Informing the measurement of wellbeing among young people living with HIV in sub-Saharan Africa for policy evaluations: a mixed-methods systematic review

Young people living with HIV (YPLHIV) in sub-Saharan Africa (SSA) are at high risk of having a poor quality of life. Addressing wellbeing explicitly within HIV/AIDS policies could assist mitigation efforts. However, guidance on wellbeing measures to evaluate policies for YPLHIV is scarce. The aims of this mixed-methods review were to identify: i) key dimensions of wellbeing and ii) wellbeing measures that align to these dimensions among YPLHIV (15–24 years) in SSA. We searched six social science and medical databases, including grey literature. We included studies that examined correlates and lived experiences of wellbeing, among YPLHIV in SSA, from January 2000 to May 2019. Two reviewers independently screened abstracts and full texts and assessed methodological quality of included articles. We analysed quantitative and qualitative data using descriptive and meta-ethnographic approaches, respectively. Thereafter, we integrated findings using a framework approach. We identified 6527 citations. Of these, 10 quantitative and 30 qualitative studies were included. Being male, higher educational status, less stigma and more social support were likely correlates of wellbeing. Themes that shaped experiences suggestive of wellbeing were: 1) acceptance and belonging— stigma, social support; 2) coping; 3) standard of living. Our final synthesis found that the following dimensions potentially characterise wellbeing: self-acceptance, belonging, autonomy; positive relations, environmental mastery, purpose in life. Wellbeing for YPLHIV is multi-dimensional and relational. Relevant measures include the Personal Wellbeing Index, Ryff’s Psychological Wellbeing Scale and Mental Health Continuum Short Form. However, psychometric evaluations of these scales among YPLHIV in SSA are needed.


Introduction
Adolescence and young adulthood are critical periods for wellbeing, an indicator of the quality of life. This life stage is important for building human and social capital, resources that can sustain wellbeing in adulthood [1][2][3]. For young people living with HIV (YPLHIV), who because of greater access to anti-retroviral therapy (ART), can expect to reach adulthood, building and sustaining wellbeing as they grow older is crucial. In sub-Saharan Africa (SSA) there are an estimated 2.8 million YPLHIV aged 15-24 years [4]. Many YPLHIV in SSA are at high risk for HIV-related mortality, morbidity [5] and poor quality of life due to the sub-optimal access to treatment and psycho-social services, including exposure to multiple stigmas [6][7][8][9]. According to the United Nations [10], the number of YPLHIV in Africa is projected to increase by 44% between 2015 and 2030. Models predict that investment in the quality of life among the growing youth population in SSA, particularly via addressing health needs, could increase labour productivity and resultantly accelerate economic growth [11]. If long-run economic growth is to be achieved, then public health policies need to also promote the wellbeing among YPL-HIV, a vulnerable and growing population in this region.
In social psychology, wellbeing denotes a sense of thriving in multiple life domains such as family, career and health [12]. Two constructs of wellbeing exist, subjective wellbeing (SWB) which focuses on people's emotional and cognitive evaluations of their lives [13] and psychological wellbeing (PWB) which emphasise positive functioning, relationships and human potential [14,15]. The wellbeing of YPLHIV in SSA is likely to be compromised given the impact HIV/AIDS has on key dimensions of wellbeing such as finding meaning, selfacceptance and maintenance of positive caregiver and peer relationships [16]. Importantly, emerging empirical evidence has shown that YPLHIV in this setting are at high risk for depression and anxiety [17][18][19] and suicidal ideation [20]. Negative mental health functioning is one of the strongest predictors of wellbeing [21], and if experienced during adolescence may lead to reduced wellbeing in adulthood [2,22], and subsequently unfavourable labour-market outcomes such as low earnings [23][24][25]. The risk of lowered wellbeing among YPL-HIV is of major concern as econometric evidence has shown that the stocks and intergenerational transfer of human capital to young people in this region has already been eroded by the HIV/AIDS epidemic [26,27].
In line with Sustainable Development Goal (SDG) 3 ("Ensure healthy lives and promote well-being for all at all ages"), wellbeing has emerged as a major policy outcome [28,29]. Substantial investment and progress has been made in reducing the HIV epidemic among young people (SDG 3.3) via scale-up of HIV testing and access to treatment [30,31]. Yet, promoting mental health and wellbeing (SDG 3.4) among young people has received very little attention. There is growing recognition for the need to include quality of life as an HIV/AIDS programme target [32,33]. Current multi-sectoral HIV programmes are focused on reducing new HIVinfections among adolescent girls and young women via improving access to health and social services, including economic opportunities [34]. However, monitoring and evaluation frameworks of these programmes do not explicitly include wellbeing as an outcome, a recommended measure in the field of economics for measuring social progress [35]. Valuing only narrow health outcomes in health economic analyses may fail to capture the broader impacts of the multi-sectoral initiatives that go beyond health [36,37]. From a public health financing perspective, understanding the impact of HIV policies on wellbeing could provide a range of ways to direct and strengthen investments for YPLHIV.
Wellbeing measures are used in economic policy evaluations as empirical approximations of individual welfare (i.e. the satisfaction (utility) gained from consuming a good or service) [38]. Of the few health-related economic studies that have examined wellbeing among children and adolescents, most have used uni-dimensional life satisfaction measures [39,40]. However, several multi-dimensional wellbeing measures developed for young people in the Global North exist in the field of social psychology, these include multi-dimensional SWB scales (i.e. Student Life Satisfaction Scale [41], Personal Wellbeing Index [42]), and PWB measures (Ryff Scale of PWB [43], Mental Health Continuum Short-Form [44]). In the field of HIV/AIDS epidemiology, health-related quality of life measures that encompass some wellbeing dimensions such as emotional and social functioning have been recommended [45] and applied among adolescents living with HIV, these include the Paediatric Quality of Life Inventory [46], KIDSCREEN [47], WHO Quality of Life [48]. However, there is a lack of guidance on which wellbeing measures, underpinned by wellbeing theory, are appropriate for health economic surveys seeking to evaluate the impact of HIV policies on the wellbeing of YPLHIV.
Wellbeing may be conceptualised, experienced, and valued differently across cultures. A better understanding of the correlates and local experiences of wellbeing among YPLHIV in SSA and how these relate to international conceptualisations of wellbeing is required in the selection of culturally sensitive, yet generalisable, measures of wellbeing for SSA. While several studies have explored wellbeing among young people in this sub-region [49,50], to the best of our knowledge, no study has systematically reviewed the evidence on what constitutes wellbeing among YPLHIV on the continent.
The objectives of this mixed-methods review are to identify and critically assess: 1) quantitative evidence on the correlates of wellbeing among YPLHIV; 2) qualitative evidence on the lived experiences of wellbeing among YPLHIV; 3) integrate these two strands of evidence and identify appropriate measures of wellbeing based on key dimensions of wellbeing relevant to YPLHIV.
The main reason for using quantitative and qualitative evidence in this review is to provide a broad perspective in order to gain a detailed and thorough understanding of the indicators of wellbeing and potential mechanisms by which factors influence wellbeing [51][52][53]. The findings from this review could guide the selection of wellbeing measures for policy evaluations focused on YPLHIV.

