Social Science & Medicine 133 (2015) 2e10

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

Intimate partner violence, modern contraceptive use and conflict in the Democratic Republic of the Congo Rachel Kidman a, *, Tia Palermo a, Jane Bertrand b a

Program in Public Health, Department of Preventive Medicine, Stony Brook University, Health Sciences Tower, Level 3, Room 078, Stony Brook, NY 11794, USA b Department of Global Health Systems and Development, Tulane University School of Public Health and Tropical Medicine, 1440 Canal Street, Suite 1900, New Orleans, LA 70112, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 16 March 2015

Intimate partner violence (IPV) has been found to be negatively associated with contraceptive use in developing countries, but evidence from Africa is mixed. This study examines whether the above association differs in conflict settings, which have the potential for both higher levels of violence and more limited access to family planning. We use nationally representative data from the Democratic Republic of the Congo to examine the relationship between individual- and community-level IPV and modern contraceptive use, and to explore whether conflict modifies the relationship between IPV and contraceptive use. Nationally, only 6% of women reported current modern contraceptive use, while 53% reported experiencing physical IPV and 32% reported experiencing sexual IPV. In multivariate models, we found that individual-level sexual IPV was positively associated with current using modern contraceptive use, but that a combined measure of physical and sexual IPV did not demonstrate a similar association. Community-level IPV was not associated with individual-level contraceptive use. Conflict exposure was neither an independent predictor nor modifier of contraceptive use. Results suggest improved access to family planning should be a priority for programming in DRC, and efforts should ensure that sufficient resources are allocated towards the reproductive health needs of women in both conflict and nonconflict regions. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Intimate partner violence Modern contraception Conflict Democratic Republic of the Congo Family planning Sexual violence

1. Introduction The highest rates of intimate partner violence (IPV) globally are found in Central Africa, where 66% of ever-partnered women report experiencing IPV (Devries et al., 2013). Negative health consequences of IPV include sexually transmitted infections, chronic pain, physical disability, psychological sequelae, and substance abuse (Campbell, 2002; Campbell and Soeken, 1999; Ellsberg et al., 2008; Heise et al., 2002; Peterman and Johnson, 2009; Rees et al., 2011). Additionally, research in developing countries has consistently found that IPV is a risk factor for mistimed and unwanted pregnancy (Gazmararian et al., 1996; Heise et al., 2002; Miller et al., 2010; Pallitto and O'Campo, 2004; Valladares et al., 2002), largely through its influence on contraceptive use (Coker, 2007). The

application of this last finding to African settings has, however, been challenged (Alio et al., 2009). Moreover, we know of no study that examines the interplay between IPV and family planning in conflict areas. This is a critical oversight: women in conflict settings typically have less access to family planning and may experience greater levels of IPV e a dangerous combination. Even for those without personal experience of IPV, heightened IPV levels may deter many women from initiating contraceptive use. Drawing on data from the Democratic Republic of the Congo, this paper fills three important gaps: first, it investigates the relationship between individual IPV and current modern contraceptive use in an African context; second, it extends such investigations to test the role of community IPV prevalence; and third, it examines whether these associations differ by conflict setting status. 1.1. IPV and family planning

* Corresponding author. E-mail addresses: [email protected] (R. Kidman), tia.palermo@ stonybrook.edu (T. Palermo), [email protected] (J. Bertrand). http://dx.doi.org/10.1016/j.socscimed.2015.03.034 0277-9536/© 2015 Elsevier Ltd. All rights reserved.

Women who experience IPV may have limited control over family planning, either because they are subjected to more frequent

R. Kidman et al. / Social Science & Medicine 133 (2015) 2e10

and forced sex or because they are unable to negotiate condom use successfully (Campbell et al., 2013; Coker, 2007). Specifically, studies show that some women fear raising the issue of contraception, lest their partner react violently (Bawah et al., 1999; Ezeh, 1993; Fort, 1989; Heise et al., 2002). Partners may interpret a request to use contraceptives as evidence of the wife's infidelity, an accusation of the husband's infidelity, or a general affront to his masculinity (Heise et al., 2002). Consistent with the above pathways, there is an extensive literature documenting the relationship between IPV and poor reproductive health outcomes in the developed world, with much of this literature focused on contraceptive use (Coker, 2007). Far fewer studies have been conducted in developing country contexts, despite the higher prevalence of IPV in these regions (Devries et al., 2013), and studies that have examined this relationship report conflicting findings. Prospective studies in India (Stephenson et al., 2013) and in urban areas of South Africa and Zimbabwe (Kacanek et al., 2013) confirm the negative relationship between IPV and contraceptive use observed in more developed countries. Other studies, however, challenge the applicability of these findings to the African context. In a South African study, women with a past history of domestic violence were more likely to ask their current partners to use a condom, though we note that IPV in the past year was not related to condom use (Jewkes et al., 2003). Similarly, a Nigerian study found women exposed to physical IPV had a greater likelihood of using modern contraception (Okenwa et al., 2011). Further, a crosssectional study that pooled data from six African countries found partner IPV was associated with greater likelihood of ever having used a method of modern and traditional contraception (Alio et al., 2009). Possible explanations posed by the authors include a desire to avoid pregnancy in unfavorable circumstances, a desire to protect against HIV with violent partners, and finally, that contraceptive use incited IPV (reverse causation) (Alio et al., 2009). Taken together, the heterogeneity of findings suggests the interplay between IPV and modern contraceptive use is highly dependent on context, and warrants unique investigation by country (Okenwa et al., 2011). Past studies have also investigated the role of community level IPV in family planning. Women living in communities in which IPV is highly prevalent may fear reprisals if they suggest contraceptive use, regardless of prior personal experiences with IPV (Hung et al., 2012; Pallitto and O'Campo, 2005). Specifically, Hung et al. (2012) suggests that community IPV levels may reflect norms around violence and unequal gender relations. By signaling that violence is a culturally accepted response, community-level IPV may operate to deter other women from using contraception. Their fear may not be unfounded: McQuestion (2003) found that women living in Colombian communities characterized by high IPV had 64% higher odds of experiencing IPV themselves, independent of individual risk factors. In addition, community IPV levels may reflect gender norms more broadly, including power imbalances that inhibit contraceptive negotiation. While further qualitative work appears warranted in this area, there is emerging evidence of an empirical association. For example, Speizer et al. (2009) reported that community prevalence of IPV was associated with lower use of contraception among South African youths. In a related investigation, Hung et al. (2012) found that community prevalence of IPV had a similar and independent correlation with birth spacing (the primary reason for contraceptive use in sub-Saharan Africa (Jansen and William, 2005)). Further, Pallitto and O'Campo (2005) demonstrated an association between community-level measures of patriarchal control in Colombia and unintended pregnancy (McQuestion, 2003; Pallitto and O'Campo, 2004), a common correlate of IPV. Thus, there is evidence that community IPV rates are associated with reproductive outcomes, though studies have yet to explicitly examine contraceptive use.

