WUNRN

http://www.wunrn.com

 

http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(14)61703-7/fulltext

  Full Text may ask for Log In. Full Text includes research References.

http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(14)61703-7/abstract

 

Prevention of Violence Against Women & Girls: What Does the Evidence Say?

Prof Mary Ellsberg, PhD

Diana J Arango, MSC, 

Matthew Morton, PhD, 

Floriza Gennari, MPH,

Sveinung Kiplesund, MPACS, 

Manuel Contreras, PhD, 

Prof Charlotte Watts, PhD

 

Published Online: 20 November 2014

 

Summary

In this Series paper, we review evidence for interventions to reduce the prevalence and incidence of violence against women and girls. Our reviewed studies cover a broad range of intervention models, and many forms of violence—ie, intimate partner violence, non-partner sexual assault, female genital mutilation, and child marriage. Evidence is highly skewed towards that from studies from high-income countries, with these evaluations mainly focusing on responses to violence. This evidence suggests that women-centred, advocacy, and home-visitation programmes can reduce a woman's risk of further victimisation, with less conclusive evidence for the preventive effect of programmes for perpetrators. In low-income and middle-income countries, there is a greater research focus on violence prevention, with promising evidence on the effect of group training for women and men, community mobilisation interventions, and combined livelihood and training interventions for women. Despite shortcomings in the evidence base, several studies show large effects in programmatic timeframes. Across different forms of violence, effective programmes are commonly participatory, engage multiple stakeholders, support critical discussion about gender relationships and the acceptability of violence, and support greater communication and shared decision making among family members, as well as non-violent behaviour. Further investment in intervention design and assessment is needed to address evidence gaps.

Introduction

Violence against women and girls is a global human rights violation and a substantial development challenge. It affects women throughout the world, and crosses cultural and economic boundaries. WHO estimates that more than 30% of women worldwide have experienced either physical or sexual partner violence.^1, 2 7% of women worldwide have experienced non-partner sexual assault.³ About 100–140 million girls and women worldwide have undergone female genital mutilation (FGM) and more than 3 million girls are at risk for FGM every year in Africa alone.⁴ Nearly 70 million girls worldwide have been married before the age of 18 years, many of them against their will.^5, 6 The effect of violence against women and girls on their health and welfare, their families, and communities is substantial.^7, 8, 9 The costs of violence against women and girls, both direct and indirect, are a staggering burden for households and economies.¹⁰

In the past 20 years, much research has been dedicated to the extent of violence against women and girls and understanding the underlying causes and risk factors associated with violence perpetration and victimisation.¹¹ There has also been enormous growth in the quantity and breadth of interventions in diverse settings, including in health care, justice systems, and social campaigns to address violence against women and girls worldwide. The first generation of interventions mainly focused on provision of support services for survivors of violence, and sought to reduce perpetrators' impunity and increase the effectiveness of the justice system. A second generation of programming, mainly in low-income and middle-income countries, has had a greater focus on violence prevention. These interventions developed organically, often linked to HIV prevention efforts, and have used many approaches. These include large-scale campaigns, sophisticated education-entertainment or so-called edutainment programmes, skills building and economic empowerment programming, community mobilisation, and participatory group education efforts, aiming to change attitudes and norms that support violence against women and girls, empowering women and girls economically and socially, and promoting non-violent, gender-equitable, behaviours. Not much research has been done to assess the effectiveness of these programmatic efforts, particularly in low-income and middle income countries.^12, 13 Despite the scarcity of empirical research, a small, but promising, body of evidence shows either significant or highly promising positive effects in reductions or prevention of violence against women and girls.

Key messages

·         •

Evidence for interventions is highly skewed towards high-income countries, and response, rather than prevention. Most research has been done in intimate partner violence, with far less evidence on how to prevent other forms of violence.

·         •

In high-income countries, response interventions have shown greater success in improvements in physical and mental health outcomes for survivors of violence and increased use of services, but evidence for their effectiveness to reduce revictimisation is weak. Much research has been done on interventions for perpetrators, with little evidence of effectiveness.

