Correlates of intimate partner violence among married women in Uganda: a cross-sectional survey

Background In Uganda, just like in many sub-Saharan countries, studies on Intimate Partner Violence (IPV) among married women are limited. The aim of this paper was to determine the correlates of emotional, sexual, physical IPV and any form of IPV among married women in Uganda. Methods The 2016 Uganda Demographic and Health Survey (UDHS) data was used, and a weighted sample of 6879 married women were selected from the Domestic Violence module. Frequency distributions were used to describe the characteristics of respondents. Chi-square tests were used to establish the association between IPV and the explanatory variables. Binary logistic regressions were used to establish the factors that were associated with IPV among married women in Uganda. Results More than half (56%) of the married women experienced some form of IPV. Sexual IPV was the least prevalent (23%) and 4 in 10 women (41 and 40%) experienced physical and emotional IPV, respectively. Factors associated with all the different forms of IPV included, age, region, witnessing parental violence, partner’s controlling behaviors, duration of the relationship, and frequency of intoxication of the male partner. Conclusion IPV among Ugandan married women is far too common. This calls for collective efforts to reduce IPV in Uganda by addressing excessive alcohol consumption, controlling behaviors, and lack of awareness of the issue. Interventions aimed at preventing perpetration and tolerance of violence in the home settings should be promoted.


Background
Globally, gender-based violence (GBV), also referred to as domestic violence, has gained momentum as a social, health and human rights issue [1,2]. Of all forms of GBV, intimate partner violence (IPV) is the most common form, which involves all physical, sexual, or psychological harms as well as controlling behaviors aggravated by a current or former partner [3,4]. The incidence of IPV is more severe in women compared to men with approximately 30% of women worldwide reporting violence by an intimate partner at some point in their life [5][6][7]. In low and middle income countries (LMICs), the prevalence of IPV is higher, with about 37% among women age 15-69 years [8].
In Uganda, IPV is still the highest contributor to GBV with a prevalence level of 40% among ever married women in the 12 months preceding the 2016 Uganda Demographic and Health Survey (UDHS), a rate which surpasses that of the rest of the world (30%) [9,10]. While IPV has generally declined among women over time, its prevalence remains unacceptably high among Ugandan women. It contributes greatly to morbidity and mortality and its consequences are dire, including unwanted pregnancies, sexually transmitted infections, miscarriages, unsafe abortions, stillbirths, premature labour, low birth weight, anxiety, depression, among others [10][11][12][13]. Additionally, national statistics reveal that more than half (51%) of the cases of violence go unreported in Uganda [9]. This is partly due to the tolerance or acceptance of violence, which is rooted in socio-cultural beliefs that men are unconditionally entitled to sex [9].
The risk factors for IPV are complex and multifaceted, which calls for sound theoretical frameworks [14,15]. Therefore, the nested ecological framework theory [16][17][18] and the social learning theory [16,19,20] were used to conceptualize this study. Multi-level hierarchical relationships at the proximal, intermediate and distal levels are explored in relation to IPV [21].
Socio-demographic factors like age, education, region and wealth status have been associated with IPV [1,13,[21][22][23][24][25][26][27]. These function through intermediate factors (partners' behavioral factors, history of violence and marital factors) to drive IPV. Intermediate factors, especially controlling behaviors, precede and catalyse IPV [25,28]. Also, the economic empowerment of women, which is assessed by a woman owning a house or land (either alone or jointly with a partner) or receiving cash payment for her work, is a predictor of IPV.
Some studies on IPV in Uganda have been done in consonance with Article 33 of the Constitution of the Republic of Uganda and the Domestic Violence Act 2010 [24][25][26][29][30][31]. However, studies focusing on "all forms of IPV" are limited. In addition, marital factors remain under-investigated. Therefore, this study considered marital factors (duration of relationship, number of co-wives, age at first marriage and parity) in its conceptualization. This study sought to establish the correlates of all forms of IPV among married women in Uganda.

Data source, study population and sample size
The study used data from the 2016 Uganda Demographic and Health Survey (UDHS), accessed with permission from DHS Program [32]. The UDHS (2016) was a cross-sectional nationally representative survey capturing national and sub-national estimates including, but not limited to, domestic violence and IPV in particular.
This survey employed a two stage stratified sampling procedure and cluster sampling design based on the sampling frame from the 2014 National Population and Housing Census [9,33]. The domestic violence module, however, was based on the shortened and modified version of the Conflict Tactics Scale (CTS) [34]. An indepth description of the sampling procedure is reported in the 2016 UDHS report [9].
This study used a sample of 9232 women selected in the domestic violence module who were age 15-49 years. From this sample, a weighted sample of 6879 married women (either married or cohabiting) were selected for the analyses. We used the domestic violence weighting variable (d005) included in the UDHS data and the Stata survey (svy) command to weight the data during the analyses in order to account for the complex survey design [35].
Variables and measures Figure 1 shows the conceptual framework used in selecting different sets of variables and measures used in the study. Categories of variables included outcome and explanatory variables.