3

1.2. IPV, family planning and conflict Ensuring a woman's access to family planning is all the more difficult in times of prolonged conflict or war (McGinn et al., 2011). This is in part due to the restricted supply of family planning services (e.g., health systems collapse, human resources flee). Even when services are available, conflict may disrupt travel routes and/ or create unsafe conditions for travel, thus creating another barrier to access (Mock et al., 2004). Prolonged conflict may also heighten the risk of IPV. Sexual violence may become ‘normalized’ in situations of conflict, thus increasing civilian rape and intimate partner sexual violence (IPSV) (for a review of conflict and IPV, see (Baaz and Stern, 2010)). Even after conflict ends, the learned behavior of returning soldiers and civilians may continue to perpetuate the epidemic of IPV. Some women who suffered war-related trauma also continue to experience psychological symptoms, which increases their risk for IPV (Saile et al., 2013). While the above pathways suggest that IPV would be elevated in conflict situations, only one empirical study has tested this hypothesis: Janko et al. (2014) found a significant, positive relationship between conflict and IPV. Finally support for this hypothesis comes from the observation that IPV prevalence estimates from conflict settings are high (Stark and Ager, 2011) as compared to global estimates (Devries et al., 2013). Further, men who have been affected by conflict (displaced, wounded, or witnessed violence) report perpetrating IPV at higher rates (Peacock and Barker, 2014). Moreover, the epidemic of sexual violence e both from combatants and partners e may fundamentally alter the context in which family planning decisions are made. For example, the desire to avoid pregnancy in such dire circumstances may outweigh the fear of reprisals if a woman is caught using contraceptives. Despite the potentially heightened IPV risk and lowered access to family planning that accompanies war, we know of no studies examining the relationship between the two in a conflict zone. 1.3. The Democratic Republic of Congo The DRC is characterized by high rates of IPV and low rates of family planning, making this a unique setting in which to study their association. Approximately a third of the women report intimate partner sexual violence, more than double the rates reported in neighboring countries (Hindin et al., 2008; Peterman et al., 2011). Contraceptive use, on the other hand, remains extremely low (20% report ever using a modern method; only 6% report currently using such), resulting in one of the highest total fertility rates globally
re du Plan et Macro International, 2008). In contrast, (6.3) (Ministe current modern contraceptive usage ranged from 13% to 33% in five neighboring countries with data available for the same time period (Macro International Inc., 2014). Moreover, by focusing our investigation within the DRC, we are able to examine the association between conflict, IPV and contraceptive use. The country has been in a virtually continuous state of conflict since 1996, despite two peace agreements (Central Intelligence Agency, 2013). It is estimated that between 3.3 and 5 million people died as a result between 1998 and 2007 alone (Gambino, 2011; International Rescue Committee, 2007). Sexual violence, perpetrated by both military and civilians (Bartels et al., 2011; Johnson et al., 2010; Wakabi, 2008), has been a common feature of the conflict: based on a 2007 survey, it was estimated that 1.69 to 1.80 million women nationally, or between 642,000 and 704,000 in conflict-affected areas of Eastern DRC, had been raped in their lifetime (Peterman et al., 2011). Further, interviews with survivors highlight brutal tactics related to sexualized violence including gang rape, rape with instruments such as guns, and forced incest (Kelly et al., 2011). This is often followed by rejection

4

R. Kidman et al. / Social Science & Medicine 133 (2015) 2e10

of victims by partners, families, and communities; stigma; and social isolation (Kelly et al., 2011; Kelly et al., 2010). Widespread perpetration of sexual violence has been highlighted by researchers and policymakers as a characteristic of DRC's conflict, but sexual violence in conflict is not unique to DRC (Wood, 2009). Thus it is important to examine linkages between conflict, gender-based violence, and contraceptive use. In DRC, while no region is totally removed from the consequences of this conflict, the violence has largely concentrated in the eastern region of the country, specifically in Province Orientale, Maniema, North Kivu, and South Kivu. However, there is substantial heterogeneity in exposure between and within provinces, as can be seen in Fig. 1. While there has been limited inquiry around conflict and IPV, previous studies have largely treated conflict as a homogeneous country-level exposure. To our knowledge, only one existing, unpublished study has capitalized on the local variability to examine the association between conflict exposure and IPV (Janko et al., 2014), and no study has examined whether conflict is a contextual modifier of the association between IPV and contraceptive use. Using nationally-representative data from the DRC, we examined the relationship between individual- and community-level IPV and modern contraceptive use. We hypothesized that:

of Planning with support from the Ministry of Health between January and August of 2007. The DHS are population-level household surveys administered by host country governments with technical assistance from ICF Macro and funding from USAID. They have been implemented in over 90 countries and routinely collect information on population, health, HIV and nutrition (www. measuredhs.com). The 2007 DRC-DHS is a multi-stage stratified cluster sample of 8886 households from 300 villages or neighborhoods (not exceeding 500 households in size). Within these households, 10,338 women between the ages of 15 and 49 were interviewed regarding a range of reproductive health topics. Additionally, one randomly selected woman from every other household was administered a module on domestic violence (n ¼ 3436); the analysis for the current study focuses a sub-sample of ever married women (n ¼ 2859). For the security of the interviewee, these domestic violence interviews were only conducted if privacy could be assured. Questionnaires were conducted in French, Kikongo, Lingala, Swahili, and Tshiluba (Ministry of Planning, Ministry of Health [Democratic Republic of Congo], and Macro International Inc., 2008).

(1) women with a personal history of IPV would be less likely to currently use modern contraceptives; (2) women who lived in communities with high rates of IPV would be less likely to currently use modern contraceptives; and (3) conflict would modify the relationship between current contraceptive use and IPV.

The primary outcome examined was current use of a modern contraceptive method. Women were first asked to list contraceptive methods that they had heard about; for methods not spontaneously mentioned, women were prompted, “Have you ever heard of [Method].” Local terms with short explanations were used for the various methods. If they had previously heard of the method, women were further asked whether they had used the method ever or currently. We classified the following as modern methods: sterilization, hormonal pills, intrauterine device (IUD), injectables, implants, condoms (male or female), diaphragm, spermicides, and emergency contraceptive pills. The two main independent variables were individual-level experience of IPV and community-level means of IPV. This study focused specifically on experiences of physical and sexual partner

2. Methods 2.1. Sample Data used in this study come from the 2007 DRC Demographic and Health Survey (DHS), which was implemented by the Ministry

2.2. Measures

Fig. 1. Conflict events in the Democratic Republic of Congo (2002e2006) included in the analyses.

