A global comprehensive review of economic interventions to prevent intimate partner violence and HIV risk behaviours

ABSTRACT Background: Intimate partner violence (IPV) and HIV are co-occurring global epidemics, with similar root causes of gender and economic inequalities. Economic interventions have become a central approach to preventing IPV and HIV.Objective/Methods: We undertook a comprehensive scoping review of published evaluations of economic interventions that sought to prevent IPV and/or HIV risk behaviours.Results: Forty-five separate analyses of interventions met our criteria. Broadly, unconditional cash transfer interventions showed either flat or positive outcomes; economic strengthening interventions had mixed outcomes, with some negative, flat and positive results reported; interventions combining economic strengthening and gender transformative interventions tended to have positive outcomes.Conclusions: The review highlighted a number of gaps. Specifically, there were limited studies evaluating the impact of economic interventions on female sex workers, young women, and men. In addition, there were missed opportunities, with many evaluations only reporting either IPV- or HIV-related outcomes, rather than both, despite overlaps.

Globalisation; health information; health determinants; health intervention; population health Background Intimate partner violence (IPV) and HIV acquisition are co-occurring global epidemics [1]. Globally the World Health Organization (WHO) [2] estimates that 30% of all women have experienced some form of sexual and/or physical violence from an intimate partner in their life. There is clear evidence that IPV is a major driver of HIV acquisition amongst heterosexual women [1,3] with studies suggesting up to 25% of all HIV acquisitions occurring amongst women are linked to their experiences of IPV [3,4]. In addition, HIV acquisition is also a cause of IPV [1]. The recognition of these linkages has led to a concerted effort by global institutions and researchers to design interventions to prevent IPV and HIV simultaneously [1,[5][6][7], with programmes such as the Determined, Resilient, Empowered, AIDS-free, Mentored, and Safe women's (DREAMS) initiative, a President's Emergency Plan for AIDS Relief (PEPFAR)-led programme in six countries across southern and eastern Africa, channelling significant money into tackling HIV acquisition through reducing women's experiences of IPV, with a particular focus on adolescents.
Studies highlight the overlapping drivers of IPV-and HIV-vulnerability, particularly poverty and gender inequalities [6,[8][9][10]. There is overwhelming evidence that gender inequalities shape men's perpetration of IPV and women's experience of IPV as well as their vulnerability to HIV in heterosexual relationships [3,[11][12][13][14]. The evidence linking poverty and HIV-vulnerability is less clear, with recent longitudinal studies suggesting poverty can either be a risk or protective factor for HIV depending on other social factors including community acceptability of violence against women [15]. There is stronger evidence about the impact of poverty on women's experiences of IPV. While it is certainly clear that IPV is a global phenomenon experienced by women in high-income as well as low-income countries [2], studies consistently show that women's recent experience of IPV is strongly associated with their experiences of poverty [16][17][18].
Less contentious is the argument that poverty, when intersecting with gender inequalities, places women in economically and socially dependent relationships with men, which increases their vulnerability to HIV and IPV. This exacerbates the challenges of negotiating condoms, leaving violent and controlling relationships and exposing them to controlling behaviours: all risk factors for HIV acquisition and experiencing IPV [8][9][10]. More recently for men, qualitative research has argued that men's partial exclusion from the capitalist economy has led men to develop identities that draw on and deploy forms of emphasised heterosexuality (occassionaly labelled a 'hyper-masculinity'), which include control and domination over women, including IPV as a way to control women, as well as seeking multiple sexual partners [19][20][21].
In recognising how poverty shapes IPV and HIVvulnerability, there has been significant research around using economic interventions to prevent HIV and IPV [6,[22][23][24]. Broadly there are three conceptual approaches. The first is a social protection framework and the direct transfer of resources to households in the form of either cash or food/vouchers [25]. Transfers can be unconditional, or conditional on recipients accessing services or similar, and target poverty as an 'upstream' driver of ill-health [26]. A second approach is located within behavioural economics and uses direct cash transfers to incentivise certain behaviours or outcomes, be this HIVtesting or remaining Sexually Transmitted Infections (STI)-free [26,27]. These assume people make decisions based on trade-offs between alternatives and that the economic incentives are large enough to change behaviours in positive directions [26]. A third approach focuses on supporting those targeted to develop their own economic assets through either extending savings and loans systems to poor populations (e.g. microfinance) or providing vocational training [22]. A subset of these include gender transformative programming at the same time as economic strengthening [6]. These combined interventions are often focused on challenging women's social and economic dependency on men.
There has been recent interest in economic interventions to prevent HIV acquisition and IPV, particularly in the context of UNAIDS' Investment Framework [28,29]. Currently there exist a number of reviews of this evidence base, but these have been limited in scope. Reviews have focused only on economic interventions [22], or only on HIV outcomes or IPV outcomes, or have been of limited geographical scope [6,24,25]. None have sought to review any economic interventions including combined economic and gender transformative interventions and include HIV and/or IPV as outcomes. In this paper we undertook a comprehensive review of economic interventions globally that seek to prevent HIV or IPV.

