Interventions in sexual and reproductive health services addressing violence against women in low-income and middle-income countries: a mixed-methods systematic review

Objectives To synthesise evidence on the effectiveness, cost-effectiveness and barriers to responding to violence against women (VAW) in sexual and reproductive health (SRH) services in low/middle-income countries (LMICs). Design Mixed-methods systematic review. Data sources Medline, Embase, Psycinfo, Cochrane, Cinahl, IMEMR, Web of Science, Popline, Lilacs, WHO RHL, ClinicalTrials.gov, Google, Google Scholar, websites of key organisations through December 2019. Eligibility criteria Studies of any design that evaluated VAW interventions in SRH services in LMICs. Data extraction and synthesis Concurrent narrative quantitative and thematic qualitative syntheses, integration through line of argument and mapping onto a logic model. Two reviewers extracted data and appraised quality. Results 26 studies of varied interventions using heterogeneous outcomes. Of ten interventions that strengthened health systems capacity to respond to VAW during routine SRH consultation, three reported no harm and reduction in some types of violence. Of nine interventions that strengthened health systems and communities’ capacity to respond to VAW, three reported conflicting effects on re-exposure to some types of VAW and mixed effect on SRH. The interventions increased identification of VAW but had no effect on the provision (75%–100%) and uptake (0.6%–53%) of referrals to VAW services. Of seven psychosocial interventions in addition to SRH consultation that strengthened women’s readiness to address VAW, four reduced re-exposure to some types of VAW and improved health. Factors that disrupted the pathway to better outcomes included accepting attitudes towards VAW, fear of consequences and limited readiness of the society, health systems and individuals. No study evaluated cost-effectiveness. Conclusions Some VAW interventions in SRH services reduced re-exposure to some types of VAW and improved some health outcomes in single studies. Future interventions should strengthen capacity to address VAW across health systems, communities and individual women. First-line support should be better tailored to women’s needs and expectations. PROSPERO registration number CRD42019137167.


INTRODUCTION
Violence against women (VAW) is a violation of global health and human rights. 1 The most common forms of VAW are intimate partner violence (IPV) and non-partner sexual violence (NPSV). One in three women worldwide have experienced physical and/or sexual violence, mostly by an intimate partner. VAW is more prevalent in low-and middle-income countries (LMICs). Exposure to VAW is associated with mental and physical health problems, including increased sexually transmitted infection (STI) and HIV, unplanned pregnancy and abortion, gynaecological conditions. 2 3 Although IPV against men is increasingly recognised within the context of both same sex and heterosexual relationships, understanding of male victimisation and its health consequences is still poorly understood 4 and there is a dearth of primary research on the healthcare response to male victims and perpetrators. 5 The healthcare system has a key role in preventing VAW because most women attend sexual and reproductive health (SRH) services at some point. 6 7 The main role of the healthcare system is to contribute towards secondary and tertiary prevention through early detection of VAW and mitigation of its impact which can prevent ill health and reoccurrence of violence. Healthcare providers (HCPs) are uniquely placed to identify victims/survivors, provide first-line support and clinical care, and connect them with other services. Healthcare systems can also contribute to primary prevention through early identification of children exposed to violence in the home and support to programmes like home visiting or early childhood development. 8 The capacity of healthcare systems to respond to VAW is defined as the cumulative availability and strength of the following building blocks from the Health Systems Wheel: (i) leadership and governance, (ii) multi-sectoral coordination, (iii) workforce development, (iv) healthcare delivery, (v) infrastructure, (vi) financing, (vii) monitoring and evaluation (WHO 2010). 9 10 The Health Systems Wheel 11 highlights key components that need to be in place to support individual HCPs and organisations to offer a comprehensive and client-centred response to VAW. It assumes that all elements of the health system -individual, organisational, contextual, and structural -impact on provision of response to VAW. The WHO guidelines for evidence-based health systems response to VAW adopted the Health Systems Wheel framework to recommend intervention activities across the health systems building blocks. 12 13 In LMICs, healthcare delivery for VAW has been implemented through integration at the level of individual HCPs, healthcare facility, and healthcare system. 14 Systematic reviews 15 16 and WHO guidelines 17 found scant evidence from LMICs on effectiveness of VAW interventions in healthcare. This study addresses the gap by answering two questions: (i) What is the evidence for effectiveness and cost-effectiveness of interventions in SRH services that address VAW? (ii) What are the barriers to the effectiveness? METHODS All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; We conducted concurrent quantitative and qualitative evidence syntheses with integration into a line of argument 18 and mapping onto a logic model. 19 The mixed-methods design allowed integration of diverse types of evidence to inform VAW research and intervention development in LMICs. Our analysis was informed by the WHO Health Systems Wheel framework for responding to VAW. 8 We defined the health systems capacity to respond to VAW as the cumulative availability and strength of the Health Systems Wheel building blocks. We looked at the capacity of the health systems at three levels: individual providers (e.g., attitudes, knowledge, confidence, behaviour, and practices), services and organisations (e.g., infrastructure, availability of supplies/medicines), community (attitudes, knowledge, practices). 13 We defined women's capacity to respond to VAW as their readiness and ability to seek help, disclose abuse, get referrals, and receive services. We followed the Cochrane 20 and PRISMA guidelines. 21 The study protocol was registered with PROSPERO (CRD42019137167).

