Domestic and intimate partner violence among pregnant women in a low resource setting in South Africa: a facility-based, mixed methods study

Rates of violence against women are reported to be highest in Africa compared to other continents [1]. In South Africa, the national mortality rate attributed to intimate partner violence (IPV) was found to be double that of the United States [2]. A World Health Organisation multi-country study on IPV found that between 15 and 71% of women reported lifetime physical or sexual violence by a partner [3]. This is an important public health concern.

Domestic violence (DV) is defined as any physical, sexual, psychological or economic abuse that takes place between people who are sharing, or have recently shared a residence [4]. While this usually takes place between intimate partners [5], in the context of a low-resource setting, where extended family members reside within the same household, DV is not limited to intimate partners. Globally, rates of DV seem to be higher in rural than urban areas, with most cases not being reported to police or healthcare providers. Thus, data reported in epidemiologic studies are likely to underestimate the prevalence [6].

A national study conducted in the USA, with data from over 34,000 participants, showed that DV is associated with physical illness including injury, chronic pain, asthma and sexually transmitted infections [7]. In an economic evaluation of the cost of IPV to the Australian health system, it accounted for 8% of the overall burden of disease for women of child-bearing age, contributing more than either raised blood-pressure, obesity or tobacco use. The association with poor mental health was one of the greatest contributors to this burden [8]. Globally, high levels of symptoms of perinatal depression, anxiety, and Post-traumatic Stress Disorder (PTSD) are significantly associated with having experienced DV [8–10]. Greater severity of traumatic experience and sexual violence are each associated with greater levels of depressive symptoms [11].

A Nigerian study reported that co-wives and step-sons perpetrate 24 and 17% of DV respectively [12]. In Bangledesh and Uganda, acid throwing is a commonly reported method of violence against women, usually perpetrated by family members other than partners [13]. In a South African study, Hoque et al. (2009) found that 12% of DV that occurred during pregnancy, was perpetrated by a family member other than a partner [14]. An earlier South African study indicated that 24% of DV during pregnancy was perpetrated by an in-law [15].

Violence during pregnancy has negative implications for both the mother and the child. In its most severe form, violence against pregnant women has been reported as a contributing cause of maternal deaths [16]. Violence during pregnancy has been associated with inadequate uptake of antenatal care, with abused women being more likely to delay seeking pregnancy care and to attend fewer antenatal visits [17]. Detrimental perinatal physical health outcomes for both mother and child have been reported. These include: low birth weight, foetal death by placental abruption, antepartum haemorrhage, foetal fracture, rupture of the uterus and premature labour [11] A strong association between suicidal ideation and IPV in pregnant women has been reported in low-income settings [18].

Many of the risk factors for violence during pregnancy have also been identified in studies of IPV and DV against women in general. However, global literature indicates that both unintended and unwanted pregnancies are each associated with experiencing violence during pregnancy [19]. A Canadian study showed that women with unintended pregnancies are three times more likely to experience IPV than those with intended pregnancies [20]. From global literature, other risk factors pertaining specifically to the antenatal period include: low socio-economic status, being young or adolescent, being unmarried, becoming separated or divorced during pregnancy, belonging to an ethnic minority, alcohol misuse by either the woman or her partner and low educational status [19, 21]. Further factors associated with IPV in Southern Africa, have been identified as: having a younger male partner, problem drinking by the partner, partner control of woman’s reproductive health, and risky sexual practices [22]. In Africa, studies on Human Immunodeficiency Virus (HIV) diagnosis and status as risk factors for IPV are inconclusive [21]. However, sexual risk factors have been positively associated with experience of IPV. These included transactional sex, having more than five lifetime sexual partners and having multiple sexual partners [21]. Further, this review shows that one of most significant predictors of violence during pregnancy is a history of abuse (defined as experiencing abuse before the age of 15, abuse in the past 12 months and abuse at any point over one’s lifetime) [21]. However, the pattern of abuse may be influenced by pregnancy itself. A global review study indicated that between 13 and 71% of women who are abused during pregnancy reported an increase in the frequency and/or severity of violence during this time [19].

