Gender Based Violence against Women in Sub-Saharan Africa: A Systematic Review and Meta-Analysis of Cross-Sectional Studies

This study aimed to systematically review studies that examined the prevalence of gender based violence (GBV) that included intimate partner violence (IPV) and non-IPV among women in sub-Saharan Africa (SSA). This evidence is an important aspect to work towards achieving the Sustainable Development Goals (SDG’s) target of eliminating all forms of violence in SSA. The Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) guidelines were followed. Ovid Medline, CINAHL, Cochrane Central, Embase, Scopus and Web of Science were used to source articles with stringent eligibility criteria. Studies on GBV in SSA countries that were published in English from 2008 to 2019 were included. A random effect meta-analysis was used. Fifty-eight studies met the inclusion criteria. The pooled prevalence of IPV among women was 44%, the past year-pooled prevalence of IPV was 35.5% and non-IPV pooled prevalence was 14%. The highest prevalence rates of IPV that were reported included emotional (29.40%), physical (25.87%) and sexual (18.75%) violence. The sub-regional analysis found that women residing in Western (30%) and Eastern (25%) African regions experienced higher levels of emotional violence. Integrated mitigation measures to reduce GBV in SSA should focus mainly on IPV in order to achieve the SDG’s that will lead to sustainable changes in women’s health.


Introduction
According to the United Nations (UN), gender based violence (GBV) is defined as "any act of gender based violence that results in, or is likely to result in, physical, sexual, or mental harm or suffering to women, including threats of such acts, coercion or arbitrary deprivation of liberty, whether occurring in public or in private life [1]." GBV occurs and is classified in various ways. It can be defined depending on the relationship between the perpetrator and victim (intimate partner violence (IPV) and non-IPV), or by type of the act of GBV, such as sexual, physical or emotional violence [2]. This definition resonates throughout this manuscript.
GBV is a global public health problem that poses challenges in human health, with a higher prevalence in developing countries [3,4]. GBV not only plays a significant component in the morbidity The fourth least developed sub-region of SSA is Eastern SSA that included Burundi, Comoros, Djibouti, Eritrea, Ethiopia, Kenya, Madagascar, Malawi, Mauritius, Rwanda, Seychelles, Somalia, Somaliland, Tanzania and Uganda [16].
GBV is reported as a common practice in SSA and sexual violence prevalence is high in some countries such as Zambia (90%) and Ethiopia (711%) [3,17]. According to the Gender Equality Index Report, which includes data on reproductive health, employment, and empowerment, 27 of the 30 countries in the world that exhibit unequitable gender indices, are in Africa [13]. Most African cultural beliefs and traditions promote men's hierarchical role in sexual relationships and especially in marriage [18]. Almost two-thirds (63%) of the African population live in remote rural settings that increases the difficulty to access basic amenities [16] and communities are disparate from the influence of central government or laws that prohibit GBV [13]. Only 22 African countries have adopted laws that prohibit GBV [14].

Information Source
A search of six electronic databases including Ovid Medline, CINAHL, Cochrane Central, EMBASE, Scopus, and Web of Science were undertaken. Relevant reference listings were checked, and grey literature was included, in addition to key research publications. Prior to starting this systematic review, the authors ensured the research question did not appear in any existing systematic reviews using Cochrane, Health Services Research Projects in Progress (HSRProj), and Prospero International Prospective Register of Systematic Reviews (PROSPERO) database registries.

Search Strategy
This systematic review was conducted using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19]. Pre-selected Medical Subject Headings (MeSH) terms and text words were used and searched in the above six databases for peer reviewed articles published between January 2008 and July 2019. The year 2008 was used as a baseline that provided increased global commitment of addressing GBV over the past decade [20]. There has been an increased uptake on the number of studies determining GBV prevalence internationally [20]. Moreover, the population dynamics have changed rapidly over the past ten years including improvements in health service access and education [21]. The search was limited to English language papers. Gender-based violence, intimate partner violence, domestic violence, spouse abuse, physical abuse, emotional violence, reproductive coercion, sexual assault, sub-Saharan countries, women aged 15-49 years, prevalence, magnitude and estimates were the key words used to conduct the search (Table S1). The specified age of 15 years was used as a baseline as most studies used Demographic Health Surveys (DHS) that focused on women aged 15 to 49 years of age.