Search strategy and selection criteria
We included primary intervention studies reported in any language with an English abstract published since 2005, the year of the first published evaluation of VAW interventions in SRH services (expert opinion from the study advisory group). We identified earlier studies through reference checking. We used terminology and definitions from WHO guidance on strengthening health systems to respond to VAW (Table 1). 13

Inclusion criteria Exclusion criteria Participants
Recipients of healthcare services -women of reproductive age (15-49 years old) AND/OR Healthcare providers -organisations (e.g., hospital, clinic, primary care centre, other service delivery points) or individuals (e.g., health care professional, community health worker, or any other person who is trained to deliver healthcare in their community). Studies which recruited only a subset of recipients or providers of healthcare services.
Female children and girls under 15 years old. While recognizing that pregnancies occur among young adolescents 10-14, most studies consider women aged 15-49 years as the main group using SRH services in LMICs. Male recipients of healthcare services.

Interventions
Any intervention addressing violence against women (VAW). These are complex interventions aimed to identify women affected by violence, provide first-line support, clinical care, and signpost, or refer to available community support services including specialist VAW services. Any definition of VAW, including any type of IPV, domestic violence and abuse, family violence or nonpartner sexual violence against a woman, including transgender women.

No intervention
Hypothetical intervention addressing VAW. We are synthesising evidence of interventions that have been tested. Female genital mutilation/cutting, trafficking. These type of VAW were addressed in recent systematic reviews. 'Honour'-based violence, forced marriage. There is an overlap between IPV, domestic violence and abuse and 'honour'-based violence and forced marriage. Therefore, we will capture relevant studies through including papers on IPV and domestic violence and abuse.

Comparators
Controlled studies: usual care, no VAW intervention, delayed VAW intervention, minimal intervention (e.g., information provision). Uncontrolled studies: group before the intervention. No control group.

Outcomes
Outcome is an event or measurement collected for participants in a study. Primary outcomes: any health outcomes for survivors of VAW (for example, re-exposure to VAW, sexual and reproductive health, mental health, physical healthy, quality of life), any harms, cost effectiveness of VAW interventions.

Inclusion criteria
Exclusion criteria Secondary outcomes: patient and provider health-related cognitive and emotional outcomes (for example, knowledge, attitudes, confidence, readiness); health-related behaviour and practices (for example, identification and disclosure of VAW, provision and uptake of referrals and SRH services). Phenomenon of interest: provider and recipient experiences of and views on VAW interventions.

