Determinants of Gender-based Violence Among Girls and Young Women Aged 11-24 Years Seeking Care at a Rural Hospital in Zimbabwe: A Case-Control Study

Background: Physical and sexual violence against women and girls has become a widespread public health challenge associated with negative physical and psychological consequences not only to the victim but to society at large. Gender-based violence (GBV) can result in unwanted pregnancies, unsafe abortions, anxiety, stress, depression, and sexually transmitted infections. The problem is more prevalent in low-income settings like Zimbabwe and this study aimed to investigate the determinants of GBV among young women and girls aged between 11- 24 years, seeking healthcare services at a rural hospital in Zimbabwe. Methods: A 1:1 unmatched case-control study was conducted at a rural hospital in Buhera District, Zimbabwe. A total of 104 cases and 104 controls were recruited using purposive and systematic random sampling methods, respectively. The data was collected using a pretested interviewer-administered questionnaire and data were analyzed using Epi Info statistical package. Bivariate analysis and multivariate logistic regression at 95% condence interval and statistical signicance were set at p-value<0.05. Results: The pooled mean age of participants was 17.6±3.8 years and 89.4% of the cases had no familial relations with perpetrators. Bivariate analysis revealed that young age (<19 years), apostolic religion, having no formal education, not married, living in a polygamous family, and having >3 siblings were associated with GBV. Cases were more likely to report to a health center and were less likely to conde in someone. The most prevalent causes of GBV were religious and cultural beliefs, forced marriages, and alcohol abuse within a family. After multivariate analysis, young age [Adjusted odds ratio (AOR): 4.8; 95% Condence interval (CI):1.3-17.2], polygamy [AOR: 13.8(3.6-53.1)], alcohol abuse in the family [AOR: 4.1(1.3-13.3)] were associated with GBV, while conding in someone was protective [AOR: 0.05 (0.01-0.2)]. Conclusion:

with GBV [2,29]. There are laws and legislation put in place to promote reporting of GBV but women are reluctant to do so due to the barriers mentioned above.
Despite the existence of the Domestic Violence Act 14/2006 law in Zimbabwe, sexual and physical violence offenses against women remain a widespread problem in the country [30]. Zimbabwe has a high prevalence of violence against women and girls with 35% of women aged between 15-49 years have experienced physical violence and 14% have experienced sexual violence at least once in their lifetime [30]. To the knowledge of the researchers of this study, there is a dearth of local empirical evidence on the factors contributing to GBV in the country. This study sought to i) identify the socio-demographic factors contributing to GBV, ii) assess the perceptions of GBV and iii)determine the family practices associated with GBV among females aged 11-24 years seeking care at Murambinda Mission Hospital, Buhera District, Manicaland Province, Zimbabwe.

Study Design
An unmatched 1:1 case-control study was carried out to determine the factors contributing to GBV among females aged 11-24 years at Murambinda Mission Hospital.

Study setting
The study was conducted at Murambinda Mission Hospital, which is 170km south-west of Zimbabwe's fourth-largest city of Mutare. The 125-bedded hospital is an a liation of the Roman Catholic Church and was established in 1968. A total of 27 satellite clinics fall into the Murambinda Mission Hospital catchment area and in 2016, there were 6,170 in-patient admissions, 47,222 out-patient attendances, 1524 babies delivered, and 248 major operations done [31]. The hospital had a staff compliment of four medical doctors, 37 trained nurses, 18 nurse aides, 17 general hands, as well as ancillary and administrative staff. The hospital serves the surrounding rural population of approximately 300,000 people. Data were collected at the Out-patient Department (OPD) which serves an average of 170 patients per day. Besides, the OPD also hosts a One-Stop-Center (victim-friendly unit) for GBV survivors which provides comprehensive psycho-social support, medical examination, con dential voluntary testing and counseling, post-exposure treatment, and follow-up support by a team of trained healthcare workers, police o cers, and victim advocates.

Study population
The study population was made up of female GBV survivors aged 11-24 years seeking healthcare services at Murambinda Mission Hospital. A case was a girl or young woman aged 11-24 years who sought GBV services at the One-Stop-Center of the hospital OPD and resided within Buhera District. A control was a girl or young woman aged 11-24 years who sought services at the hospital OPD presenting with other health complaints which were not GBV-related and lived within Buhera District.
Sample size and sampling procedure Using a 1:1 unmatched case -control study, the sample size was calculated using Epi Info TM 7.2.1.0 (CDC) Stat-Cal Function at 95% con dence interval and 80% power. Assuming alcohol abuse was a signi cant risk factor for GBV with an Odds Ratio of 2.25 and 30.1% of the controls exposed [32]. A minimum sample size of 102 cases and 102 controls was calculated giving a total sample size of 204. GBV is a rare complaint; thus, the ideal sampling method for selecting cases was using the purposive sampling method. The systematic random sampling method was used to select controls attended at the hospital OPD using a sampling factor of 3.