·         •

In low-income and middle-income countries, there is increasing emphasis on prevention of different forms of violence against women and girls, including intimate partner violence, non-partner sexual assault, female genital mutilation, and child marriage. Assessments of programmes indicate that it is possible to prevent violence, with some interventions achieving large effects in programmatic timeframes. Successful programmes engage multiple stakeholders with multiple approaches, aim to address underlying risk factors for violence including social norms that condone violence and gender inequality, and support the development of non-violent behaviours.

·         •

The specialty of violence prevention is at an early stage. Further investment is needed to expand the evidence base for what interventions are effective in different contexts, assess a broader range of intervention models, and explore issues of intervention cost, sustainability, and scalability.

In this Series paper, we review available evidence for what works to reduce the prevalence and incidence of violence against women and girls (panel). The studies cover a range of interventions, and many forms of violence against women and girls, ranging from violence in armed conflict and intimate partner relationships, to FGM and child marriage. We used a broad focus to allow cross-learning across interventions and types of violence.

Panel

Systematic review of reviews on interventions to reduce violence against women and girls¹⁴

We used the results of a 2014 systematic review of reviews¹⁴ to identify assessments of interventions to reduce all forms of violence against women and girls. The review of reviews identified 58 reviews and 84 rigorously evaluated interventions (using experimental or quasi-experimental methods) that aimed to reduce one or more forms of violence against women. We examined the studies identified in the review of reviews and extracted relevant information including sample size, outcomes, and effect sizes. From these, we identified 21 studies with significantly positive results. We also searched relevant electronic databases and supplemental sources (search terms available in the appendix) and did outreach to more than 30 experts in the specialty to identify recently published and unpublished studies that had not been identified through the review of reviews. Through this process, we identified six more rigorously evaluated studies with significantly positive or highly promising results. Our Series paper summarises the findings from more than 100 reviews and evaluations.

From the systematic review of reviews,¹⁴ evidence for effective interventions was highly skewed towards high-income countries. More than 80% of the rigorous evaluations were done in six high-income countries (Australia, Canada, Hong Kong, New Zealand, the UK, and the USA), comprising 6% of the world's population. The USA alone accounted for two thirds of all the intervention studies. The search strategy included all forms of violence against women and girls mentioned in the definition established by the UN General Assembly (1993),¹⁵including child and forced marriage, child sexual abuse, female genital mutilation, femicide, intimate partner violence, non-partner sexual assault, and trafficking. However, rigorous intervention evaluations were only identified for four types of violence: intimate partner violence, non-partner sexual assault, female genital mutilation, and child marriage.

Intimate partner violence was the subject of more than two thirds (58 of 84) of the rigorously evaluated interventions, followed by non-partner sexual assault with 17 studies and nine studies addressing harmful traditional practices (either female genital mutilation or child marriage). Only one study addressed multiple forms of violence (intimate partner violence and female genital mutilation). No studies meeting our inclusion criteria were related to trafficking or child sexual abuse. Among the interventions to prevent non-partner sexual assault, most were implemented with college students; no studies addressed sexual violence in conflict settings.

The types of violence against women and girls studied varied according to geographic location. In high-income countries, most of studies (51 of 66) dealt with intimate partner violence, followed by non-partner sexual assault with 15 studies. By contrast, half of the studies in low-income and middle-income countries (nine of 18) addressed child marriage or female genital mutilation, followed by intimate partner violence (seven), with non-partner sexual assault and multiple types of violence each represented by one study.

Among the 84 studies with available data, about two thirds (52) focused on responses to violence against women and girls at the individual level, and the remaining 32 interventions focused on prevention at the community or group level. Interestingly, the proportion of studies focusing on prevention was much lower in high-income countries (16 of 66) compared with low-income and middle-income countries where nearly all of the studies (16 of 18) focused on prevention. Most of the interventions targeted women alone (38) or women and men (17). 22 studies targeted only men, most of which were interventions for men who assault women (18).

When we synthesise the findings, we use the terms prevention and primary prevention to refer to interventions that work with individuals or communities irrespective of their history of violence. These interventions seek both to prevent violence from occurring in individuals who have not experienced it before and to reduce reoccurrence in those who have already experienced or used violence. We use the term response and secondary prevention interchangeably to refer to interventions that specifically target either women who have already experienced some form of violence or male perpetrators, with the aim of reducing revictimisation or recidivism.

