Postpartum physical intimate partner violence among women in rural Zambia

Abstract Objective To examine the demographic characteristics and mental health of women in rural Zambia who experienced physical intimate partner violence (IPV) postpartum. Methods The present secondary analysis was conducted using baseline data from an impact evaluation of a maternity waiting home intervention in rural Zambia. A quantitative household survey was conducted over 6 weeks, from mid‐April to late May, 2016, at 40 rural health facility catchment areas among 2381 postpartum women (13 months after delivery; age ≥15 years). Results A total of 192 (8.1%) women reported experiencing any type of physical IPV in the preceding 2 weeks; 126 had experienced severe physical IPV (had been kicked, dragged, beat, and/or choked by a husband or partner). High levels of depression were recorded for 174 (7.3%) women in the preceding 2 weeks. Being a female head of household was associated with an increased likelihood of experiencing severe physical IPV (aOR 2.64, 95% CI 1.70–4.10). Women with high depression scores were also at an increased risk of experiencing any physical IPV (aOR 17.1, 95% CI 8.44–34.9) and severe physical IPV (aOR 15.4, 95% CI 5.17–45.9). Conclusion Future work should consider the implications of government and educational policies that could impact the screening and treatment of pregnant women affected by all forms of physical IPV and depression in rural Zambia.

disease and premature ageing, which in turn can lead to increased morbidity and premature death. 4,5 The prevalence of IPV during pregnancy is 2%-57% across African countries, with a meta-analysis yielding an overall prevalence of 15.23%. 6 Experiencing IPV during pregnancy and in the immediate postpartum period is of particular concern given the adverse health effects for both mother and child. The intergenerational effects of IPV are also well established; IPV during the perinatal period has been linked to negative developmental and health effects among offspring, 7 including future episodes of mental health problems. 8 Witnessing IPV can also affect the attitudes and behaviors of children, increasing the likelihood that they will either perpetrate or experience IPV as adults. 9 Further complicating the issue is the bidirectional association between IPV and mental health. Depression has been linked to IPV both as a consequence of experiencing such behavior and as a risk factor. 10 The published literature on IPV in Sub-Saharan Africa has primarily focused on patriarchal gender norms and demographic factors as potential predictors of IPV and has not explored the role of mental health. 11 However, a South African study 12 found that high levels of depressive symptoms were associated with an increased risk of experiencing IPV among postpartum women.
In Zambia, 47% of women and 32% of men agreed that a husband is justified in beating his wife for at least one specified reason. 13 In addition, the overall prevalence of IPV reported among women aged 15-49 years who were currently or previously married was 47%, with 10% of these women experiencing IPV during pregnancy. 13 Previous research on IPV in Zambia has mainly sought to under- Despite the frequency of IPV during the perinatal period, only limited investigation has been conducted among reproductive-aged women living in rural areas of Sub-Saharan Africa, including Zambia.
In addition, little evidence exists regarding the prevalence of IPV and its relationship with depression among postpartum women in Zambia, particularly among those living in remote areas.
An understanding of the driving forces behind IPV in Sub-Saharan Africa is imperative to meet international benchmarks, including the United Nations Sustainable Development Goal 5 (gender equality). 15 Given that the perinatal period is critical for maternal and child health, we evaluated the demographic and mental health risk factors associated with IPV in three districts of Zambia (Eastern, Southern, Luapula) currently receiving increased programmatic infrastructure surrounding maternal health.
The aims of the present study were to determine the prevalence of postpartum physical IPV among women in rural Zambia and to evaluate demographic characteristics and mental health status (e.g. depression) within this population.

