Background

In sub-Saharan Africa, consistent use of condoms among adolescents is low [1]. This is likely a contributory factor for the high prevalence of sexually transmitted infections (STIs), including HIV, and adolescent pregnancy in the region [1,2,3,4], which has persisted despite more than two decades of research and programming. As many as 20% of women in sub-Saharan Africa have their first child by the age of 19 [5], and around half of these pregnancies are unintended [6]. In 2017, adolescent girls accounted for 25% of all new global HIV infections, despite comprising only 10% of the population [7]. Further, in the Southern African region in particular, there are stark gender disparities in relation to HIV risk: risk among adolescent girls aged 15–19 is six times higher than among their male counterparts [8]. Further, a recent systematic review has identified persistent perceived and experienced barriers to STI services for adolescents [9].

Much research has attempted to uncover determinants of condom use among adolescents that might be targeted by sexual and reproductive health (SRH) programmes. A complex array of individual-, interpersonal- and structural-level influences are at play [10,11,12,13,14], including lack of desire to use condoms, lack of local access to condoms, gender inequalities and social norms restricting condom use, and age-disparate and transactional sexual relationships [2, 15,16,17,18]. These factors, taken together, may help explain why SRH programmes that focus only on individual determinants of condom use and ignore wider interpersonal and structural determinants have had limited impact [19]. One recent meta-analysis [20] found individual-level social-cognitive determinants accounted for only 15–30% of the variance in adolescent condom use intentions and behaviours. These findings indicate that interpersonal and social dynamics, plus context-dependent cultural and structural conditions, are key factors associated with condom use that must be explored further both theoretically and empirically [20]. In order to craft more effective condom use interventions for adolescents, programme developers must adopt a multilevel approach, including a combination of structural and behavioural interventions that target both males and females [7, 13]. Systematic reviews can aid this process by examining multilevel barriers and facilitators of condom use in particular geographic regions. Results of such work would be better positioned to guide future intervention planning and empirical work on this important subject. Qualitative studies from Southern Africa are particularly important for deriving deeper and more nuanced understandings of the barriers to condom use, and can assist researchers and practitioners develop, adapt, or scale-up interventions in the region [13, 21].

This review aims to describe the individual-, interpersonal- and structural-level determinants of condom use among male and female adolescents in Southern Africa and then develop an integrated conceptual framework. The findings provide clear recommendations for future research aiming to test causal pathways explaining condom use among adolescents, and for SRH programme development and adaptation in the Southern African region.

Methods

The synthesis reported here is part of a larger research study to adapt an SRH intervention for adolescents in Southern Africa [22]. The project includes two linked reviews that used a common preregistered protocol [23] but reviewed qualitative and quantitative literature separately. In this paper we include qualitative studies only, in order to promote a deep exploration of salient individual and interpersonal barriers and facilitators of condom use among adolescents, as well as broader social, cultural and structural dynamics. Findings from a review of quantitative studies are reported elsewhere [24].

For the present review, the ENTREQ (Enhancing Transparency in Reporting the Synthesis of Qualitative Research) [25] guidelines informed reporting of the analysis. This review aimed to answer the following research questions: 1) What are the individual-, interpersonal- and structural-level barriers and facilitators of condom use among adolescents in Southern Africa? 2) How can these multilevel barriers and facilitators be conceptualised within an integrated framework?

Guiding frameworks

‘Best-fit’ framework synthesis [26, 27] was used to guide the data analysis. The approach provides a method for incorporating relevant theories within a framework analysis and is particularly suited to the development of new conceptual models relating to health behaviours. The approach involves identifying an a priori theoretical framework based on one or more published theories or models against which review data are coded.

The current review is part of a study that involves the adaptation of an SRH intervention named If I Were Jack (JACK) for the Southern African context [28,29,30]. JACK was developed in consultation with stakeholders for use in the UK and Ireland. Its theory of change is underpinned by social cognitive theories of behaviour change plus gender-transformative theories and understandings of broader socio-cultural influences and underlying values (especially relating to religiosity and social class) [28,29,30]. Given the focus of the broader study, we chose to use the intervention theory of change [28] to inform the ‘best-fit’ framework.