Methods
The quantitative review was prepared in accordance with Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) checklist [54] (Additional file 1). Whereas the qualitative review adhered to the Enhancing transparency in reporting the synthesis of qualitative research (ENTREQ) statement [55] (Additional file 2). Reporting of this review was informed by mixedmethod review guidelines [56].

Epistemological perspective and study design
We adopted a relational wellbeing perspective as increasing research from the fields of social psychology, anthropology and international development have highlighted that relationships are central to the experiences of wellbeing in low-and middle-income countries, particularly studies from SSA [57][58][59]. We drew on the relationality meaning model as proposed by Wissing [60] as relationships and connections have been shown to be central to formation of wellbeing among young people [61][62][63][64][65]. According to the relationality meaning model, relationships, with self (intrapersonal) and others (interpersonal) are at the centre of a meaningful life, and play an important role in the connections between people and context (i.e. social, cultural, ecological, physical, spiritual) [60]. This model has a strong PWB orientation and emphasises three key elements: 1. meaning of life (experienced in the connectedness to something larger than life and the realisation of values); 2. meaning in life (experienced in the belongingness and relatedness to other people); and 3. meaning to life (expressed via actions such as expressions of affection, longing for belongingness, building relationships). This model hypothesises that meaning of life facilitates the construction of meaning in life which in turn directs activities to provide meaning to life [60]. We specifically applied elements of this model to understand the pathways to wellbeing in order to elucidate dimensions that are relevant in capturing the wellbeing among YPLHIV.
The mixed-methods approach used was a segregated (convergent) design [66]. We selected this design because we regarded quantitative and qualitative research as complementary. However, we conducted the quantitative and qualitative review separately as different approaches were used and the review criteria differed. For the quantitative review, we followed systematic review methods used in evidenced-based medicine [67]. Whereas for the qualitative review, we used a flexible meta-ethnography design [68]. Quantitative and qualitative studies were retrieved, analysed and synthesised separately, prior to the final synthesis [66].

Eligibility
The criteria for inclusion of studies are summarised in Table 1.

Studies
We included published and non-published quantitative and qualitative studies. We selected observational quantitative studies reporting on primary or secondary data analysis from cohort or cross-sectional datasets. We deemed any qualitative study design, data collection technique (e.g. group interviews, in-depth interviews, participant observations) or analytical approach (e.g. thematic analysis, framework analysis) eligible for the qualitative arm of this review. Furthermore, we included mixed-method studies that satisfied corresponding inclusion and exclusion criteria, separately determining eligibility for quantitative and qualitative components. In addition, we scanned the bibliographies of relevant quantitative or qualitative reviews to identify potentially eligible primary studies not yielded by the primary search.