R. Kidman et al. / Social Science & Medicine 133 (2015) 2e10

violence; questions were based on a modified version of the Conflict Tactics Scales (Kishor and Hindin, 2004). Women were asked whether their current or most recent husband/partner had ever done the following: push, shake, throw something at her; slap her, twist her arm; punch her with his fist or something that could hurt her; kick, drag or beat her up; try to choke or burn her on purpose, threaten or attacked her with a knife, gun or other weapon; physically force her to have sexual intercourse; or force her to perform other sexual acts. Women answering affirmatively to any of the above were classified as having a history of any IPV with that partner; women reporting either of the last two experiences were also classified as having a history of sexual IPV. Data on individual IPV histories (using the definitions above) were aggregated across communities (defined as the villages or neighborhood primary sampling unit) to create non-self clustered means for any IPV and sexual IPV (i.e., means were calculated for each woman such that her experience was excluded from the calculation of her non-self clustered community mean in an effort to preserve the exogenous nature of these community-level variables; see (Balk, 1994) for more information). For the purpose of describing the sample, communities were divided into quartiles of IPV prevalence based on these cluster means. A community-level measure of conflict exposure was created using data from the Armed Conflict Location and Events Dataset (ACLED) (Raleigh et al., 2010). ACLED contains geocoded data on individual conflict events within the DRC from 1997 to 2013. We focused on events from 2002 to 2006 (the five years preceding the DHS survey), as these represent the most salient exposure for the respondents. A five year period was chosen to reflect that the impact of conflict could manifest for several years following exposure (e.g., through normalization of violence or mediated through psychological sequalea). Additional analyses were, however, conducted limiting exposure to the single year preceding the survey to determine whether findings were sensitive to the time frame. Fig. 1 shows the location of conflict events included in our analyses. We linked this data to the geocoded DHS, and classify communities by the number of conflict events or fatalities that occurred within 100 km of the community centroid. In the analyses presented, we define conflict exposure as 5 or more conflict events within 100 km of the community centroid; this corresponds to the 50th percentile. Sensitivity analyses demonstrated consistent results when five-year conflict was measured with alternative thresholds for dichotomization, as a continuous count of conflict events, or by using fatalities (data not shown). Additional controls included individual-, household-, and community-level characteristics. Age in years was modeled continuously. Marital status was categorized as currently married or in union (reference) and formerly married (widowed, divorced, or separated); never-partnered women were not eligible for the violence module and thus were excluded from our analysis. Education was classified as no education or incomplete primary education (reference), complete primary education, and some secondary or higher education. Household wealth was categorized into quintiles. Distance to a health facility was a dummy variable indicating whether the woman reported she considered the distance to be a “big problem.” We also controlled for whether the respondent lived in an urban or rural area and their region of residence. 2.3. Statistical analyses Descriptive statistics account for the for complex survey design and sample weights. We ran logistic regressions predicting current modern contraceptive use; these controlled for the aforementioned individual-, household-, and community level characteristics and

5

we adjusted standard errors for clustering at the community (the DHS primary sampling unit) level. We began by examining the relationship between individual- and community-level experience of any IPV and contraceptive use (Model 1). This model was then extended to examine the influence of conflict exposure on contraceptive use (Model 2). Finally, to examine whether relationships varied in strength by conflict exposure, we included an interaction between conflict and individual-level IPV (Model 3). A second set of models was then constructed focusing more narrowly on sexual IPV (Models 4e6). All analyses were performed using Stata Version 11 (College Station, TX). 3. Results 3.1. Current modern contraceptive use Only one fifth of all women had ever used a modern contraceptive, and only 6% were currently doing so (Table 1). Modern contraceptive use was slightly higher in conflict clusters as compared to non-conflict clusters (8% versus 4%). A closer examination by province reveals that women in South Kivu and North Kivu e two of the four provinces still engaged in widespread conflict e had the highest rates of current modern contraceptive use (15e16%). Current use conformed to the usual sociodemographic patterns: use was higher among those with more education and more wealth (Tables 1 and 3). We did not find differences by urban residence. 3.2. Intimate partner violence Overall, 59% of the women surveyed reported experiencing at least one form of IPV (i.e., sexual or physical abuse; Table 1) with their current or most recent partner. While most of these women reported physical violence (52.8%), sexual violence was also highly prevalent (reported by 32.4% of total women; these categories are not mutually exclusive). Based on descriptive statistics, higher levels of wealth appear to be protective, but higher levels of education do not (Table 1). Currently married women had higher rates of lifetime IPV compared to formerly married women. These relationships, however, do not reach significance in adjusted models (Table 4). 3.3. Intimate partner violence and modern contraceptive use In bivariate analyses, there was no statistical difference in current modern contraceptive use by individual IPV history or by community IPV level (Table 2). Similarly, in the fully-adjusted multivariate analyses, women who had experienced any IPV (combined measure) with their current or most recent partner or lived in communities with higher levels of IPV prevalence were no more likely to be using modern contraceptives (Table 3; Model 1). We also ran a second set of analyses focusing more narrowly on sexual violence. In bivariate analyses, 7% of women who reported sexual IPV also reported current condom use; this compares to 5% of women who report no sexual IPV experience, but the difference is not statistically significant (Table 2). In multivariate analyses, women having ever experienced sexual IPV had 68% higher odds of currently using a modern method (Model 4: adjusted OR 1.68; 95% CI 1.20, 2.35). Once again, women in communities with higher rates of sexual IPV were not significantly less likely to be using a modern method of contraception than women in communities with lower rates of sexual IPV, however there was a consistent trend across models (ORs ranged from 0.67 to 0.70).

6

R. Kidman et al. / Social Science & Medicine 133 (2015) 2e10

Table 1 Weighted sample characteristics, Democratic Republic of Congo, 2007 (n ¼ 2855).

Total Age 15e24 years 25e34 years 35e49 years Marital status Currently married Formerly married Education None/incomplete primary Complete primary education Some secondary or higher education Wealth Poorest Poorer Middle Richer Richest Residence Rural Urban Distance to health facility Big problem Not a big problem Conflict zone Conflict cluster Non-conflict cluster Province Bandundu Bas-Congo Equateur Kasaï Occidental Kasaï Oriental Katanga Kinshasa Maniema North Kivu Province Orientale South Kivu