Study type
Primary intervention studies of any design. Primary studies that used quantitative designs such as randomised controlled trials, controlled and uncontrolled before-after studies, interrupted time series studies, cross-sectional studies.
Primary studies that used qualitative designs such as ethnographic research, interview or focus-group based studies, case studies, process evaluations and mixed methods designs. We include these studies if they had used qualitative methods for data collection and analysis and reported quotes from participants. Mixed-methods studies.
Systematic reviews. We used systematic reviews to identify potentially eligible primary studies.  independently assessed eligibility. Disagreements were resolved through consensus or third opinion (LJB). NVL checked references and citations.

Data analysis
NVL adapted the Cochrane Effective Practice and Organisation of Care (EPOC) data extraction form. 22 We collated multiple reports from the same study and used the most detailed report as the primary source for extracting study results. The included studies were divided among reviewers who worked in pairs, one to extract data and another to check. The pairs reconciled data extraction through discussion. We extracted study details on setting, study design and aim, sample size, participants characteristics, intervention characteristics and theories, and outcomes relevant to our review questions. For each quantitative outcome, we extracted type of measure and effect estimates as reported in the primary study. If authors did not report intervention effects, we extracted the postintervention point estimate. If a follow-up measure was reported repeatedly, we extracted the latest measure. We judged intervention effectiveness by improvement in any primary or secondary outcome listed in the individual studies (Table 1). We used authors' interpretation of their findings based on statistical significance or 95% confidence intervals (CI) and categorised effect estimates as improvement, mixed effect, or null effect. We ascribed a mixed effect when one or more, but not all measures of the same outcome changed under the same intervention (e.g., reduction in physical and sexual but not psychological IPV, improvement in some coping behaviours but not in others). NVL asked corresponding authors to check data extraction forms for their studies and provide missing information; nine responded.
Reviewers assessed the quality of the primary studies as part of data extraction. For randomised controlled trials (RCTs), we used the Revised Cochrane risk-of-bias tool for randomized trials (RoB 2). 23 For quasi-experimental studies we adapted the criteria listed by the EPOC Group. 24 For qualitative studies we adapted the Critical Appraisal Skills Programme (CASP) Qualitative Checklist. 25 We did not exclude studies based on their methodological quality.
We summarised interventions by mapping them onto the Health Systems Wheel 8 and models of health system responses to VAW in LMICs. 14 It was not possible to conduct a meta-analysis of quantitative outcomes due to the heterogeneity of the interventions, the outcomes, and their measurement. We undertook a narrative quantitative synthesis 26 and thematic qualitative synthesis, 27 summarised quantitative and qualitative syntheses in tables, and integrated them through a line of argument 18 and mapping onto a process-oriented logic model. 19 Reviewers (NVL, MC, LJB) drafted the logic model by mapping primary and secondary outcomes in the hypothesised logical order of occurrence and refined it through three iterative cycles of revisions.
1. Direct effects result from intervention activities producing structural changes at service level and changes in health-related cognitive and emotional outcomes among HCPs and women indicating improvement in the health systems capacity and women readiness to respond to VAW.
2. Intermediate effects result from direct effects producing changes in health-related behaviour and practices indicating improvement in the health systems capacity and women readiness to respond to VAW. HCPs identify women affected by VAW and provide first-line support; women disclose VAW, use offered support, develop adaptive coping strategies.

Health outcomes result from intermediate effects producing changes in women health and safety
indicating improvement in their readiness to cope with VAW. Women use adaptive coping strategies and safety behaviours; these lead to reduction in re-exposure to VAW and better health.

PATIENT AND PUBLIC INVOLVEMENT
No patients or members of the public were involved in this study.