| MATERIALS AND METHODS
The present secondary analysis included baseline data from an impact evaluation of a maternity waiting home (MWH) intervention in rural Zambia. 16 A cross-sectional quantitative household survey was conducted over a 6 week period between mid-April and late May, 2016, at 40 rural health facility catchment areas before the establishment of a maternity waiting home intervention designed from formative research in seven Saving Mothers, Giving Life districts in Zambia. 17,18 Saving Mothers, Giving Life is a country-wide initiative to improve maternal morbidity and mortality. Maternity waiting homes are physical structures built near rural health facilities that provide women with a place to stay before and after delivery. The districts targeted in the present study were Choma, Kalomo, Lundazi, Mansa, Nyimba, Pemba, and Chembe. Ethical approval for the household survey was obtained before data collection from the institutional review boards of the University of Michigan (Ann Arbor, USA) and Boston University (Boston, USA), as well as the research ethics committee of ERES Converge, Lusaka, Zambia. Informed consent was obtained from all participants before data collection.
A multistage random sampling approach was used to select a representative sample of women living in remote dwellings within the 40 health facility catchment areas. First, all villages were geocoded to identify those located at least 9.5 km (rounded up to nearest kilometer) from the catchment area health facility by the most direct travel routes using ArcGIS Online (Esri, Redlands, CA, USA). Villages were then randomly selected from this sample using probability proportionate to population size. A maximum of 10 clusters was selected per each health facility catchment area. Second, all eligible households within the selected villages were listed with the assistance of village leadership and community members. Systematic random sampling was then used to select every nth household from the list to approach for participation until the required sample size for that village was reached-full details of the process have been published previously. 16 Eligible women were from unique households, had delivered within the preceding year, and were aged 15 years or older. Participants were excluded if they were unwilling or unable to provide informed consent.
Although the survey asked about deliveries within the past 12 months, birthdates are often approximated in Zambia; therefore, the range of the present sample went up to 13 months. Households were defined as usually cooking together based on their cultural background. If there was more than one eligible woman in the household, one of them was randomly selected for inclusion in the present study.
A team of Zambian research assistants, literate in both the appropriate local languages (Bemba, Nyanja, Tonga and Tumbuka) and English, were trained in human participant protection and data collection methods during a 5-day training program. Data were captured electronically using SurveyCTO Collect version 2.212 (Dobility, Cambridge, MA, USA), which was installed on encrypted tablet devices.
Participants were initially contacted through personal visits from the research assistants. They were then invited to select a space where they felt comfortable and could speak in private. All survey questions were read aloud to the participants by the research assistants; each survey took approximately 45 minutes to complete. Participants received a small token of appreciation, valued at approximately US $2, for their time.
Household and individual sociodemographic variables assessed were household size, marital status (married or cohabitating, divorced, separated, widowed, and never married), number of deliveries, number of wives shared with a husband, and age. These variables were selected on the basis of past research showing them to be predictors of physical IPV in Sub-Saharan Africa. 13,19 Head of household was ascertained by the question "Are you the head of household?" Women who responded "no" were then asked, "What is your relationship to the head of household?" The scale used to assess depression comprised four items ask- Data on recent physical IPV were collected by asking how often women had been pushed, shoved, or slapped by their husband or partner in the previous 2 weeks. They were also asked how often they had been kicked, dragged, beaten, or choked by their husband or partner in the previous 2 weeks (classified as severe physical IPV). Participants could select from four categories that ranged from "never" to "almost always." The data were analyzed using Stata version 14.0 (StataCorp, College Station, TX, USA). Descriptive statistics and adjusted odds ratios (aORs) were calculated to examine the associations between sociodemographic characteristics, depression, and physical IPV (none vs at least one incident). List-wise deletion was used to account for the small amount of missing data. Binary logistic regression models were used to estimate aORs and 95% confidence intervals (CIs) while controlling for the sociodemographic characteristics. The Stata robust cluster estimator was used to account for clustering within each of the seven districts. P<0.05 was considered statistically significant.

| RESULTS
The response rate among the 2741 women invited to participate in the present study was 86.9%, giving a final sample size of 2381.
Of the women who were eligible to participate but who did not respond, 280 (10.2%) were unavailable owing to their work in the fields during harvest, 60 (2.2%) refused participation, and 20 (0.7%) withdrew after beginning the survey or else had incomplete surveys and so were dropped from the analysis.

| DISCUSSION
The present study found prevalence rates for any physical IPV and severe physical IPV of 8.1% and 5.3%, respectively, among a group of postpartum women living in rural Zambia. The risk of experiencing such behavior was increased by being a female head of household or high levels of depression.
Most IPV prevalence studies have used the past 12 months to indicate "recent" experiences of such behavior. 12,22 By contrast, the present study examined IPV during the past 2 weeks. Despite this marked difference in time frame, the current prevalence rates were comparable to previous reports of past year physical and/or sexual IPV in the postpartum period (5.2%-10.5%). 12,22 The present study found that postpartum women in femaleheaded households were more likely to report physical IPV, specifically severe physical IPV, in the preceding 2 weeks than those in maleheaded households. This observation was in agreement with a study conducted in Haiti, which found that women in communities with a high proportion of female-headed households showed increased risk of sexual IPV. 23  as less between-community variation in financial status exists in rural Zambia, the relationship between IPV and head of household has been measured at an individual level rather than the population level. 13 Nontraditional gender roles might also influence the association between head of household and IPV. Previous work in Zambia found that the most common reasons given by reproductive-aged women for justifying IPV were when women had transgressed from their expected gender roles; for example, equal autonomy related to household decisions. 25 The present study also found an association between physical IPV and depression, although a causative link could not be established however, there has been little change in the rates of IPV. 13 The results of the present study indicated a need to focus on both IPV and men-

AUTHOR CONTRIBUTIONS
MLM-K contributed to the design of the study, data analysis and interpretation, and writing the manuscript. NS contributed to designing the study, data collection and interpretation, and writing the manuscript.
CJB, SMM, and JRL contributed to designing the study, the interpretation of data, and writing the manuscript. PTV contributed to the analysis and interpretation of data, and writing the manuscript. GM contributed to designing the study, data collection, and writing the manuscript. All authors provided final approval of the version to be published and agreed to be accountable for the accuracy and integrity of the work.

CONFLICTS OF INTEREST
The authors have no conflicts of interest.