Methods
We conducted a comprehensive review of published and grey literature. We searched the formal academic websites PubMed, Web of Science and EbscoHost and searched grey literature using Google Scholar. The search terms covered (1) IPV, (2) HIV, (3) economic interventions (the full search string is given in the Appendix). We also used a snowballing technique to search reference lists and review articles to identify other studies.
Articles were included in the review if: (1) they had a quantitative evaluation of an intervention, whatever the study design (including cross-sectional, quasi-experimental and randomised control trial [RCT]); (2) they reported an outcome of either IPV or HIV risk behaviour -HIV risk behaviours had to be behavioural such as condom use at last sex, transactional sex and number of partners; (3) the intervention included an economic component; (4) they were published in English between 1 January 2000 and 1 January 2015.
Exclusion criteria were: (1) qualitative evaluations; (2) they only reported knowledge and/or intentions around HIV (rather than a behavioural outcome); (3) they fell outside of the date range for publication; (4) they were reviews synthesising other studies. There were no exclusion criteria based on quality of study design. The lack of inclusion of qualitative research means that there cannot be a focus on the processes of change in the review.
Initial searches were conducted on relevant databases with all references downloaded to EndNote 7 for review. Articles were initially screened based on title by the first author. A more comprehensive review of abstracts was then conducted. Full texts were read if clarity was further needed (see Figure 1). Data was extracted into a spreadsheet to ensure consistency of reporting.
Where a number of separate analyses sought to understand the impact of the same intervention (for instance multiple analyses of the Child Support Grant [CSG] in South Africa), these are reported separately. We provide a narrative review of the evidence.