RESULTS
All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; Searches identified 6082 records, we assessed 313 full text reports and included 32 28

Quality appraisal
Most quantitative studies were at high risk of bias (online supplementary file 3). Of 12 RCTs, seven had high risk of bias from deviations in intervention adherence, 32

Types of interventions
All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; All interventions were complex healthcare interventions, 61 however only two 30 60 included components across all domains on the Health Systems Wheel 8 ( Table 2) and only four were theoretically informed. 32 34 50 59  1. Response to VAW during routine SRH consultation (n=10). These interventions aimed to strengthen health system capacity to respond to VAW through integrating identification and first-All rights reserved. No reuse allowed without permission.
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; line support into routine SRH consultations. 28 32 33 38 39 45 46 54 57 58 This comprised training for HCPs in VAW screening, basic psychosocial counselling, and linkage to VAW resources. Training aimed to improve HCP knowledge, attitudes, and practices on VAW. Identification and response by trained HCPs aimed to increase women's readiness to respond to VAW. Duration of the integrated SRH-VAW consultation ranged between 7 32 and 30 minutes. 38 46 2. Response to VAW during routine SRH consultation plus community engagement (n=9). These interventions aimed to strengthen health system capacity to respond to VAW across SRH service and surrounding communities. 30  3. Response to VAW in addition to routine SRH consultation (n=7). These interventions aimed to strengthen women's readiness to respond to VAW. 29 35 40 47 50 52 55 Study personnel screened women attending routine SRH services and delivered the interventions to self-selected women with experience of VAW. This comprised more intensive support through specialist psychosocial counselling, 29 35 47 50 coping skills training 52 55 and psychoeducation. 40 52 55 The average number of sessions was three (range 1 to 7) with each session lasting from 30 29 35 47 to 90 minutes. 40 52 55 Interventions were delivered face-to-face individually, 29 35 40 47 50 in a group, 55 and mixed format. 52

Interventions effects and outcomes
The logic model displays all outcomes of interest in the three intervention categories (Figure 2). The arrows illustrate the hypothesised flow of change from intervention activities through health-related direct and intermediate effects to health outcomes.
None of the primary studies reported outcomes at service level. Most studies that evaluated interventions that aimed to strengthen health system capacity to respond to VAW across SRH service and community measured direct and intermediate effects on HCP and women's knowledge, attitudes, and behaviour. In contrast, all studies that evaluated interventions that aimed to strengthen women's readiness to cope with VAW reported their health outcomes, but only a few looked at preceding changes in women's cognition, emotions, and behaviour ( Figure 2, Table 3). Note. HCP health-care providers, VAW violence against women, SRH sexual and reproductive health., RCT randomised controlled trial. Bold indicates studies that reported sample size calculation.
All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Intermediate effects on behaviour and practices. Changes in professional behaviour were measured through the rates of VAW screening, provision of referrals to support services, and post-rape care.
Changes in women's behaviour were measured through VAW disclosure, uptake of referrals and other services. The overall evidence was uncertain across all three intervention categories with RCTs and non-randomised evaluations reporting improved, mixed, and null effects.
Health outcomes. Only half the studies reported measures of health and re-exposure to VAW; two interventions reported no harm resulting from taking part in interventions, 32 38 and 10 led to some health improvement. The overall direction of effect on any outcomes of interest was towards improvement in the routine SRH-VAW consultation category and in the additional response category.
In contrast, most interventions in the plus community engagement category reported mixed or null effect on women's health and re-exposure to VAW. We found that although some interventions did not reduce re-exposure to VAW, none reported violence escalation. Of 10 studies that measured reexposure to VAW, six found a reduction, 29 32 34 40 46 47 two reductions in some violence types but not in others, 55 59 and two reported no change. 38 60 Of 26 studies, only two reported changes across all three domains of the logic model, one from the routine SRH-VAW consultation category 38 and one from the plus community engagement category. 60 Four evaluations of the additional response category reported changes across two domainsintermediate effects on behaviour and practices and women's health outcomes. 29 35 50 52 These six studies were consistent with our hypotheses. If intervention improved women's safety behaviour and use of support services, their health improved. 50 Mixed or null effect on HCP and women's cognitive and behavioural outcomes suggested some explanation for no change in re-exposure to VAW. 38 60 Contradicting direct and indirect effects and outcomes 29 52 indicated possible barriers on the pathway from intervention activities to outcomes.