In South Africa, high levels of violence occur within a context of multiple contributing social dynamics. These include prominent patriarchal norms where masculinity is associated with defence of honour, harshness and risk taking [23]. Poverty and gender inequalities contribute to the structural determinants of violence [24].

The aim of this study is to determine the associations between mental illness, demographic, psychosocial and economic factors with experience of IPV among pregnant women in Hanover Park, Cape Town. We also aimed to explore the contextual factors associated with violence in the homes of those women who received mental health counselling.

In our sample of pregnant women, 15% were experiencing IPV and this was associated with food insecurity, unemployment, unmarried, but stable relationship status, past experience of abuse and discontent with the current pregnancy. Current mental health diagnosis and a self-reported history of mental illness were also significantly associated with IPV. Domestic violence within the household was not limited to intimate partners and, domestic violence in this context was often perceived as ‘normal’ behaviour by the participants.

The prevalence of IPV in this study was 15% using the CTS2 tool, which is considerably lower than the finding from a survey of pregnant women attending a public sector South African antenatal service. That study reported 38% had experienced abuse from a partner at some point in their lives, with 35% reporting domestic violence during the current pregnancy [15]. It is possible that women who experience mental health problems or domestic violence may be more likely to decline participation in a study of this nature. However, this does not explain the discrepancy in the findings of the two studies [15]. The authors of the Mbokota study, both health professionals, indicated that they used directed interviews to investigate the presence of domestic abuse. It is possible that this method was perceived by participants as a more confidential space than a survey questionnaire administered by a research assistant, and thus increased disclosure. Other literature supports increased detection via face-to-face screening [52] and further increased identification of DV when this was done by a health professional [53]. The prevalence in our study is the same as the overall prevalence yielded by a meta-analysis review on IPV during pregnancy in Africa [21]. However, while the case study sample size was very small, the prevalence of DV was higher than that reported by the survey sample: 32% as opposed to 15%. Several factors could contribute to this difference. The case study notes were from women who had received counselling for mental health problems and would therefore have been a high risk group. The association between mental illness and IPV has been reported in South Africa [54–56], as well as in global studies [8–10]. Further, women may have been more open to disclosure of DV during counselling than in a survey. Other studies have demonstrated that asking open-ended questions about DV are more likely to elicit disclosure [57, 58]. In addition, women from our case note sample reported violence within the household, of which, 55% was perpetrated by a non-intimate partner. Another South African study reported high levels of domestic violence during pregnancy, with 24% of the abuse being perpetrated by the mother-in-law of the pregnant woman [15]. This underscores the need to screen for domestic violence as perpetrated by other members of the household, and not only by intimate partners.

We found that women who were unemployed were more likely to have reported experiencing violence from their partners. While the settings are not the same, these findings contrast with a study conducted in Ugandan that found no difference in the experience of abuse between women who were unemployed and those who were employed in either the formal or informal sectors [50]. Further, we found that food insecurity was associated with IPV. Shamu et al. assert that, in Africa, the feminization of poverty means that many poor women rely on their partners for household maintenance and access to pregnancy care. Men who are perpetrators of physical violence exploit this economic vulnerability by abusing their partners [21]. This economic exploitation is a further abuse, in and of itself [4, 59].

We identified intimate relationship status and perceived support as significant factors associated with abuse. Women in stable but unmarried relationships were more likely to have reported experiencing intimate partner violence than those who were married. This is supported by findings from a United States (US) based study on low-income pregnant women [60]. Further, our findings concur with evidence from low and middle income countries that a lack of perceived support from a significant other, is significantly associated with IPV [19, 61]. Interventions targeting domestic violence may be enhanced by linking the survivors of violence to supportive networks and by assisting them in identifying and maximising any existing supportive relationships. We showed a significant association between those women who were not pleased about the current pregnancy and IPV. This is supported by findings in a review that consolidated findings from global literature [19] and has been found in a US study among pregnant women in a community setting [20]. We postulate that an unwanted pregnancy reflects dysfunction in interpersonal relationships and or compromised socio-economic status, both of which are themselves factors associated with domestic violence. Further qualitative investigation may elucidate the complexity of these interactions.