Intervention evidence from high-income countries

Introduction

In practise, although reduction of some form of violence against women and girls was a stated aim of all of the studies identified through the systematic review of reviews (panel),¹⁴ most studies identified from high-income countries focused on responses to violence. We also identified evidence from prevention programmes for school and university sexual violence.

Women-centered interventions for survivors of violence

We reviewed 22 rigorously evaluated interventions that provided services to women who experienced intimate partner violence.^{16, 17, 18, 19} These interventions, often referred to as women-centred, use a combination of strategies, including psychosocial support, advocacy and counselling, and home visitation to provide women with resources and support to reduce their future risk of violence, and to improve their physical and psychological health and wellbeing. Most of the interventions take place in health-care services such as family planning or antenatal care, in which women with histories of intimate partner violence are identified through routine inquiry.^{19, 20, 21, 22}Basic psychosocial support by health providers usually includes danger assessments, safety planning, information about rights and available resources, and referral to specialised services.

As described by García Moreno and colleagues²³ in the second paper in this Series, there is evidence that some health-sector-based interventions can have some positive outcomes for women and their children such as reductions in depression.^{16, 24, 25, 26, 27} However, only two studies report significant decreases in violence. Randomised control trials done in Washington, DC, and Hong Kong in pregnant women with histories of intimate partner violence showed significantly lower rates of violence revictimisation among women who received a psychosocial intervention, compared with women in control groups.^28, 29 Two other intervention models, involving advocacy and home visitation interventions, have also had promising results to reduce intimate partner violence victimisation.^30, 31, 32 These interventions include psychosocial support and the provision of additional assistance by a trained layperson, to help women identify and access services. Usually, these studies have a longer duration and greater intensity than have health services-based interventions alone. For example, Hawaii's Healthy Start Programme was designed mainly to prevent child abuse and neglect and to promote child health and development in newborn babies from families at risk of poor child outcomes. A 3 year follow-up study showed lower rates of intimate partner violence victimisation in mothers given the intervention compared with controls (appendix).^33, 34

Interventions for perpetrators

Although several high-income countries have implemented extensive court-mandated programmes to reduce recidivism in male perpetrators, there is little evidence of programme effectiveness. Of 18 rigorous studies identified through Arango and colleagues'¹⁴ systematic review of reviews (panel), only two studies reported any significantly positive results (appendix).^14, 35, 36 Interventions for men who assault their female partners typically involve some type of group education lasting from 8 weeks to 24 weeks. Common approaches include the Duluth Model, a feminist approach that engages men in discussions around power and control, as well as cognitive behavioural therapy and anger management, both of which mainly focus on the use of violence itself, rather than on underlying beliefs.³⁷ Some newer approaches have also been tested, such as combining these interventions with substance abuse programmes, couples therapy, or culturally adapted programmes for specific populations. The findings from these studies have been inconclusive.^38, 39

Reports about interventions for men who assault their female partners indicate a general decrease in recidivism in men who complete the full training. However, there are important methodological weaknesses in the available evidence base. Most studies reviewed the histories of men who were court-mandated to such treatment as a result of a domestic violence arrest, and compare recidivism (measured either as new arrests, or spousal reports of violence) among men completing the programme to men who dropped out or never attended at all. Overall, these programmes have very high dropout rates, with few consequences for failure to complete the programme. Since men who drop out are likely to be less motivated to change than are those who complete the programme, it is not possible to identify how much of the change can be attributed to the intervention itself.^{39, 40, 41,42}

School-based interventions

Most prevention programmes for intimate partner violence and non-partner sexual assault in high-income countries are school-based group training interventions. Evidence from these programmes has not been encouraging, but there have been a few exceptions. The Healthy Relationships programme in Canada was tested in two settings: one with male and female high school students and the other in the community with male and female at-risk young people. Both studies showed significant reductions in both perpetration and victimisation of dating violence in both boys and girls in the intervention groups compared with the control groups (appendix).^43, 44 Studies of two well known interventions, Shifting Boundaries and Safe Dates, reported a reduction in dating violence in adolescents. Neither investigators reported results separately by sex of the victim or perpetrator, and so it is not clear whether the effect was similar for boys and girls.^45, 46

Only two of 17 rigorously assessed school-based interventions to reduce non-partner sexual assault had significantly positive results.^14, 47, 48 Both were done in the USA, in female college students, and focused on sexual assaults by acquaintances or so-called date rape. It is not yet clear whether these programmes could be meaningfully applied to other settings or populations.^49, 50, 51 Some of the interventions with null findings were very brief (for example, a 1 h educational session), which likely contributed to the absence of positive findings.