Response to VAW during routine SRH consultation
Of ten evaluations, two RCTs 32 38 and one UBA 46 studies reported conflicting findings on re-exposure to VAW; none measured women's health ( Figure 2, Table 3, Supplementary file 4). These three interventions did not lead to escalation of violence. There was some evidence for the reduction in HIV-disclosure-related violence at up to 2-month follow-up 32 and risk of becoming a victim of femicide at 3-month follow-up 46 possibly through some improvement in HCP's and women's cognition and practice. Two RCTs reported that integrated HIV-IPV consultation caused no harm to women. 32 38 However, all studies were at high risk of bias.
All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; An RCT 38 and UBA 57 in the Kenyan hospital with on-site GBV centre reported convergent findings.
The UBA study of an integrated HIV-IPV consultation with assisted referral to GBV centre reported 8% IPV disclosure rate, 75% referrals provision and 40% uptake. 57 The RCT of an integrated HIV-IPV consultation with referral to in-service GBV specialist found increased rates of IPV screening but no effect on provision of referrals. The 29-minute integrated HIV-IPV consultation increased women's knowledge about VAW and IPV disclosure, but had no effect on their attitudes, readiness to address VAW, uptake of referrals and re-exposure to IPV. 38 Another RCT of a 7-minute integrated HIV-IPV consultation over the phone reported that 62% of women used a safety plan and 80% employed at least one safety strategy, however their use of SRH services and perceived risk and safety did not change. Despite no effect on women's behaviour, the trial reported a four-fold reduction in HIV-disclosure-related violence (OR 4.37; 95% CI 1.46-13.44). 35 One UBA found that a 30-minute integrated HIV-IPV consultation contributed towards a reduction in the risk of femicide (mean difference 3.2, SD 3.56; 95% CI 2.43-3.98). 32 Other randomised and non-randomised evaluations of varied one-off training for HCPs reported conflicting findings on their knowledge, attitudes, and readiness to address VAW. Non-randomised studies reported increased IPV screening rates, 39 low VAW disclosure rates, 45 and mixed effect on provision of post-rape care. 28 54 Qualitative evaluations confirmed that training increased HCP ability to respond to VAW during routine SRH consultations. 38 45 54 57 One evaluation of HCP training on post-rape care described a potential mechanism of impact on HCP negative attitudes by separating personal beliefs about victims from the provision of clinical care. 54 Women found that integrated HIV-IPV consultation improved their knowledge about IPV. They benefited from emotional support and felt empowered. 33 38 57 Response to VAW during routine SRH consultation plus community engagement Three cluster RCTs reported conflicting findings on women's SRH and re-exposure to VAW. The overall effect was uncertain ( Figure 2, Table 3, Supplementary file 5). 34 59 60 A Nigerian RCT at low risk of bias evaluated universal home visits that discussed DV and other risk factors with pregnant women and their spouses. The trial reported no effect on women's use of SRH services, reduction in the proportion who experienced physical DV (RD 0.064 (95% CI 0.045-0.084), and mixed effect on pregnancy and birth indicators. 34 A Tanzanian RCT at high risk of bias evaluated integrated HIV-VAW consultation, onsite and external referrals, community, and couple education. Intervention had a mixed effect on women's attitudes about VAW and gender roles, no effect on rates of enquiry and referrals, mixed effect on women's use of SRH services, and null effect on re-exposure to IPV (OR = 0.85, 95% CI 0.62-1.16). 60 A Ugandan RCT at high risk of bias evaluated integrated HIV-IPV consultation, onsite referral, and community education. The intervention had no effect on women's safety behaviour and null effect on SRH. Re-exposure to physical and sexual IPV reduced (relative All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; prevalence risk ratios (PRR) of 0.74 (95% CI 0.63-0.86), 0.75 (95% CI 0.62-0.90), respectively), but psychological IPV and HIV did not change. 59 Non-randomised studies reported more positive effects on HCP knowledge, attitudes, preparedness, 30 and provision and use of post-rape care. 31 42 53 They also reported high rates of IPV screening and provision of referrals and clinical care by HCPs vs low uptake of referrals and other services by women. 49 56 Qualitative evaluations confirmed that VAW training transformed HCP attitudes towards patients and their own work and improved their diagnostic and counselling skills. HCPs appreciated the intervention and expressed a willingness to continue VAW work. 30 49 56 57 Women felt empowered and supported by HCPs. 30 33 45 49 Community engagement raised awareness about SRH-VAW services. 56

