Why interventions to prevent intimate partner violence and HIV have failed young women in southern Africa

Abstract Introduction Adolescent girls and young women aged 15 to 24 years have some of the highest HIV incidence rates globally, with girls two to four times more likely to be living with HIV than their male peers. High levels of intimate partner violence (IPV) experienced by this age group is a significant risk factor for HIV acquisition. While behavioural interventions to prevent IPV and HIV in southern Africa have seen some success in reducing self‐reported experiences of IPV, these interventions have largely failed to achieve similar outcomes for young women. Discussion We identify three main reasons for the failure of IPV/HIV interventions for many young women in southern Africa. First, interventions are usually developed without the meaningful involvement of both young women and young men. Youth input into research design is largely focused on user testing or consultation of targeted groups, involving relatively low levels of participation. Second, interventions are focused on addressing individual risk factors rather than broader social and structural contexts of being a young woman. “Risk factor” interventions, rather than supporting women's agency, can pose a major barrier for supporting changes in behaviour among young women because they often fail to dislodge well‐entrenched gender and age‐related inequalities. Third, current intervention models have not adequately accounted for changes in gender norms and relationships across southern Africa. Individuals are getting married later in life (or not at all), new technologies are transforming romantic interactions and opening new opportunities for violence, and discussions about women's rights are both challenging gender inequalities and reinforcing them. Conclusions In order to move beyond the status quo of current approaches, and to support real innovation, IPV/HIV prevention interventions need to be co‐developed with youth as part of a meaningful participatory process of research, intervention design, youth involvement in development and implementation. This process of co‐development needs to be radical and break with the current focus on adapting existing interventions to meet the needs of young people, which are not well understood and often do not directly reflect their priorities. Broader social contexts and compound lenses are needed to avoid narrow approaches and to accommodate evolving norms.


| INTRODUCTION
Adolescent girls and young women aged 15 to 24 years, particularly those out of school, have some of the highest HIV incidence rates globally, with girls two to four times more likely to be living with HIV than their male peers [1]. High levels of intimate partner violence (IPV) experienced by this age group is a significant risk factor for HIV acquisition [2,3]. IPV places young women at greater risk of depression [4], anxiety [5], and harmful alcohol use [6], which in turn increases their risk of HIV [7,8]. While IPV increases the susceptibility of young women to forced sex by male partners [9], far more problematic and insidious are connections between men's use of violence against female partners and men's attempts to control women's autonomy and assert male power, reducing young women's self-efficacy and ability to negotiate safe sex [10,11].
Young women's susceptibility to HIV is linked closely to their male partner's risk behaviours. Men who are violent tend to have other high-risk behaviours including substance abuse, and multiple and concurrent sexual partners, and are more likely to be living with HIV [12,13]. IPV often belongs to a wider context of violence for young men, including witnessing violence within their household or in broader society as children, making them more likely to perpetrate violence as young adults [14]. Exposure to violence for both adolescent boys and girls contributes to lower academic performance in school [15] and internalized stigma among youth living with HIV [16], Mannell J et al further contributing to a cycle of violence and poverty that reduces individual opportunities and increases HIV risk.
While behavioural interventions to prevent IPV and HIV in southern Africa have seen some success in reducing selfreported IPV experience and perpetration, and HIV risk behaviours, these interventions have largely failed to achieve similar outcomes for young women [17][18][19][20][21][22]. For instance economic interventions for womenparticularly when combined with gender transformative interventionshave shown promise at reducing IPV and HIV risk, but three reviews have shown that these positive outcomes have typically been amongst older women, in more stable (often rural) settings, and not among adolescents [17,18,22]. While a recent systematic review of HIV prevention interventions highlighted the lack of consistent positive findings by evaluations among young women [23]. This is compounded, by the fact that there is little age-disaggregated information within evaluations; a systematic review of adolescent focused IPV and HIV prevention interventions found only six rigorous evaluations, of which only one showed reductions in physical IPV for young women in school, another showed an impact on young men's perpetration of IPV, and the other four, did not disaggregate by age [24].
These failures have knock-on effects for HIV-related biomedical interventions with adolescent girls and young women. Reviewing oral antiretroviral pre-exposure prophylaxis (PrEP) for young women in sub-Saharan Africa, Celum and colleagues [25] point to the limited overall impact among women under the age of 21 linked primarily to poor engagement and adherence. They argue the failure of PrEP interventions for young women is the result of the lack of consideration of local contextual factors, including social norms around sexuality and broader structural barriers making it difficult for young women to adhere to PrEP. Similar findings of low adherence in younger women were described in the topical microbicide and vaginal ring trials [26,27].
A radical new approach to behavioural interventions that prevent HIV and IPV is urgently needed to address the context-specific needs of young women living in southern Africa, including issues of economic and relationship insecurity and the implications these have for HIV prevention. In this commentary, we discuss some reasons behind the failure of behavioural IPV and HIV risk prevention interventions for young women with the purpose of identifying a potential way forward for future intervention development.

