked for accuracy. Any disagreements were resolved by discussion among reviewers. No records were excluded based on quality alone, but the quality assessment was used to inform the narrative synthesis of included studies.

2.7 Data analysis and synthesis

Due to the heterogeneity of included study designs and outcome measures, a narrative synthesis was used to synthesize data, which was guided by the Economic and Social Research Council (ESRC) guidance on the conduct of narrative synthesis.¹⁹ This was undertaken collaboratively between authors (L. R., H. B. and R. P.) and the resultant presentation of results was discussed among the wider study team. We used thematic analysis to map data to our three research questions deductively, arranging evidence with similar lines of evidence to ensure reliability. We then conducted analysis inductively to systematically generate theory about wellbeing and mental health in response to each question ensuring that each piece of information was relevant to the synthesis. Quantitative and qualitative data were synthesized simultaneously, and we used textual description, grouping and tabulation methods for preliminary synthesis and to explore patterns across studies. Included studies were tabulated in terms of study characteristics and extracted data. Initial inductive coding was undertaken at the point of extraction to characterize data in relation to descriptive categories. Differences in identified categories in relation to the country of origin, age and gender of included participants and other relevant contextual information were considered next. Finally, the draft synthesis was considered in light of the quality appraisal results (see below).

3.1 Study characteristics

A PRISMA study flow diagram in Figure 1 describes how studies were selected for inclusion in the review. A total of 20 papers describing 19 studies were included comprising 11 qualitative studies, 4 quantitative studies, primarily survey designs and 5 mixed methods studies. There were no experimental studies and one quasi-experimental study that reported qualitative data relevant to our review question among the selected studies. A total of 22,257 children and adolescents were included across studies though most individuals (n = 17,854) came from one study that analysed the network structure of adolescent well-being using a psychometrically tested tool in a representative sample across multiple sites across China.²⁰ Quantitative evidence came from validation studies assessing the equivalence of a Swedish measure of mental health²¹ and novel psychological wellbeing scale development and validation.²² Sharma et al.²³ used a rudimentary adaptation of the mental health literacy scale to evaluate help-seeking preferences and priorities among Indian adolescents and Davids et al.²⁴ used the health-promoting lifestyle questionnaire to examine mental health behaviour and psychological wellbeing. There were no studies published in the least developed nations, 9 reported research conducted in upper middle-income countries and 11 were conducted in lower middle-income countries as per OECD classification. There were insufficient data to evaluate conceptualisations across WHO regions.

Participants in this review had a mean age of 16.6 (range: 6–26) and the majority were adolescent and young person populations. Four studies included children under the age of 10 alongside adolescents.^{20, 25-27} Apart from three studies,^28-30 two of which reported on the same sample, all studies drew their sample from school-going populations. Two studies were non-English (Spanish and Portuguese) and data in these studies were extracted by affiliated researchers. Study characteristics are detailed in Table 2.

3.2 Quality appraisal

The quality of studies was varied and scores on the MMAT ranged from 0% to 100%. All but one of the qualitative studies achieved a score of 80%–100%, indicating these studies were of high quality. Fewer quantitative studies achieved high scores within their category of assessment with half showing a risk of measurement and response bias. The quality of mixed methods studies was also varied and few of the mixed methods designs addressed issues of divergence between the findings from quantitative and qualitative methodologies, nor did they adequately identify the explanatory or exploratory nature of the chosen design. We extracted qualitative data from mixed methods studies; with the exception of one study [22] none of the available studies provided quantitative data relating to the research questions. The few cross-sectional studies that were included were examined for their contribution to the synthesis to determine the weight of evidence from these sources. Quality appraisals are detailed in Table 3.

To evaluate the robustness of our synthesis, we examined the construction of meaning for concepts within the analysis contained in each of our research questions when (a) studies with the lowest quality score were removed and (b) studies with both moderate and low scores were removed. We then assessed the contribution of each individual piece of evidence to the consistency of descriptions within our synthesis to examine whether different pieces of information were compatible with the overall synthesis. We removed irregular pieces of information that were not congruous with our interpretations. Removing the results from low-quality studies (n = 3) did not influence the synthesis and the removal of both weak and moderately weak quality studies (n = 6) demonstrated some bearing on the quantity of evidence that supported the synthesis. As such the synthesis findings were drawn mainly from interpreting and integrating the findings from well-conducted and reported qualitative studies. These are presented in Table 4.

