Intimate partner violence and current modern contraceptive use among married women in Uganda: a cross-sectional study

Introduction This paper examined the relationship between Intimate Partner Violence (IPV) and current modern contraceptive use (MCU) among married women in Uganda. Methods We used the 2011 Uganda Demographic and Health Survey (UDHS) data, selecting a weighted sample of 1,307 married women from the domestic violence module. Chi-squared tests and multivariate complementary log-log (clog-log) regressions were used to examine the relationship between IPV and current MCU, controlling for women's socio-demographic factors. Results Significant predictors of current MCU (25.3%) among married women were: women's reported ability to ask a partner to use a condom, number of living children and wealth index. The odds of current MCU were higher among women who could ask their partners to use a condom (aOR = 1.87, 95% CI: 1.26-2.78), had more than one child (aOR = 2.05, 95% CI: 1.07,3.93) and were from better wealth indices for example the richest (aOR = 2.52, 95% CI: 1.25-5.08). IPV was not associated with current MCU independently and after adjusting for women's socio-demographic factors. Conclusion In Uganda's context, IPV was not associated with current MCU. Interventions to promote MCU should enhance women's capacity to negotiate MCU within union and target women of lower socio-economic status.


Introduction
Modern contraception is essential for averting maternal deaths through reducing unintended pregnancies and unsafe abortion [1].
Whereas contraceptive prevalence rate (CPR) at the global level was 57% in 2012, sub-Saharan Africa (SSA) lagged behind with a CPR of 24% [2]. In particular, Uganda's modern contraceptive prevalence of 26%, is the lowest in the East African region [3]. Contextual gender relations between men and women influence the latter's reproductive health behavior including fertility regulation or contraception [4]. One of the ways in which men's behaviours influence women's reproductive health outcomes is through intimate partner violence (IPV). Generally, IPV is associated with poor reproductive health outcomes among women [5,6]. Intimate partner violence (IPV) refers to physical, sexual or psychological/emotional violence perpetrated by a husband or intimate partner [7][8][9]. It includes acts of physical and psychological abuse, forced intercourse and other forms of sexual coercion and controlling behaviors. The prevalence of intimate partner physical, sexual and emotional violence in the last 12 months preceding the 2011 survey in Uganda were 25%, 21% and 33 respectively. About 45% had experienced any of the three forms of violence in the last 12 months from an intimate partner [3]. The association between IPV and modern contraceptive use (MCU) in Africa has mixed findings: some studies report that IPV is associated with increased or decreased contraceptive use, while others report no relationship at all [7]. IPV has been reported to increase MCU in Africa [10,11] and Nigeria [12]. Elsewhere in sub-Saharan Africa (Cameroon, Kenya, Malawi, Rwanda, Uganda and Zimbabwe using a pooled sample), IPV increased the odds of female controlled modern contraceptive use [10]. Additionally, a systematic review based on longitudinal studies of IPV and contraception also found a significant relationship between IPV and contraceptive use [13]. The explanation for the significant association is that women in such contexts do not wish to have children in conditions that are not conducive [10].
Concerning specific forms of IPV, a study in the Democratic Republic of Congo established that intimate partner sexual violence (IPSV) predicted contraceptive use rather than the combination of intimate partner physical (IPPV) and sexual violence. In the same study, IPSV was a stronger predictor of contraceptive use [7,10]. In Guinea, women who experience IPV resort to use of injectable, which is not conspicuous to their partners [14]. Other studies have reported that IPV decreases the odds of MCU [15]. IPV was significantly associated with inability to negotiate condom use [16,17]. In addition, experience of IPV limits control over timing of sexual intercourse, thus limiting effectiveness of some contraceptives such as the condom [7,18,19]. A pooled analysis of seven countries in central and west Africa reported that highly tolerating perceptions to domestic violence reduced the odds of using modern contraception [20]. In Egypt, women who experienced physical violence were less likely to use MCU [21]. On the other hand, some studies have found IPV not associated with MCU for example in the Democratic Republic of Congo (DRC) [7], Zambia and Malawi [22]. Since the evidence about IPV and modern contraceptive use is not consistent in Africa, several studies recommend conducting a country specific analyses due to differences in societies and cultures [11]. In Uganda, several studies have assessed predictors of contraceptive use in general [10,[23][24][25][26]. One of the studies investigated the association between lifetime modern contraceptive use using the 2003-2006 Uganda Demographic and Health Survey (UDHS) data [10]. A multi-country study using DHS data in 13 African countries (including Uganda) focused more on the interaction between physical IPV and MCU [27]. Therefore, this study aimed at investigating how a combination of the three measures of IPV (physical, sexual and emotional violence) are associated with current MCU.

Methods
Data source: We used the 2011 Uganda Demographic and Health Survey (UDHS) data, with permission from the Demographic and Health Survey (DHS) Program website [28]. Data were collected using a cross-sectional nationally representative survey that employed a stratified two-stage cluster sampling design based on the sampling frame from the 2002 Population and Housing census.
Details of the sampling procedure can be accessed in the 2011 UDHS report [3]. A total 2,056 (unweighted cases) ever-married women were selected for the domestic violence module. From this sample, we extracted a weighted sample of 1,307 currently married women. A woman was considered currently married if she was either married or living with a partner [3]. We used the domestic violence weighting variable (d005) found in the UDHS individual women's dataset and the Stata survey (svy) command to apply individual weights during the analyses. Survey weighting was necessary to account for the complex survey design [29].