Response to VAW in addition to routine SRH consultation
This intervention category had the most robust evidence from six RCTs 35 40 47 50 52 55 and one UBA study 29 (Figure 2, Table 3, Supplementary file 6). The studies reported conflicting results with more evidence for a reduction in re-exposure to VAW at up to 6-month post-intervention and improvement in health possibly through improvement in women's safety behaviour. There was some evidence that longer interventions produced better outcomes. 29 40 47 Psychosocial counselling. Three RCTs of counselling sessions for pregnant women with experience of VAW reported conflicting results with no evidence for a dose-response. 35 47 50 The overall effect was towards reduction in re-exposure to violence and improvement in health outcomes. An adequately  50 The embedded qualitative study confirmed that women felt empowered, supported, and valued by the counsellor. 51 In contrast, a Peruvian RCT (with some bias concerns) of a 30-minute counselling session with a resource card and external referral had no effect on women's safety behaviours, health, use of community resources. 35 A Kenyan RCT of up to three 30-35-minute counselling sessions with resource card, safety planning and external referral reduced depression (MD = 7.12; 95% CI 6.21-8.03) and re-exposure to IPV (MD = 13.51; 95% CI 9.99-17.02). 47 Similarly, an Indian UBA evaluation of two or more 30-45-minute psychosocial counselling sessions found that most women had increased awareness of and readiness to address VAW. Physical violence and health problems decreased. 29 Coping skills training. Two RCTs with high risk of bias evaluated more intensive training interventions and found mixed effects on behaviour and mixed and null effects on VAW and health. 52 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; https://doi.org/10.1101/2021.11.18.21266456 doi: medRxiv preprint 55 An Iranian RCT of four 90-minute group sessions reported a reduction in re-exposure to physical IPV (RR 0.78; 95% CI 0.63-0.83) and psychological IPV (RR 0.73; 95% CI 0.64-0.83), but null effect on sexual IPV (RR 0.87; 95% CI 0.69-1.09). 55 A South African RCT of seven 90-minute sessions reported null effect on coping behaviour, use of SRH services, post-traumatic stress disorder (PTSD) and HIV viral load among HIV positive women with a history of sexual violence. 52 However, the embedded qualitative evaluation found that training increased women's knowledge about VAW impact and improved their self-esteem, coping and communication skills. 44 Psychoeducation. An Iranian RCT with high risk of bias of four 90-minute sessions with pregnant women reported reduction in scores of IPV and psychological distress. 40 41

Cost effectiveness outcomes
No studies evaluated cost effectiveness of VAW interventions in SRH services. One study of an integrated HIV-IPV consultation paid HCPs $6 per day for identifying patients experiencing VAW and referring them to the on-site GBV clinic. 57 One evaluation of post-rape service improvement with community engagement reported resource costs. 42 43 Seven studies across all three intervention categories mentioned intervention costs but did not report actual data. 30 38 47 50 53 56 57

Barriers to intervention effects and outcomes
Supplementary file 7 summarises factors that women and HCPs perceived as barriers to intervention implementation and impact. We developed three analytical themes cross-cutting through individual, community, and system levels.

