Introduction

Intimate partner violence (IPV) is defined as behavior within an intimate relationship that causes physical, sexual, or emotional harm, and is the most common form of violence against women (VAW) [1, 2]. It is estimated that 30% of women experience IPV in their lifetime, making it a global health concern [2, 3]. In Kenya, the 2014 Demographic and Health Survey showed that 41% of ever married women reported lifetime IPV, while 23% reported IPV in the past year [4].

According to the ecological framework of violence against women, a combination of individual, relationship, community, and societal factors play a role in IPV among women [5, 6]. Individual risk factors include age, pregnancy, alcohol, and a prior history of violence. The risk of IPV is generally higher in younger women [6, 7]. There are conflicting findings on whether pregnancy increases or reduces the risk of IPV [8, 9]. A review of studies in developed countries showed higher risk, while a review of African studies showed lower risk of IPV for women during pregnancy [10, 11]. There is evidence suggesting that alcohol use among women predisposes them to IPV [7, 12, 13]. Previous history of violence has also been consistently associated with higher risk of IPV [7, 14].

Relationship-level risk factors for IPV include marital status and male partner controlling behaviors [5, 7]. Some studies show that married women are at increased risk of IPV, while others show reduced risk compared to women who are single or divorced [7]. A literature review of studies conducted in the United States found that separated or divorced women were at higher risk of IPV compared to married women [7]. The reviewed data suggested that women may experience IPV during times of relationship stress, such as negotiations with a male partner about children or property [15]. Married women also experience IPV, particularly in the setting of ongoing marital conflict [6, 7]. Women who have male partners with controlling behaviors are at increased risk of IPV [12, 16].

Community-level risk factors for IPV include poverty and lack of family support [6, 17]. Women experiencing poverty may be financially dependent on their partners and lack alternative means of survival, making it difficult to leave abusive relationships. Additionally, those who lack family support are likely to be dependent on their male partners, and not report abuse [17]. At a societal level, violence against female partners is viewed as the norm in some African cultures, and women are expected to endure abuse from their partners without complaining or seeking help [17, 18]. Lack of legislation categorizing IPV as a criminal offence may also perpetuate IPV [6, 17].

Numerous studies have suggested that human immunodeficiency virus (HIV) infection predisposes women to IPV [12, 19, 20]. Women in serodifferent relationships may experience IPV after disclosure, where lack of trust and concerns regarding infidelity as a possible source of the HIV infection may lead to conflict [21, 22]. In addition, women with HIV in serodifferent partnerships may suffer from IPV-related outcomes such as depression, post-traumatic stress disorder, alcohol and substance use, all of which may interfere with their adherence to antiretroviral therapy (ART) and attendance of clinic appointments, hindering efforts to manage HIV and reduce transmission risk [23].

Improving our understanding of the correlates of IPV in women with HIV in serodifferent relationships can help to identify women at risk, inform interventions to reduce IPV, improve HIV-related health outcomes, and reduce the risk of transmission to partners and infants [12]. This study aimed to determine the prevalence and correlates of IPV in the past year among women with HIV in serodifferent relationships in Kenya.

Methodology

Results

The study was conducted between March 2013 and March 2016. During this time, a total of 216 WLHIV were enrolled for care at the Couples Counseling Center. Of these, 172 (79.6%) were screened for participation in the research cohort. The main reasons for couples not being screened were 25 (56.8%) declined, 14 (31.8%) were living away from the study area and could not make study appointments, and 5 (11.4%) were considered to be too sick to enroll. Of the WLHIV screened, 159 (92.4%) were enrolled in the study. Their baseline characteristics are presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristic of the study sample (N = 159).

https://doi.org/10.1371/journal.pone.0272640.t001

Participants’ median age was 33 years (interquartile range [IQR] 29–38), and more than two thirds had attended school for more than eight years (111, 69.8%). The majority (142, 89.3%) were currently married, and few were pregnant (10, 6.3%). Three quarters of the participants (119, 74.6%) did not drink alcohol. Among the 40 participants with any alcohol use, AUDIT scores were consistent with minimal alcohol use in 32 (80.0%) women, while 8 (20.0%) had AUDIT scores suggestive of moderate/severe drinking. While most participants reported that their partners would be excited (99, 62.3%) if they became pregnant, some reported that they would not care (33, 20.8%), and a few reported they would be upset (17, 11.3%). Partner controlling behavior was reported by more than a third of the participants (66, 41.5%).