Sample distribution

Current modern contraceptive use

Any IPV experience

Sexual IPV experience

%

SE

%

SE

%

SE

%

SE

100.0

0.0

6.0

0.8

59.3

1.9

32.4

2.2

28.0 37.1 34.8

1.6 1.7 1.5

4.5 6.7 6.5

1.0 1.2 1.5

60.0 60.8 57.1

3.2 2.8 2.7

31.3 32.9 32.7

3.1 3.1 3.1

87.8 13.2

1.0 1.0

6.0 6.1

0.8 1.9

63.0 34.5

1.9 5.4

34.0 21.5

2.3 5.6

56.7 7.6 35.7

2.6 0.1 2.4

3.2 5.2 10.7

0.8 1.7 1.5

59.2 61.1 59.0

2.4 4.6 2.5

33.9 31.4 30.2

3.0 5.7 2.6

19.6 23.1 20.4 18.7 18.3

2.1 2.0 1.7 2.1 2.5

2.7 3.2 4.9 6.5 14.1

1.0 1.2 1.3 1.5 2.5

65.3 58.8 60.3 57.3 54.3

3.4 3.0 3.7 4.1 3.1

40.0 32.1 33.3 32.4 23.6

5.1 3.0 4.2 3.8 2.4

59.4 40.6

4.1 4.1

3.8 9.3

0.9 1.4

59.6 58.8

2.6 2.6

32.2 32.7

2.9 3.3

40.4 59.5

3.1 3.1

4.8 6.9

1.2 1.0

62.0 57.4

3.4 2.0

31.7 33.3

2.4 3.3

51.5 48.5

5.0 5.0

7.9 4.4

1.3 0.9

50.8 66.3

2.5 2.3

50.8 39.3

2.5 3.2

14.0 4.1 13.1 9.7 17.2 10.3 12.3 3.3 3.5 13.5 4.5

3.7 1.0 3.0 2.4 2.9 2.2 2.3 0.8 1.1 3.5 1.6

4.4 8.5 5.7 2.0 1.5 5.5 11.4 7.6 14.9 3.7 15.5

2.5 2.0 2.5 1.3 0.8 1.6 2.2 2.2 4.3 1.4 5.1

64.3 47.5 78.1 69.7 61.6 58.4 47.8 61.9 52.9 45.9 47.3

3.8 3.9 4.8 6.6 4.0 4.6 3.2 4.8 5.8 5.3 7.6

42.8 14.8 51.3 45.5 31.3 23.0 21.2 30.7 32.6 22.1 19.7

7.2 3.3 5.4 7.3 4.5 7.2 2.7 7.1 4.2 3.0 3.8

3.4. Intimate partner violence, modern contraceptive use and conflict We expected that IPV would be higher among women exposed to greater conflict, however this pattern did not emerge for either the Table 2 Weighted bivariate associations between IPV and current modern contraceptive use, Democratic Republic of Congo, 2007 (n ¼ 2855). Current modern contraceptive use % All women Any IPV (physical or sexual) Yes 6.0 No 6.1 Sexual IPV Yes 7.4 No 5.3 Community-level IPV Quartile 1 (lowest) 7.1 Quartile 2 6.3 Quartile 3 7.5 Quartile 4 (highest) 3.0 Community-level sexual IPV Quartile 1 (lowest) 6.5 Quartile 2 7.5 Quartile 3 5.3 Quartile 4 (highest) 5.0

p-value

SE

1.0 1.2

1.0

1.5 0.9

0.2

1.5 1.3 2.0 0.9

0.1

1.4 1.4 1.4 1.7

0.6

combined or sexual IPV measures; we confirmed this null association adjusting for known sociodemographic correlates (i.e., age, education, marital status, and urban residence; Table 4). Similarly, conflict did not emerge as a significant independent correlate of contraceptive use in the primary analyses (Table 3; Models 2 & 5). Conflict was a significant predictor of contraceptive use in sensitivity analyses that limited exposure to the single year preceding the survey, however this observation was true only for the model that simultaneously controlled for sexual IPV. Finally, regardless of the time frame used, we found no evidence that conflict was a statistically significant modifier of individual IPV impact, measured either as any IPV (Model 3) or sexual IPV (Model 6).

4. Discussion In developing countries, evidence overwhelmingly supports a relationship between individual experiences of IPV and reduced contraceptive use. Evidence from Africa is split, however, with some studies reporting increased contraceptive use among women who experience IPV (Alio et al., 2009). We investigated this relationship in the DRC, a country marked by conflict and sexual violence. We found no difference in current modern contraceptive use by individual IPV history defined as combined physical and/or sexual IPV, but we did find a positive relationship between experience of sexual IPV and current contraceptive use.

R. Kidman et al. / Social Science & Medicine 133 (2015) 2e10

7

Table 3 Logistic regression of current contraceptive use and intimate partner violence, Democratic Republic of Congo, 2007 (n ¼ 2855). Model (1)

Model (2)

Model (3)

Model (4)

Model (5)

Model (6)

OR (95% CI)

OR (95% CI)

OR (95% CI)

OR (95% CI)

OR (95% CI)

OR (95% CI)

IPV experience Ever experienced any IPV 1.05 Community-level mean of IPV 0.90 (continuous) Ever experienced sexual IPV Community-level mean of sexual IPV (continuous) Conflict experience Conflict Conflict*IPV Conflict*sexual IPV Covariates Marital status (ref ¼ currently married) Formerly married 1.15 Age 1.01 Education (ref ¼ none/incomplete primary) Complete primary education 2.35** Some secondary or higher education 2.59*** Wealth (ref ¼ poorest) Poorer 0.90 Middle 1.54 Richer 2.36* Richest 3.47** Urban 1.20 Distance to health facility ¼ big 0.85 problem Province (ref ¼ Kinshasa) Bandundu 1.66 Bas-Congo 1.81* Equateur 1.70 Kasaï Occidental 0.48 Kasaï Oriental 0.38 Katanga 1.10 Maniema 1.80 North Kivu 3.14*** Province Orientale 1.33 South Kivu 1.54

(0.75e1.47) (0.40e2.04)

1.05 (0.75e1.47) 0.91 (0.40e2.07)

0.84 (0.46e1.52) 0.93 (0.41e2.10) 1.68** (1.20e2.35) 0.67 (0.26e1.72)

1.17 (0.68e2.00)

0.94 (0.47e1.90) 1.45 (0.71e2.94)

1.68** (1.20e2.35) 0.68 (0.26e1.77)

1.40 (0.75e2.63) 0.70 (0.27e1.81)

1.18 (0.68e2.03)

1.07 (0.58e1.96) 1.33 (0.63e2.79)

(0.69e1.92) (1.00e1.03)

1.15 (0.69e1.90) 1.01 (1.00e1.03)

1.14 (0.69e1.89) 1.01 (1.00e1.03)

1.19 (0.72e1.97) 1.02 (1.00e1.03)

1.19 (0.72e1.95) 1.02 (1.00e1.03)

1.18 (0.72e1.95) 1.02 (1.00e1.03)

(1.27e4.34) 2.34** (1.26e4.32) 2.34** (1.27e4.33) 2.36** (1.28e4.37) 2.35** (1.27e4.35) 2.35** (1.27e4.33) (1.65e4.06) 2.57*** (1.64e4.05) 2.57*** (1.64e4.05) 2.60*** (1.66e4.07) 2.58*** (1.65e4.05) 2.58*** (1.65e4.04) (0.43e1.91) (0.79e3.02) (1.09e5.13) (1.41e8.56) (0.69e2.08) (0.59e1.24)