Current mental health problems were significantly associated with IPV in the bivariate analyses in this study. This association is supported by studies in both high and low-income settings [8–10, 18, 62] and by a systematic review and metanalysis [63] . Further, a South African, clinic-based study by Mbokota et al. indicated that 78% of the sample who experienced domestic violence during pregnancy experienced psychological problems [15], this is supported by our multivariable analyses, which indicated that women who reported experiencing violence were 2.4 times more likely to experience a mental health problem. However, the Mbokota study [15] failed to indicate how psychological problems were assessed and did not present diagnostic data for specific disorders. Chen et al. provide a systematic review on sexual abuse (only) for any women in any setting, with diagnostic psychiatric data. They found significant associations between a history of sexual abuse and diagnoses of depression, anxiety, eating and sleeping disorders and Post-traumatic Stress Disorder (PTSD). However, few of these studies were from low-income settings [64]. A community-based, South African study on pregnant women, linked depressive symptoms, obtained from screening data, with IPV and alcohol abuse [65]. The association of IPV with the use of alcohol and other drugs, either by the pregnant woman or by members of her household, is supported by other research from the US with urban, minority women [66, 67]. Our study adds to the literature from low-income settings by demonstrating associations with current mental health problems as assessed by diagnostic interview. Additionally, our finding that a history of mental health problems is significantly associated with IPV is widely supported by others’ evidence [8–10, 18, 62]. Although many of these studies, like ours, were not designed to demonstrate causal relationships, the evidence suggests that mental ill-health and violence against women exist in a vicious cycle, maintained by inequitable gender norms, low relationship power, poverty and the societal acceptability of violence [11]. This is supported by work conducted on social determinants of mental ill-health [68]. Machisa et al. described the structural pathways to IPV for a sample of South African women where binge drinking, depression, PTSD and lower relationship power mediated the relationship of prior childhood abuse and recent IPV [11]. Not excluding other factors, we hypothesise that the experiences typical of common mental disorders, i.e. low self-esteem, social withdrawal and a sense of helplessness [69, 70] may confer particular vulnerabilities to violence for pregnant women with these mental health problems. Mental ill-health, possibly through the same mood and cognitive features described above, also compromises women’s access to social and emotional resources [71, 72] which further maintains their vulnerability to victimisation. The co-existence of mental ill-health and violence against pregnant women has implications for the design of interventions for women who experience domestic violence.

In a review of studies on IPV from African countries, a history of experiencing abuse (abuse before the age of 15, abuse in the past 12 months and abuse in lifetime), is strongly associated with IPV during pregnancy [21]. While our qualitative sample size was low, and should not be considered as representative, the findings support this association, with examples of currently abused women reporting having been abused during childhood or by previous partners. For example, “Abused as a child, raped, sodomized/abused by ex-husband”.

The qualitative data from this study provide examples of how domestic violence was considered normative in this sample. This is reflected in the examples: “He hit her against her head and hit her with a fist against her stomach. This is how they normally handle conflict. It doesn’t seem strange/abnormal that they are so violent with each other.” and “She and her husband often get into physical fights with each other. An argument inevitably leads to fighting.” Several studies from low income settings have produced findings indicating the influence IPV has on social norms. Women who experienced family violence as a child are more likely to perpetrate violence [73], and find wife beating acceptable [74–77]. A WHO World Report on Violence and Health indicates that social norms and values play a powerful role in how violence is perceived, condoned, inhibited and responded to [78].