Participants
We selected studies focused on young people, defined as older adolescents and young adults between the ages of 15-24 years, who were living with HIV [69]. We focused on this condition only as HIV/AIDS remains the leading cause of mortality among young people in this region [70,71]. We included studies if at least the average age of the sample was within our age-range or results could be extractable for the age-range.

Outcome measures
Drawing from the field of social psychology, we conceptualised wellbeing as subjective and psychological. Thus, our outcome measures included any dimension of these two constructs: 1) subjective-positive and negative affect (mood states, emotions), life satisfaction [13]; 2) psychological-self-acceptance, positive relations, environmental mastery, autonomy, purpose in life, personal growth [43,72]. Given the strong correlation between mental health and wellbeing [21,72,73], we included mental health as one of our outcome measures. We applied a biomedical definition of mental health (i.e. the absence or presence of symptoms of mental illness) [74,75].
We included: quantitative studies that assessed factors associated with any dimension of wellbeing or mental health using statistical regression techniques; qualitative studies that explored lived experiences of wellbeing or experiences related to dimensions of wellbeing or mental health.

Study setting
We included studies conducted in any setting (e.g. household, clinic, school) in a country within the sub-Saharan African region, as per the World Bank country classification [76]. We chose to focus on SSA only as this region accounts for the highest number of YPLHIV [77]. Furthermore, young people living in this region have the highest risk of HIV acquisition [78].

Time and language
We restricted the search to studies conducted between 01 January 2000 and 11 May 2019. This timeline covers the key post-ART periods in SSA: 1) ART introduction (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007); 2) expanded ART (2008-2010); and 3) scaled-up ART (2011-2019) [79]. We considered the post-ART period more relevant to the current international HIV policy landscape as treatment for all HIVpositive individuals is now the recommended approach [80]. No language restrictions were placed on the search.

Search strategy
We searched electronic databases using either compound search strategies containing subject headings that were supplemented with text terms (Additional files 3, 4, 5 and 6) and JEL codes (Additional file 5), or simple Boolean logic search strategies with keywords (Additional files 7, 8, 9 and 10). Our search terms were aligned to dimensions of the subjective and psychological wellbeing constructs (e.g. negative and positive affect, life satisfaction, self-acceptance, social relationships) [13,72], with a specific focus on relational dimensions. In addition, our search included terms related to the attitudes and symptoms of common mental disorders (i.e. depression, anxiety) among YPLHIV [81].
Study records-data management, selection process, data collection process We imported search outputs into an EndNote (X 8) library and removed duplicate references. Thereafter, we imported this library into an online systematic review manager (Covidence systematic review software, Veritas Health Innovation, Australia). DG and IDA dual screened titles and abstracts independently, and 10% of these were dual screened by GF. AW assisted DG with the dual screening of grey literature. DG subsequently resolved conflicts via discussion with the dual screener, obtained the full text for all potentially eligible abstracts and applied the inclusion criteria to these studies. GF, JS, CM randomly checked all full texts. Less than 5% of abstracts were non-English language studies. For these specific studies, we obtained a translated electronic English version of the study.

Data items
Once consensus was reached on eligible studies, DG entered relevant data from potentially eligible studies onto electronic data extraction forms. The design of these forms were informed by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist [82] and Consolidated criteria for Reporting Qualitative Research (COREQ) checklist [83], for quantitative and qualitative studies, respectively.
We extracted information on the following variables from each included study: author, year published, study setting, country income classification [76], design, outcome definition, analytical techniques, participant demographics, treatment status, disclosure status. For quantitative studies, we specifically extracted data from regression models (e.g. sample size, measures of associations, confidence intervals and p-values). For qualitative studies, we extracted primary themes or first order constructs (i.e. participants' understanding as reported in the study via verbatim quotes or authors description), and secondary themes or second order constructs (i.e. authors' interpretations of participants' understandings). Co-authors randomly checked data extractions.
Outcomes a) Quantitative: explanatory variables associated with wellbeing or mental health, and wellbeing and mental health measures. b) Qualitative: meanings and manifestations of wellbeing, any dimension of wellbeing or mental health

Quality appraisal
All potentially eligible studies were critically appraised by assessing their methodological quality. For quantitative studies, we applied a risk of bias tool, adapted from the Cochrane guidance on assessing risk of bias in non-randomised studies [67]. DG and GF selected items for this tool based on their relevance to observational study designs in epidemiology and psychology. Studies were categorised into three groups, depending on the level of bias: low, medium or high risk of bias. This was assessed by evaluating measures applied to reduce the following biases: 1) selection biasrandom sampling techniques; 2) information biastraining of interviewers in the administration of the scale, translation of the scale to local language, assessing the validity and reliability of scale for given population; 3) confoundingadjustments for potential confounders. For qualitative studies, we applied a quality assessment tool adapted from the COREQ checklist [83] and key studies [84,85]. DG and JS selected the following indicators for this tool based on its relevance to qualitative research design 1) rigoruse of appropriate theoretical frameworks, sufficient data collected; 2) sincerityself-reflexivity, transparency about the methods and challenges; 3) credibility triangulation of data, inductive nature of derived themes.