Results
In total we identified 45 separate analyses of interventions meeting our criteria (see Figure 1). Given 14  the heterogeneity in interventions we categorised interventions into three categories based on their economic intervention type: (1) evaluations of cash transfer interventions (n = 13; see Table 1); (2); evaluations of economic strengthening interventions (n = 11; see Table 2); and (3) and evaluations of economic strengthening and gender transformative interventions (n = 21; see Table 3). Overall, only eight interventions were reported in the grey literature.
(1) Cash transfer interventions Thirteen separate analyses of six different cash transfer interventions were identified. They were separated into whether they subscribed to a broad social protection approach, or used a behavioural economics approach as their theoretical underpinnings [26].
(2) Social protection Ten analyses of social protection interventions were identified on eight different interventions ( Table 1); five focused only on child outcomes (all from Africa) and five focused only on adult women's outcomes (all but one from Latin America).
(3) Child outcomes Five analyses focused on two different interventions. In South Africa the CSG had three different analyses [30][31][32] and the Kenyan Cash Transfer for Orphans and Vulnerable Children (Kenya CT-OVC) had two different analyses from one study [33,34]. The Kenya CT-OVC [33,34]was an RCT, while one study of the South African CSG used a prospective cohort, with one analysis using propensity score matching to adjust for selection bias [30,31], while the other study of the South African CSG was a cross-sectional design [32].
All five social protection cash transfer evaluations focused on child outcomes (under 18) reported on HIV risk behaviours; none reported on under-18s' experiences of IPV. All analyses of the South African CSG reported positive outcomes for children. Cluver et al. [30] reported a significant reduction in transactional sex and age-disparate sex for girls, but no impact for boys, amongst households receiving the CSG. Similarly cross-sectional analysis of the CSG by UNICEF reported a significant reduction in sexual activity, delayed sexual debut and reduced pregnancies for both girls and boys whose families received the CSG [32]. Cluver and colleagues [31] undertook a separate analysis of cash combined with care (defined as either a supportive teacher or positive parenting at home), and found significant reductions in HIV risk behaviour for girls and boys, where families received the CSG. In Kenya, Handa et al. [33] reported a significant delay of sexual debut for children receiving cash; however, Rosenberg et al. [34] reported no impact of the transfer on transactional sex or relative partner age. They do suggest that there were marginal reductions in transactional sex if the girl was younger or in school [34].
(4) Adult outcomes Five separate analyses of four different broad-based social protection interventions were identified; all reported adult women's outcomes. All focused exclusively on women's self-reported experiences of IPV, without consideration of HIV outcomes, with four from Latin America and one from Kenya. Two analyses focused on Mexico's Oportunidades, which is a conditional cash transfer linked to accessing health and education services [37,38]; one focused on Bono de Desarrollo Humano (B.D.H), a state-run unconditional cash transfer in Ecuador [35]; and one focused on a humanitarian relief intervention in Ecuador, which operated for six months [75]. The final study was of an unconditional cash transfer called 'GiveDirectly' run by an NGO in rural Kenya, which provided either a lump sum or regular transfers of a smaller amount over nine months [39]. Study quality was generally high with randomised designs in Kenya and Ecuador [35,36,39]. Two studies, both from Mexico, were cross-sectional in design [37,38].
All analyses focused on women's self-reported experience of IPV. Two studies reported reductions of physical and sexual IPV. In Kenya the GiveDirectly intervention showed a 30-50% reduction in various forms of physical violence and a 50% reduction in rape (p < 0.1) and 60% reduction in other forms of sexual violence (p < 0.05) [39]. Similarly, in Ecuador, the short-term transfer intervention showed reductions in controlling behaviours and women's experiences of IPV [75].
In contrast, other studies reported mixed results. In Mexico, one analysis of Oportunidades showed mixed outcomes for women receiving a cash transfer: while it reduced physical IPV by 40% (p < 0.01), it increased women's experiences of emotional violence by 3 to 5 percentage points, but this change was not statistically significant [35]. In the BDH evaluation no impact was seen on reducing violence, while an increase in male controlling behaviours was reported (p < 0.05) [35]. Another study showed no long-term impact on IPV; however, analyses suggested this was because women could leave abusive partners because of receipt of the transfer [38].
(5) Cash transfers as incentive for behaviour change Three cash transfer interventions were based in behavioural economics, aiming to shift health behaviours and thus HIV-or STI-status. Two studies were conducted in sub-Saharan Africa (SSA) [40,41] and one in the USA [42]. In general study design was rigorous, using RCTs in two studies [40,41] and a quasi- experimental design in the other [42]. In Tanzania remaining STI-free was incentivised with cash and assessed through testing every 4 months over a 12month period [40]. In Malawi a study focused on incentivising maintaining HIV status over 12 months for women and men (if HIV-negative at start, maintaining this; if HIV-positive, they automatically received the cash transfer at the end) [41]. Finally, in the USA, a study provided cash transfers on progression through a series of health checks and training programmes [42]. All three studies focused on women and men, 16 years old and older.