3.3 Conceptualisation of mental health and wellbeing

Ten studies evaluated adolescents' views about what constitutes positive mental health, psychological wellbeing or health, social and emotional wellbeing^{20-22, 24, 25, 31-35} and explored factors that influenced the relationships and construction of the meaning of wellbeing.^{25, 31}

A range of factors was adduced by young people to explain what it means to have mental health and wellbeing. Optimism was described as a consistent marker of psychological wellbeing among qualitative and quantitative studies encompassing themes of positive thinking, cheerfulness, happiness and good mood. Evidence obtained from several transparent and systematic qualitative analyses showed that having a positive outlook, positive thoughts and emotions,^33-35 cheerfulness and optimism³³ were central to young people's notions of mental health and wellbeing. Quantitative evidence provided firmer evidence that having a positive outlook was prioritized over other features of mental health perceived by adolescents. Data from 17,854 randomly selected Chinese primary and secondary level students aged 6–18 provided longitudinal evidence that optimism is a central aspect of wellbeing for young people.²⁰ Similarly, data from a sizeable Brazilian sample also ranked cheerfulness high on the importance of mental health constituents.²¹ Qualitative analyses from individual studies explicated the varying facets of personal situations through which demonstrating positivity pervaded life situations. Keeping the focus on the positive aspects of situations and of other people,³³ positive interactions with others and having harmonious personal relationships are principal to concepts of well-being.^{21, 31, 33} Optimism and positivity were also considered to mean how one relates to oneself, having a good attitude towards oneself³³ and self-acceptance²² were viewed as crucial for emotional wellbeing.

Older adolescents included in qualitative studies to assess views about what constitutes mental health in Indonesia, Kenya, Ghana and Thailand viewed daily functioning and performing expected roles inherent to definitions of mental health. Ably performing daily activities and tasks as a fundamental condition³¹ and coping with everyday challenges^{31, 33-35} were further emphasized as key indicators of being mentally healthy. Quantitative data supported this viewing functioning in a narrower sense, confined to academic performance. In this latter study, mental health concepts were obtained from 1168 Brazilian adolescents. These were sampled purposively from different socioeconomic sectors, using an adapted measure and ranking potential components of well-being by importance.²¹

Self-agency emerged as a pertinent constituent of mental health and wellbeing, comprising multiple facets, including personal attributes and competency for making decisions, having a central purpose to guide life decisions and being given the responsibility to make one's own decisions. Evidence drawn from both qualitative and quantitative studies illustrated that having the necessary personal attributes to make one's own choices leading to competent decision-making and being granted the freedom to do so by others signified the key elements of autonomous behaviour. Quantitative survey data from 1635 Mexican adolescents sampled purposively showed both future plans and having a purpose in life were key factors characterizing psychological wellbeing.²² Quantitative data from South African students randomly sampled from rural areas showed that intrinsic life goals were significantly positively correlated with psychological wellbeing.²⁴ Having the freedom and ability to make one's own choices about own actions and enterprise^{22, 31, 35} were highly valued among respondents in a large cohort of adolescents in Brazil. Survey data demonstrated that having personal control of life choices²¹ is an essential aspect of mental health explicated by in-depth qualitative interviews extending control to having command of one's problems, emotions, stress and personal limits.³⁵ Personal attributes for self-agency included maintaining balance for good decision-making linked with well-being,³⁴ being principled and showing moral awareness, behaving normally²¹ and being ‘sober-minded’ in judgements.³⁴

There were consistent findings that mental health was a significant concern and demonstrated clearly in qualitative analyses among samples of Indonesian, Vietnamese and Chinese adolescents^{30, 32, 35} and Indian adolescents advocating the need for school-based mental health services.³⁶ This was corroborated by quantitative data from Brazilian adolescents with the majority espousing the importance of mental health.²¹ Conversely, one study demonstrated adolescents in Kenya recognized that understanding of positive mental health was limited.³⁴ Evidence from qualitative analyses consistently indicated that adolescent perspectives of mental health were often conflated with negative emotions and mental ill-health and adolescents sometimes used others with mental illnesses as a frame of reference to describe their own views and perspectives of health states.^{25, 30, 32, 35}

There was also some discrepancy about whether adolescents believed that to attain wellbeing, mental illnesses must be absent. For example, qualitative inquiry of Kenyan adolescents views of mental health and illness showed they believed one could possess attributes of good mental health while having a diagnosed mental illness.³⁴ Quantitative data indicated the opposite and adolescents indicated that reliance on mental health professionals was contrary to attaining good mental health.²¹

3.4 Risk and protective factors for mental health and wellbeing

Further to conceptualisations of positive mental health, 12 studies investigated factors that lead to either positive or negative impacts on psychological, emotional wellbeing or mental health and protective factors for maintaining wellbeing.^{25-29, 31, 33-38} Individual, school and community level factors were identified from the perspective of CYP in included studies.