Data collection and analysis
Pretested, structured, and interviewer-administered questionnaires were used to collect data from the study participants while an unstructured questionnaire was used for key informants whose input was used for data triangulation. The key informants were the Nurse-in-Charge of the OPD, two nurses working in the One-Stop-Centre Clinic, a police o cer from the Victim-Friendly Unit, and the District Medical O cer. The study participant questionnaire was translated to a local language Shona by two linguistic experts from Africa University before the two versions underwent validation by two psychotherapists. The data collection tools were pretested at Hauna District Hospital using 21 cases and 21 controls who met the inclusion criteria. Hauna District Hospital is also a rural hospital in the same province as Murambinda Mission Hospital.
The researchers sought consent from the potential participants after they had received all the necessary health services and the doctor had certi ed that they were stable and t for interview. The interviews were conducted in a private room and the hospital primary counselors were on alert in case a participant needed further counseling services. The researcher continued to visit the hospital OPD until the calculated minimum sample size of cases and controls was satis ed. Data were collected by EC during the period January 2018 to June 2018. To prevent missing data, the researchers double-checked the answered questionnaire soon after the interview before the participant had left the interview room.
Data from the paper-based questionnaires were entered into Microsoft Excel where it was cleaned before it was imported into Epi Info™ version 7.2.1.0 (CDC) statistical package for analysis. Statistical analysis included frequencies and means of the socio-demographic characteristics. Bivariate analysis was done to identify the association between the dependent variable (GBV) with the independent variables. The variables that were found to be statistically signi cant at p<0.05 during bivariate analysis were entered into a logistic regression model to control for confounding.

Ethical considerations
Permission to carry out the study was sought from the Ministry of Health and Child Care through the o ce of the Provincial Medical Director and the District Medical O cer while ethical clearance was obtained from Africa University Research Ethics Committee (Approval Number: AU313/18). Written informed consent was obtained from all study participants and for the minors under the age of 18, written consent was obtained from the parent or guardian after the participant had verbally assented.
Girls under the age of 18 years were interviewed in the presence of their parent/s or guardian/s. Participants were assured of con dentiality and numerical codes were used to identify respondents instead of writing their real names.

Results
A total of 104 cases and 104 controls were interviewed and the response rate for cases was 92% while that of controls was 95%. The pooled mean age was 17.6±3.8 years (Cases: 15.3±3.1 and Controls: 19.9±3.0 years) while the pooled mean number of siblings of participants was 5.8±2.4 (Cases: 6.5±2.1 and controls: 4.9±2.5). While 62.5% (n=65) of the cases had a history of GBV, only two of the cases were physically challenged, and approximately 89.4% (n=93) of the cases had no familial relationships with the perpetrators. When given an option of where to report to following a GBV incident, the majority of participants would report to a health facility (89% of controls and 97% of controls) and not the police. Table 1 shows the socio-demographic characteristics of the study participants. Perceptions of GBV among study participants Study participants were asked for their perceptions on action to be taken after the abuse incident plus where and when GBV occurred in Murambinda. Table 2 shows the participant's responses to questions assessing their perceptions on GBV aggregated according to case status. All participants reported that rape was regarded as a form of sexual violence in their families. Sexual touching/kissing and attempted sex were regarded as a form of sexual violence by the families of 16.8% (n=35) and 12% (n=25) of the participants, respectively. The most common causes of GBV in the study community were evil spirit cleansing and forced marriages as reported by 87% (n=181) and 84.1% (n=175) of the participants, respectively. Most of the study participants cited HIV and STIs (70.7%, n=147), unwanted pregnancy (66.8%, n=139), and psycho-social trauma (36.1%, n=75) as the common negative impacts of GBV on families. The major reasons for not reporting were protecting religious and family beliefs, and poverty as reported by 71.6% (n=149) and 65.4% (n=136) of the respondents, respectively. Table 3 shows family practices associated with GBV. After bivariate analysis, all variables which were statistically signi cant at p<0.05 were imputed for logistic regression analysis, and the ndings are displayed in Table 4. Interviews with the key informants also identi ed the causes of GBV in Buhera District as poverty, patriarchal nature of the communities, and restrictive religious belief. Girls married early with the help of the nancial security that comes with having a male intimate partner taking care of their nancial needs. The majority of families were a liated with the apostolic sects whose clandestine and conservative actions made it di cult for the women's rights groups to penetrate the communities. The authoritative leadership of the apostolic sects warned their female members to refrain from GBV-related discussions which were viewed as 'secular'.