Data analysis-synthesis
We conducted a simple descriptive analysis of key quantitative findings. We categorised studies by country income classification and SSA sub-region as the quality of life among YPLHIV may vary depending on the country's healthcare resources and HIV-epidemic pattern specific to that sub-region. For the qualitative studies, we applied meta-ethnographic analytical methods [86], as implemented by Atkins, Lewin [87]. We specifically used the reciprocal translation analytical approach to develop themes. This approach entailed analysing and synthesising participant views (first-order construct) and authors' interpretation (second order constructs) to develop third-order constructs. We implemented this approach by comparing first and second order constructs across studies that were homogeneous in terms of design elements (setting, population and period of ART rollout) and chronologically ordered studies based on publication date [87]. Subsequently, we matched themes across papers ensuring the third order captured similar themes from various studies. We then tabulated translations by highlighting key third order constructs derived and supporting quotes and narratives. Finally, we interpreted the themes across studies to develop a line-of-argument synthesis describing how all themes interacted to shape wellbeing. We evaluated the quality of the qualitative evidence synthesis using the GRADE-CERQual (Confidence in Evidence from Reviews of Qualitative research) approach [88]. This approach includes an assessment of the methodological limitations, coherence, relevance, and adequacy. Lastly, we integrated the main quantitative and qualitative findings by mapping key correlates and themes that emerged from the data to subjective and psychological wellbeing dimensions [13,72]. Thereafter, we interpreted our data drawing on the relationality meaning model [60].

Screening protocol
The electronic database search yielded a total of 7563 citations, and our grey literature and manual search yielded 771 citations ( Fig. 1). After removal of duplicate entries, 6527 studies were evaluated using title and abstracts; 5909 citations were excluded as these were based on populations that were not in line with the reviews' study population (i.e. caregivers, older adults, elderly, pregnant women) or outcome (i.e. drug effectiveness, access to education or sexual reproductive health services). The remaining 618 potentially eligible studies were retrieved for full-text assessment (356 quantitative review, 262 qualitative review). Of the 356 citations identified for the quantitative review, 10 studies were included, and 346 studies excluded mainly due to the lack of agestratified analyses or focus on a biomedical outcome measure. Of the 262 citations identified for the qualitative review, 30 studies were included, and 232 studies were excluded mainly because of: 1. a narrow focus on barriers to disclosure or adherence; 2. examined outcomes not related to subjective or psychological wellbeing dimensions (e.g. feasibility and acceptability of a biomedical intervention or health service programme); 3. included adolescents or young adults in the sample but did not provide quotes from participants that spanned the age-range 15-24 years.
Overall study characteristics Tables 2 and 3 summarize the main characteristics of the 40 studies (10 quantitative and 30 qualitative). These studies were equally representative of the eastern and southern sub-regions of SSA, with nearly 50% conducted in low-income country settings (n = 19). The majority were conducted during the scaled-up ART phase. Participants were mainly between 15 to 19 years of age and sampled from urban public healthcare facilities. None of the included studies defined wellbeing in terms of subjective or psychological wellbeing as defined by Diener and Ryan [89] and Ryff and Keyes [72], respectively. However, these studies examined dimensions related to these constructs such as mental health, relationships and self-acceptance.

i. Demographics
Being female was strongly correlated with poor mental health functioning in four studies (note, the outcome definition was different in each of the four studies) [92,93,96,99]. The largest gender effect was observed in Malawi, with females almost eight times as likely compared to males to exhibit higher depression-related symptoms scores (β = 2.13, 95% CI [0. 82-3.43], p = 0.002) [93]. In two urban studies from low income settings, there was between an 8 to 23% increase in depression scores with age [93,99].
ii. Standard of living Educational attainment was found to have a protective effect on mental health functioning in three low income countries [93,95,99]. In the Ugandan study, those with secondary school attainment were five times more likely to have better mental health than those with no education (adjusted odd ratio (aOR) = 5.3, 95% CI [1. 86-15.41], p < 0.00) [95].
iii. Psycho-social HIV-related stigma was strongly positively associated with poor mental health functioning in six studies [91,92,94,96,98,99]. Among these studies, the largest effect of stigma was documented in Zambia, which found that the odds of having higher depressive symptom scores was almost three times higher for YPLHIV who experienced stigma than in those who did not (aOR = 2.99; 95% CI [1.07-8.41], p = 0.01) [94]. Having someone to talk to or feeling satisfied with health services or the social support received promoted positive mental health functioning in four studies (Namibia (n = 1) [92], Uganda (n = 2) [95,96], Zambia (n = 1) [94]. In an Ethiopian study, Abebe, Shumet [98] reported that those with low social support were 2.74 times more likely to develop depressive symptoms than those with strong social support (95% CI [1.42-5.27], p = < 0.01). Moreover, poor adherence was positively correlated with depressive symptoms in three studies [97][98][99].