There was mixed evidence on the impact of incentivising HIV-status or STI-status. In Tanzania, adjusted analysis showed significant impacts of high-value incentives on remaining STIfree [40]. In Malawi, however, there was no impact on maintaining HIV-status [41]. But, when women and men were given the incentive, HIV risk behaviours reduced for women and increased for men [41]. In the USA, the combination of incentives linked to accessing services and progression in training programmes reduced the likelihood that participants had sex [42].
(1) Economic strengthening interventions Eleven studies sought to assess the impact of economic strengthening interventions on IPV and HIV-prevention (Table 2). Ten were from Bangladesh and one from South Africa. All studies focused on the impact of microfinance and used cross-sectional study designs. Given study designswith many looking at population-level datait is difficult to untangle whether similar interventions were assessed. In Bangladesh Bajracharya [44] used propensity score matching on cross-sectional data to overcome bias. All studies included measures of IPV experienced by women; one study also included measures of HIV risk behaviours [53].
Overall there was no clear evidence on the impact of microfinance on women's experiences of IPV. Two studiesone from Bangladesh and one from South Africasuggested women's participation in microfinance had no impact on IPV [44,53], while two studies from Bangladesh suggested involvement in microfinance reduced IPV [50,52], with three others suggesting a potential initial increase in women's experience of IPV, and a reduction in risk over a longer time period [43,45,51]. In contrast, the other four studies suggested involvement in microfinance increased IPV, particularly under specific circumstances, including living in a conservative area [48], being wealthier [46,47,49] and residing in urban areas [49], suggesting a significant role was played by contextual factors. The only study which looked at HIV risk behaviours cross-sectionally in South Africa found no impact on these [53].
(1) Economic strengthening and gender transformative We identified 21 interventions, each with their own separate analysis, which combined economic strengthening interventions with gender transformative components, primarily group-based discussions, and couples interventions (Table 3). Three main types of economic interventions were identified: microfinance/Village Savings and Loans Association (VSLA), formal savings, and vocational training. We report under each of these types separately.
In general, despite variations in reported outcomes, there was a trend towards positive outcomes. For HIV-related outcomes, six reported significant improvements in condoms use [23,58,60,[62][63][64], while two reported non-significant improvements [59,77] and one reported no change [56]. Other changes in reported HIV risks included a reduction in STIs [62] and a reduction in the number of sexual partners [63,64], all in sex worker populations.
(1) Savings and gender transformative Only one discrete intervention was identified that reported a behavioural outcome in two studies [65,66]. This focused on young women and included a pilot from two countries (Kenya and Uganda) using a pre-test, post-test design [66] and a subsequent larger study from Uganda, with three arms: full treatment, economic-only treatment and comparison group [65]. The full intervention provided safe social spaces, reproductive health training, financial education and opened a savings account for children [65,66]. Only IPV outcomes data were reported.
The outcomes of this saving and gender transformative intervention were mixed. In the pilot there were limited effects seen in Kenya, but in Uganda there was a significant decrease in being indecently touched by anyone (whether a partner or non-partner) in the past six months [66]. In the subsequent controlled study, the full intervention showed no impact on sexual violence; however, in the economic-only arm, there was a significant increase in indecent touching by anyone in the past six months (p < 0.01) [65].
(1) Vocational training and gender transformative Eight interventions provided vocational training combined with gender transformative training [59,[68][69][70][71][72][73][74]. Four were in SSA, two in Asia and two in the USA. One study was an RCT, although randomisation occurred at the individual, rather than group, level [67]. All others were quasi-experimental, although some did include elements of randomisation. All eight reported on a variety of HIV-related outcomes, however only three reported on IPV as well [67,68,73].
There were significant variations in HIV-related outcomes. Seven studies reported positive outcomes in terms of HIV-related ones, although the specific impact varied. In three studies condom use improved significantly [67,72,73]. In five studies, there were significant changes in other types of sexual behaviour or partner types, suggesting a reduction in HIV risk behaviour [59,68,70,71,74]. Only one study reported entirely no impact on behavioural measures related to HIV risk [69].
In the three studies reporting IPV, all reported reductions in IPV. Dunbar et al. [67] reported a non-significant reduction in IPV (IOR = 0.10 vs COR = 0.63; p = 0.06) for SHAZ!, while Jewkes et al. [68] reported a significant reduction in any sexual IPV experienced by women (37%; p = 0.033), but did not see a reduction in men's perpetration of IPV for Stepping Stones and Creating Futures. The MEN Count intervention had too small a sample at baseline to report IPV trends, but suggested there was a non-significant reduction [73].