Benjamin (2014) draws on the concept of dysfunctional community syndrome, where violence takes many different forms and occurs increasingly over generations and at increased intensity, to describe the setting in Hanover Park [79]. Benjamin and Crawford-Browne (2011) refer to observed patterns of behaviours in clients from Hanover Park who attended counselling over a number of years, and were exposed to continuous trauma due to their violent environment. These clients had an inability to regulate emotion which was demonstrated by heightened or flattened affect in response, high levels of aggression, a reduced capacity for empathy and an inability to regulate impulsivity. There was also the tendency to minimise the impact of trauma [80]. The behaviours they describe seem to indicate how acceptable violence is in this environment, and is reflected in the examples provided in the qualitative descriptions of our study.

In the context of normalised violence, gender-based power disparities and poverty typical within the South African settings such as this [23], it is thus not surprising that domestic violence is perpetrated both by intimate partners as well as by other members of the household. Our finding that 55% of the case study sample had experienced domestic violence by a member of the household who was not an intimate partner concurs with other studies that have disaggregated the perpetrators of domestic violence in samples with pregnant women [12, 14, 15].

This demonstrates the need for intervention design to take account of motivations for violence and to deconstruct relationships of power and coercion. Further, our study reinforces the imperative to change societal norms regarding the acceptability of violence, and the need for preventive work and well as programmes that separately target men, women and children.

Our study experienced several limitations in design. Firstly, while the counsellor explored experience of domestic violence in her initial assessment, there was no routine structured format for this and therefore she may not have elicited information around frequency, progression, intensity or perception of violence. Further, the source of the qualitative data was the notes from one counsellor and thus represent her opinions and judgements. There is therefore the potential that some elements were unduly emphasized while others may have been minimized. This potential bias is mitigated by the fact that the counsellor had a four-year university qualification in counselling (Bachelors of Psychology), a formal registration with the national health professions council, had nearly a decade of experience working with clients experiencing high levels of violence, and received weekly individual clinical supervision by a clinical psychologist.

In addition, the definitions of violence across the quantitative and qualitative data differ, one taking into account intimate partner violence only, and the other using the broader definition of domestic violence. The cross-sectional nature of the quantitative data limits an investigation into the progression of violence before, during the pregnancy and after the birth, particularly as the RFA item pertaining to past abuse was unspecific. Further, it precludes interpretations concerning the directions of associations and causality. The study sample, was taken from one site and is relatively small and hence should not be generalised to other settings. All data collected were self-reported and we did not verify information such as asset ownership and prior self-reported mental health diagnosis. There is therefore room for recall bias and misreporting of information. Further, although uptake of antenatal services are high in South Africa (97%) [81], data were collected from service users of a clinic. This means that non-users of antenatal services, who might be the most vulnerable to domestic violence, would have not have been included in the sample.

The findings from this study have implications for service design and implementation. Detection of DV may be problematic, and should not be limited to IPV. Service designers need to be mindful of the possibility of under-detection. Further research may be needed to determine the best methods for this. Women who are more vulnerable to abuse may simultaneously have diminished access to resources due to poverty. Further, interventions should take into account the effects of complex household dynamics on relationships and the implications for safety for women and children.

When detecting domestic violence, we recommend simultaneous screening for mental disorders and the inclusion of mental health care into the design of interventions against domestic violence. In particular, interventions should take in to account the impact that mental distress may have in creating barriers to uptake of any services that are provided. For pregnant women, we would thus recommend social and mental health support for IPV and DV to be provided on-site and integrated into routine antenatal care offered by health centres. Additionally, broad-based interventions that assist with access to social grants or employment skills could assist with breaking the financial dependence on violent partners or other household members [82].

This study contributes towards a greater understanding of the risk profile for IPV amongst pregnant women in low income settings. Adversity including poverty and mental ill-health are closely associated with IPV during the antenatal period. We advocate for increased, holistic and contextually responsive interventions to assist those experiencing domestic violence from partners as well as other members of the household.