The principal source of stress identified from qualitative analyses related mainly to interpersonal relations with close family and wider social networks. Strictness, teasing, arguments and quarrelling with family, friends and teachers were cited as a major source of stress and threat to wellbeing for adolescents.^{31, 36} Fighting with friends was a significant pressure identified in several reports in Ghana and India.^{28, 29, 31, 36} In terms of peer relations, a number of studies found that perceived mistreatment, including being teased, insulted, ignored, rejected or betrayed, negatively affected psychological wellbeing^{25, 26, 31} (Nastasi & Borja, 2015). More extreme concerns within wider peer networks were experiences of being isolated, excluded and bullied or even beaten and were perceived as negative impacts in both India and Kenya.^{25, 28}

Family conflict and disagreements, particularly about education and romantic relationships were considerable wellbeing stressors.^{32, 36} Pleasure-seeking was an identified risk factor among these children and adolescents, specifically becoming over-involved in hedonistic pursuits could lead to delinquent behaviours like following media personalities, gaming, internet use and cigarette smoking. Young people also reported awareness of generally increasing negative emotions and distortions in their sense of self during adolescent years as a specific threat to their wellbeing.³⁴

Academic pressure is viewed as a consistent and considerable source of stress among adolescents in qualitative analyses. With the exception of younger groups,²⁸ students overwhelmingly reported academic overload, especially with homework, being unable to complete work and projects on time and exam pressure.^{26, 29, 34-36} Anxiety about securing a job once education was completed was particularly salient among older adolescents.³⁶ Students also reported strictness, especially among teachers, reportedly fearing teachers' reactions to them, the effect on their confidence, ability to express themselves and ill-treatment negatively affecting their reputation among peers.^{31, 35} Unfairness and cruelty were words associated with how teachers treated students.³⁵

Gender differences were evident in qualitative data reporting specific factors affecting females wellbeing. There were relatively fewer data on males views and perspectives. One study constructively illuminated the complexities of specific socioeconomic environments but the analysis was limited to 37 females in one setting in India without comparison with males.28 A second qualitative analysis of 15 Indian adolescents corroborated that parents mistreatment of adolescent girls, in contrast to males, left them feeling less valued³⁷ and they perceived a lack of fairness and equality that was significant for emotional wellbeing. Females, in Adelson et al.'s^{28, 29} study, faced higher expectations, having to assume caring responsibilities, such as looking after a sibling while parents were at work, in addition to schoolwork. Particularly after menarche, female freedom was restricted; being told how to behave by parents, siblings and especially older brothers²⁸ and this formed a significant source of stress impacting wellbeing. Females feared being threatened, grabbed or even sexually assaulted on the streets.²⁹ This impacted their family relationships and they experienced significant fear, anger and embarrassment as a result.²⁸ Specifically, they would be blamed or held responsible by others, especially the men in their families, if they were victims of such an attack in public.

Fewer studies addressed the question about which factors protect children and adolescents mental health and wellbeing. While interpersonal difficulties comprised the principal threat to wellbeing among selected studies, the support received through these relationships was reported to safeguard against developing emotional problems in several qualitative studies.^{29, 33, 35} Relationships with peers and family, particularly parents, were viewed as important; children and adolescents perceived promoting and maintaining social wellbeing as strongly linked with education from peers, socializing and having positive peer and supportive relationships with family, particularly parents.²⁵ Effective coping during stressful times is not only closely linked with receiving positive support from others but also the quality of relationships becomes important. Family relationships were considered especially supportive if close family relations were both understanding and open-minded³⁵ and displayed warmth.³³ Children and adolescents identified encouragement, advice, religious and spiritual support administered by close family and peers as key to managing stress.³¹ One study described how deficiencies in familial and close interpersonal relationships lead to the loneliness that contributed to poor mental health³⁸ irrespective of the number of contacts in one's network.