Discussion
This is the first study to undertake a comprehensive review of any types of economic intervention which report either an HIV-related and/or IPV behavioural outcome. We identified 45 separate analyses of interventions that sought to use economic interventions (either on their own or in combination with other interventions) to prevent HIV risk behaviours and/or IPV. Interventions varied substantially in how they conceptualised the role of economics in preventing HIV or IPV, in study design, and in measurement. As such there was too much heterogeneity to undertake a meta-analysis of outcomes. However, the broad scope of the review enables some assessments of the field of interventions to be made.
The provision of cash through broad-based social protection programmes appears to have broadly positive outcomes for children targeted through these interventions. In the five separate analyses that looked at the impact of providing cash transfers to families with children, the results were positive in four of the studies in reducing HIV risk behaviours of children [30][31][32][33] and showed no impact in the fifth [34]. This is highly suggestive that, following a range of other reviews, the provision of broad-based social protection mechanisms as an approach to improving child health has multiple and overlapping positive outcomes [26,78]. However, no study looked at the potential impact of social protection on reducing children's vulnerability to IPV as they moved into adolescence.
For adults, however, interventions that solely strengthened economic well-being, through either cash transfers, or involvement of adults in economic strengthening activities such as microfinance or VSLA approaches, showed mixed outcomes, with studies reporting increases, decreases and no impact on HIV risk behaviours and IPV. The 'mixed' findings of these interventions can be explained in a number of ways. First, some studies have emphasised that as women gain economic autonomy and power in relationships more generally, they may face a 'male-backlash' [79], as men start to feel their 'authority' is challenged. In India, a prospective cohort study of married women found that those who secured work in the study period were more likely to experience IPV than those who remained unemployed [80]. Second, studies reviewed highlighted that the impact of economic strengthening interventions on IPV may be contextually specific, with contextual factors including urban or rural residence [49] and whether the community was more liberal or conservative [48] shaping outcomes. Similarly, Heise and Kotsadam [81] in their crosscountry analysis highlight that women's work is protective for IPV in contexts where many women work, but increases IPV-vulnerability if few women work.
In contrast, the interventions which combined economic and gender transformative interventions showed positive or flat results, and no negative findings. This highlights the fact that women's experiences of HIV-and IPV-vulnerability are often shaped at the intersection of gender inequalities and economic marginalisation [79] and that successfully working to reduce these risks needs to combine economic and gender transformative interventions [82].
The review highlighted a lack of evidence around effective interventions for a number of specific populations: in particular female adolescents, female sex workers and men. In general, interventions tended to focus on adult women, with only 14 out of the 45 analyses including women under the age of 18 (of which 5 were child outcomes for broad-based social transfer programmes). The lack of inclusion of women and girls under the age of 18 has been noted in wider reviews of women's economic empowerment interventions [83]. The lack of inclusion of younger women emerges from a confluence of factors including lack of recognition of their vulnerability to HIV and IPV, ethical challenges of working with younger people, and resistance to seeing adolescent girls as autonomous actors.
Only three interventions specifically focused on female sex workers, despite global evidence on the overwhelming burden of HIV and violence in this population [84,85]. Overall, sex worker interventions showed positive results, and were often embedded in wider sex worker mobilisation interventions, rather than being 'standalone' interventions. The potential of wider structural and social change through community mobilisation, as with sex worker mobilisation, is an important avenue for further consideration. In addition, there remains an ongoing debate about definitions of sex work and transactional sex, especially for women under 18, where women who do not identify as sex workers may be excluded from sex worker mobilisation and interventions.
The lack of research around interventions for these two populations (women under 18 and female sex workers) is critical not only because of the high burden of HIV and IPV they face, but also because it is likely they face specific programming challenges. Indeed, the first generation of interventions with adolescent girls and young women showed many of the challenges, with flat or underwhelming outcomes [59,61]. Yet as lessons have been learnt and interventions have been modified, programmes with adolescent girls are increasingly showing positive outcomes [67,68]. Similarly, a review by the Global Network of Sex Work Projects of economic interventions for female sex workers in Africa highlighted the importance of involving sex workers in the design and implementation of projects [86], suggesting that there needs to be significant investment in developing and evaluating economic interventions for sex workers and that there is no simple transferability of interventions from one setting to another.
There was also a lack of focus on the potential and pitfalls of including men in economic and gender transformative interventions. Only 11 interventions out of 41 directly targeted men as recipients of the intervention and/or evaluated the impact on them. That there is little focus on including men in evaluations is a challenge for three reasons. First, men are already included in largescale public works programmes across the world that seek to build their economic livelihoods and little is known about the impact of involving these men on HIV risk behaviours and IPV perpetration. Second, studies working with men for gender equality continually raise the challenge of poverty and the barriers this causes for transformation of masculinities [87] and as such, there may be significant benefit in including men in combination gender transformative and economic strengthening interventions. Third, given the potential for a 'male-backlash' around women's involvement in economic strengthening interventions, understanding men's responses remains critical. Further research and theorising need to be done to understand how working with men on strengthening livelihoods can be done in a way that promotes gender equality and does not lead to the exclusion of women [88].
The review also highlights missed opportunities in understanding the impacts of interventions at the intersections of HIV and IPV; only 6 out of 45 studies measured both HIV risk behaviours and IPV outcomes. Given the clear evidence of how IPV and HIV have common risk factors, the failure to include both sets of measures is a major evidence gap. Moreover, studies also point to the need to include a wider range of measures. For instance some studies pointed to the potential of economic-based interventions to either increase or decrease male controlling behaviours [35,36,68], an important factor for HIV acquisition and IPV [89], but this measure was not common across many studies. In addition, economic violence was only measured in a limited number of studies [55], yet given the underlying aim of some economic interventions is to increase women's economic autonomy in relation to male partners, this is similarly a missed opportunity.
This review has a number of limitations. There is a widely known publication bias whereby positive results are much more likely to be published [90], which while somewhat mitigated by the inclusion of grey material, is unlikely to have been totally overcome. Additionally, as any quantitative evaluation was included, irrespective of the quality of the study design, comparing different studies to each other may misrepresent interventions' true effectiveness. Finally, as the review was not a full systematic review following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, there are likely to be differences between this review and a PRISMA-guided one.
This comprehensive review of economic interventions highlights the large evidence base that exists around reducing HIV risk and IPV globally through tackling poverty through a variety of interventions. It is highly suggestive that broad-based cash transfer interventions have widespread positive benefits for women who receive them, as well as their children targeted. In addition, it emphasises the positive outcomes of combining economic strengthening and gender transformative interventions. However, it also highlights the need for further research on this topic, including research on specific populations, female adolescents, female sex workers and men, if a full understanding of the benefits of these interventions is to be achieved.