Given that a further aim of the study was to examine the relevance of the If I Were Jack Theory of Change to the Southern African context (i.e., to examine if it was appropriate given the likelihood of contextual differences between the UK and Southern Africa), we also included in the a priori framework theoretical models that conceptualise a broad range of environmental and individual-level influences on behaviour. These included the established Social Ecological Model [31, 32], which acknowledges that individual decision-making is shaped by a dynamic interplay between features of the environment at intrapersonal, interpersonal, organisational, community and society level, as well as an integrated social-cognitive model of condom use among adolescents in sub-Saharan Africa [20], which highlights individual level factors that relate to past condom use behaviours, including attitudes about condoms and condom use, and perceptions of norms, risks, behavioural control and barriers to condom use. Together, these three models were used to generate a number of a priori themes against which the extracted data were coded (see Table 1).

Table 1 A priori themes reflecting theoretical determinants of condom use among adolescents from three ‘best-fit’ models
Full size table

Criteria for study inclusion

Primary or secondary qualitative studies published in English in peer-reviewed journals between January 2000 and August 2019 were included in the review. The inclusion of only English publications was due to a lack of resources for translation. We included all study designs that directly reported on relevant determinants of condom use in adolescents and were conducted in Southern African Development Community (SADC) countries (Angola, Botswana, Comoros, Democratic Republic of Congo, Eswatini, Lesotho, Madagascar, Malawi, Mauritius, Mozambique, Namibia, Seychelles, South Africa, Tanzania, Zambia and Zimbabwe). We also included studies with participants of any sex, sexual orientation and gender identity aged 13 to 19. We chose this age range as the intervention that is the focus of the broader study is aimed at teenagers aged 13–16. If a study included participants outside this age range, we sought to examine outcomes for those participants within the 13–19 age range only. If this was not possible, we included the full study if the mean age, or at least half the participants, fell within this age range. We included only studies that reported outcomes relating to condom use by adolescents. These criteria informed the development of the search terms and screening of records.

Search strategy and screening methods

We searched four databases - MEDLINE, PsycINFO, Embase as recommended by Cochrane [33] - supplemented with the Web of Science to ensure any possible bias in database search algorithms was minimised. The search terms used are detailed in Additional File 1. Duplicates were removed prior to review. As per our review protocol [23], we independently screened titles, abstracts and full text articles in duplicate, applying the study eligibility criteria. If there were multiple published reports for a study, we merged the information from these publications to provide a single record of the study’s data. We used the software programme Rayyan [34] for record management and screening. The inclusion and exclusion of records were recorded in a PRISMA flow diagram (Fig. 1) [35].

Fig. 1
figure 1

PRISMA Flowchart

Full size image

Data extraction

We captured all necessary information from the included studies using a pre-designed data extraction form. For each eligible study, one review author (SG) extracted data, including primary data extracts, themes presented by authors and author commentary about those data and/or themes. This extraction was checked by a second author (SR or CL) for a random sample of 25% of the included studies (6 studies in total) after data from the first ten studies had been extracted Discrepancies were identified, recorded and resolved through discussion. Study characteristics entered onto the form include: study setting, inclusion and exclusion criteria, sample characteristics, research design and analysis methods, type and details of intervention, as well as narrative direct quotations or summaries of quotations relating to determinants of condom use.

Data consisted of adolescent self-reports, or participants’ perceptions of determinants of adolescent condom use. Outcomes were categorised as behavioural outcomes (i.e., condom use, access to condoms) or cognitive outcomes (i.e., outcome relating to attitudes, beliefs or knowledge about condom use). Within each sub-category, we further determined whether the outcomes assessed were individual-, interpersonal- or structural-level determinants.