Quality of studies
Five of the ten quantitative studies were judged as having low quality as findings were subject to a high risk of bias [90-92, 94, 99] (Table 5). The main quality concerns in these studies were the lack of reporting on psychometric properties of the scale and standardisation of scale administration, including the inadequate reporting and interpretation of statistical analyses. Only one study psychometrically evaluated the chosen scale on a similar study population and found it to have good validity and reliability [93]. None of the studies reported information on the cultural validity of the chosen scale/s. All studies were subject to selection bias as samples were drawn from healthcare facilities, primarily using non-random sampling techniques.

Qualitative studies-characteristics and data synthesis
Of the 30 qualitative studies (Table 3), most were undertaken in the eastern sub-region (N = 14): Uganda (n = 11), Kenya (n = 2), Tanzania (n = 1) and the southern sub-region (N = 14) (South Africa (n = 9), Namibia (n = 1), Swaziland (n = 1), Zambia (n = 2), Zimbabwe (n = 1), with one study conducted in both regions (Tanzania and Botswana). The remaining study was conducted in West Africa (Nigeria n = 1). Only six studies focused on a specific gender (females n = 5, males n = 1). Whilst none of the 30 studies specifically aimed to examine lived experiences of subjective or psychological wellbeing they did, however, explore experiences related to dimensions of PWB or mental health. The aim of most of the studies was to understand the psycho-social challenges experienced among YPLHIV. Five studies examined broader life experiences associated with wellbeing using ethnographic methods. Data were collected mainly via focusgroup discussions and in-depth interviews. Several studies included perspectives from caregivers (n = 9) and healthcare workers (HCWs) (n = 7), with only one study that included perspectives from educators.

Specific analytical themes
Three key themes emerged across all studies: 1) acceptance and belonging, 2) coping; 3) standard of living (Table 6, Fig. 2). These themes shaped experiences suggestive of wellbeing as detailed below.

i. Acceptance and belonging a) HIV-related stigma and discrimination
All the studies highlighted the role HIV-related stigma and discrimination played in reinforcing social isolation and poor mental health functioning among YPLHIV. Perceived contributors included forms of internalised stigma, characterised as fear of the rejection and isolation if HIV status was accidentally disclosed to peers [100], partners [101][102][103][104] or parents [103,105,106]. Hence, keeping one's HIV status undisclosed was often described as an everyday lived experience [107][108][109][110]. Studies described how internalised stigma impeded ability to build "healthy identities" [104,108,111] and access social support [101]. Gendered social expectations appeared to facilitate internalised stigma and consequently limited disclosure. For example, in Uganda, a young woman living with HIV feared loss of social value if her intimate relationship was curtailed and hence opted for non-disclosure of her HIV status [101]. Similarly, young  -Non-disclosure to parents-fear of loss of rights and entitlements ▪ "For me my father is alive and I am the heir, but if he knows that I am positive he might remove the heirship from me thinking that I will die before him. I must first weigh the possible outcome of disclosing and to whom."   ▪ "At home they help me with everything and give me all the support I need. It helps me get through knowing they love me. We take our pills the same times so we always remind each other. When she takes hers, she calls me to take mine" (16 year-old female, South Africa) (Jena, 2014) ▪ "I believed that when one is positive he/she can die any time. I also felt am worthless in this world. Later my brothers came and assured me that there was no need of worrying much because they were there for me. They told me that they will take care of me." -Unsupportive family networks-impact on coping, selfacceptance, social-acceptance ▪ "My auntie told me that I do not belong to the family, because of my condition and I was always segregated from other family ▪ "He did not see eye to eye with his sister-in-law who did not like the fact that Mpendulo was HIV positive...in one incidence the sisterin-law found him eating food from a plate that was not designated for him. She scolded the boy for using the plate; stating that...he will infect her children with HIV. That angered Mpendulo a lot. He said he felt unwelcomed and not wanted. -Multiple losses and complicated grieving-longing for relationships they never got to experience -Longing for fathers-shaped by deep cultural expectations ▪ "It's that every child wants to have a dad and a mother at the same time … growing up having a dad and a mother because it's really sad seeing some of my friends having their families and telling me they went out with their dads, then I knew that I didn't have a dad … so many children do suffer from that thing because  Table 6 Key themes that shaped experiences suggestive of wellbeing among YPLHIV in SSA, as derived from the qualitative meta-synthesis ▪ "It is because (crying) I sometimes get short of the money. .. it is sometimes so difficult for me to come and collect her medication because of the lack of money. .. I am unable to buy the right food for her because she has a special diet since she is sick" [Caregiver, men living with HIV in Uganda felt unable to uphold masculine socio-cultural identities and thus chose not to disclose their HIV status to their parents in fear of the "loss of rights" to endowments [105,106]. Most studies described the manifestations of HIV-related stigma in terms of experienced stigma, which was encountered in relationships with mothers [112,113], family members [101,114], learners and educators (102Mutumba et al., 2015,107,115), and HCWs [115] [113]. Authors also described how stigma intersected with gendered attitudes towards sexuality and self-acceptance among young women [101,[115][116][117], including cultural norms such as respect for elders [113,118,119].