High-level policy commitment and legislative reform

Although many of the programme evaluations described above did not show reductions in violence against women and girls during the relatively short periods of study follow-up, the potential cumulative effect of these interventions should not be overlooked. According to the US Bureau of Justice, the rate of intimate partner violence in the USA fell by 53% between 1993 and 2008 and the number of intimate partner homicides of women decreased by 26%. Many experts attribute this decline to the Violence against Women Act (VAWA), first authorised by Congress in 1994, which provides funding for many of the programmes mentioned above.⁵² The Act originally authorised US$1·6 billion in funding in 5 years and has been reauthorised three more times since then. A study of more than 10 000 jurisdictions between 1996 and 2002 showed that jurisdictions that received VAWA grants had significant reductions in the numbers of sexual and aggravated assaults compared with jurisdictions that did not received VAWA grants.⁵³

Promising practices in low-income and middle-income countries

Legislative and justice sector responses

Until recently, programmes in low-income and middle-income counties to prevent violence against women and girls followed the tendency of those in high-income countries to focus mainly on increases in women's access to justice through better legislation and training of judges and police and to provide survivors of violence with coordinated emergency services. Although the number of countries with domestic violence legislation has grown exponentially as a result (from four to 76 between 1993 and 2013),⁵⁴ implementation is a serious problem. Most domestic violence laws are not accompanied by budget allocations and there is often resistance to the laws from male-dominated judiciary and police.^55, 56, 57

One of the most prominent public policies to address violence against women and girls in low-income and middle-income countries is the establishment of specialised police stations for women and girls, particularly in Latin America and south Asia. In Latin America, 13 countries have women's police stations, and in Brazil alone there are more than 300 such stations.⁵⁸ They vary a great deal according to the type of services they provide and the quality of these services. Although they have undoubtedly raised visibility around the issue of violence against women and girls, and have led to increased reporting of violence in some settings, there is little evidence as yet for effectiveness. Qualitative research suggests that training and improved legislation alone do not improve outcomes for women or reduce violence at a community level, and that system-wide changes are needed to improve the enforcement of laws.¹²

Health sector approaches and one-stop centres

As discussed in the second paper in this Series,²³ the health sector in low-income and middle-income countries has been slower to engage on the issue of violence against women and girls. One common approach has been the establishment of one-stop centres, which aim to provide comprehensive care for survivors of violence against women and girls. Many of the centres are located in hospitals, such as the Thuthuzela care centres in South Africa, the family support centres in Papua New Guinea, and the Malaysian one-stop centres.⁵⁹ In Latin America, they are frequently stand-alone centres run by women's rights activists, and, in some cases, by the national or municipal governments—eg, Ciudad Mujer (city of women) in El Salvador or the Centros Emergencia Mujer (women's emergency centres) in Peru. Most one-stop centres provide services for both intimate partner violence and sexual violence. However, in much of sub-Saharan Africa, the demand for sexual assault services and access to post-exposure prophylaxis to prevent HIV infection after rape has spurred the creation of post-rape care centres in many hospitals, which are not necessarily linked with services for intimate partner violence.⁶⁰ As with the women's police stations, there is enormous variation in the level of funding, accessibility, and quality of services provided, and little evidence exists for their effectiveness to reduce violence against women and girls or to mitigate the negative consequences for survivors.