Outcome variables
The outcome variables were the three different forms of IPV (emotional, physical and sexual). In addition, an aggregate measure of IPV, which combined all the three forms of violence was generated. The dependent variables were based on the following set of questions asked to women in the survey. Women indicated whether their husbands/partners had ever or did; a) Hit, slap, kick or do anything else to hurt them physically? b) Force them to have intercourse or perform any other sexual acts against their will? c) Say something to humiliate them in front of others, threaten to hurt them or someone they care about, insult them or make them feel bad.
The response expected was either 'yes' or 'no'; with 'yes' to the questions a, b, and c implying experience of physical, sexual and emotional IPV respectively and 'no' implying no experience of IPV. In addition, a 'yes' to any of the three questions a, b and c implied experience of any IPV and a 'no' implied no experience of any IPV.

Measures of explanatory variables
The independent variables were classified into two broad categories, background factors and intermediate factors.
Background factors included socio-demographic factors and economic empowerment, while intermediate factors included partners' behavioral factors, history of witnessing parental violence and marital factors.
The socio-demographic characteristics included women's age, place of residence, women's education level, region, wealth status index, partner's education, parity and age difference. Economic empowerment included female ownership of property (house and land) and type of earning from a woman's work. It was obtained by merging women responses to questions: does a respondent: a) own a house? b) own land [either alone or jointly with a partner for both questions (a) and (b)] and c) the type of earning from her work. Analysis dichotomized question c into paid (cash only, cash and in kind, and in kind only) and not paid. A 'yes' to any of the three questions a, b and c implied that a woman was considered empowered and a 'no' implied non empowerment. Responses to these questions were recoded into two categories (0 = Not empowered, 1 = Empowered).
Partners' behavioral factors comprised of partner's controlling behaviors, partner's alcohol consumption and frequency of a husband / partner being drunk. To measure the partner's controlling behaviors, female participants were asked, "Does your partner ever or did; a) Prohibit you to meet female friends? b) Limit you contact your family? c) Insist on knowing where you are at all times? d) Is jealous if you talk with other men? and e) Frequently accuses you of being unfaithful?" These were merged into one variable called the "partner's controlling behaviors." Any affirmative response (yes) to any of the above questions implied presence of partner's controlling behaviors and no to all the questions implied nonexistence of such behaviors. The partner's alcohol consumption was measured by responses to the question, "Does your partner drink alcohol?" and it had a binary outcome (0 = No, 1 = Yes). Frequency of a partner being drunk was a follow-up question to those respondents whose partners indicated that the partner drank alcohol.
History of violence comprised witnessing parental violence and was measured by whether the respondent's father ever beat her mother. It had a binary outcome (0 = No, 1 = Yes). The study also considered marital factors which included duration of relationship, number of Addiction to local beer (which is part of culture in some regions in Uganda) leads to drunkardness, which is a precursor to the experience of IPV. Therefore, region (a socio-demographic characteristic) can function through harmful use of alcohol (intermediate factor) to drive IPV co-wives, parity and age at first cohabitation/marriage. Both number of co-wives and age at first marriage had binary outcomes, duration of relationship was categorised as 0 = 0-4 years, 1 = 5-9 years, 2 = 10-14, 3 = 15-19 and 4 = 20+ years. Parity was categorised as 0 = None, 1 = 1-4 and 2 = 5+. In this study, the term "partner" included husbands and also partners in cohabiting relationships.

Statistical analyses
Data analysis was done using Stata version 14. Descriptive statistics (specifically frequency distributions) were used at univariate analysis to describe the characteristics of the respondents. Pearson chi-square tests were used to test initial associations. Finally, multivariable logistic regressions were used to assess the association of the explanatory factors on the experience of emotional, sexual, physical IPV and any form of IPV as an aggregate measure of the three variables. Also, goodness of fit of the different models was tested and checked by running the link test.