b) Social support
The importance of relationships and social connections was expressed in several studies. Shabalala, De Lannoy [114] notes that for YPLHIV in eSwatini (formerly called Swaziland) strong relationships with caregivers were linked with "being accepted, being connected and welcomed". Mutumba, Bauermeister [102] describes how pill taking between HIV-positive parents and YPLHIV in Uganda fostered bonding "whenever she took hers, I also took mine". In contrast, in three studies, young women expressed the lack of support from caregivers [101,112,120]. Trusting relationships with family members were frequently described as important [101,102,109,121,122]. In a Kenyan study, authors indicate that the acceptance received from siblings served to "validate the participant as a human being" [109]. Unsupportive family networks were also described in the literature in the context of poor mental health functioning and reduced wellbeing (i.e. lack of selfacceptance) [101,114].
Furthermore, studies described how YPLHIV drew on their friendship networks to manage treatment adherence, seek advice and feel accepted [104,109,123]. Several participants expressed gratitude for the informational and emotional support received from HCWs [107,109,119,124]. Furthermore, in settings were YPLHIV had access to support groups, feeling "comforted" [103], "normal" [103] "welcomed" [114], "open to share" [121], and "not alone" [111] were dominant in the narratives. Caregivers in a South African study highlighted their lack of social support and how it hindered their ability to cope and subsequently support their children [104]. In three studies, YPLHIV articulated a deep longing for their deceased parents which authors indicated signified the importance parents played in nurturing belonging and its link with creating meaning in life [104,114,117,120]. Moreover, the yearning for one's father was interpreted as being underpinned by "cultural expectations of being cared for and finding a rightful place in their father's home" [114].

ii. Coping
A key positive factor that YPLHIV utilised for coping was religion and faith, especially when support was limited. Religion and faith were primarily expressed as "belief in God" [100,109,111,112,116,121,123,125]. Most authors perceived religion as bringing comfort and hope, as well as meaning and purpose in their lives. In addition, in a Namibian study, interpreting HIV acquisition as a "test or punishments from God" helped a young woman to accept her HIV status [116]. Another major positive coping strategy that YPLHIV applied was future goal setting such as desire for marriage and children [107,119,125,126]. Educational aspirations were perceived as bringing a sense of purpose or social value to participants' lives [107,122]. In contrast, YPLHIV also utilised negative coping strategies such as social withdrawal [107,108,111], self-blame [102,123] and anticipation of death [103,104,117].

iii. Standard of living
Most YPLHIV resided in non-nuclear and skippedgeneration households that had encountered multiple losses. These households were often described by authors as lacking economic security and social protection. In Kenya, Zimbabwe and South Africa, caregivers and community HCWs reported how food insecurity challenged ART adherence and positive mental health functioning among YPLHIV [104,110,111,127]. Young men living with HIV often reported difficulty in establishing economic security as the sole reason for delaying sexual debut, marriage and having children [100,103,126]. In Uganda, two studies noted how young men had "shattered dreams" [112] as household poverty traps prevented young men from completing their schooling, achieving their career aspirations and entering into the formal labour market. A young man's inability to contribute to his household made him feel "unwanted" and "hurt" [114].

Quality of studies
Twenty-seven out of 30 qualitative studies were of medium (n = 19) to high quality (n = 8) ( Table 7). The remaining three studies were of low quality largely due to insufficient information on the methods and participant sample, including lack of description in the analysis. For most studies included in this synthesis, findings were presented clearly with concrete detail and discussed in relation to other literature and theories.
Based on the CERQual assessment, we report moderate confidence in most themes (Table 8).