Violence prevention programmes

There has been a much greater emphasis on violence prevention in low-income and middle-income countries. Many models of violence prevention emerged from HIV programming and the growing recognition that gender inequality and violence underpin many women's vulnerability to HIV. As we describe below, prevention programmes use a wide range of approaches, including group training, social communication, community mobilisation, and livelihood strategies. Most interventions use more than one approach, and many target underlying risk factors for violence, such as poverty, women's economic dependence on men, low education, and inequitable norms for male and female behaviour. Whereas women and girls were originally their focus, programmes are now also target men and boys or both men and women. Programmes are moving from trying to achieve change in groups of individuals to trying to achieve change at a community level.¹¹

Group-based training interventions to empower women and girls

Most violence prevention programmes in low-income and middle-income countries use participatory group training, which consists of a series of educational meetings or workshops with targeted groups of individuals. The goal of such programmes is not only to prevent violence against women and girls, but also to address underlying expectations about male and female roles and behaviour, and to support the development of new skills for communication and conflict resolution through a process of critical reflection, discussion, and practice. There is a wide range of training durations, target groups, and components. Violence against women and girls prevention components are often embedded in programmes that aim to improve sexual and reproductive health, or livelihood programmes such as microfinance or vocational training.

Two successful programmes in Uganda and Kenya sought to empower adolescent girls through training in life skills, self-defence, and vocational training (table 1).^61, 64 Findings from randomised control trials showed significant improvements in knowledge and behaviour in sexual and reproductive health in girls in the intervention group, and large reductions in coerced sex (in Kenya, sexual assaults decreased by 60% in girls in the intervention group compared with those in the control group). Training programmes for girls have also had some success in lowering rates of child marriage, although they are more likely to combine direct activities for girls with community level activities. Two programmes, one in India⁶³ and one in rural Ethiopia (the Berhane Hewan programme)⁶² used a comprehensive set of activities including intensive life skills training for unmarried girls, community conversations, mentorship, and community service activities to encourage parents to keep girls in school and to delay marriage. The Berhane Hewan programme also provided support for basic school supplies and an economic incentive (a goat) for families whose daughters were still unmarried by the end of the programme. Both programmes showed some success in delay of the age of marriage by 1 or more years. The programmes yielded additional benefits by addressing of several drivers of early marriage, resulting in increased knowledge and skills in the girls and changes in attitudes in the community towards child marriage.

Group training that targets men and boys

As presented by in the third paper in this Series by Jewkes and colleagues,⁷⁰ there is a diverse range of interventions involving boys and men in violence prevention, although the evidence of their effectiveness is still limited. One successful programme, Yaari Dosti, was carried out in two sites in India.⁶⁶ The intervention was based on programme H, which was developed in Brazil,⁷¹ and investigators aimed to reduce male-perpetrated violence against women and girls by transforming gender inequitable norms through group training and social communication programmes. Young men in the intervention groups in Mumbai and Gorakhpur were about five times and two times, respectively, less likely to report perpetration of physical or sexual partner violence in the previous 3 months than were participants in the comparison sites.

Other similar programmes targeting young men have been implemented globally, including the young men's initiative in the Balkans,⁷² Parivartan (targeting cricket coaches in India),⁷³ and the male norms initiative in Ethiopia.⁷⁴ Assessments of these interventions indicate promising outcomes in changes to young men's attitudes towards gender equality and the use of violence, but they did not result in significant behavioural changes. It is not clear why Yaari Dosti was more successful than the other interventions, but it could be related to the intensity and duration of the intervention, or that the other studies were underpowered. More research is needed to understand what elements of the interventions with men and boys are key to achieve behavioural changes.⁷⁵

Group training with men and women: synchronising gender approaches

In response to the increasing recognition that both men and women should be engaged in efforts to prevent violence against women and girls, more programmes are using gender synchronised approaches that intentionally reach out to both men and women in a coordinated way. Stepping Stones is a widely adapted programme that uses participatory learning approaches with both men and women to build knowledge, risk awareness, communication, and relationship skills around gender, violence, and HIV. A cluster randomised trial of young men and women in South Africa showed that at 2 years after the intervention, men's self-reported perpetration of physical and sexual intimate partner violence was significantly lower than were those from men in control villages. The programme also achieved a significant reduction in infections with herpes simplex virus 2 in both men and women. No differences were noted in women's reports of victimisation from intimate partner violence between the intervention and control villages.⁶⁵