Adolescents believed internal factors could potentially buffer against the ill effects of negative factors affecting mental health, such as having a spiritual mindset, feeling of being empowered or having self-esteem.^{25, 36} Personal attributes were also important for maintaining wellbeing, such as self-acceptance,²² being comfortable with oneself³⁴ and taking pride in oneself.³³ Overall, protective factors were understudied in the studies selected for this review and views about the origin and nature of psychological attributes were unexplored. Few wider societal factors were considered by adolescents to either improve or diminish one's wellbeing. Qualitative analyses in Indonesia, India and Mexico and quantitative estimates of stressors among 68 adolescents in Mexico found negative wider societal impacts were prominent and adolescents reported violence and crime were aspects significantly impinging mental health.³⁵ Antisocial behaviour is a specific concern threatening adolescents safety, including theft and gun crime.^{26, 29}

3.5 Self-help strategies

Reported self-help interventions incorporated both individual strategies and leveraging support from lay networks. One study looked at the relationship between such strategies and mental health outcomes and found that mental health behaviour was not a significant predictor of positive or negative affect.²⁴ The majority described self-help strategies and nonprofessional support sources alongside beliefs about the effectiveness of these.

Several self-help strategies were considered important to managing positive mental health and promoting wellbeing; the most identified strategy was using distraction from stressful situations by engaging in valued activities. Activities identified included music,^{26, 29, 33, 38} exercise and sports,^{23, 33} watching television or reading a book,^{26, 27, 38} interacting with pets,²⁶ prayer, religious worship or spiritual practices,^{24, 33, 35, 39} housework/cooking,²⁹ gardening,³³ meditation,^{23, 37} writing³³ and going online.³⁸ Overall, evidence to support the effectiveness of any individual strategy or comparative evaluations of different strategies were lacking and the evidence to support passive activities and recreation to enhance wellbeing is weak.

Children and adolescents in qualitative analyses identified the following groups as important sources of social support: parents^{27, 29, 31, 33} mothers in particular,^{29, 31} family members,^{26, 28, 29} peers,^{26-29, 31, 33} siblings^{27, 33} teachers,^{26, 27, 33} church leaders³⁹ and neighbours/community members.^{26, 29} Friends were considered particularly useful through their role in the facilitation of distraction from stressful situations by encouraging the valued activity described above.²⁷ However, all identified sources of support were also considered to be potential sources of stress (see Section 3.4) highlighting the complexity of relationships with lay support networks for this group.²⁶

In reporting preferences for self-help strategies, differences relating to gender and age were limited in included studies. No differences were found between genders in stress response and self-help choices to promote wellbeing in a qualitative analysis.²⁷ This was supported by findings in another study comparing mental health management behaviours between genders using quantitative data relating to 243 South African adolescents.²⁴ Two qualitative analyses provide evidence that for females, greater importance is placed on maternal relationships and the support and advice received through these bonds as well as having close female confidantes among other kin relations.^{29, 31} A separate study comprising 68 Mexican students found females had a higher number of supportive relationships than males but not a significantly higher proportion, indicating the size of males' wider supportive networks to be smaller when compared with female counterparts.²⁶ Younger children were identified as having fewer self-help strategies and older children were more likely to describe aggressive physical and verbal reactions as effective ways of coping with stress²⁷ but investigation of any age differences that may exist were limited to this one study.

4.1 Strengths and limitations

Children's and adolescents' understanding of how to obtain and maintain good mental health is understudied in both low- and high-income settings^{47, 48} and there are significant gaps in the literature to contrast in different contexts. We did not exclude any study based on the quality, which may affect our interpretations; however, we employed robust systematic review methods, ensuring the integrity of the synthesis. We did not use an exhaustive list of search terms related to the phenomenon of mental health and wellbeing though we recognize that terms are used interchangeably in some circumstances and lack clarity and shared understanding such that research in this area can be difficult to synthesize. Our broad approach taken in our search strategy mitigates this to some degree. Nonetheless, the synthesis is dependent on existing evidence and as stated, there are significant gaps in research in these areas across LMICs. A further gap exists across demographic groups within the country. All but two studies researched school-going populations and while these are important sites for delivering effective health promotion interventions,^{9, 10, 14} sizeable and varying portions of children and adolescent populations in low resource settings are absent from school or attend sporadically, thus this marginalized group are underrepresented in empirical evidence.