Measures of outcome variable: Modern contraceptive use (MCU)
was generated out of variable V313 "current contraceptive use by method type". It was recoded as a binary outcome (1 = "yes" or 0 = "no"). The first "no method", the second "folkloric method" and third "traditional method" categories were coded 0 (No) while "modern method" was coded as 1 (Yes). Modern contraceptives included the pill, IUD, injections, diaphragm/foam/jelly and condom, female sterilization, male sterilization and implants. Lactation amenorrhea method cases (5 in number), which were part of the outcome variable were dropped from the analysis [3]. Questions addressing intimate partner sexual violence (IPSV)were whether the respondent had ever been: a) Physically forced by the partner to have sexual intercourse with him even when she did not want to; b) Physically forced to perform any sexual acts against her will; c) Forced her with threats or in any other way to perform sexual acts she did not want to Intimate partner emotional violence (IPEV) questions were whether the spouse or partner: a) Said or did something to humiliate her in front of others; b) Threatened to hurt or harm her or someone she cared about; c) Insulted her or made her feel bad about herself. An affirmative response to a question was followed with a question about the frequency of the act in the 12 months preceding the survey. A "yes" answer to any of the items in each category constituted physical or sexual or emotional violence [3]. The binary responses (yes or no) from the three categories of questions on physical, sexual and/or emotional violence), were merged into an aggregate measure of IPV [3], coded as 0 = no, and 1 = yes [9]. Women's sexual empowerment was measured by (aOR) and their 95% confidence intervals. All the analyses were weighted to account for complex survey design, clustering and stratification [29]. In order to explore IPV further, we tested for multi-collinearity among variables (results not presented) and found that the mean variance inflation factor (VIF) was 1.52, where household wealth index had the highest (1.98). When the VIF is equal or greater than 5, then multi-collinearity is a problem in the model which means that there was no problem with multicollinearity among the explanatory variables. However, a correlation test showed a strong positive correlation between age group and the number of living children. We tried to remove age group and kept the number of living children in the model, there was a negligible change in model results (results not presented).
Therefore, we put back the variable age group in the model.

Results
Descriptive characteristics of the women: Results in Table   1 show that one in four (25%) of the women considered for the domestic violence model used modern contraceptives. Seven in ten Page number not for citation purposes 4 (70%) were below 35 years. Geographical regions were proportionately represented ranging from 19% for Northern to 28% for Central Uganda. The majority (84%) were rural residents and Christians by religion (Catholics 40% and Protestants 29%). Over half (55%) of the women were married. The majority had two or more children (80%), primary or no formal education (78%) and belonged to middle or poor wealth quintiles (59%). Just over half (53%) were engaged in agriculture as their main occupation (Table   1). With respect to sexual empowerment, 75% indicated that they could ask their partner to use a condom. Overall, nearly six in ten (59%) married women experienced any form of IPV (IPPV 41%, IPSV 17% and IPEV 42%) in the last 12 months preceding the survey. Current MCU prevalence was 25% among married women in Uganda. More than half (51%) of the women used injections compared to 18% who used condoms. One in ten women (12%) had undergone female sterilization.

Discussion
This paper aimed at investigating the association between IPV and current modern contraceptive use among married women in Uganda. Only a quarter (25%) of the married women were using modern contraceptives. This was slightly below the national contraceptive prevalence rate of 26% [3]. Nearly six in ten (59%) women reported to have experienced IPV (physical, sexual or emotional) in the last 12 months preceding the survey. Current MCU was associated with women's ability to ask partners to use condoms, number of living children, wealth index but not with IPV.
Experiencing IPV in the last 12 months was not associated with current MCU: both at univariable level (Table 2) and multivariable analysis (Table 3). In this study, the results confirmed absence of a direct association between IPV and current MCU among married women in Uganda. As highlighted in the introduction section, findings on the association between IPV and MCU in SSA are mixed [7,27]. Several studies using DHS data in African countries have reported similar scenarios, with no significant association between IPV and current MCU. IPV was not associated with MCU in the Democratic Republic of Congo [7], Zambia and Malawi [22]. A multicountry study including Uganda also found no association between IPV and MCU, except that its analysis was focused on intimate partner physical violence only [27]. A qualitative study suggested that gender-based violence in general was associated with use of family planning [26]. This paper aimed to include emotional, sexual and physical violence in the measurement of IPV. Forced sex and negotiation for condom use are the critical pathways for the association between IPV and MCU [7]. In this study, intimate Page number not for citation purposes 5 partner sexual violence was low (27%) while women's ability to negotiate for condom use was high (75%) among married women in Uganda (Table 1). However, studies which have reported association between IPV and MCU have used life time (ever use of) modern contraceptive use [7,10], yet our study used current modern contraceptive use. IPV by a current partner is a best assessed with current MCU [7,10,13,16,30]. Sexual empowerment measured by respondents' ability to ask a partner to use a condom, was significantly associated with MCU. This finding is in agreement with some study in Uganda [23] and in Ghana [31].
The variable is closely related to MCU since the male condom is a contraceptive (as well as a sexually transmitted infections preventive device). Condom use (18%) was the second most popular method used by married women in Uganda, after injections (51%) as indicated in Table 1.
In Uganda's context, women's socio-economic status (measured by household wealth index) seemed to be more important in predicting current MCU, than IPV. From Table 3 other studies should be taken with caution because some studies used different methods [7,30], datasets, settings, period [10,30,33,35] and pooled data prior to analysis [10]. In some cases, all rather than modern methods of contraception were considered. Some studies focused on IPPV only not all the three measures of IPV [27].

Conclusion
Our findings show that women's reproductive rights expressed by a woman's reported ability to ask a partner to use a condom was a