0.90 1.54 2.34* 3.41** 1.20 0.86

(0.43e1.91) (0.79e3.01) (1.08e5.07) (1.39e8.35) (0.69e2.09) (0.59e1.25)

0.91 1.55 2.33* 3.38** 1.21 0.86

(0.43e1.92) (0.79e3.03) (1.08e5.04) (1.38e8.28) (0.69e2.10) (0.60e1.26)

0.89 1.55 2.36* 3.45** 1.20 0.85

(0.42e1.89) (0.79e3.04) (1.08e5.18) (1.39e8.58) (0.69e2.09) (0.59e1.25)

0.89 1.55 2.35* 3.39** 1.21 0.86

(0.42e1.89) (0.79e3.03) (1.08e5.11) (1.37e8.34) (0.69e2.10) (0.59e1.25)

0.89 1.55 2.32* 3.34** 1.21 0.87

(0.42e1.90) (0.79e3.03) (1.06e5.09) (1.35e8.28) (0.69e2.11) (0.60e1.26)

(0.77e3.60) 1.91 (0.74e4.92) 1.84 (0.70e4.84) 1.68 (0.78e3.65) 1.94 (0.74e5.08) 1.89 (0.71e5.04) (1.05e3.11) 2.07* (1.01e4.25) 2.01 (0.97e4.18) 1.83* (1.05e3.18) 2.10* (1.00e4.42) 2.05 (0.97e4.30) (0.90e3.21) 1.95 (0.85e4.43) 1.94 (0.85e4.41) 1.66 (0.86e3.17) 1.90 (0.83e4.34) 1.91 (0.84e4.35) (0.15e1.46) 0.53 (0.16e1.75) 0.52 (0.16e1.73) 0.46 (0.15e1.45) 0.52 (0.15e1.73) 0.51 (0.15e1.70) (0.14e1.08) 0.39 (0.14e1.11) 0.38 (0.13e1.09) 0.38 (0.13e1.07) 0.38 (0.13e1.09) 0.38 (0.13e1.07) (0.55e2.21) 1.14 (0.55e2.34) 1.12 (0.54e2.31) 1.10 (0.54e2.21) 1.14 (0.55e2.34) 1.13 (0.54e2.34) (0.87e3.74) 1.85 (0.87e3.95) 1.82 (0.85e3.87) 1.76 (0.86e3.61) 1.82 (0.86e3.82) 1.80 (0.86e3.76) (1.78e5.53) 3.10*** (1.76e5.46) 3.07*** (1.75e5.41) 3.07*** (1.67e5.66) 3.03*** (1.64e5.59) 2.95*** (1.60e5.45) (0.59e3.02) 1.34 (0.59e3.05) 1.34 (0.58e3.06) 1.34 (0.58e3.05) 1.35 (0.59e3.09) 1.34 (0.58e3.07) (0.67e3.54) 1.52 (0.66e3.50) 1.52 (0.66e3.49) 1.54 (0.67e3.55) 1.52 (0.66e3.51) 1.51 (0.65e3.49)

***p < 0.001, **p < 0.01, *p < 0.05; 95% confidence intervals in parenthesis. Note: Inteff command was run for interaction effects of conflict and IPV variables; no significant interactions were found.

4.1. Intimate partner violence and modern contraceptive use These findings represent an important contribution to the previous literature on IPV and contraceptive use in Africa, which to Table 4 Logistic regression of intimate partner violence and conflict (N ¼ 2855), Democratic Republic of Congo, 2007. Lifetime IPV

Lifetime sexual IPV

OR (95% CI)

OR (95% CI)

Conflict 0.87 (0.66e1.14) Marital status (ref ¼ currently married) Formerly married 0.32*** (0.25e0.42) Age 0.99* (0.98e1.00) Education (ref ¼ none/incomplete primary) Complete primary education 1.08 (0.80e1.47) Some secondary or higher 1.09 (0.88e1.34) education Wealth (ref ¼ Poorest) Poorer 0.94 (0.74e1.18) Middle 0.94 (0.72e1.22) Richer 1.04 (0.76e1.42) Richest 1.19 (0.80e1.76) Urban 1.04 (0.79e1.37) Distance to health facility ¼ big 1.03 (0.85e1.24) problem

0.94 (0.65e1.36) 0.47*** (0.34e0.66) 0.99 (0.98e1.00) 0.95 (0.67e1.33) 0.96 (0.75e1.22)

0.91 0.79 0.95 0.97 1.10 1.12

(0.71e1.17) (0.60e1.05) (0.67e1.35) (0.62e1.54) (0.79e1.53) (0.91e1.37)

***p < 0.001, **p < 0.01, *p < 0.05; 95% confidence intervals in parenthesis. Note: regressions also control for province.

date has been mixed. Seemingly contradictory findings may be due to differences in methodological approaches; alternatively they may indicate important heterogeneity in the association due to unmeasured or country-level contextual factors. In this section, we briefly review the former; we then turn our attention to the topic of conflict as a contextual determinant. As mentioned in the introduction, cross-sectional studies have largely reported positive associations between IPV and contraceptive use, whereas several prospective studies have reported a negative relationship. Our findings may provide some insights into why these literatures diverge, and push the field to more carefully consider how it operationalizes both the variables of interest and their association. In a cross-sectional study of 10 countries, Alio et al. (2009) found that IPV was positively associated with ever having used contraceptives. Our study reports a null finding; however it differs in approach: our indicator of modern contraceptive use refers to current use, rather than use at any time over the lifecourse. This proxies the causal hypothesis more appropriately: IPV with the current partner is linked to contraceptive choices with the same partner. The choice of indicator qualitatively changes the conclusion: when we reran our analyses using the lifetime measure of modern contraceptive use, we duplicated the positive association with IPV (adjusted OR 1.40; 95% CI 1.45, 1.71; data not shown) found by Alio et al. This disconnect strongly suggests that reverse causation is a possible explanation for the Alio et al. findings, particularly in light of research