Eligibility Criteria
The following eligibility criteria were used to include studies in the systematic review: (i) Studies that reported the prevalence of GBV that focused on either or a combination of IPV, non-IPV, physical, sexual or emotional violence; (ii) Sample size greater than 300; (iii) Females within the age range of 15-49 years of age; (iv) Studies conducted in SSA countries [16] including countries in Western, Central, Eastern and Southern countries (see study setting for the list of countries); (v) Published in English from 2008 to 2019; (vi) Only quantitative studies.

Exclusion Criteria
i.
GBV studies with no prevalence reported for example, studies that focused on factors associated with GBV; GBV consequences; ii.
Sample size less than 300; iii.
Qualitative studies not included as the main objective was to generate a pooled prevalence of GBV using the meta-analysis; iv.
Studies conducted outside SSA; v.
Studies published before 1st of January 2008; vi.
Studies published other than English; vii.
Study participants less than 15 years of age or greater than 49 years of age.

Quality of Study
The quality of the studies that met the inclusion criteria was appraised using a Critical Appraisal Skills Programme (CASP) checklist for cross-sectional studies [22]. The following criteria were the key questions derived from the CASP to appraise the quality of the studies:

1.
Did the study address a clearly focused issue? 2.
Were the participants of the study recruited in an acceptable way? 3.
Was the outcome accurately measured to minimise bias? 4.
Was the sampling appropriate for the study? 5.
What are the results of the study? 6.
How precise are the tools used to measure the results? 7.
Do you believe the results? 8.
Can the results be applied to the local population? 9.
Are the results of the study relevant and fit with other available evidence? 10. What are the implications of this study for practice?
The two independent reviewers rated the quality of each study by screening and considering the findings in relation to current practice or policy or relevant research-based literature and whether the findings can be transferred to other populations. The quality of each paper was rated using a ten-point scale using the CASP measurement criteria, 0 (none of the quality measures met) to ten (all quality measures met). The quality of the paper was based on the sum of points awarded. Studies were rated as poor quality (score ≤ 6); medium quality (7)(8); and high quality (≥9) (See Table S2 in the Supplementary Materials.

Data Extraction
Endnote was used to manage search results. The authors reviewed the titles, abstracts, and keywords of every article retrieved by the search according to the selection criteria developed that included author, country, population/study subjects, study design, sample size and key findings and quality of the paper. The full texts of the articles were retrieved for further assessment if the information suggested that the study met the selection criteria or if there was any doubt regarding eligibility of the article based on the information in the title and abstract. Outcome data were extracted from studies using a tailored data extraction form adopted from various literature.

Data Analysis and Synthesis
This study was based on secondary data analysis. The syntax "metaprop" in Stata version 16.0 [23] was used to generate forest plots for each of the Figures S1-S8. Each forest plot showed the prevalence of an indicator in individual authors and countries and its corresponding weight, as well as the pooled prevalence in each sub-region and its associated 95% confidence intervals (CI's). A test of heterogeneity of the DHS and other data sets were obtained for the different authors and countries that showed a high level of inconsistency (I 2 > 50%) thereby warranting the use of a random effect model in all the meta-analyses. Sensitivity analyses were conducted to examine the effect of outliers by using a method similar to that employed by Patsopoulos and colleagues [24] which involves comparing the pooled prevalence before and after elimination of one author or country at a time. Subgroup analysis was conducted by Eastern Africa, Western Africa and Southern Africa based on the UN classification [16]. The findings of the systematic review are synthesized and presented in summary form in Table 1.