Prevalence of IPV
Almost an equal number of women experienced both physical and emotional forms of IPV (41 and 40% respectively), while sexual IPV was the least common (23%). Overall, the majority (56%) of the women experienced any form of IPV. Table 1 presents the background characteristics of respondents. More than half (76%) of the women resided in rural areas, had primary education (60%), were economically empowered (86%), and their partners had mostly a primary education (53%). Fifty percent of women reported an age difference of 0-5 years with their partners. There was nearly an even distribution of women by region. The Northern region had the fewest (20%) women, while the Central region the most (28%) representation. Table 2 presents other IPV related factors. The majority (71%) of women experienced partner controlling behaviors, married after 18 years of age (52%), never had co-wives (74%) and had partners that never drank alcohol (57%). About 2 in 5 women witnessed parental violence (36%). Table 3 presents cross tabulations for associations between the different explanatory variables and experience of the different IPV forms. Except for partner's age differences, all other factors were significantly associated with either emotional, sexual, physical or any IPV. Remarkably, women who were over the age of 35, resided in a rural area, married before age 18, represent the poorest wealth quintile, or personally witnessed parental violence also experienced all forms of IPV. There was an almost even distribution of IPV by region. Women whose partners had controlling behaviors, consumed alcohol and had more than one wife experienced more IPV, while those whose partners had a secondary education or above had the least prevalence of IPV. Wealth status was strongly associated with physical IPV and any IPV. Specifically, women in the poorer, middle, richer and richest wealth quintiles had lower odds (OR = 0.79; 95% CI: 0.63-1.00, OR = 0.71; 95% CI: 0.56-0.91, OR = 0.67; 95% CI: 0.51-0.89 and OR = 0.53; 95% CI: 0.39-0.72 respectively) of experiencing physical IPV compared to those from the poorest wealth quintile.
Partners' controlling behaviors were strongly associated with all forms of IPV. Women who reported partners' controlling behaviors had higher odds of experiencing emotional IPV (OR = 5. 24