8

R. Kidman et al. / Social Science & Medicine 133 (2015) 2e10

showing men may react violently to the clandestine or explicit use of contraception (Kaye, 2006; Koenig et al., 2003). Both our study and that by Alio et al. (2009) use cross-sectional data. Other studies have been better able to parse out the temporal relationship using prospective data, and their findings mirror those in the more developed countries: in India, women experiencing IPV were less likely to subsequently adopt contraceptive use (Stephenson et al., 2013); in South African and Zimbabwe, recent IPV was associated with contraceptive non-adherence (Kacanek et al., 2013). Taken together, these two bodies of research suggest that the association between IPV and contraceptive use may operate in both directions: contraceptive use incites IPV; IPV discourages contraceptive use. The null findings from our study on the combined measure of IPV and current contraceptive use may thus reflect simultaneity bias e however, we are clearly limited in our causal inference by the cross-sectional nature of the available data. Further study in the DRC and elsewhere should utilize longitudinal data collection to better investigate the bi-directional causal nature of this relationship. As noted above, the time frame used to capture contraceptive use (ever or current) qualitatively changes the findings. Likewise, findings were highly sensitive to the definition of IPV. We report null findings when using a combined measure of physical and sexual IPV, however a significant positive relationship with contraceptive use emerged when sexual IPV was examined in isolation (with women experiencing no IPV or physical IPV only as the combined reference group). This reinforces the need to better standardize measures between studies; it also highlights the potential for different conceptual pathways. It may be that for individuals experiencing sexual violence, the fear of unintended pregnancy outweighs concerns about physical reprisals. Alternatively, this may again reflect reverse causation: victims of sexual violence may actually have better access to contraceptives than their peers. Resources earmarked for sexual violence survivors combined with a lack of basic health services and widespread poverty in the DRC may mean that rape victims gain better access to services (Eriksson Baaz and Stern, 2010). In fact, DRC was the top recipient of aid by government donors for sexual and gender-based violence-related projects in 2012, with 39.4 million USD; the next highest recipient (Uganda) received less than half of this amount (Development Initiatives, 2014). Furthermore, this aid is not evenly distributed throughout the country and much of it is focused in the eastern, conflict-affected part of the country. The directionality and differential influence of sexual versus physical IPV is an area in need of further qualitative research.

review found almost no research on this topic (Stark and Ager, 2011). Only one study explicitly tested differences in IPV between conflict and non-conflict periods (in Timor); they too found no quantitative difference among populations exposed to conflict in East Timor (Hynes et al., 2004). However, findings from a recent, unpublished study demonstrated a positive relationship between conflict and IPV in Rwanda (Janko et al., 2014). Clearly, this is an area ripe for investigation. We similarly expected lower levels of modern contraceptive use in conflict regions (due to deteriorating health infrastructure and supply disruptions), but found no association. Across the DRC, there is a weak health care system infrastructure, with roots in decentralization under Mobutu in the 1970s. It was highly dependent upon a large missionary presence, USAID and other funding into the 1980s, then subsequently affected by a deep economic crisis in the 1980s, and finally, by 1991, suffered withdrawal of most donor organizations and widespread pillaging of the health care system (Gambino, 2011). Through today, although donor organizations including USAID have returned, there are continued low levels of public investment in health care. As of 2008e9, only 0.2% of funding for reproductive health services (of which only a small proportion goes towards family planning) was coming from the government (Health Systems 20/20 Project, 2011). Against such a backdrop, conflict may play little role in determining contraceptive use; similar analyses should be undertaken to examine the role of conflict in settings with more robust health care systems. We do, however, note that there appear to be substantial disparities in family planning use by province, with a marked concentration of current contraceptive use in South and North Kivu. This likely reflects greater access in these provinces, though the only available data on such was collected a few years after the DHS data: 70% of health facilities in South Kivu and 56% in North Kivu offered the pill in 2011; these represented the highest rates in the
re country and compared to a national average of only 34% (Ministe  Publique and Organisation de Sante  Mondiale, 2012). de la Sante Similarly, South Kivu and Maniema had the highest rates of condom availability at health facilities. We note that all three provinces e South Kivu, North Kivu and Maniema e are characterized by high levels of conflict. Taken together, this evidence might suggest that some women in conflict zones may have better access to family planning services than other women in non-conflict zones, a factor our data does not capture. Regardless of the mechanism, the conclusion is the same: focusing aid solely on areas most affected by violent conflict will exclude large populations in need of reproductive health services.

4.2. Intimate partner violence, modern contraceptive use and conflict

4.3. Limitations

A unique contribution of our study is our consideration of the ecological context: we examine both the potential influence of community-level IPV and the influence of conflict on contraceptive use. With regard to the former, our results demonstrated that community-level IPV (both the combined measure and sexual violence alone) was not associated with current contraceptive use at the individual-level. With regard to the latter, our study highlights two surprising findings. First, we expected that IPV would be higher in conflict areas, and thus would explain a greater proportion of contraceptive use at the population level. We did not find evidence to support this hypothesis: conflict had no significant association with IPV. This finding held regardless of how we characterized conflict (i.e., by conflict events or by fatalities, as a continuous or dichotomous measure) or IPV (i.e., a combined measure or restricted to sexual IPV). While conflict is assumed to heighten IPV prevalence, a recent

There are notable limitations to our findings related to the DHS data. As mentioned earlier, we use cross-sectional data, and thus our findings cannot discern causal relationships. Moreover, our investigation is restricted to women who are or have been partnered. Other groups of women may experience high rates of IPV, but are not captured in the DHS. For example, we cannot speak to whether the dynamics are similar among couples who are dating but not co-habiting; neither can we speak to experiences of women engaged in transactional sex. The latter may have particular relevancy for conflict settings, as women may be forced into such relationships as a survival strategy. The conflict setting may also pose additional challenges to survey implementation; however there is no indication that this reduced the response rate and thus generalizability within the DRC. The overall response rate for individual women was 97%; this did not vary substantially between provinces (rates ranged from 94% in Kasaï Occidental to 99% in Nord-Kivu)
re du Plan et Macro International, 2008). (Ministe

