Comparison with other studies
The lifetime prevalence of the different IPV types associated with terminated pregnancy among women found in this study is comparable to findings in studies in Bangladesh and Canada [45, 46], but comparatively lower than those in a study in Kenya . These differences may be methodological and linked to varied understanding of the term “terminated pregnancy”, and under-reporting of both terminated pregnancy and IPV within the different country contexts. Women who had terminated a pregnancy were more likely to have experienced physical IPV, sexual IPV, and “any IPV” compared to women who had no prior experience of these IPV types, which is consistent with findings from studies in other national contexts [19, 24, 26, 47–49], and validates the first hypothesis in this study. These findings may be partly attributed to the characteristics of the women that had terminated a pregnancy, given that the women were more often rural residents, did not use contraceptives, were younger than their spouse, earned less than their spouse, and had a spouse who was more likely to have controlling behaviour (corresponding to several explanatory factors listed in Table 1). As such, some of these women may have been economically dependent on their spouse, and lacked greater control over their fertility by not being empowered enough to refuse sex or negotiate the use of contraceptives for fear of abuse [50, 51]. The women may also have been more prone to experience IPV due to the general acceptance of male dominance within the societal contexts within which they live in, which enforce more traditional gender roles. It is also plausible that women who are exposed to IPV might have less control over the choice of contraceptive method, and not taking contraceptives for fear of further abuse , and are more likely to seek termination in the event of pregnancy [24, 26, 46]. Our findings however contrast with those in a study based on men’s self-reported perpetration of abuse in Bangladesh , which showed a non-significant association between “any IPV” and terminated pregnancy. Emotional IPV was however not significantly associated with terminated pregnancy after adjusting for confounders in this study, which is in line with a study from Cameroon . Further in depth research is needed into the association between physical and non-physical or non-sexual IPV and pregnancy outcomes.
Of the measures of relationship control, controlling behavior by husband/partner was consistently the strongest factor associated with all the IPV types and terminated pregnancy, which corroborates findings from a recent study , and indicates that male partner controlling behaviour plays a significant role in the occurrence of IPV. This may stem from the need for abusive male partners to enforce power and control within the relationship and during pregnancy. This finding is consistent with those from the Nigerian , and other contexts documenting women being coerced by abusive partners to have abortions [54, 55]. We presume that abusive and controlling male partners may influence their female partner’s non-use of contraceptives, and in the event of pregnancy, may exhibit behaviours (including violent acts) to control pregnancy outcomes in an attempt to induce abortion or coerce women to terminate the pregnancy . This stresses the need for further population-based research investigating forms of reproductive coercion and associations with IPV. Likewise, longitudinal studies are needed to examine the interrelation between reproductive coercion, IPV and unintended/terminated pregnancy, possibly specifying the appropriate chronological order and patterns, so as to better inform the design of interventions to reduce both the risk for unintended/terminated pregnancy and IPV victimization.
In related findings, we found a positive association between women justifying wife beating and IPV (physical; sexual; and “any”), consistent with another other studies [7, 33]. This symbolizes women’s acceptance of violence towards them within societies that tolerate conventional gender role attitudes that permit men to discipline their wife/partner for various transgressions. These attitude of accepting wife beating increases the likelihood of IPV, and within the context of abusive relationships, may be accompanied by pregnancy coercion and termination when they occur in pregnancy . The finding that having decision-making autonomy was associated with IPV (physical; and sexual) may represent the indirect causal mechanisms through which IPV affects women’s reproductive health outcomes [25, 57]. These collectively indicate the skewed power imbalances in intimate relationships within patriarchal societies where women’s subordination and submission to men is expected, accepted, and in many cases, attractive to some men , thus supporting the second hypothesis in this study, and validates those reported by others [54–57]. Our finding therefore calls for policies that would enforce women’s reproductive autonomy (i.e. women’s ability to make independent decisions about their reproduction).
The significance of intra-relationship economic power in the dynamics of abuse and terminated pregnancy is evident in the finding that women who earned more than their spouse, and women who were more educated than their spouse were more likely to experience sexual IPV; likewise, women who earned more than their spouse, and women whose spouse did not contribute to household earnings were more likely to experience emotional IPV (though the association between emotional IPV and terminated pregnancy was statistically non-significant), and women whose spouse did not contribute to household earnings were more likely to experience “any IPV”. Some of these findings have been previously reported [27, 58], although only in relation to the association between relationship characteristics and IPV. Differences in relationship characteristics have not been fully explored in relation to such outcomes as terminated pregnancy, and to our knowledge, this is the first study to examine this relationship. Plausible explanations are that women who are earn more or are more educated than their male partner in this context may be regarded as challenging existing gender norms, which increases the likelihood of exposure to abuse and the possibility, and also shown in other studies [59, 60]. This calls for behavioural change interventions promoting joint decision-making within intimate relationships as an attractive strategy for increasing women’s views within the marriage whilst encouraging men to settle household disputes through negotiation, and not violence.
Among the socio-demographic explanatory factors, the finding that women 25–34 years were more likely to report physical IPV, and women 24 years or younger were more likely to report sexual IPV in comparison to women 35 years or older is in agreement with findings from other studies [26, 61], which indicate that younger age at pregnancy is associated with higher risk of partner violence. Younger women tend to be less educated, which may translate to limited economic opportunities, increased vulnerability, and economic dependence on the male partner, and submission to male dominance and abuse . Alternative explanations may be under-reporting of abuse among women 35 years or older who tend to be more educated, have higher socioeconomic status within and beyond the household, and greater reproductive autonomy, and possibly not desiring more children, wanting instead to pursue a career. These findings substantiates those from a study documenting high rates of abuse in intimate relationships among women (especially younger women) presenting for sexual and reproductive health services , and underscores the potential opportunity provided by fertility or family planning clinics in initiating screening and providing intervention measures aimed at adverse reproductive outcomes (unwanted/terminated pregnancies, morbidity/mortality) and partner abuse.