Ethical Statement
This review used secondary data available in the public domain including the six electronic databases for the systematic review and the DHS dataset that are publicly available. Therefore, ethical approval was not required for this study because the data included in this analysis contained no identifying information and is publicly available and ethical approval has already been obtained by the original author or by the DHS program.

Results
A total of 4931 articles were found in the initial search from all databases. After removal of duplicates, 3275 remained for screening. Screening by title led to the exclusion of 3021 articles. Further reading of abstracts for 245 full-text articles led to the exclusion of another 187 articles. Twelve grey literature articles were included. Finally, 58 articles met the inclusion criteria ( Figure S1).

Description of Included Studies
Fifty-eight articles were reviewed for data analysis and interpretation.
The majority (95%) [4,9,15,21, of research articles included in this review were cross-sectional and the remaining (5%) were cohort studies [86][87][88]. Only cross-sectional studies were used to estimate the pooled prevalence rates. Overall, the total sample sizes ranged from 300 to 86,024 women of reproductive age (Table 1).
Overall, 58 cross-sectional studies investigated the prevalence of IPV either in the woman's lifetime or over the previous year. Four studies reported non-IPV [4,27,60,80]. The studies that focused on IPV included 23 that reported physical violence, 18 that reported sexual violence and 20 studies that reported emotional violence. A relatively larger number of studies were found from Nigeria, South Africa, Kenya, Ethiopia and Uganda ( Table 1). The assessment of the studies' quality found that 30 (52%) were very good, 22 (38%) were medium quality and six (10%) studies were deemed low quality. The details of this assessment are provided as a supporting document (Table S2).

Prevalence of IPV among Women Aged 15-49 Years of Age
The prevalence of IPV in various SSA countries was sourced from 25 studies. The findings showed the prevalence ranged from as low as 13.9% (95% CI 10.8, 17.6%) [86] in a study conducted on perinatal women with depression symptoms in South Africa to as high as 97% (95% CI 94.6, 98%) [37] in a study conducted among rural women in Nigeria. The overall meta-analysis estimate for prevalence of IPV was 44.4% (95% 38.4, 49.8%) ( Figure S2).

Prevalence of Intimate Partner Physical or Sexual Violence among Women Aged 15-49 Years of Age Using DHS Data (2008-2019)
Additional information was sourced from the most recent DHS reports that were conducted in SSA countries from 2008 to 2019 [101]. Only 29 sub-Saharan countries from the DHS reported on GBV. The prevalence focused on physical or sexual violence committed by a husband or partner against women [101]. We found that prevalence ranged from as low as 6.4 % in Comoros to 51% in Cameroun [101]. The meta-analysis showed a pooled prevalence of 31.3% (95% CI 26.3, 36.3) with heterogeneity detected among various surveys and countries ( Figure S3).

Prevalence of Past Year IPV among Women Aged 15-49 Years of Age
A total of 18 studies investigated experiences of IPV over the past year among 24,941 women. The highest prevalence of IPV was found among women engaged in commercial sex work 78.7% (95% CI 75.2, 81.8%) in Kenya and in Nigeria (52.5%) (95% CI 46.7, 58.2%) [66]. Furthermore, a meta-analysis was estimated at 35.5 % (95% CI, 27.2, 44.12) ( Figure S4). The sub-region analysis showed the highest pooled estimates in Eastern Africa (38.93%), followed by Western Africa (32%). Limited studies were sourced in South and Central SSA countries on IPV over the past year. Another sub-group analysis over the past year's prevalence of GBV among pregnant and non-pregnant women showed the prevalence of experiencing any form of GBV amongst pregnant women was 30.5% (95% CI 21.2, 39.6) compared to non-pregnant women 39.8% (95% CI 26.98, 52.69) ( Figure S4).