When asked about violence since the age of 15 years by someone other than the current partner, 23 (14.5%) reported physical violence and 12 (7.6%) reported sexual violence. Intimate partner violence in the past year was reported by 47 (29.6%, 95% CI 22.9–37.2%) participants. Of these, 32 (68.1%) reported emotional violence, 27 (57.4%) reported physical violence, and 27 (57.4%) reported sexual violence. These percentages add up to more than 100%, as participants could report experiencing more than one type of IPV.

In univariate analyses, pregnancy (prevalence ratio [PR] 2.61, 95% CI 1.61–4.23), alcohol use (minimal drinking PR 2.34, 95% CI 1.47–3.73, moderate/severe drinking PR 2.04, 95% CI 0.64–4.29), any male partner controlling behavior (PR 2.48, 95% CI 1.50–4.12), history of physical violence by someone other than the current partner (PR 2.51, 95% CI 1.61–3.90), and history of sexual violence by someone other than the current partner (PR 1.79, 95% CI 0.96–3.34) were associated with a higher prevalence of IPV in the previous year (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Correlates of any IPV in the past 12 months by the current partner.

https://doi.org/10.1371/journal.pone.0272640.t002

Each additional year of age (PR 0.97, 95% CI 0.93–0.99) was associated with a lower prevalence of IPV in the past year. In the multivariate model, pregnancy (adjusted prevalence ratio [aPR] 2.14, 95% CI 1.09–4.20), alcohol use (minimal aPR 1.91, 95% CI 1.10–3.33, moderate/severe aPR 1.17, 95% CI 0.53–2.59), male partner controlling behavior (aPR 2.09, 95% CI 1.24–3.51), and history of physical violence by someone other than the current partner (aPR 1.93, 95% CI 1.22–3.05) remained significantly associated with higher prevalence of IPV in the past year.

Past physical and sexual violence by someone other than the current partner were collinear, so the less strongly associated sexual violence variable was not included in the multivariate model. Finally, in the multivariate analysis, the association between age and IPV was substantially attenuated (aPR 0.99, 95% CI 0.95–1.03).

Discussion

This study examined the prevalence and correlates of IPV among Kenyan women with HIV in serodifferent relationships. Nearly one third of women had experienced IPV in the past year. The prevalence of IPV was substantially higher in women who were pregnant and in those who reported any alcohol use. In addition, controlling behavior by the current partner and history of violence by someone other than the current partner were associated with a higher prevalence of IPV.

While few studies have examined the prevalence of IPV in women with HIV in serodifferent relationships, a number of studies provide estimates of IPV prevalence in African women with HIV enrolled in care. Two studies from Uganda found prevalence of past-year IPV in the range of 30%, which is similar to the prevalence identified in this study of Kenyan women [32]. In contrast, a third study from Uganda found a prevalence of past-year IPV of 7.5% among women with HIV in care [33].

Differences in the prevalence of IPV in different studies need to be interpreted in the context of the specific instruments used to collect data [34]. The current analysis used the WHO questionnaire with behaviorally specific questions on acts of IPV, which has been found to improve disclosure rates [35]. In contrast, the Ugandan study with lower prevalence used the conflict tactics scale, which does not include questions about specific types of physical or sexual IPV and does not include measures of emotional IPV [33].

In this cohort of Kenyan women with HIV in serodifferent relationships, pregnancy was associated with a higher prevalence of IPV in the past year. During pregnancy, women’s sexual behavior may change as a result of reduced sexual desire associated with hormonal changes and difficulty in sexual positioning due to increased abdominal size, leading to marital conflict and IPV [36, 37]. A previous cohort study conducted among serodifferent couples in Africa found a weaker association between pregnancy and IPV [38]. Direct comparison of these studies is difficult, as the cohort study combined both non-pregnant and pregnant follow-up time for women who were pregnant at any time during follow-up.