Discussion
This study sought to identify the socio-demographic factors contributing to GBV, assess the perceptions of GBV, and determine the family practices associated with GBV among females aged 11-24 years seeking care at Murambinda Mission Hospital, Buhera District, Manicaland Province, Zimbabwe. After controlling for confounding variables, a young age, living in a polygamous family, and alcohol abuse in the family were found to be strong predictors of GBV in the present study. Forced marriages, religious and cultural beliefs were also perceived as causes of GBV while the women and girls preferred to report to the hospital than to the police.
The study ndings revealed that the mean age of cases was lower than that of controls and young girls (11-19 years) were nearly ve times more likely to be victimized when compared to young women (20 to 24 years of age). Adolescent girls are at a higher risk of becoming GBV victims [28,33] and this can be attributed to limited experience in relationships particularly with male partners who are older than them [34]. In intergenerational relationships, young girls lack the power to make autonomous decisions; hence they are forced into sexual acts which they usually regret later in life. This was also reported in a similar study that found out that for almost half of female childhood (under the age of 18 years) sexual victims, the perpetrators were ve or more years older [35].
One of the most common causes of GBV identi ed by the participants in this study was religion and culture-related evil spirit cleansing. Although there has been a growing upsurge in international efforts to end violence against women, many groups lobbying for women's rights usually withhold their denouncing of GBV when it is associated with the 'culture' or 'religion' of a certain community [36]. These groups tend to play down violence against women and girls as an authentic practice justi ed for preserving the cultural, religious, and traditional sovereignty of the communities. Legitimizing violence based on culture and religion conversely deprives women and girls of their self-determination rights as well as control of their bodies and sexuality. Since those who employ culture and religion to justify GBV do it because of their interpretation of one 'authentic' view oblivious of alternatives, it is critical to demystify such beliefs by exposing the alternatives to them [36]. Many societies have normalized the attitudes and systems that elevate religious doctrine endorsed by male authorities and these patriarchal structures have also regularized violence in the society [37]. Therefore, to reduce GBV emanating from misguided religious beliefs, religious institutions must improve women's representation in leadership positions while communities should be empowered and resourced to host multi-faith women networks [38]. These networks can be used for interactive communications and discussions on GBV issues among both men and women in the communities. Religious leaders also have to amplify their voices and exercise their in uence to communicate a vivid message against GBV.
The researchers also noted that forced marriages were a common source of GBV in this community.
Globally, over 1.5 million girls younger than 15 years are married annually [39] and there were an estimated 15.4 million people who were living in forced marriages in 2016, 88% of these were women and girls [40]. There are a variety of reasons why forced marriages take place, and these include economic incentives, con ict settlement, and preserving family 'honor' by forcing girls or women who have been raped to marry the rapist [41]. Forced marriages emanate from the perspective that girls are property that can be exchanged with such views being reinforced by culturally accepted mores and precedents. Education on the importance of gender equality and empowering the girl child can be used to defeat the age-old patriarchal structures.
The major reason why some women in this community did not report perpetrators was that they wanted to protect religious beliefs and poverty forced girls into early marriages. In Zimbabwe, cultural beliefs are girded by the male-dominated in uence which seeks to protect the family name, even if that meant sacri cing a family member's dignity. Social norms that are intolerant to extramarital pregnancy while blaming women and girls for becoming pregnant outside marriage diminishes the victim's will to report cases due to self-blame. A study by Perrin et al., (2019), in Somalia and South Sudan, noted the stigma associated with being a member of the family where the emphasis was placed on addressing GBV within the family rather than reporting to authorities [42]. This was done to protect the family name over the safety and wellbeing of the victim. Orienting women and girls on self-preservation can help them understand that violent behavior is wrong and should be denounced at all costs.
Women living in a polygamous family or marriage were more likely to become victims of GBV. The social complexity of polygamous/polygynous marriages exposes the women involved to GBV [43]. Polygyny is very common in sub-Saharan Africa [44] and a recent study noted that the practice is associated with intimate partner violence of a prevalence of 30.7% [45]. This could be attributed to polygamous families having poor social interaction as a result of co-family rivalries over resources [46]. Although GBV within polygamous families is di cult to tackle, such families must be educated on the importance of peaceful resolution of con icts and shun violence. The high rates of GBV among polygamous families in Buhera District warrants further qualitative inquiry to establish speci c solutions for this speci c population.