Prevalence of Sexual IPV among Women Aged 15-49 Years of Age
Seventeen studies showed an overall prevalence of violent experiences and seven studies found experiences of sexual violence over the past year. Overall, pooled prevalence of sexual violence was 18.61% (95% CI 15.21, 22.00) with a high disparity among studies detected ( Figure S6). The highest prevalence report was found in women in Northern Uganda (50%) (95% CI 46, 53%) [29], followed by a study conducted amongst women (39.7%) (95% CI 32.2, 47.2) in the Democratic Congo [31]. The lowest prevalence was found in Ghana (4%) (95% CI 3.1, 5.1) [37] and Nigeria (6.6%) (95% CI 6.3, 6.9) [45] amongst women of reproductive age. Similarly, a study conducted amongst nurses in Ethiopia showed one in 25 nurses (3.8%) had an experience of sexual violence (95% CI 2.5, 5.6) [47]. Eastern African women experienced relatively more sexual violence compared to other sub-regions. Among the seven studies with women experiencing sexual violence over the past year, violence ranged from the highest in Nigeria (42%) [66] and Ethiopia (31%) [39] to the lowest being 2% (95% CI 1.1-3.6 %) in a study conducted among HIV infected pregnant women in South Africa [93]. The results show there were no differences in lifetime and past year sexual IPV experiences ( Figure S6).

Prevalence of Emotional IPV among Women Aged 15-49 Years of Age
There were 57,434 study participants included in the analysis. The prevalence of emotional violence was the highest among health care workers in Ethiopia (53.1%) (95% CI 48.7%, 57.4) [100] to Rwanda 9.7% (95% CI 8.8, 10.7) [56]. In particular regions, one in three women in most parts of Western Africa were emotionally abused by their partner. For instance, two studies conducted amongst women aged 15-49 years of age in Nigeria indicated the prevalence rate of emotional violence experienced was 44.4% (95% CI 40.9, 47.9) [58] and 34.7% (95% CI 29.5, 40.2) [28]. The most common type of violence was purported to be emotional violence in these countries in comparison to other regions. Sub-group analysis was conducted based on timing of the violence and found a pooled overall prevalence of emotional violence of 29.36% (95% CI 24.77, 33.9) and past one-year prevalence rate of 21.42% (95% CI 17.58, 25.26) ( Figure S7). The test of heterogeneity and publication bias was detected (I 2 = 98.9% and 88.6%, Egger's test = 0.205).
Six studies have demonstrated the magnitude of emotional violence over the past year. The highest prevalence was found among female sex workers (31.9%) (95% CI 26.7, 37.1) [71] in Nigeria, followed by a study in Ghana (24.6%) (95% CI 20.5, 29.2) [98]. Correspondingly, a study conducted in Ethiopia showed one in five pregnant women experienced IPV over the past year [52].