R. Kidman et al. / Social Science & Medicine 133 (2015) 2e10

There are also limitations inherent in our measure of conflict. The use of geocoded micro data on conflict events in the DRC allowed us to create a detailed exposure variable specific to each local community. This is an advance over broad characterizations of conflict based on province, both because the larger geographical units mark heterogeneity within their borders, and because there may be economic, social and political differences between provinces that confound the primary association of interest (between IPV and contraceptive use). Furthermore, the use of micro data allowed us to run sensitivity analyses to test different methods of modeling conflict (e.g., dichotomized by alternative thresholds; using events versus fatalities) and to confirm a consistent pattern of results. However, the ACLED relies on daily incident reports from secondary sources and is thus subject to missing data on events that went unreported; there may also be measurement error in the exact location and fatalities associated with each event (Raleigh et al., 2010). Finally, we note that our findings may not be generalizable outside of the DRC. Our findings emphasize the incredibly low access to family planning in the DRC. It is plausible that IPV may be a less influential determinant of contraceptive use in contexts where family planning is all but non-existent; thus we may see larger effect sizes in other developing countries. We also note that the political and aid landscape is changing rapidly in the DRC, and that our findings reflect the situation in 2007. The DRC conducted another Demographic and Health Survey in 2013e14. When this data is released, it will be possible to compare findings from 2007 with 2013e14. 5. Conclusion Despite the above limitations, this paper makes critical contributions to the field. Evidence linking IPV to reproductive outcomes in developing countries is scant; the only existing papers on contraceptive use in sub-Saharan Africa are in conflict with each other, as well as with evidence from more developed countries. We examine this relationship using national data from the DRC, extending previous work to a high-conflict setting. We find that a history of sexual IPV is associated with women currently using contraceptives, though a combined measure of sexual and physical IPV does not demonstrate a similar relationship. Contrary to expectation, we found the greatest need in non-conflict provinces e these communities demonstrated a higher IPV prevalence and lower rates of modern contraceptive use e though this relationship is not evident when conflict is characterized by community exposure. Thus, international and humanitarian aid flowing into the DRC should ensure that sufficient resources are allocated towards the reproductive health needs of women in both conflict and nonconflict regions. References Alio, A.P., Daley, E.M., Nana, P.N., Duan, J., Salihu, H.M., 2009. Intimate partner violence and contraception use among women in Sub-Saharan Africa. Int. J. Gynecol. Obstet. 107 (1), 35e38. Baaz, M.E., Stern, M., 2010. The Complexity of Violence: a Critical Analysis of Sexual Violence in the Democratic Republic of Congo (DRC). Sida Working Paper on Gender Based Violence. The Nordic Africa Institute, Uppsala, Sweden. Balk, D., 1994. Individual and community aspects of women's status and fertility in rural Bangladesh. Popul. Stud. 48 (1), 21e45. Bartels, S.A., Scott, J.A., Leaning, J., Kelly, J.T., Mukwege, D., Joyce, N.R., VanRooyen, M.J., 2011. Sexual violence trends between 2004 and 2008 in South Kivu, Democratic Republic of Congo. Prehospital Disaster Med. 1 (1), 1e6. Bawah, A.A., Akweongo, P., Simmons, R., Phillips, J.F., 1999. Women's fears and men's anxieties: the impact of family planning on gender relations in northern Ghana. Stud. Fam. Plan. 30 (1), 54e66. Campbell, J.C., 2002. Health consequences of intimate partner violence. Lancet 359 (9314), 1331e1336. Campbell, J.C., Lucea, M.B., Stockman, J.K., Draughon, J.E., 2013. Forced sex and HIV

9

risk in violent relationships. Am. J. Reprod. Immunol. 69, 41e44. http:// dx.doi.org/10.1111/aji.12026. Campbell, J.C., Soeken, K.L., 1999. Forced sex and intimate partner violence effects on women's risk and women's health. Viol. Against Women 5 (9), 1017e1035. Central Intelligence Agency, 2013. The World Factbook 2013e14. Retrieved March 7, 2014, 2014, from. https://www.cia.gov/library/publications/the-worldfactbook/geos/cg.html. Coker, A.L., 2007. Does physical intimate partner violence affect sexual health?: A systematic review. Trauma Viol. Abuse 8 (2), 149e177. http://dx.doi.org/10.1177/ 1524838007301162. Development Initiatives, June 2014. Trends in Donor Spending on Sexual and Gender-based Violence (SGBV). United Kingdom. Devries, K.M., Mak, J.Y.T., Garcia-Moreno, C., Petzold, M., Child, J.C., Falder, G., Lim, S., Bacchus, L.J., Engell, R.E., Rosenfeld, L., Pallitto, C., Vos, T., Abrahams, N., Watts, C.H., 2013. The global prevalence of intimate partner violence against women. Science 340, 1527e1528 (28 June 2013). Ellsberg, M., Jansen, H.A.F.M., Heise, L., Watts, C., Garcia-Moreno, C., 2008. Intimate partner violence and women's physical and mental health in the WHO multicountry study on women's health and domestic violence: an observational study. Lancet 371 (9619), 1165e1172. Eriksson Baaz, M., Stern, M., 2010. The Complexity of Violence: a Critical Analysis of Sexual Violence in the Democratic Republic of Congo (DRC). Sida & The Nordic Africa Institute, Uppsala, Sweden. Ezeh, A.C., 1993. The influence of spouses over each other's contraceptive attitudes in Ghana. Stud. Fam. Plan. 163e174. Fort, A.L., 1989. Investigating the social context of fertility and family planning: a qualitative study in Peru. Int. Fam. Plan. Perspect. 88e95. Gambino, A.W., 2011. Background Case Study: Democratic Republic of the Congo. World Development Report 2011. The World Bank, Washington, DC. Gazmararian, J.A., Lazorick, S., Spitz, A.M., Ballard, T.J., Saltzman, L.E., Marks, J.S., 1996. Prevalence of violence against pregnant women. JAMA: J. Am. Med. Assoc. 275 (24), 1915e1920. Health Systems 20/20 Project, 2011. National Health Accounts 2008e2009 Executive Summary. Abt Associates Inc., Bethesda, MD. Health Systems 20/20 project. Heise, L., Ellsberg, M., Gottmoeller, M., 2002. A global overview of gender-based violence. Int. J. Gynecol. Obstet. 78 (Suppl. 1), S5eS14. Hindin, M., Kishor, S., Ansara, D., 2008. Intimate Partner Violence among Couples in 10 DHS Countries: Predictors and Health Outcomes. Macro International Inc, Calverton, MD. Hung, K.J., Scott, J., Ricciotti, H.A., Johnson, T.R., Tsai, A.C., 2012. Community-level and individual-level influences of intimate partner violence on birth spacing in Sub-Saharan Africa. Obstet. Gynecol. 119 (5), 975e982, 910.1097/ AOG.1090b1013e31824fc31829a31820. Hynes, M., Ward, J., Robertson, K., Crouse, C., 2004. A determination of the prevalence of gender-based violence among conflict-affected populations in East Timor. Disasters 28 (3), 294e321. International Rescue Committee, 2007. Mortality in the Democratic Republic of Congo: an Ongoing Crisis. Retrieved March 7, 2014, from. http://www.rescue. org/sites/default/files/migrated/resources/2007/2006-7_congomortalitysurvey. pdf. Janko, M., Bloom, S.S., Spencer, J., 2014. Community Exposure to Violent Conflict Increases the Risk of Intimate Partner Violence in Rwanda. Paper presented at the Population Association of America Annual Meeting, Boston, MA. Jansen, I., William, H., 2005. Existing demand for birth spacing in developing countries: perspectives from household survey data. Int. J. Gynecol. Obstetrics 89, S50eS60. Jewkes, R.K., Levin, J.B., Penn-Kekana, L.A., 2003. Gender inequalities, intimate partner violence and HIV preventive practices: findings of a South African cross-sectional study. Soc. Sci. Med. 56 (1), 125e134. http://dx.doi.org/10.1016/ S0277-9536(02)00012-6. Johnson, K., Scott, J., Rughita, B., Kisielewski, M., Asher, J., Ong, R., Lawry, L., 2010. Association of sexual violence and human rights violations with physical and mental health in territories of the Eastern Democratic Republic of the Congo. J. Am. Med. Assoc. 304 (5), 553e562. Kacanek, D., Bostrom, A., Montgomery, E.T., Ramjee, G., de Bruyn, G., Blanchard, K., van der Straten, A., 2013. Intimate partner violence and condom and diaphragm non-adherence among women in an HIV prevention trial in Southern Africa. JAIDS: J. Acquir. Immune Defic. Syndromes. Kaye, D.K., 2006. Community perceptions and experiences of domestic violence and induced abortion in Wakiso District, Uganda. Qual. Health Res. 16 (8), 1120e1128. Kelly, J., Betancourt, T., Mukwege, D., Lipton, R., VanRooyen, M., 2011. Experiences of female survivors of sexual violence in eastern Democratic Republic of the Congo: a mixed-methods study. Confl. Health 5 (1), 1e8. Kelly, J., VanRooyen, M., Leaning, J., Cragin, W., 2010. “Now, the World is without Me”: an Investigation of Sexual Violence in Eastern Democratic Republic of Congo. Harvard Humanitarian Intitiative, Cambridge, MA. Kishor, S., Hindin, M.J., June 2004. Profiling Domestic Violence: a Multi-country Study. Measure DHSþ ORC Macro, Calverton, MD. Koenig, M.A., Lutalo, T., Zhao, F., Nalugoda, F., Wabwire-Mangen, F., Kiwanuka, N., Gray, R., 2003. Domestic violence in rural Uganda: evidence from a communitybased study. Bull. World Health Organ. 81, 53e60. Macro International Inc, 2014. MEASURE DHS STATcompiler. Retrieved December 3 2014, from. http://www.measuredhs.com. McGinn, T., Austin, J., Anfinson, K., Amsalu, R., Casey, S., Fadulalmula, S., Yetter, M.,