Overall synthesis
Both our quantitative and qualitative evidence suggest that social networks were at the core of wellbeing for YPLHIV, which supports a relational construct of wellbeing among YPLHIV. Drawing on the relationality meaning model [60], our findings suggest that social relationships were critical in fostering belongingness and connectedness. This in turn contributed to a sense of meaning in and to life, and ultimately wellbeing. In addition, socio-cultural norms and values were important to the wellbeing of YPLHIV, as these helped realise the meaning of and in life. In contrast, our synthesis also revealed that in certain instances, social networks and gender norms embedded within these networks compromised wellbeing. Based on our evidence, dimensions that potentially constitute wellbeing for YPLHIV can be mapped onto the following PWB dimensions: 1) selfacceptanceinternalised stigma, externalised stigma, social acceptance; 2) belongingfamily connectedness; 3) autonomydisclosure, sexual intercourse; 4) positive relationssocial support; 5) environmental masterypositive and negative coping; 6) purpose in lifereligious activities, educational aspirations .

Discussion
Through this mixed method review we sought to identify key dimensions of wellbeing among YPLHIV in SSA by synthesising the evidence on correlates and experiences of wellbeing. The goal of this review was to inform wellbeing measurement selection for health economic surveys that seek to evaluate the impact of HIV policies on wellbeing. Findings from our review indicate that key dimensions which define wellbeing for this population include social support, family belonging, self-acceptance, coping and purpose. These results suggest that multidimensional wellbeing measures with a strong focus on social relationships may be appropriate for this setting.
Economic studies from high income settings have found robust positive associations between social ties (e.g. family, peers, neighbours) and SWB (life satisfaction, happiness) [128]. However, our findings suggest that social networks have both benefits and liabilities for wellbeing among YPLHIV in this setting, which is consistent with recent evidence that highlights the positive and negative relational mechanisms of social networks on SWB [129], particularly among adolescents [130]. Our quantitative synthesis showed a strong positive correlation between HIV-related stigma and poor mental functioning which may likely reduce overall wellbeing as shown in previous studies among adults living with HIV in SSA [131] and high income countries [132]. Drawing on our qualitative review findings together with the relational meaning model [60] and Goffman's accounts of stigma [133], internalised stigma may reduce wellbeing by lowering self-worth. In addition, externalised stigma encountered within social networks may reduce connectedness of self with family members and peers [102,112,134]. Together, these processes are likely to challenge self-acceptance. The effects of stigma on wellbeing are likely be more pronounced for sexually HIV-positive young women, who, due to the intersection of stigma with sexual norms, may harbour heightened feelings of shame and rejection [113,116]. These findings suggest that wellbeing measures which focus on self-acceptance and self-worth may be appropriate for this population.
Our synthesis also highlighted the importance of social relationships in bringing meaning to life for YPLHIV. However, caregivers often limited the decision-making of YPLHIV with regards to disclosure [102] and sexual intercourse [119], which may have reduced the ability of Table 7 Quality appraisal $ of full-text studies included in the qualitative synthesis (N = 30). These studies examined lived experiences related to dimensions of wellbeing or mental health among YPLHIV in SSA $ = The quality appraisal tool was adapted from the COREQ checklist (Tong et al., 2007) and informed by studies in this area (Tracy, 2010;Mays and Pope, 2000).  YPHIV to build relationships and reciprocate affection. In addition, the absence of relationships, particularly with fathers, may have reduced belongingness [114]. Family belonging has been shown to mediate pathways to SWB [135] and PWB [136] among adolescents in developed settings, possibly via promotion of meaning in life [137]. Overall, these findings highlight the relevance of selecting wellbeing measures with dimensions designed to capture family belongingness and autonomy. Our review revealed a link between strong social support networks and positive mental health among YPL-HIV. It is plausible that these strong social networks also promote PWB as shown in a previous adolescent study from Brazil [138]. Accounts from our qualitative synthesis suggest that YPLHIV drew on the emotional support from family, peers and HCWs members during challenging life events (i.e. disclosure, adherence challenges, rejection) which may have helped maintain positive mental health and potentially PWB [102,103,109]. However, in instances where supportive social networks were limited, we found that YPLHIV drew on negative coping mechanisms (e.g. social withdrawal, self-blame) [104], which likely reduced their PWB [139]. Our findings suggest that social support positively impacts PWB by adding meaning to life, specifically on an intrapersonal level such as improving one's sense of environmental mastery (e.g. ART adherence), a dimension of PWB. Our findings suggest that wellbeing measures should also encompass dimensions related to positive relations and environmental mastery.
Our review showed that socio-cultural values, norms and beliefs were critical for creating purpose in life among YPLHIV, which is also considered important for positive youth development [140]. Results from our qualitative synthesis suggest that religious beliefs, values and practices, helped YPLHIV understand the meaning of life, particularly in relation to their HIV status [116], which is agreement with previous adolescent studies [100,125,141,142]. Educational and employment aspirations were commonly reported in our qualitative evidence [107,122,126], and may have helped promote wellbeing by creating a sense of purpose in life [43]. In addition, the desire for economic security among males [103,114,126] also suggests that their wellbeing may be rooted in their ability to fulfil socio-cultural roles such as breadwinners, and thereby add purpose to their lives. These findings suggest that wellbeing measures that consider domains such religion living standards or dimensions such as purpose in life might be valuable for this population.