Some prevention methods used in non-conflict settings are now being adapted to conflict and post-conflict settings. Two studies from Côte D'Ivoire looked at the incremental effect on intimate partner violence when gender dialogue groups were added to an economic empowerment group savings programme for women. One of the studies showed a reduction in physical intimate partner violence in couples who attended more than 75% of the meetings, whereas the second study showed improvements in men's attitudes towards violence but no significant behavioural changes.^76, 77

Community mobilisation

By contrast with group-training programmes, which seek to reduce violence in a targeted group of individuals, community mobilisation interventions aim to reduce violence at the population level through changes in public discourse, practices, and norms for gender and violence. Community mobilisation approaches are typically complex interventions that engage many stakeholders at different levels (eg, community men and women, youth, religious leaders, police, teachers, and political leaders). They use many strategies, from group training to public events, and advocacy campaigns such as the 16 Days of Activism Against Gender Violence (Nov 25–Dec 10).

The interventions often make use of social media, including mobile phone applications, such as Hollaback, Circle of Six, and Safetipin in India, to provide information about violence and neighbourhood safety, and to help women to report violence or to receive emergency help from friends and authorities.¹⁰ Community activists have partnered with innovative edutainment programmes such as Soul City, Sexto Sentido, and Bell Bajao, in the development of high-quality communication materials such as posters, street theatre, and radio and television programmes. Although there is no evidence that social communication programmes alone can prevent violence, rigorous assessments have shown significant changes in knowledge and use of services, attitudes towards gender, and acceptance of violence against women and girls, which can provide crucial support for local efforts.^{10, 78, 79, 80}

Because of their complexity, community mobilisation programmes are challenging to evaluate, and very few rigorous assessments have been done. As described in the fourth paper of this series by Michau and colleagues,⁸¹ a small cluster randomised trial of the SASA! programme in Kampala, Uganda, showed highly promising (although non-significant) results, by reducing community prevalence of physical partner violence by 54% (table 2).⁶⁷ A similar programme in Rakai, Uganda, showed not only reductions in physical and sexual partner violence, but also reduced incidence of HIV/AIDS.⁶⁹ This model is now being adapted in other settings throughout sub-Saharan Africa and in Haiti.

Community mobilisation approaches have also been used successfully to reduce FGM and child marriage. Use of the Tostan model,⁶⁸ developed in Senegal, has been replicated in several countries in sub-Saharan Africa, with community-based education programmes that address a range of issues, including health, literacy, and human rights. Through these programmes, villagers identify priority issues for community action, and both FGM and intimate partner violence emerged as key issues. In many cases, villages have taken pledges to renounce FGM and to encourage neighbouring villages to do the same. A quasi-experimental assessment of the programme in Senegal noted that women in the intervention villages reported significantly less violence in the preceding 12 months than did women in the comparison villages.⁶⁸ Also, mothers of girls aged 0–10 years less frequently reported that their daughters had undergone FGM in the intervention villages.⁸⁶

Economic empowerment

Studies around the world have consistently shown associations between intimate partner violence and poverty at both a household and community (correlated with country wealth) level^87, 88although the directionality and mechanisms for these associations are not clear. These findings have led some development practitioners to argue that increasing of women's economic opportunities should be a key strategy to reduce violence. However, the evidence for women's economic empowerment and its effect on violence is mixed, with research suggesting that increased access to credit and assets could either decrease or increase women's risk of intimate partner violence, depending on the context in which the women live.^89, 90, 91 Increased access to assets could reduce a woman's risk of violence in many ways; potentially allowing financial autonomy enabling women to leave a violent relationship. It could also increase a woman's value to the household, and increase a woman's relative bargaining power within the relationship. More broadly, reductions in household poverty could reduce economic stress and so reduce potential triggers for conflict.

To test whether adding a gender training and HIV prevention component to microfinance programmes for women could contribute to reductions in intimate partner violence, investigators for the IMAGE study⁸⁵ combined livelihood and empowerment strategies to address gender issues, HIV, and violence in women living in rural South Africa. The intervention combined microfinance with ten participatory training and skills-building sessions on HIV, cultural beliefs, communication, and violence. After 2 years, a cluster-randomised trial showed a 55% reduction in reports of physical or sexual partner violence from women, with economic assessments that suggested that the intervention is cost-effective.⁸⁵ IMAGE is being scaled up in South Africa and is being expanded to Tanzania and Peru.