Quality appraisal

One reviewer (SG) assessed the quality of each included study using the Joanna Briggs Institute (JBI) critical appraisal checklist for qualitative studies [36], which yielded an overall classification of high or low quality for each study. Studies were classified as being of low quality if more than three domains (out of a possible 10) were missing. Assessments were checked by a second reviewer (CL). Although no studies were excluded from the review based on quality, they did widely vary, so quality classifications were included in the interpretation of findings.

Data synthesis

‘Best-fit’ framework synthesis [26, 27] was used to guide the analysis because it provided a method for incorporating the identified theories within a framework analysis and provided a framework for development of a new conceptual model. It incorporates both deductive and inductive analysis. Qualitative data extracts were coded against the 12 a priori concepts from the ‘best-fit’ model using NVivo 11. We entered concept headings (see Additional File 2) into NVivo and data were coded deductively under the relevant headings. All data fit within the a priori framework. Subsequently, we revisited the evidence to explore the relationships between a priori concepts. Using thematic analysis, we clustered and synthesised the 12 a priori concepts into a final set of six themes and used them to develop a conceptual model of the barriers and facilitators of condom use among adolescents in Southern Africa. The analysis was led by one author (ÁA), with discussions held with the broader team at each stage until consensus was reached.

In the results section below, all research participant statements (quotes) appear in italics to easily distinguish them from study author views and statements, which are presented in italics without quotation marks. Also, any words removed from a quote to enhance readability are shown by ellipses (…) and added words are provided in square brackets.

Results

Search results

We identified 2102 records, of which 780 were duplicates. Following screening of the remaining 1322 records, 23 qualitative studies were included in the review (see PRISMA flowchart Fig. 1).

Characteristics of the selected studies

The characteristics of the included studies are summarised in Table 2. Studies from 11 countries in Southern Africa were included. Sample size was reported in all but one of the studies and ranged from 16 to 255, and resulting in a reported total of 2270 participants. Two studies included participants from four countries outside of Southern Africa, however, only data from included countries was extracted. Age range was reported in 22/23 studies and ranged from 10 to 24 years old, with mean age within review parameters. Reporting of participant sex/gender in the included studies was sporadic (n = 12; 57.1%). Eight studies were single sex (n = 5 included girls only and n = 3 included boys only). The remaining four studies that reported sex of participants included roughly an equal number of females and males.

Table 2 Included study characteristics
Full size table

Quality of included studies

Overall, the quality of studies was found to be high, with notable exceptions.

Five of the 10 domains were scored “Yes” (present) across all 23 studies and covered issues related to statements about research methodology and question; correct methodology employed; representation of data and analysis; interpretation of results; and representation of participant voices. Only one study (Bosmans et al., 2006) received a “No” (not present) for conclusions flowing from data interpretation and analysis. Two domains were slightly more mixed: 6/23 studies (26%) did not have congruency between the stated philosophical perspective and the research methodology and another six did not report ethical research criteria and evidence of ethical approval by an appropriate body.

In contrast with these overall positive findings, two of the domains in the assessment tool, related to theoretical and/or cultural positionality of the researcher, and accounting for the influence of the researcher on research itself, were each lacking in 22 studies, with only one study providing both types of statements (Mwalabu et al., 2017) [54].

Synthesis of the evidence

The findings provided evidence to support all 12 a priori themes in the original ‘best-fit’ model depicting the multi-level determinants of condom use among adolescents in Southern Africa (see Table 3). Further thematic analysis resulted in six inductive themes that revealed factors that can act as barriers or facilitators of condom use at the three levels in the model. A summary of the themes and sub-themes along with examples of supporting data are presented in Table 4 and a synthesis of the barriers and facilitators and their overlapping nature is presented in the conceptual model (Fig. 2).