Interestingly, we also found a protective effect of low socio-economic position i.e. women with primary education being less likely to report physical IPV, emotional IPV, and “any IPV”, and the likelihood of emotional IPV, and “any IPV” being lower among unemployed women. Though counter intuitive, this finding contrasts those in other studies [28, 64], in which women with lower levels of education (primary or none) were shown to have an increased likelihood of IPV compared to women with higher education. Findings in our study could have resulted from under-reporting of IPV and terminated pregnancy by less educated women, given that several studies have indicated this to be the case among women in Nigeria [33, 65]. As such, less educated women may be even more likely to under-report IPV due to cultural hindrances and beliefs, and a greater adherence to traditional gender norms. This emphasizes the complexity of the association, and the paradoxical effect of women’s education on IPV. Accordingly, the greater personal and economic independence associated with women with higher education may translate into perceptions of powerlessness among some men, who may become abusive, as has been reported in other studies [53, 66, 67]. Further research is warranted on how educational attainment in individual and community-levels is associated with IPV.
The non-use of contraceptives among women who reported physical IPV and sexual IPV is in line with findings in other studies [24, 68, 69], but contrasting with others [45, 52, 70, 71]. Women in violent relationships may have reduced control over the choice and use of reproductive services, including family planning or other fertility control measures for fear of further violence from abusive partners. Alternative explanations may be related with the current economic hardship in Nigeria which may reduce the desire of some men for more children; a woman’s non-use of contraceptive may further stress abusive men to control pregnancy outcomes either directly through IPV or indirectly by coercing abused women to terminate the pregnancy . Rural women were more likely to have experienced all the IPV types, which is consistent with previous studies . Although rural areas are not homogenous, apart from running the risk of under-reporting, they tend to be communities with more traditional gender views [57, 72]. However, a study in Nigeria indicated the higher likelihood of pregnancy terminations in urban areas ; differences in findings may be methodological, as our study also accounted for variations in community-level characteristics. Geographic and social isolation among rural women may also limit opportunities for IPV-exposed rural women to seek much needed social and reproductive health services [18, 32, 73].
Support in part for our third hypothesis is evident in the finding that women’s mean educational level for the community was associated with a higher likelihood of physical IPV, emotional IPV, “any IPV”, and terminated pregnancy. A reasonable explanation may be that women’s increased education might improve their socio-economic status and economic resources to levels equaling or surpassing that of their male partners. Abusive men within communities tolerant of IPV may perceive such women as opposing traditional values that threaten men’s role as “bread winners” of the family, resulting in higher likelihood of abuse, and terminated pregnancy. Similar findings have been previously reported [50, 53], and stress the need to change traditional gender norms based on social learning processes and imitation that could be achieved by repeated social interactions and exchanges defining the parameters of acceptable behavior within communities.
The association between women’s low level of justification of wife beating within the community, physical IPV, “any IPV”, and terminated pregnancy reflects community-level norms and attitudes that accept or are indifferent to IPV and gender inequality; this is consistent with prior findings [28, 29, 53, 74]. That age at first marriage aggregated at the community level was associated with lower likelihood of physical IPV, Sexual IPV, and terminated pregnancy is not unexpected. Early marriage is frequently associated with increase fertility , early childbearing and adverse outcomes for the woman, her fetus or offspring . Early motherhood often terminates women’s educational possibilities, and reduces their employment opportunities, making them socio-economically dependent on their male partner. Such young women are less likely to report IPV, and in the event of pregnancy, may lack sufficient reproductive autonomy to terminate an unwanted pregnancy. Since customs surrounding marriage in Nigeria, such as the age of marriage and selection of a spouse depends to a great extent on societal norms and traditions, age at first marriage could be described as a product of society. In the light of the mounting evidence against it, there is a need for culturally-sensitive interventions aimed at changing such practices.
In addition, living in a community with levels of contraceptive use below the median for the community was associated with a higher likelihood of women’s exposure to physical IPV, sexual IPV, and terminated pregnancy; conversely, residing in a community with contraceptive use at levels above the median for the community was associated with lower likelihood of women’s experience of physical IPV, sexual IPV, “any IPV”, and terminated pregnancy. The inverse association between contraceptive use and IPV is indicative of its complexity, and reiterates our findings at the individual level. However, it is plausible that fear of abuse may constrain the ability of economically disadvantaged women to effectively negotiate or use contraception . Although we cannot infer that the direction of causality flows from domestic violence to contraceptive behavior due to the cross-sectional design of this study, contraceptive use has been reported in a study in Uganda to lead to abuse , due to the belief that women's discrete use of contraceptives was an indication of sexual promiscuity, and a justifiable reason for abuse. Another study in South Africa found that young women attending family planning clinics were often physically abused by their partners . We suggest that there is a need to actively refute common erroneous local beliefs.
Finally, the significant variance between neighbourhoods in the likelihood of terminated pregnancy associated with all the forms of abuse indicates that the total variance in terminated pregnancy in association with these forms of IPV could be explained at the neighbourhood level.