Cash transfers

Although not designed to address violence against women and girls specifically, cash transfer programmes can contribute to reductions in both intimate partner violence and child marriage. Studies of unconditional cash transfer programmes in Kenya⁸³ and Ecuador⁸⁴ reported, in addition to large economic and nutritional benefits to households, significant reductions in rates of intimate partner violence in both settings (table 2). The study from Kenya noted that large transfers were associated with significant decreases in cortisol concentrations in both men and women, suggesting that the reduction in intimate partner violence might be partly due to drops in household stress. In Ecuador, the investigators reported that the transfers did not lead to increased decision-making power for women in the household, and concluded that the effect on intimate partner violence could be due to reduced stress.

Financial or material incentives have also been used with promising results to reduce child marriage. The incentives include school uniforms, livestock, or cash transfers.^62, 82 Usually, these incentives are conditional on the girl staying in school or staying unmarried until the age of 18 years, although a programme in Malawi showed promising results in keeping girls in school and delaying marriage through unconditional cash transfers.⁹² An innovative programme established in 1994 in the State of Haryana, India, used savings bonds as an incentive to encourage parents not to marry their daughters before they were aged 18 years. Preliminary findings from continuing assessment indicate that beneficiary girls have achieved higher educational attainment compared with non-beneficiaries (table 3).⁹³

Discussion

In view of evidence for the high prevalence and severe health outcomes of violence against women and girls, it is troubling that rigorous data for what works to prevent violence are still scarce. Available intervention research is highly skewed towards studies done in high-income countries, and it largely focuses on response rather than prevention. Our Series paper suggests that, despite the crucial value of provision of timely and appropriate services to survivors of violence, little evidence exists that such programmes alone can lead to significant reductions in violence against women and girls.

The evidence base is limited by several methodological weaknesses. Many of the studies had very small sample sizes (commonly with few clusters in randomised controlled trials). For this reason, some of the null findings reported probably result from underpowered studies rather than a definite absence of intervention effect. There is also a very wide range of outcome measurements and timeframes, which makes comparisons difficult. Of concern, many studies did not control for potential confounding factors, which might result in some bias in the results. Most of the assessments identified did not include a long follow-up period, if any, making it difficult to establish whether changes are sustained over time.

There are several areas in which the evidence base is small or non-existent. We found no rigorous assessments of interventions to prevent trafficking, and a few evaluations from humanitarian and emergency situations. Few assessments have been done in indigenous or ethnically diverse populations or in older populations. With a few exceptions, the evaluations in this review did not measure cost-effectiveness of interventions, which is a pivotal decision point for those who wish to implement and adapt an intervention, particularly in low-resource settings. There is little documentation on how interventions can be adapted to different settings.

Despite the shortcomings of the available evidence base, some promising trends emerge. Several studies show that it is possible to prevent violence against women and girls, and that large effect sizes can be achieved in programmatic timeframes. Multisectoral programmes that engage with multiple stakeholders seem to be the most successful to transform deeply entrenched attitudes and behaviours. Strong programmes not only challenge the acceptability of violence, but also address the underlying risk factors for violence including norms for gender dynamics, the acceptability of violence, and women's economic dependence on men. They also support the development of new skills, including those for communication and conflict resolution. Some of the studies showed potential benefits from integration of violence prevention into existing development platforms, such as microfinance, social protection, education, and health sector programming, which could allow scalability. Community mobilisation models also provide a means to achieve measurable community level effects. Importantly, there are several positive examples of impact from low-income and middle-income countries that could potentially be transferred to high-income countries.

Overall, the findings point to the imperative of greatly increasing investment in violence research and programme evaluation, particularly in low-income and middle-income countries. Alongside programmatic investment, it will remain important to support rigorous evaluations and guide international efforts to end violence against women and girls. As the specialty continues to develop, importance should be given to learning more about the costs of programmes and identification of models of intervention that can be delivered to scale.