Table 3 A Priori Themes – Individual, interpersonal and structural level determinants of condom use among adolescents in Southern Africa
Full size table
Table 4 Synthesis themes and sub-themes
Full size table
Fig. 2
figure 2

Multilevel model of barriers and facilitators of condom use among adolescents in Southern Africa

Full size image

Barriers and facilitators of condom use among adolescents in southern Africa

Theme 1: unequal gender norms and restrictive masculinities

Pervasive unequal gender norms and restrictive masculinities were identified barriers to condom use among adolescents in Southern Africa in the reviewed studies, with some emerging evidence of shifting norms. Examples of these barriers include stigmatisation of adolescent women who use condoms and unequal gendered norms relating to sexual decision-making, responsibility and pleasure that favoured young men. While it was reported that condom use among adolescent women was highly stigmatised and neither acceptable nor expected of women of ‘good’ moral character [38, 40, 42, 46,47,48,49, 51,52,53], it was also evident that there was widespread concurrence with the notion that ‘real’ men do not use condoms [40, 46, 52, 53].

Reflecting these unequal gender norms, adolescent boys and young men in several studies reported a belief that girls who carry or use condoms are ‘easy’, untrustworthy and likely suffering from a sexually transmitted infection (STI) [38, 42, 46]. Similarly, studies reported adolescent girls’ fear of embarrassment or judgment if they sought to obtain condoms, or if they carried or requested to use condoms [44, 46, 48, 51,52,53]. For example, one participant in a Zimbabwean study indicated that if she requested for her male partner to use a condom “he [would] think that you were up to no good, you were having sex with many people previously.” [48].

Several studies noted that adolescent boys and young men played the central role in sexual decision-making in the study countries [46,47,48, 53, 59]. When adolescent girls and young women took ‘responsibility’ for condom use, they also take took ‘blame’ for engaging in taboo behaviours. The author of one Mozambican study notes:

‘Girls have to decide for themselves if they will take the responsibility for carrying condoms. In doing so, they can avoid risk-behaviour and lessen the male’s sense of duty. At the same time, girls have to face embarrassment when the initiative of carrying condoms comes from the female partner.’ [40]

Unequal gender norms as a barrier to condom use were also evident in widespread reporting of sexual pleasure as a privilege of men [39, 40, 42, 44, 45, 47, 56, 57, 61]. A contradiction in this regard is highlighted in a Mozambican study: although young men believed condoms reduced pleasure, they would still use condoms with sex workers because of the perceived heightened risk of HIV. The author notes ‘Condoms only appear to interrupt pleasure in close and steady relationships, and not with people considered more likely to be HIV infected’ [47].

Several studies suggested a shift in thinking that might facilitate condom use, with young men in one study reporting that they viewed using condoms as a sign of respect for their partners [37], and others reporting shifts in thinking in relation to gendered norms, specifically female agency as a facilitator of carrying and using condoms [45, 46, 59]. There were no clear indications or explanations offered for why or how these shifts were occurring.

Theme 2: other social norms

The included studies also highlighted the existence of other social norms that acted as barriers to condom use. These included 1) the stigmatisation of adolescent sexuality and non-traditional family planning and SRH education and 2) social norms that appeared to act as moderators of risk perception.

Social norms that favoured traditional methods of family planning or SRH education that was not evidence-based were reported as barriers to condom use. Several studies mentioned norms promoting traditional practices and methods of avoiding HIV and pregnancy (which generally involved avoiding sex outside of marriage) rather than embracing contemporary methods, which were often viewed more negatively [38, 43, 44, 53, 54, 59].

Social norms reflecting negative perceptions of adolescent sexuality were suggested in several studies; stigma attached to adolescent sexual activity was reported as a barrier to both male and female adolescents’ ability to obtain condoms [51, 52, 59]. The pervasive nature of these norms was evident in organisations that prevented young people from accessing free condoms or judgement-free SRH services [38, 46, 50,51,52, 54, 55], and inadequate provision of accurate SRH education in schools or communities [39, 46, 47, 52, 55, 57]. A respondent from one Zimbabwean study noted, “How can you take a condom from the collection points at the clinics when all patients (some friends, relatives, and teachers, religious and traditional leaders) in all age groups will be staring at you?” [52].