| DISCUSSION
Through our collective experience working on behavioural interventions to prevent IPV and reduce HIV risk across sub-Saharan Africa (see Table 1), and engagement with the reviews described above, we have identified three main reasons for the failure of these interventions for many young women. First, interventions are usually developed without the meaningful involvement of both young women and young men. Second, interventions are largely focused on addressing individual risk factors rather than supporting spaces for young women's agency. Third, current intervention models have not adequately accounted for changes in gender norms and relationships across southern Africa. We outline each of these reasons in turn and discuss how they contribute to failed interventions.

| The absence of meaningful involvement of young people in designing interventions
Innovation in interventions with the aim of meeting the needs of young women requires the direct involvement of young women and men in the design of interventions, however, this often does not happen in a meaningful way. Where there is input in intervention design, it is largely focused on user testing or consultation of targeted groups, involving relatively low levels of participation. For example in Stepping Stones and Creating Futures (SSCF), which involved two of the authors (AG & SW)a curriculum-based intervention to reduce IPV and HIV among young people in urban informal settlements in South Africathe livelihoods manual (Creating Futures) was developed using a log-frame, and then tested with 20 young men and 20 young women through a five-day workshop. On the basis of the workshops, the manual was revised tested again with new groups, and finalized before piloting the intervention [41]. While this form of involvement of young people in the design of interventions is a strong move in the right direction and is a dominant approach in programmatic design [42], it is a far cry from the meaningful involvement of young people as active participants in the research process as a means of bringing local perspectives into the way research itself is conceptualized and how it informs intervention design [43].
New approaches to intervention development such as "human centred design" offer potential approaches which centralize participants' experiences and perspectives [44]. However, in a recent review of human centred design in global health, few examples went beyond undertaking focus groups and some basic ethnographic research, to meaningfully involve those who the intervention is designed for in the development, piloting and refining of the intervention [45]. Engagement needs to go beyond simplistic incorporation of "users" views.
The lack of meaningful involvement of key populations in the development or adaptation of interventions contributes to interventions not resonating with the current priorities of young people. For example the implementation of IPV prevention interventions targeting adolescent girls and young women in rural, deprived areas of KwaZulu-Natal, South Africa, with persistently high HIV incidence illustrated challenges of scaling up curriculum-based interventions [46]. One key challenge was poor uptake and completion rates, particularly by those young women out of school or who reported recent migration. While young women welcomed the focus on their needs and the opportunity to learn about health, they described a disconnect between the focus of the programme (individual-level risks that they did not always identify with), in contrast with the community-wide risks and anxieties they faced. Their anxieties related to livelihoods, lack of opportunities, hope and even recreation, and significant worries about future fertility and ability to bear children. Adapting content to relate to the social context and needs of the young women and the delivery model to fit into their daily routines and life structure may help overcome some of these barriers to exposure and thus impact.
Without the meaningful involvement of young people in intervention implementation design, intervention delivery fails to adapt to the structural barriers of time, space and convenience that constrain the daily lives of young women. This limitation is evident in group-based curriculum-driven interventions, which involve sessions held at regular intervals across a number of weeks. While this mode of delivery is based on theories of adult learning, which assume learning happens through reflection and "testing" of strategies, with time between sessions allowing this to happen [47], in contexts where young women are out of school, mobile and with multiple competing priorities, adhering to regularly scheduled activities can be impractical. In low-income settings, young women who are not in school are often involved in a repetitive process of seeking piece work, which may be sporadic, require long hours and be located long distances from home. Similarly, young unemployed women are often expected to care for any children in her extended family.
Meaningful involvement of young people in the development of interventions, including in how they are designed is critical for overcoming these challenges. By meaningful involvement we refer to approaches such as the co-development of interventions, where those targeted by interventions are the ones who design interventions, with academics and practitioners supporting young people in this process [48]. When sufficient time is allowed, young people can be supported to come to analyse their own, and their peers', lives and identify strategies and intervention models that resonate with their own life worlds and experiences, rather than ones mediated by researchers.