Acceptability of VAW. Four evaluations of interventions on response during routine SRH consultation
and response with community engagement described attitudes accepting violence and patriarchal gender norms as major barrier to behaviour change. 33 45 54 56 Fear of negative consequences. Eight studies across all three interventions categories identified fear of negative consequences as a barrier to identification, disclosure, and engagement in VAW interventions. 30  Limited readiness. Evaluations reported limited readiness for engaging in VAW interventions at system and individual levels. In evaluations of response to VAW during routine SRH consultation 28 33 38 45 57 and response with community engagement, 30 49 HCPs consistently mentioned chronic problems with staffing, inadequate funding, no private space, lack of support from leadership, and high demand for basic SRH services without the additional VAW work. Readiness gaps at system level included the lack of services to refer to, poor referral systems and untrained staff in other agencies. Screening identified many IPV-positive women and specialist services could not address the increased demand. 30 33 45 56 Across all intervention categories, HCPs and women described barriers at societal level that prevented women from accessing SRH services, using referrals, and participating in All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; psychosocial interventions. Work-related conflicts, no money for transport, and financial dependence on husband were mentioned most frequently. 38 44 45 53 Finally, two evaluations of response to VAW during routine SRH consultation explored reasons for low uptake of referrals to specialist services.
Some women had expectations that could not be met by the current services. Instead of referral, they wanted HCPs to talk to their partners about stopping the abuse. 33 Some women wanted to receive all SRH and VAW services on the same day. 38 57

DISCUSSION
We conducted a mixed-methods systematic review of studies from LMICs on the effectiveness and barriers to strengthening SRH services response to VAW. We grouped 26 heterogeneous complex interventions into three categories: (i) response to VAW during routine SRH consultation, (ii) response to VAW during routine SRH consultation plus community engagement, and (iii) response to VAW in addition to routine SRH consultation. We mapped outcomes on a process-oriented logic model illustrating the hypothesised changes from intervention through direct and intermediate effects on health-related cognition, emotions, and behaviour to health outcomes. We cannot conclude which intervention was the most effective in improving any of these effects and outcomes due to heterogeneity of the interventions and measures at varying time points. Overall, ten interventions did not escalate violence and two reported no harmful events. We found mixed effects on women's health and re-occurrence of VAW across all three categories, with studies reporting conflicting findings.
Evaluations of the varied responses to VAW during routine SRH consultation found reduction in HIVdisclosure-related IPV and potential risk of becoming a victim of femicide, but no effect on IPV in the past month. Some of these effects could be attributed to improvement in HCPs' readiness, screening, and provision of first-line support for VAW. For women, these effects could be attributed to increased knowledge about VAW and disclosure of violence. Response to VAW during routine SRH consultation plus community engagement had uncertain evidence with single studies reporting improvement, mixed effect, and no effect on re-exposure to violence and SRH possibly through some improvement in provision and use of SRH services. More intensive psychosocial interventions delivered to women with experience of VAW in addition to routine SRH consultation had the most robust evidence for reduction in re-occurrence of violence and improvement in health outcomes possibly through an improvement in safety behaviours. We identified individual, community, and system-level barriers that could disrupt the pathway from intervention activities to outcomes: (i) attitudes and social norms that accept and normalise violence, (ii) fear of negative consequences and (iii) limited readiness of individuals, health systems and society to address VAW. No studies reported cost-effectiveness analysis.