There was also evidence suggesting that social norms may act as a moderator of risk perception relating to HIV. In one Tanzanian study [57], it appeared that some young people had normalised HIV as something they should not fear because it was so prevalent in their communities. A participant notes: “[Young people] know everything [about HIV] and may pass people in the street and even say ‘so and so is infected’. But the boys say, why are you afraid of HIV and not flu? They have become unafraid because it’s so normal.”

Relatedly, and more common, were indications that adolescents associated HIV with those of ‘bad’ moral character and were therefore themselves immune to it because their ‘steady’ sexual partners were of ‘good’ moral character, like themselves [37, 41, 42, 44, 46,47,48,49, 54, 57, 59]. This was illustrated by a participant in one South African study: “It’s if you have two girlfriends, your steady and your secret lover. You can never use a condom with your steady but you can use one with your secret lover because you don’t know if she has a disease.” [46].

Theme 3: political and economic climate

Political buy-in to the promotion of adolescent SRH was implicated as a potential facilitator of condom use among adolescents in some studies. These included policy-led provision of free condoms and SRH services in resource-poor and rural settings in South Africa and Madagascar [44, 46]. One Zimbabwean study author [52] noted that a lack of adolescent SRH legislation or national policy was a barrier for educators wishing to incorporate RSE into school curricula. Conversely, a government-backed national media campaign promoting condom use in Zimbabwe was reported as a facilitator [52].

Poverty and socioeconomic status were also noted as potential moderators of adolescent knowledge, attitudes, behavioural control and perceptions of risk. A common finding was the impact of poverty that led young women having to forego using condoms in age-disparate and transactional relationships [41, 42, 44, 48, 59]. Conversely, higher socioeconomic status and associated future orientation emerged as facilitators of condom use, and these factors appeared to mediate adolescents’ perception of risk. Several studies indicated that not being able to finish school due to pregnancy or HIV was the primary consideration among adolescents, with younger adolescents and those of higher socio-economic status reporting more strongly that they feared the consequences of unprotected sex [37, 40, 43, 49, 56, 57]. One Mozambican study author indicated ‘Many boys expressed the idea that it was important to use condoms because ‘life must be protected when you have something to live for, something you want to do in your life’. [43].

Theme 4: community-based resources and influences

Service- and community-level barriers and facilitators were reported across the 23 studies and included SRH services, religious organisations, schools, and the physical features of communities that influenced both access to condoms and the spaces in which young people have sex.

Accessible adolescent SRH services

A lack of adolescent-friendly, community-based SRH services frequently was presented as a barrier to condom use [38, 46, 50,51,52, 54, 55]. Some adolescents reported that they did not know where to obtain condoms in their communities, and others reported that although they had this knowledge they did not feel confident doing so because they feared judgment from clinic staff [38, 44, 51, 57]. Reflecting the influence of social norms, including gender norms, several studies reported that health professionals did not distribute condoms freely, privately or without judgment because they did not want to encourage sexual activity in adolescents [38, 46, 51, 54, 55]. Negative experiences at clinics were reported particularly by young women [46, 51, 55]. Conversely, one Tanzanian study reported the facilitative effect of positive attitudes about condom use from health professionals and emerging changes in social norms [50], and another study in Botswana reported positive experiences at a clinic where ‘boxes of condoms were kept at [an open] window and no questions were asked [of adolescents who took them]’ [51].

Sex education in schools and communities

Several studies mentioned the absence of sex education in schools and communities as a barrier to condom use [39, 46, 47, 52, 55, 57]. Inadequate or inaccurate SRH knowledge was common [38, 39, 47, 48, 50, 55, 57], mainly because participants had not received comprehensive SRH education [40, 50].

Discomfort among teachers presenting SRH material was noted by one Tanzanian study [57] as a possible reason for a lack of sex education provision, and a lack of SRH resources for teachers was noted by a Zimbabwean study [52]. Other studies indicated that information provided by existing programmes was inaccurate, such as giving young people false statistics relating to the efficacy of condoms [41, 59]. Adolescents in Zambia and Malawi reported learning about SRH primarily from peer discussion [39, 55]. Some studies noted the provision of condoms in South Africa [46] and SRH programmes [40] in schools in Mozambique as facilitators of condom use.