Discussion
The aim of this study was to establish the correlates of IPV among married women in Uganda. The prevalence of IPV among these women remains relatively high (56%) compared to an average of 37% for other LMICs [8]. From the results, woman's education, region, wealth status, age, witnessing parental violence, partner controlling behaviors, duration in relationship, frequency of partner being drunk, age at first marriage and parity were significantly associated with IPV in various ways.
Women's education was significantly associated with sexual IPV as women with primary education had increased odds of experiencing sexual violence compared to those with no education. A similar finding was reported in Ethiopia [36], India [37], South Africa [38] and the WHO multi-country study [13]. Perhaps women with no education are more submissive to their partners and less likely to face IPV compared to those with some education. Also, it could be that women with a primary  education are more empowered to recognize and report (during the survey) their experience of IPV as compared to those with no education. While the partner's education was a significant predictor of domestic violence in the WHO study [13], it did not predict IPV against married women (either married or cohabiting) in Uganda. It is possible that mass media "edutainment" strategies have not yet changed social norms or influenced community responses and individual attitudes towards IPV. Education initiatives like Safe Dates in the USA have been applauded for effectively reducing the perpetration of IPV [39,40], however the impact of similar initiatives like life-skills in Uganda is yet to be experienced. Region was significantly associated with all forms of IPV. Particularly, women in the Northern and Western regions of Uganda had higher odds of experiencing emotional, physical and any IPV. Women in Eastern and Western Uganda were also more likely to experience sexual IPV compared to those in Central Uganda. This finding is consistent with other studies by Karamagi, Tumwine [41], Wandera, Kwagala [26] and Annan and Brier [42] where they found that structural factors, like gender inequality, devastating poverty, alcoholism and police corruption, helped sustain IPV. The social acceptance of violence as a tool to resolve conflicts in relationships among some Ugandan societies alongside weak implementation of community sanctions against IPV could explain this finding.
Witnessing parental violence was significantly associated with all the forms of IPV. Women who witnessed their fathers beat their mothers were more likely to experience all the forms of violence compared to their counterparts. Visual learning is an effective mind mapping tool that promotes a community of practice. Also, the social learning theory's argument that perpetration and acceptance of violence are a learned behavior could help explain this finding. This learned behavior is seen among men who perpetrate violence because they witnessed their fathers' violent actions towards their mothers and among women who accept violence because they saw their mothers being abused by their fathers. Recent studies in Uganda and other parts of the world have reported a similar trend [21,26,43,44].
Women whose partners had controlling behaviors were more likely to experience all forms of IPV and similar findings have been reported in Uganda [24][25][26] and in the WHO multi-country study [13]. The patriarchal societies that emphasize male dominance in the family and the dowry that men pay could help explain this finding.
Duration of relationship was related with all the forms of IPV. Women in a relationship for longer durations had increased odds of experiencing violence compared to those with shorter durations. This finding contradicted Urquia, O'Campo [45] who indicated that as much as marital duration was significantly associated with IPV, it is only women with shorter marital duration who experienced more IPV than those with longer marital duration. The explanation could be that women with longer marital duration remain in abusive relationships for the sake of their children and the fear of humiliation that society bestows to those that separate or divorce, in addition to the potential difficulties in starting new relationships.
Once one's desire to drink becomes uncontrollable, alcoholism becomes inevitable and this catalyzes the experience of violence. Similarly, the frequency of a partner being drunk was another predictor of emotional, sexual, physical and any IPV. Women whose partners were often or sometimes drunk had increased odds of experiencing IPV compared to their counterparts whose partners were never drunk. Perhaps men who perpetrate violence while drunk lose control over their behaviors or use alcohol as an excuse for their behavior. Literature that supports this finding indicated that harmful use of alcohol and/ or problem drinking was a precursor to the experience of IPV [30,46,47].
Age of the woman was a predictor of emotional, sexual, physical and any IPV and it had a protective effect. Women in the age bracket 25-34 years were all less likely to experience IPV compared to their counterparts  [22], the WHO multi-country study [48] and Uganda [41,49] support the relationship between age and IPV. Age at first marriage was significantly associated with sexual IPV, with women who married after 18 years of age being more likely to experience sexual violence compared to those who married before the age of 18. The legal age of marriage in Uganda is 18 years old. Studies in Canada [45] and India [50] were in agreement with this study's finding. Perhaps these women have a relatively low relationship power and cannot make certain independent decisions.
Wealth status was significantly associated with physical IPV. Our finding is consistent with the pattern reported in India [22,51] and Uganda [24,29] where women with higher income and those from higher socio-economic status have decreased odds of experiencing violence. However, this study also indicated that women in the poorer and middle wealth quintiles also had lower odds of experiencing physical violence. This seems to contradict other studies [13,21] that reported otherwise.
Closely related to the above, economic empowerment was protective against sexual IPV since empowered women were less likely to experience sexual violence. Our findings are consistent with previous studies [24,25,52] that indicated that empowered women contribute financially to household needs, get involved in decision making and have lower odds of experiencing intimate partner violence.
The number of children (parity) a woman has was associated with emotional, physical and any IPV. Women with one or more children had increased odds of experiencing emotional, physical and any IPV. This finding is in consonance with findings in the DHS Analytical Studies [21], India [53] and Uganda [25].
Concerning the different models, the three models on emotional, sexual and physical IPV culminated into correlates specific to emotional, sexual and physical violence, respectively. However, the generalised model on any IPV lead to correlates and also knowledge that can be generally applied to new situations and environments.

Study limitations
The key limitation was using cross-sectional data. First, we cannot ascertain causality among key variables. In addition, self-reporting of different forms of IPV is associated with social desirability biases and underreporting. Also, there are a number of other key predictors that have been established in other social settings [1,13,28]. Such predictors include marital unhappiness, communication issues, personality, stress, hot button issues, long standing mental health, attitudes, and provoking acts by partners, among others. Unfortunately, due to lack of data on these predictors, they were excluded from analysis. Notwithstanding the limitations, this study provides a robust estimation of the different forms of IPV among women in a relationship age 15-49 years using a nationally representative sample.

Conclusion
Different forms of IPV among women in a relationship had various predictors. The cross cutting correlates for the individual forms of IPV (emotional, sexual and physical) included age, region, witnessing parental violence, partner's controlling behaviors, duration of the relationship, and frequency of male partners being drunk. Parity was a correlate of both emotional and physical IPV. The unique predictors of sexual IPV were economic empowerment and age at first marriage. The overall (generalised) model on any IPV was associated with age, region, wealth status, partners' controlling behaviors, frequency of partner being drunk, witnessing parental violence, duration of relationship and parity.
In a bid to lessen IPV, we recommend a qualitative research approach to help gain a deeper understanding of the region specific issues (structural factors) that underpin regional IPV variations. Also, an investigation into the willingness of men to partake in Adult Education and Lifelong Learning opportunities as a formula to reduce controlling behaviors, alcohol consumption and the number of women they marry is needed. More resources and program interventions should be mobilized by policy makers, public health experts and researchers towards the problem of IPV. Furthermore, more data is required in this area to set up evidence-based strategies that respond to and prevent IPV.