| Interventions are developed based on an analysis of risk factors
There is a strong evidence-base outlining HIV risk factors for young women in southern Africa, including IPV [13,49], transactional sex [50], multiple partners [51], alcohol and substance use [52], and poor mental health [7,8]. Interventions are often designed to reduce these risk factors, [53] for example to reduce transactional sex or risky behaviours, interventions focus on building young women's access to savings or reducing alcohol consumption. An alternate approach has been to focus on "structural drivers" of HIV risk [54]. These recognize that risk is not produced by the individual, but rather by contexts that determine the limits, and options, for specific practices/behaviours [54]. Recognizing this is an important step away from individualized understandings of risk for HIV and IPV [55]. However, ultimately, such approaches still aim to reduce risk, rather than a more radical approach to intervention conceptualization.
Interventions focused on tackling "risk factors" whether at the individual or structural level, rather than supporting women's agency, can pose a major barrier for supporting changes in behaviour among young women because they often fail to recognize how gender and age inequalities intersect. For example, interventions working on IPV and HIV prevention may fail to acknowledge the ways in which some women perceive violence as an acceptable aspect of a loving relationship [28], or how others respond to violence by protecting their abusers in order to avoid further discrimination from family and community members [38].
It also neglects the fact that sex is an activity that is often pleasurable and fun. Thus, a simple focus on where "risk" comes from fails to adequately understand the multiple factors shaping women's decision-making and ability to change their lives.
Rather, we suggest there needs to be a radical shift away from "risk factor interventions" focused on individual risk behaviours and risk factors, towards identifying the spaces that exist for young women's agency to address these risks and supporting women's agency. This requires a shift in thinking about how interventions are developed to identify and strengthen what Campbell and Mannell refer to as "distributed agency, " essentially the range of agentic actions young women can take, even in contexts of oppressive relationships [39]. If interventions to prevent IPV and HIV acquisition are aiming to "strengthen" agency, then an understanding of where young women can assert agency from their own perspectives (e.g. by manipulating a boyfriend to give them money but avoiding sexual encounters), needs to be central to the development of interventions rather than simply a focus on risks they face. Working to enhance women's own strategies, and mitigate any negative outcomes of these actions, must be central to the work of interventions [55].

| Gender norms and relationships are rapidly changing in southern Africa
Gender norms and intimate relationships are not static social phenomena, but change over time [56]. For example women are increasingly taking on traditionally male roles as the head of households (some by necessity and others by choice) coinciding with a number of young people not getting married or marrying much later in life [57]. Discussions about women's rights have taken centre stage, often conflicting with notions of gender in ways that challenge gender norms but can also reinforce them [58]. Conversations about same-sex relationships and transgender rights are also happening, however, the specific needs of these groups are still being neglected in ways that exclude some of the most vulnerable young women and men from interventions [59].
Moreover, the use of new technologies including cellular phone use and social media has increased substantially across southern Africa [60], creating new spaces for health intervention and social interaction [61]. However, this has also created new opportunities for gendered forms of violence to occur. For example in addition to offering new communication opportunities, the widespread penetration of cell phones in Zambia has provided opportunities for new forms of controlling behaviours by husbands and the justification of violence by boyfriends through monitoring women's cell phone use [62]. These changes in gender norms and the advent of new technologies are increasing the complexity of choices available to young women in ways that affect potential IPV and HIV risk reduction interventions.
The interventions we are using to address IPV and HIV risk not only often fail to acknowledge these new manifestations of gender norms but may also reinforce old ones. Economic interventions that provide alternative livelihoods for women to support them in leaving violent relationships have often reinforced women's roles as homemakers through sewing or handicraft interventions [63]. In addition, supporting women to work and have livelihoods is often in addition to women's other responsibilities, including unpaid care work and household labour. The gender norms that condone young women's financial dependence on men's productive work and women as unpaid labourers have remained largely unchallenged.