Participants who lived in families where alcohol was abused were more likely to become victims of GBV.
This was also the case in a similar study conducted in Zambia where alcohol drinking behavior increased the odds of GBV by more than double [32]. Alcohol interferes with cognitive and physical functions of the body diminishing self-control, and ultimately leaving the person incapable of negotiating con icts and confrontations in a non-violent manner [47]. Problematic masculine alcohol intake is associated with increased sexualized behavior [48]. Alcohol abuse also culminates into nancial di culties, in delity, and childcare problems which can potentially lead to intimate partner violence and child abuse [49]. In Buhera District, traditional alcohol is brewed in the communities where unemployed men go to buy and drink the cheap brew as a way of killing time. As a result, these men may physically or sexually assault their partners or other women as a way of venting their frustrations. Abusing alcohol has also been linked to risky sexual behavior which can result in the exposure of the GBV victims to sexually transmitted infections such as HIV/AIDS.
The study participants viewed contracting HIV and STIs, and unwanted pregnancy as the most devastating consequences of GBV. Research into the high prevalence of HIV among girls and young women in sub-Saharan Africa reported that the risk and vulnerability to HIV among this population are deep-rooted in violence against women [50]. According to the UNESCO Report (2013), the risk of HIV among women increases three-fold in women who have experienced GBV when compared to those who have not [50]. Forced sexual intercourse with an HIV-infected individual increases trauma-causing lacerations and abrasions which provide easy entry of STIs/HIV. Although most STIs can be treated, there is no treatment for HIV/AIDS, a disease often stigmatized in many communities. Unwanted pregnancies, particularly as a result of sexual violence, can lead to long-term poor mental health outcomes [51]. Women and girls should know the importance of the prompt visit to the health facility after they have been sexually abused so that they receive post-exposure prophylaxis of HIV as well as legal abortion services.
The majority of participants in this study stated that following a GBV incident, they would report to a hospital or clinic rather than the police. This nding was consistent with other studies which cited that raped women were highly unlikely to report to the police [52]. In Zimbabwe, reporting violence cases to law enforcement authorities may be hampered by persistent socio-economic and political instability which saw women suffering violence and torture at the hands of security forces along with the police departments being perceived as politically in ltrated by individuals in leadership positions [53,54]. As a result, citizens may have reservations about reporting when the authorities who are supposed to protect them, persecute them instead. Likewise, literature has also noted how con ict and humanitarian emergencies expose women and girls to the risk of GBV [55,56]. Thus, the health facility becomes an appropriate and convenient reporting place for GBV survivors rather than the police, because the One-Stop-Centre establishment provides comprehensive aftercare medical, counseling, legal services, and even the police, under one roof.
Participants believed that it was important to con de in another trusted person after experiencing GBV. Women deride reporting to authorities like police due to the fear of either being blamed or not believed; thus, victims choose a trustworthy con dante, most likely a close friend, who can offer emotional support [52,57]. Blaming GBV survivors seeking solace was associated with increased rates of re-victimization over a period of twelve months [57]. Despite the limited effect of the positive con dante reactions, the opposite, which negative reactions such as blame and rejection, can cause psychological harm to survivors [58]. Summarily, sincere responses from friends and trusted others matter deeply.

Limitations
When interpreting the ndings of this study, one should consider the possibility of potential social desirability bias due to the sensitive nature of the topic and self-reporting of participants. Besides, this study may have missed women who were GBV survivors who did not report to the OPD at Murambinda Mission Hospital, consequently, this can have a major bearing on the generalizability of the ndings.

Conclusions
Gender-based violence is an important public health challenge in Buhera District. The religious and cultural beliefs, as well as family structure, are major areas to consider when intervening against violence against women in this population. To eradicate GBV, trained health personnel, religious, and traditional leaders could be used as champions to intensify the message of gender equality and the use of con ict resolutions that do not involve violence. Education on GBV should integrate children of school-going age, faith-based organizations, and community members in crafting sustainable interventions. Furthermore, the researchers recommend a qualitative, community-based prospective study that will cover the entire district guided by ndings from this study. Availability of data and materials

Abbreviations
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request