Strengths and limitations
All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; This review is methodologically strong. It involved a multidisciplinary team of researchers from LMICs and the UK with content and methodological expertise in health systems response to VAW and global health. We followed the Cochrane method and included studies of any design reported in peer-reviewed and grey literature in any language with English abstract. This comprehensive approach ensured inclusion of the most relevant studies from the field and reduced the potential for bias/errors. The uncertain evidence for the two SRH consultation-based intervention categories is consistent with other evidence for a healthcare response to VAW in LMICs 15 62 and to VAW among pregnant women. 16 63 64 The evidence for the effectiveness of longer psychosocial interventions as an addition to routine SRH consultations is in line with a recent meta-analysis which found that psychosocial interventions in healthcare settings and communities in LMICs led to a 25-27% reduction in IPV. 65 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; An important finding on the direct effects of interventions is that increasing awareness about VAW and relevant procedures often did not lead to a shift in judgemental attitudes towards victims, the major barrier to changes in professional and patient behaviour and practices. An exception was studies in the plus community engagement category, one reporting improvement in HCP attitudes and one partial shift in women's attitudes and gender norms. These findings can be explained by the community and societal roots of gender norms and attitudes of HCPs and women which are best addressed at community and societal levels. Our findings suggest that a shift in individual's attitudes potentially leading to behavioural change can be achieved through service-based plus communitybased education. This finding is consistent with a review of evidence on what works to prevent VAW in LMICs. The review found good evidence for community activism approaches to shift harmful gender attitudes, roles and social norms. 66 An important finding is that routine integrated SRH-VAW consultations with referral or signposting to VAW/GBV specialist or other services did not increase women's use of these services. This gap between availability and acceptability of referrals to other services has several explanations. Our themes of barriers to intervention impact suggested that HCP response may not have matched women's needs and expectations, or the VAW services were not accessible, or contextual factors prevented women from accessing them. This finding suggests that integrated SRH-VAW consultation and VAW services require better tailoring to women's needs and expectations. This should be based on understanding what women want and need and what is feasible. A recent qualitative metasynthesis found that after disclosing IPV to HCPs, women wanted assistance with documentation of injuries, insurance issues and help with connecting to community services more than referrals to IPV services. 67 The feasibility and acceptability of HCPs engaging with men who use violence needs further exploration in LMICs.
Finally, most interventions in the first two categories targeted the behaviour of individual HCPs rather than the SRH service or health system. The expectation was that trained HCPs would integrate VAW work into their clinical practice without structural changes to the environment, support from leadership, supervision, monitoring and incentivisation. Most qualitative evaluations described passionate HCPs who were enthusiastic about helping patients experiencing violence. However, some studies reported HCPs concerns about unrealistic expectations and limited health system readiness for embedding VAW work in routine practice. This finding is in line with other studies on health systems readiness for responding to VAW. 9 10 The obstacles to integrating a VAW response in SRH services overlap with those reported in the systematic review of barriers and facilitators to integrating health systems responses to IPV in LMICs. 68 This review's results are relevant for practitioners and policy makers in LMICs. The logic model approach allowed us to: (i) illustrate the hypothesised cause-result pathway, (ii) map evidence from All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; primary studies for the direct and intermediate effects and outcomes, (iii) identify barriers that can disrupt the trajectory of changes. It allowed us to present evidence in a format understandable to end users: people who develop, deliver, evaluate and fund VAW interventions in LMICs. We need more methodologically robust evaluations of interventions for strengthening the capacity of the health systems, communities, and individual women to respond to VAW with measures throughout the pathway from intervention activities to women's outcomes. Absence of evidence on the costeffectiveness of VAW interventions in SRH services is another gap. Finally, very few interventions have been evaluated in LMICs outside Africa.

CONCLUSION
We found that interventions to improve response to VAW in SRH services did not escalate violence.
Some interventions that strengthened capacity of HIV and ANC services increased identification and reduced some types of IPV. Some interventions that strengthened capacity of HIV and ANC services and communities improved use of SRH services and reduced re-exposure to some types of VAW.
Several studies identified a gap between provision and uptake of referrals to VAW services suggesting that first-line support should be better tailored to women's needs and preferences. Most additional psychosocial interventions that strengthened women's readiness to respond to VAW  All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 18, 2021. ; AR was funded by the NIHR Applied Research Collaboration West (NIHR ARC West). The views expressed in this article are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.

DISCLAIMER
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.  preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.