Religious influences

Religious organisations and their representatives were mentioned by several studies as barriers to condom use [38, 41, 43, 45, 53, 54]. The studies reported that religious leaders encouraged abstinence and monogamy, discouraging condom use because of beliefs that they encourage sexual promiscuity and/or because the use of birth control is not in line with their faith.

In one Malawian study, health professionals reported experiencing conflict between promoting sexual wellbeing and conforming to religious norms [54],while a Congolese study noted that acceptance of religious norms had negatively influenced the provision of SRH [38], although the quality of this study was impacted by a lack of clarity regarding the flow of conclusions from analysis: “Nor were adolescents given full information, even about the menstrual cycle, on the grounds that it might encourage sexual liberties” [38].

Access to condoms and spaces for sex in the community

Two studies noted that a barrier to condom use was that young people did not carry condoms with them and therefore did not have them readily available when needed [46, 57]. Relatedly, a Tanzanian study [58] reported that time and space are often barriers to young people so they choose to have sex in alleyways, cemeteries, toilets in bars, or unfinished houses. The authors suggest that due to fear of being interrupted young people are in a hurry and consider condom use a waste of time. “[They] have sex in an unfinished house (…) they are not using condoms because they are in a hurry and they do not have time.” [58] Similarly, in a South African study [46] adolescents reported having sex at home when their parents were out and not wanting to waste time by using a condom. In urban settings, organisations within communities including bars [38, 52], guesthouses [58], and shops [44, 46, 52] where young people had access to condoms were often mentioned as facilitators of condom use.

Theme 5: interpersonal influences

Across studies, interpersonal networks including sexual partners, peer influences and parent/caregiver communication presented as barriers and facilitators of condom use.

Trust and transaction in sexual relationships

Heavily influenced by unequal gendered norms, condom use within a sexual relationship was seen as dependent on the status of the relationship, with condom use in casual relationships more widely accepted than in committed relationships. For people in committed relationships, not using condoms appeared to be linked to trust, faithfulness and respect for one’s partner [37, 38, 40, 46, 47, 52, 53] and those who requested condoms were assumed to be ‘sick’ or untrustworthy, especially women [42, 48]. One study suggested an alternative perception, reflecting emerging changes in gender norms, of how condom use in committed relationships could be seen as a sign of faithfulness when a male partner uses a condom because his female partner does not want to get pregnant and therefore drop out of school [37]. Some studies indicated that age-disparate and transactional relationships acted as barriers to condom use, with older men reportedly less willing to use condoms and young women powerless to demand they do so, especially when the older man provides them with material support [41, 42, 44, 48, 59].

Peer influences

Some studies mentioned the importance of peer influences on condom use, noting that negative peer norms relating to condom use acted as a barrier, particularly for young men [46, 47]. It also appeared that peer norms also negatively affected condom use among females. One author of a South African study indicates: ‘Young women argued that for a steady partner to insist on condom use is seen as indicating a lack of respect and trust that could destroy one’s reputation within the peer group. If a boy wants to use a condom she will say it is because he disrespects her, because he wants to use ‘a plastic’.’ [46] One Botswanan study reported male peers acting as facilitators of condom use by sharing their unused condoms [51].

Parent/caregiver communication

A lack of communication and guidance from parents and primary caregivers about SRH was indicated as a possible barrier to condom use by four studies [40, 46, 51, 57], particularly because of parental discomfort discussing sexual matters and adolescent perceptions that parents would disapprove of condom use [51].

Theme 6: adolescent attitudes about condoms

Attitudes that condoms negatively affecting pleasure or sexual satisfaction were noted as barriers to their use [39, 40, 42, 44, 45, 47, 53, 56, 57], with adolescent men claiming that condoms were too tight, painful, reduced their ability to maintain an erection, delayed orgasm and reduced sensation. Some said that using condoms reduced the ‘thrill’ of sex [53, 56, 57]. Reflecting the possible unequal impact of gender norms in relation to attitudes about condoms, there were no reports from adolescent women regarding reduced pleasure.