10

R. Kidman et al. / Social Science & Medicine 133 (2015) 2e10

2011. Family planning in conflict: results of cross-sectional baseline surveys in three African countries. Confl. Health 5 (1), 1e8. http://dx.doi.org/10.1186/17521505-5-11. McQuestion, M.J., 2003. Endogenous social effects on intimate partner violence in Colombia. Soc. Sci. Res. 32 (2), 335e345. http://dx.doi.org/10.1016/S0049089X(02)00062-5. Miller, E., Decker, M.R., McCauley, H.L., Tancredi, D.J., Levenson, R.R., Waldman, J., Silverman, J.G., 2010. Pregnancy coercion, intimate partner violence and unintended pregnancy. Contraception 81 (4), 316e322.
re de la Sante  Publique and Organisation de Sante  Mondiale, 2012. CarMiniste  de la Me
re, du Nouveautographie des Interventions et Intervenants de la Sante  et de l'Enfant y Compris la Planification Familiale en RD Congo. Ministe
re de ne  Publique and Organisation de Sante  Mondiale, Kinshasa, DRC. la Sante
re du Plan et Macro International, 2008. Enque ^te De mographique et de Ministe  , Re publique De mocratique du Congo 2007. Ministe
re du Plan et Macro Sante International, Calverton, Maryland, U.S.A. Ministry of Planning, Ministry of Health [Democratic Republic of Congo], & Macro International Inc, 2008. Demographic and Health Survey 2007. Kinshasa, Democratic Republic of Congo and Calverton, Maryland, USA. Mock, N., Duale, S., Brown, L., Mathys, E., O'Maonaigh, H., Abul-Husn, N., Elliott, S., 2004. Conflict and HIV: a framework for risk assessment to prevent HIV in conflict-affected settings in Africa. Emerg. Themes Epidemiol. 1 (1), 6. Okenwa, L., Lawoko, S., Jansson, B., 2011. Contraception, reproductive health and pregnancy outcomes among women exposed to intimate partner violence in Nigeria. Eur. J. Contracept. Reprod. Health Care: Off. J. Eur. Soc. Contracept. 16 (1), 18e25. http://dx.doi.org/10.3109/13625187.2010.534515. Pallitto, C., O'Campo, P., 2004. The relationship between intimate partner violence and unintended pregnancy: analysis of a national sample from Colombia. Int. Fam. Plan. Perspect. 165e173. Pallitto, C., O'Campo, P., 2005. Community level effects of gender inequality on intimate partner violence and unintended pregnancy in Colombia: testing the feminist perspective. Soc. Sci. Med. 60 (10), 2205e2216. http://dx.doi.org/10. 1016/j.socscimed.2004.10.017. Peacock, D., Barker, G., 2014. Working with men and boys to prevent gender-based violence principles, lessons learned, and ways forward. Men Masc. 17 (5),

578e599. Peterman, A., Johnson, K., 2009. Incontinence and trauma: sexual violence, female genital cutting and proxy measures of gynecological fistula. Soc. Sci. Med. 68 (5), 971e979. Peterman, A., Palermo, T., Bredenkamp, C., 2011. Estimates and determinants of sexual violence against women in the democratic Republic of Congo. Am. J. Public Health 101 (6), 1060e1067. http://dx.doi.org/10.2105/AJPH.2010.300070. Raleigh, C., Linke, A., Hegre, H., Karlsen, J., 2010. Introducing acled: an armed conflict location and event dataset special data feature. J. Peace Res. 47 (5), 651e660. Rees, S., Silove, D., Chey, T., Ivancic, L., Steel, Z., Creamer, M., Forbes, D., 2011. Lifetime prevalence of gender-based violence in women and the relationship with mental disorders and psychosocial function. J. Am. Med. Assoc. 306 (4), 513e521. Saile, R., Neuner, F., Ertl, V., Catani, C., 2013. Prevalence and predictors of partner violence against women in the aftermath of war: a survey among couples in Northern Uganda. Soc. Sci. Med. 86 (0), 17e25. http://dx.doi.org/10.1016/j. socscimed.2013.02.046. Speizer, I.S., Pettifor, A., Cummings, S., MacPhail, C., Kleinschmidt, I., Rees, H.V., 2009. Sexual violence and reproductive health outcomes among South African female youths: a contextual analysis. Am. J. Public Health 99, S425eS431. http:// dx.doi.org/10.2105/AJPH.2008.136606. Stark, L., Ager, A., 2011. A systematic review of prevalence studies of gender-based violence in complex emergencies. Trauma Viol. Abuse 12 (3), 127e134. http:// dx.doi.org/10.1177/1524838011404252. Stephenson, R., Jadhav, A., Hindin, M., 2013. Physical domestic violence and subsequent contraceptive adoption among women in rural India. J. Interpers. Viol. 28 (5), 1020e1039. http://dx.doi.org/10.1177/0886260512459379. ~ a, R., Ho €gberg, U., Persson, L.Å., 2002. Physical Valladares, E., Ellsberg, M., Pen partner abuse during pregnancy: a risk factor for low birth weight in Nicaragua. Obstet. Gynecol. 100 (4), 700e705. Wakabi, W., 2008. Sexual violence increasing in Democratic Republic of Congo. Lancet 371 (9606). Wood, E.J., 2009. Armed groups and sexual violence: when is wartime rape rare? Polit. Soc. 37 (1), 131e161.