Overall, our findings support the use of multidimensional relational wellbeing scales aligned to life domains which are important to YPLHIV (i.e. family, religion, education, living standards). Subjective wellbeing scales that consider these domains include the Personal Wellbeing Index [42] and the Student Life Satisfaction Scale [41]. The Personal Wellbeing Index has been applied in South Africa's national social attitudes survey of individuals aged 16 years and older [143]. The Personal Wellbeing Index has exhibited good scale reliability among adolescents in developing countries [144] whereas the Student Life Satisfaction Scale has shown favourable psychometric properties among children in South Africa [145]. Broader wellbeing scales such as Ryff's PWB scale and the Mental Health Continuum Short-Form have shown strong alignment with themes that emerged in our synthesis (i.e. social support, belonging, purpose in life, self-acceptance). These PWB scales have shown good validity among adolescent populations in the North [146]. In addition, the Mental Health Continuum Short-Form has shown strong alignment with conceptualisation of a good life among adolescents in South Africa [65]. Further studies are needed to validate these wellbeing measures among YPLHIV in SSA, paying careful attention to translational issues [147,148]. These proposed scales are relatively brief and can be either self-or interviewer-administered in national economic or HIV surveys, including programme evaluations. In addition, key dimensions of wellbeing in this setting (e.g. positive relations, acceptance, coping) could be used as key output measures in monitoring and evaluation frameworks of multi-sectoral HIV policies and programmes to indirectly assess wellbeing impacts. Our findings also highlight the importance of including social support, stigma, gender and living standards as explanatory variables in econometric models examining the wellbeing effects of HIV/AIDS policies among YPLHIV.
Key strengths of this review include: 1) use of mixedmethods, with qualitative data used to explain patterns in the quantitative synthesis; 2) use of a search strategy with a comprehensive definition of wellbeing that allowed us to examine wellbeing literature from multiple disciplines; 3) inclusion of both peer-reviewed articles and grey literature; 4) quality appraisal of each included study and quality assessment of the synthesised evidence.
This review is subject to the following limitations: 1) key population groups such as LGTBQI, sex workers and injecting drug users were excluded, as their lived experiences are likely to be different given the higher levels of stigma they possibly encounter relative to the general population of YPL-HIV [149][150][151]; 2) several applicable regional conferences lack online abstract and thus our review may have missed potentially relevant material; 3) our review focused only on YPLHIV in SSA and thus findings may not be generalisable to other regions. However, results from recent studies suggest that there are similarities between YPLHIV in SSA and other developing nations within the South-East Asia Region and Region of the Americas in terms of correlates of mental illness [81,152] and adherence [153,154]. Hence, results could be relevant to settings with similar disease and socioeconomic profiles. Several gaps in our evidence base were identified. First, no study directly assessed correlates of wellbeing (using a wellbeing scale) or evaluated experiences of subjective or psychological wellbeing. Second, there were no data from Central Africa, with only one study from West Africa. Third, our evidence is not generalisable to YPLHIV in the community given that most studies sampled individuals from healthcare facilities. Fourth, variability in outcome definition and scale choice made it difficult to compare between quantitative studies and did not allow for a meta-analysis. Fifth, the lack of age-stratified data limited our ability to compare correlates and wellbeing by relevant age-bands (15)(16)(17)(18)(19) vs. 20-24 yrs.). Sixth, weaknesses in the study design and analytical techniques used in the quantitative evidence limited out ability to draw out any causal interpretations. Lastly, for the qualitative review, data saturation among 20-24-year-olds and YPLHIV in rural settings was not reached, leaving gaps in our understanding of how these factors may have shaped wellbeing.

Conclusion
The aim of this mixed-methods review was to identify dimensions of wellbeing among YPLHIV and measures that align to these dimensions for application in HIV policy evaluations. Understanding the wellbeing effects of HIV/AIDS policies could help steer policies in the direction that meets the broader needs of YPLHIV. This review has shown that social support was a key correlate of poor mental health and that social relationships shaped positive lived experiences. In view of the negative association between poor mental health functioning and measures of wellbeing found in the literature, it is plausible to posit that in this population wellbeing is multi-dimensional and that relational dimensions frame wellbeing among YPLHIV. Multi-dimensional wellbeing scales with a strong relational focus that could be applicable for this group include the Personal Wellbeing Index, Ryff's PWB and Mental Health Continuum Short-Form. Future studies should go beyond the investigation of mental health and examine wellbeing, based on definitions grounded in theory, to provide more accurate data on the wellbeing effects of policies. However, psychometric evaluations of these scales in this population in SSA are warranted, together with validation of these scales against adolescents and young people's subjective experiences in SSA.