Discussion
This review incorporated all forms of GBV, including physical, sexual and emotional violence and IPV and non-IPV. The findings showed the pooled prevalence of GBV was high in SSA countries. This high pooled prevalence included almost half of the women experiencing IPV and a considerable number of females being abused by non-IPV. Emotional IPV violence was the most common type of violence in SSA. GBV was more prevalent in the sub-regions, in Western and Eastern Africa as compared to southern regions of SSA countries. Methodological quality of cross-sectional studies was appraised. We used only cross-sectional studies because we only found three cohort studies and/or randomized controlled trials.
Overall, a high pooled prevalence of IPV among women in SSA was found as compared to the global estimate which was conducted in 56 countries in 2013 [4] and SSA countries [5,21,102,103]. The findings of this review are comparable or slightly higher to studies conducted in 14 SSA countries [27,102]. The higher prevalence of IPV in our study could provide a better overview compared to previous studies where the number of countries involved were relatively small. Most importantly, this high prevalence might be due to the prevalence of gender inequality in regions for reasons including prerogative perceptions to males, tolerant attitudes in the community to IPV, poor education of women, female disempowerment and limited law enforcement in SSA [3,4,21,102,104,105].
Further analysis of the pooled prevalence rate over the past year revealed that more than two out of five women have reported experiencing IPV in SSA countries. This figure is consistent with a study conducted in other SSA countries [104] and more than five percent greater than the global lifetime prevalence of IPV (30%) [4]. This figure could be even higher, in reality, due to the underreporting associated with GBV [7] because of factors associated with fear of stigma, women preferring to keep quiet and fear of divorce, amongst many other reasons [6,7,14,17].
One of the interesting findings from this study is that the proportion (18%) of women affected during their lifetime and over the past year's experiences of IPV were exactly the same as shown in Figures S2 and S4. This finding reflects that women in SSA countries are being subjected to experiences of violence continuously compared to other areas [14]. Overall, IPV in SSA countries is the most prevalent and challenging public health issue. The social context of the region is very complex and strong ties, extended family size and large communities of relatives are quite common that might expose women to potential perpetrators [106]. The prevention and management of GBV makes it more difficult in SSA countries.
The finding of pooled prevalence of IPV of the DHS was very high. There were statistical differences compared to the pooled prevalence of IPV computed from the electronic sourced articles. Moreover, the pooled prevalence from non-DHS studies found in electronic databases and DHS reviews were statistically different (p < 0.01). Our systematic review focused on IPV that included any of the combinations of physical, sexual or emotional violence or coexistence while DHS data focused on either physical or sexual violence among married women. In addition, DHS only explored married women, while in our study we used any population group in the age range of 15-49 years of age.
In this review, the pooled prevalence of all types of GBV, physical, sexual and emotional violence were consistently higher in SSA countries as compared to many other regions in the world [25,102,103]. Emotional violence was the most prevalent reported type of violence. Sexual experiences are reported not as frequently in many African countries for numerous reasons. The pattern of sexual violence is lower than emotional and physical violence, which might be related to victimized women being unlikely to report an attack due to fear of discrimination, feeling shame, and not being able to identify as well as physical violence [7,14].
One of the unexpected findings among health care providers was the highest prevalence rates (53%) of emotional violence and lowest prevalence rates (5%) of physical violence being reported in Ethiopia [100]. This high prevalence of emotional violence may be related to less satisfaction of service users due to long waiting times and less experienced health workers working in the health facilities. The majority of health care providers in the studies were females and this may be a reflection of gender inequality in the work areas. Most importantly, there is a lack of violence tracking or reporting mechanisms when it occurs among service providers, specifically focusing on emotional violence in the health care system [100,107]. Alternatively, the low prevalence of physical violence may be due to nurses having an understanding of the local context of GBV and being more likely to notify cases that would prevent perpetrators committing acts of violence [9,11]. Additionally, perpetrators may be unlikely to attack nurses at places such as a hospital or health centre where many other patients are receiving care from nurses.
The sub region analyses found that Eastern Africa (42%) including Ethiopia and Uganda were the most affected by all forms of IPV [29,47], followed by Western Africa (41.7%). In line with our findings, the two regions that experienced high prevalence rates of IPV in comparison to other African regions [7,25] was also consistent with other studies conducted in SSA countries [4,5]. In Eastern Africa, physical and sexual violence prevalence rates were worse and emotional violence prevalence rates were more common in Western Africa. This finding is consistent with findings of other studies [2, 6,15,27]. This might be attributed to factors such as socioeconomic class, women's disempowerment, community acceptance for wife beating and the type of community in which the study was conducted [4,6,27,29,108,109].
Alternatively, in Southern regions of Africa, the educational qualifications are relatively much better when compared to Eastern African countries [110]. A study conducted in South Africa found a combined intervention of economic intervention and education reduced IPV prevalence rates by 55% over a period of two years. Therefore, education differences could explain the differences of IPV prevalence in the two regions [110].
The pooled non-IPV prevalence (14%) experiences were very high. The pooled non-IPV prevalence experiences were slightly higher than the three studies that were conducted internationally, which was 11% [4,5,27]. The highest non-IPV prevalence may be related to political instability and war violence. For example, in Somalia the non-IPV prevalence was found to be 16.5% [80] which is mainly related to political instability and migration of the region. Moreover, some basic services are lacking, for example, health services, water and education. As a result, women are forced to travel long distances, which puts women more at risk to be subjected to violence as compared to those who have easy access and less travel time to those services.