| CONCLUSIONS
Addressing the three challenges of IPV and HIV interventions we have highlighted in this commentary requires a bold new approach. Specifically, behavioural interventions for young women to prevent IPV and HIV need to be co-developed with youth as part of a long-term participatory process of research, intervention design and implementation. This process of codevelopment needs to be radical and break with the current focus on existing interventions to support real innovation. It will require that academics and practitioners take a step back and reorient their role from designing interventions, into one of supporting young people to generate their own ideas and approaches to interventions, recognizing that young women are best placed to understand their gendered world, identify how to transform their lives, and the delivery mechanisms to do so. Such an argument resonates with other approaches such as developing youth "counter-publics" [64] and emancipatory community mobilization [65]. Re-envisaging intervention development, however, will require sustained commitment from government, donors and funders to realize its potential. Without this commitment, we run the risk of continuing to miss the mark on preventing IPV and HIV risk for southern Africa's young women.

A U T H O R I N F O R M A T I O N
JM is a Lecturer in Global Health at the University College London (UCL) and currently leads a programme of research on intimate partner violence (IPV) in high-prevalence settings using participatory research approaches. Her research includes the methodological development of participatory approaches for trials of IPV interventions, and the theoretical development of a "distributed" approach to the agency of women and girls in responding to IPV in the Peruvian Amazon, Afghanistan, India and Rwanda. SW is the Capacity Development Manager working on the global programme What Works to Prevent Violence Against Women and Girls (VAWG), and Co-Investigator of the Stepping Stones and Creating Futures intervention trial, working with young women and men from urban informal settlements in South Africa. Her work involves supporting southern researchers and practitioners to adapt, implement and critically reflect upon interventions to reduce VAWG. She is also completing a PhD reflecting on the role of agency in their experiences and decision-making, as well as pathways to change. MS holds a Faculty position at the Africa Health Research Institute (AHRI), KwaZulu-Natal South Africa, leading a programme of HIV prevention in adolescents and youth. The reflections in this Commentary are based in part on her work in evaluating the impact of programmatic scale up of combination prevention for adolescent girls and young women and using this evidence with young people to co-develop new approaches to improve uptake and retention in combination prevention intervention, such as peer navigation, the use of peers or social networks to deliver HIV selftesting for linkage to PrEP, and exploring the role of digital health to de-centralize care. JS is Professor of Anthropology and Health at the London School of Hygiene and Tropical Medicine and faculty lead for Social Science and Research Ethics at the Africa Health Research Institute, KwaZulu-Natal, South Africa. With others at AHRI she has been working to build a portfolio of research and public engagement activities with and for young people. LS was a founder member and driver on the Coalition for Children affected by AIDS, JLICA and Know Violence. Her research has explored risk and resilience in families infected and affected by HIV. She has been involved in a large number of studies exploring psychosocial provision for children and youth, examining mental health and evaluating interventions in trials in Lesotho, South Africa, Zambia, Malawi and Zimbabwe. She is currently co-director of the new Accelerate Hub which will be looking at synergies for prevention for adolescents in Africa with Prof Lucie Cluver and Dr Chris Desmond. AG is a Senior Specialist Scientist at the South African Medical Research Council. He is the PI of the Stepping Stones and Creating Futures intervention trial, working with young women and men from urban informal settlements in South Africa. He has a particular interest in understanding how interventions are delivered, and how young people draw on interventions to make changes in their lives.