It also appeared that attitudes about condoms were strongly influenced by inaccurate knowledge. Studies reported that adolescents did not use condoms because they believed they were ineffective against STIs/HIV or pregnancy because they had ‘little holes’ in them allowing sperm (and HIV infection) to pass through [38, 39, 41, 45, 52, 53, 57]. Some adolescents reported that free condoms obtained at clinics were particularly defective [51, 52]. Related to this, some studies reported common attitudes that condoms cause disease illness such as cancer, rashes, sores and stomach pains [43, 51, 53, 56, 57].

Discussion

This review aimed to synthesise qualitative evidence relating to the individual-, interpersonal- and structural-level barriers and facilitators of condom use among adolescents in Southern Africa. We identified and mapped relevant qualitative literature onto an a priori framework incorporating three theoretical models – the social ecological model [31, 32]; an integrated model of condom use among young people in sub-Saharan Africa [20]; and the If Were Jack theory of change [28,29,30] – and then used inductive thematic analysis to identify the key barriers and facilitators of condom use which we incorporated into a new conceptual model (Fig. 2).

The findings provided evidence to support all 12 a priori themes in the original ‘best-fit’ model depicting the multi-level determinants of condom use among adolescents in Southern Africa. The combined three-model framework was able to capture more determinants than any one of the models could accurately achieve on its own, thus suggesting that a multilevel conceptual model as outlined in Fig. 2 may be more appropriate.

In line with previous reviews [14, 62, 63], the findings indicate that young people in Southern Africa face a range of barriers to condom use operating at all levels of their social ecological milieu. Results also highlight complex interactions among six key themes and illustrate that both positive and negative influences are reinforced at different levels. In line with the UN Sustainable Development Goals and their focus on multilevel determinants of health, this review offers understanding and insights into how common themes such as gender inequality and poverty exert their influence on condom use across social ecological levels.

The findings of our review are consistent with other studies [14, 18, 62,63,64] whose authors have identified a need to move beyond individual-level behaviour change frameworks and programmes by incorporating broader understandings of the structural-level barriers and facilitators of condom use among young people in Southern Africa. We advance previous findings by offering a conceptual framework of the key barriers and facilitators at society, community, interpersonal and individual levels that might act as a guide for decision-makers, programme developers and researchers wishing to optimise intervention impact. We are also optimistic about the fact that we have uncovered positive findings relating to shifts in unequal gendered norms, with key messages relating to female agency in sexual decision-making emerging as key.

The findings suggest that valuable and limited resources should be best used in coordinated effort to increase condom use by implementing multilevel interventions at all levels of the social ecological system. This could be informed by emerging broad-based guidance for improving uptake and access to contraception [64]. Programming efforts should, at the very least, ensure efforts are made to start addressing the widespread influence of gender inequalities and restrictive gendered and social norms that appear to operate at all levels of the social ecological system. While this approach is not a panacea—especially in challenging political and economic climates which are characterised, at times, by a lack of policy or policy implementation relating to gender equality —it might help shift attitudes and initiate change slowly across the community, organisational, interpersonal and individual levels [21].

The findings of this study indicate the need for easily accessible and free condoms, especially in resource-poor and rural settings. Importantly, they equally emphasise the need for evidence-based adolescent SRH strategies that provide health professionals and educators training and resources needed to provide youth-friendly SRH services and education. The findings also highlight the influence of social norms reflecting negative perceptions of adolescent sexuality, and non-traditional sex education and methods of avoiding HIV, STIs and unintended pregnancy. The strength of these prevailing norms suggest that programme developers need to involve local stakeholders in co-design and co-production processes. Interventions that can engage wider community members where these negative norms are played out are also critical. Community mobilisation interventions have shown positive impacts in relation to other SRH interventions, particularly those that involve comprehensive SRH education and positive role modelling by community members such as community elders, religious leaders, healthcare professionals and educators [65,66,67]. Coupled with this approach is the need for individual-level interventions targeting health professionals, educators and community members that promote youth-friendly, evidence-based comprehensive SRH education and practice [68]. Decision-makers should also be lobbied to support such interventions with policy-led adolescent SRH strategies and to provide resources for their implementation with youth [69,70,71].