Policy Implication
Findings reported in this study provide vital evidence to inform policy and guide health practitioners to respond and prevent violence in alignment with the SDG's target by 2030. The aftermath of GBV has large ramifications for women's health. It will be a challenge to achieve the SDG's target to eradicate IPV by the year 2030, unless there is a timely intervention and policy designed for SSA regions. Governmental policies top priorities should focus on prevention of GBV, especially with the high prevalence of both IPV and non-IPV in all regions of SSA countries. This strategy needs to be supported by a legal framework to accommodate social support that includes educational and economic growth and provision of health information and services. All SSA countries need to develop an immediate action plan to support the challenges that women are facing with GBV. This review has added evidence to the current existing knowledge in the literature and has provided a stimulus for future research on the dynamics of GBV in SSA countries.

Strengths and Limitations of this Review
This is the first systematic review and meta-analysis to quantitatively summarize the prevalence of GBV that includes IPV and non-IPV that extends to SSA countries. A rigorous search was conducted from many electronic databases and selected nationally representative data sets (DHS) were used for most studies. A quality assessment was conducted with two independent reviewers conducting the quality screening. Only studies with adequate samples greater than 300 for representativeness were included in the review.
Despite the rigorous process of the systematic review, the searches only included articles published in English. The heterogeneity in our review could have been due to various factors such as different recall periods, underreporting, contextual differences including conflict, cultural differences and the quality of tools used to assess GBV. The generalizability of some small-scale studies is limited as studies may overestimate or underestimate GBV depending upon the context of the study. In addition, the number of studies on non-IPV were limited and it was difficult to identify the broader picture of GBV in the region. Furthermore, this review only included quantitative studies, most of which were cross-sectional. Therefore, qualitative studies were not included which may provide further information on the attitudes of women and communities about GBV that could indicate higher prevalence rates of GBV.

Conclusions and Recommendations
GBV against women is a pertinent health challenge in SSA countries. GBV that includes IPV and non-IPV are prevalent in SSA. More than two-fifths (44%) of women aged 15-49 years of age in SSA countries experienced some form of IPV and almost a fifth (14%) experienced non-IPV. All types of IPV (physical, sexual and emotional violence) are common experiences among women in SSA countries, with emotional violence being the most prevalent. Women living in Eastern and Western African regions experience the highest levels of GBV.
The need for an integrated mitigation measure to reduce GBV needs to be considered as a top priority in line with the SDG target in 2030 to reduce all forms of violence in SSA countries. Hence, government and private organisations should understand and address the problem of GBV. All organisations can allocate resources and design appropriate interventions that includes law enforcement to ensure social support is provided for women in the quest to eradicate GBV. In addition, more research is required to provide information on the dynamics of communities, the context, and associated factors of GBV and the subsequent effects of women's reproductive health and beyond. Furthermore, more studies on IPV in SSA are required, especially in areas where political instability and war are on the increase.

Supplementary Materials:
The following are available online at http://www.mdpi.com/1660-4601/17/3/903/s1, Figure S1: PRISMA flow chart for selection of studies on prevalence of GBV; Figure S2: overall pooled prevalence of IPV; Figure S3: pooled prevalence of physical or sexual violence committed by husband/partner among ever-married women age 15-49 years of old using latest DHS surveys; Figure S4: past one year pooled prevalence of IPV; Figure S5: pooled prevalence of physical IPV; Figure S6: pooled prevalence of sexual IPV by timing; Figure S7: overall pooled prevalence and recent past year prevalence of emotional IPV; and Figure S8: pooled prevalence of non-IPV. In addition, Table S1: search strategy, and Table S2: quality assessment of included studies.