Alcohol use was associated with a higher prevalence of IPV in this cohort of women with HIV in serodifferent relationships. Even minimal alcohol use, which was by far the largest category other than abstinence, was significantly associated with IPV. Moderate or severe alcohol use was uncommon, and had a smaller point estimate for association with IPV, but with wide confidence intervals. It is interesting that even minimal alcohol use was associated with IPV, a finding with several possible explanations. A bi-directional association between alcohol use and IPV is plausible, as women may use alcohol to cope with IPV [13]. Additionally, women who use alcohol are more likely to have partners who use or abuse alcohol, putting these women at a higher risk for IPV [13]. Further, women who consume alcohol may experience IPV when the male partner perceives their alcohol use as inappropriate and the women may be unable to negotiate non-violent solutions to conflict, or may initiate conflict which results in IPV [13].

The association between male partner controlling behavior and IPV observed in this cohort is consistent with the findings of prior studies [6, 12, 39]. Some authors have argued that men with controlling behavior dominate the relationship, leading to conflict in which male partners exert their power and control in the form of IPV [39].

In the present analysis, a history of physical violence by someone other than the current partner was associated with a higher prevalence of IPV in the past year. This finding is consistent with previous studies showing that women with a history of violence are at elevated risk of to IPV because they are more likely to meet abusive partners, and use alcohol [13, 14]. Another possible explanation is that women exposed to violence as punishment during childhood/adolescence may perceive that this is a normative way of solving conflict, making them more likely to submit to IPV [40].

The high prevalence of IPV in this cohort of women with HIV in serodifferent relationships underscores the need to roll out interventions to address this important public health concern. Several interventions aimed at reducing IPV have been evaluated [41–43]. In general, the most effective approaches have been multifaceted interventions designed to change entrenched gender norms [42, 43].

Strengths of this study included the use of questions adapted from the WHO VAW survey. This is a standardized instrument used successfully in a range of populations of women in Africa, facilitating comparison to other studies [27]. In addition, potential correlates of IPV were measured using validated tools such as the AUDIT score to measure alcohol use [26]. Furthermore, the present analysis focused on IPV in the past year, emphasizing that IPV is a continuing problem for women with HIV in serodifferent relationships.

This study had some limitations. First, we did not include male partners. Therefore, only limited male partner characteristics (e.g. male controlling behavior) were collected from the female participants, and data such as alcohol use by the male partner was not collected. On the other hand, enrolling male partners could have resulted in women being less willing to participant and/or report experiences of IPV if they knew that their male partners were participants. Second, this study design was not able to ascertain temporal and causal relationships between some of the correlates and IPV. For instance, this study did not examine the temporal relationships between pregnancy and IPV. In this context, it is not possible to distinguish between the hypotheses that pregnancy led to IPV versus sexual IPV resulting in pregnancy. Nevertheless, this analysis shows that women with HIV in serodifferent relationships may be at risk for IPV during pregnancy. Third, there may have been underreporting of IPV due to recall bias or social desirability bias. To address this concern, questions specific to behavior were asked to facilitate disclosure [27, 35]. Fourth, this study was conducted in a cohort that was recruited in Nairobi, Kenya, and may not be generalizable to all African WLHIV in serodifferent relationships. The findings are likely to be most generalizable to other WLHIV in serodifferent relationships in urban settings in Kenya and the surrounding region. Finally, it should be noted that the study was not intended to examine whether being in a serodifferent relationship was associated with a higher prevalence of IPV compared to couples where both partners are seronegative or seropositive.

In conclusion, this study identified a high prevalence of IPV in the past year among women with HIV in serodifferent relationships. Women who were pregnant and those who had experienced prior violence had a higher prevalence of IPV. These data highlight the need to screen for IPV during clinic visits, and to offer evidence based interventions for reducing IPV to affected women in serodifferent relationships.

Supporting information

Acknowledgments

We would like to appreciate the study participants who shared their personal experiences with us.