Centrally, comprehensive sex education delivered to adolescents in school and community settings should include gender transformative components; directly challenge stigma and myths surrounding condoms and condom use; and harness the facilitating effects of promoting female agency in sexual decision-making [21, 72]. These curricula should prioritise key issues such as male roles and responsibilities in relation to avoiding HIV and unintended pregnancy; sexual pleasure (particularly gender disparities in relation to expectations of pleasure and the contradictions relating to pleasure in committed versus casual relationships); and negotiating condom use (particularly in age-disparate and transactional relationships). Programmes that promote the roles and responsibilities of sexual partners, peers and parents/caregivers as key interpersonal influences will also be necessary [70].

Strengths and limitations of the review

Core strengths of this review include the systematic process by which we integrated and analysed rich contextual data from a broad range of qualitative studies conducted with adolescents in Southern Africa, and used them to develop an integrated conceptual framework that might be of value for intervention developers and decision-makers.

Limitations include the possibility that we did not identify potentially relevant articles despite our comprehensive search. In particular, due to resource constraints, we excluded studies not reported in English as well as grey literature. We may therefore have missed studies that were relevant to the review. In particular, over the past two decades, ministries of education and health in many Southern African countries have worked to implement comprehensive relationships and sexuality education policies and programmes in schools and communities that have not been captured by this review. Further, while we report findings from 11 countries in Southern Africa, five other countries in the SADC region included in our search (Angola, Comoros, Lesotho, Mauritius and Seychelles) were not represented in the review. These omissions mean that the findings may not be entirely representative of the experiences of young people across the region. Additionally, due to uneven reporting of sample composition and age range, it was difficult to determine what findings might apply to sub-populations of adolescents. Further, because some factors were only reported on by studies in some of the included countries, we cannot assume that the findings are generalizable to all countries or populations in Southern Africa.

Finally, despite promising findings regarding quality in other domains, the lack of reflexivity related to theoretical and/or cultural positionality of the researcher is important to address in qualitative research, especially regarding sensitive content such as condom use and sexual practices. While bias can be mitigated in straightforward ways in quantitative research, researcher positionality is often not clearly stated, which can negatively influence the rigour and accuracy of the findings. Indeed, while we ourselves are an international, multiracial, mixed gender, multidisciplinary research team, we recognise that our interpretation of the findings represents but one among other possible interpretations. The data synthesis and drafting of this paper was led by the first author, a white, Irish, female, PhD, who positions herself as a social psychologist heavily influenced by interpretivism and realism. While we are confident that reflexive journaling as part of the analysis process, as well as team discussions drawing out various interpretations of the findings, have increased rigour, unconscious bias may have been inadvertently introduced.

Conclusion

Our findings suggest that programmes should address the wider structural influences through both individual- and structural-level interventions, in addition to targeting individual-level socio-cognitive factors. It is especially important to understand the key barriers and facilitators of condom use that act as moderators on the pathway towards consistent condom use among adolescents. In Southern Africa, as in many other regions of the world, these barriers include unequal gender norms and restrictive masculinities that stigmatise adolescent girls; social norms that restrict positive sexualities among adolescents and perpetuate negative perceptions of condoms and condom use; and the pervasive impact of poverty on access to condoms, SRH education and the power to negotiate condom use freely. SRH programming targeting barriers and facilitators of condom use at multiple levels is urgently needed in Southern Africa. Such programmes should pay particular attention to the key gaps in knowledge and evidence-based interventions highlighted by this review, for example: promotion of female agency in SRH decision-making; increasing positive parent-child communication about SRH; increasing accessible youth-friendly SRH services; and encouraging the widespread adoption and implementation of strategies to improve SRH.