Acknowledgments
Thanks are due to Charlotte Watts (LSHTM, DfID) for input on the original research and report on which this paper is based.

Author contributions
AG, AKW and JJ conceptualised the study. AG conducted the study and wrote the first draft. AKW and JJ contributed sections, revised sections for intellectual content and edited the paper. All authors approved the paper for publication.

Disclosure statement
No potential conflict of interest was reported by the authors.

Ethics and consent
Not required.

Funding information
This document is an output from What Works to Prevent Violence? A Global Programme, which is funded by UK Aid from the UK Department for International Development (DFID) for the benefit of developing countries. A version of this paper was presented at the Greentree II Consultation (May 2015), a high-level meeting in New York, USA on GBV and HIV, with support by UK Aid from the DfID, through the STRIVE research consortium. However, the views expressed and information contained in it are not necessarily those of or endorsed by DFID, which can accept no responsibility for such views or information or for any reliance placed on them.

Paper context
Intimate partner violence and HIV are co-occuring global epidemics. Reviews have not explored the impact of economic interventions on both of these. We undertook a comprehensive review of these interventions. The review highlights that economic interventions combined with gender transformative components have stronger outcomes than economic interventions alone; there is a lack of meaningful inclusion of sex workers, young women and men in evaluations; and there are missed opportunities for measuring HIV risk and intimate partner violence.