Plain English summary

Intimate partner violence (IPV) can increase maternal health concerns. This study assessed the prevalence of intimate physical and/or sexual IPV among mothers who gave birth in the past 2 years in Bihar, India. In addition, it examined the association of IPV with increased risk of poor birth outcomes (e.g., miscarriage and stillbirth) ever or of pregnancy and delivery complications in the most recent birth. Almost half of these mothers (43%) were physically and/or sexually abused by their husbands in the past year, that is, in the perinatal period for this sample. Further, a history of such violence increased health risks, including miscarriage, stillbirth, and maternal complications.

The study also explored the intersectionality of IPV and poverty with pregnancy and birth outcomes. The association of IPV with miscarriage and stillbirth was less likely to hold true for the richest and poorest women, possibly due to the extremes of high resources (for the richest) and deprivation (for the poorest) being more important contributors to fetal health and survival. However, IPV was associated with labor and delivery complications regardless of women’s income level. Notably, mobile phone ownership was protective for stillbirth.

Overall, these findings highlight the importance of maternal health interventions to address potential health concerns related to the physical and sexual intimate partner violence women face, as a means of improving maternal and infant health outcomes. However, they also suggest the potential utility of asset ownership in conjunction with health approaches, to produce better health outcomes from an empowerment perspective.


Intimate partner violence (IPV) against women is a pervasive human rights and public health concern [1, 2]. Globally, one third of women (35%) have faced physical and/or sexual IPV at some point in their lives [2, 3]. Robust multi-country analyses and meta-analyses document associations between IPV and poor reproductive health outcomes such as miscarriage and abortion at a global scale [1, 3,4,5]. This is particularly important for South Asia, as more than 1 in 4 of the world’s pregnancies [6], and thus a likely disproportionate burden of poor pregnancy outcomes, occurs in the region, and South Asia maintains some of the highest rates of IPV seen globally [2, 3].

Studies from India, South Asia’s largest and most populous nation, also indicate high prevalence of poor reproductive health outcomes, with little improvement in the past decade [11]. State and multi-state level data from women in India indicate a significant association between IPV and abortion [7, 8]. Studies from within the states of Uttar Pradesh and Maharashtra indicate that IPV is also associated with maternal health complications [9, 10]. Nationally-representative data disaggregating miscarriage and abortion are not yet available in India to assess potential associations between IPV and these outcomes, and research on IPV and miscarriage and stillbirth outcomes in India could not be identified in the literature. Much of this work is older, and state level analyses have not included a focus on Bihar, a populous Indian state that has both higher prevalence of IPV (43% for Bihar vs. 29% for India) [11] and maternal mortality ratio (208 for Bihar vs 167 for India, per 100,000 births) [12], relative to India as a whole.

Higher vulnerability to IPV and related health outcomes in Bihar may be a consequence of poorer development and infrastructure in the state [11]. We also see poorer antenatal care utilization and use of important nutritional supplements during pregnancy, as well as higher unmet need for family planning, for Bihar relative to India [11], and these factors are both linked to IPV and can compromise reproductive and maternal health outcomes [1, 3]. These are also tied to poverty in India, which is additionally linked to increased risk for IPV [13]. Hence, poverty may help explain or even exacerbate these associations between IPV and poor reproductive health outcomes. Exploration of this issue analytically has received inadequate attention from the research literature.

The present study builds upon existing evidence by assessing the relationship between lifetime experience of IPV and reproductive health outcomes (miscarriage, stillbirth, and abortion) and maternal health complications among women in Bihar, India. Previous studies have been limited by reliance on older data and have not delineated the association of IPV with different types of pregnancy outcomes such as stillbirth and miscarriage. For maternal complications, we have disaggregated prolonged labor from other complications in pregnancy or delivery given findings from a prospective multicountry study inclusive of India that documents that these complications are on the causal pathway for increased risk for maternal mortality, stillbirth, and neonatal mortality [14]. Secondarily, we also contribute to the literature by exploring the intersectionality of IPV and poverty for women in Bihar, in terms of their potentially greater combined impact on these same reproductive and maternal health outcomes. Findings from this work can guide interventions that seek to simultaneously advance maternal and child health and gender equality.


Study design

We conducted a cross-sectional analysis of follow-up data from the Ananya evaluation, collected via household surveys from March to June 2014 with women who had given birth in the past 23 months in Bihar, India. Ananya was a public health program that supported a combination of supply-side and demand generation efforts to increase maternal and child health care utilization via the public health system in Bihar; the program and corresponding evaluation were initiated in 2012 to address persistent challenges in reproductive, maternal, newborn and child health in the state [15]. Details on the program and evaluation findings are published elsewhere [15, 16]. Ananya was not designed to identify or intervene on issues of IPV, but IPV was included in the 2014 follow-up survey.

Sampling and procedure

A sample of mothers representative at the state-level who delivered in the past 23 months was included in the current study; this included the eight districts in Bihar in which the Ananya program was implemented as well as the remaining 30 districts assigned to the standard of care control condition for the evaluation. Details on the study’s multi-stage sampling and recruitment procedures are available elsewhere [15, 16]. The 2014 follow-up survey had a response rate of 87%, and yielded 11,408 interviews with mothers who had children aged 0–11 months, and 2455 interviews with mothers who had children aged 12–23 months.

Given the sensitive nature of the information, all data were collected by trained female study staff, after obtaining written informed consent. The staff were trained in sensitive assessment according to WHO guidelines [17], ensuring complete privacy during questions and a debriefing in cases where trauma or retraumatization were indicated. Referrals were provided to local domestic violence service. If indications of immediate lethal risk was indicated, the survey was to be stopped and immediate connection to protections was to be provided. However, no such incidents occurred. Ethical approval for the original evaluation study was provided by the Screening Committee of Government of India’s Ministry of Health and Family Welfare. Ethical approval for this analysis was provided by the University of California, San Diego (UCSD). The current analysis was restricted to women with singleton births and without missing values for any variables of interest (n = 13,803, a loss of only n = 63 participants in the dataset).


The independent variable for this analysis was self-reported lifetime experiences of physical and/or sexual intimate partner violence in their current marriage. Physical violence was comprised of items on whether the participants had experienced any of the following from her husband: being slapped, having an arm twisted or hair pulled, being pushed with a fist, being shaken or having something thrown at them, being kicked, dragged or beaten up, or choking or burning. Sexual violence was measured using a single item asking “Did your husband ever physically force you to have sexual intercourse with him even when you did not want to?” Responses were yes/no for the timeframes of ever and past 12 months. Both any IPV ever and IPV recent measures were created from these items, but only the IPV ever measure was included as the independent variable given that multiple outcomes of focus were also in this timeframe. These IPV items were validated in the multi-country study on Violence Against Women by the World Health Organization and are routinely included in the Demographic Health Surveys (DHS), including in India’s DHS, known as the National Family Health Survey (NFHS) [1, 13].

Dependent variables included the following reproductive health outcomes: miscarriages, stillbirths, abortions; these were assessed via single survey items on whether these ever occurred. Maternal health complications were specific to the index childbirth that allowed them to meet eligibility criteria, which was the most recent childbirth occurring in the past 23 months, and these were divided into labor complications and symptoms of other pregnancy or delivery complications. Items on labor complications assessed prolonged labor over 12 h and pre-term labor, defined as labor pain in the seventh or eighth month of pregnancy. Items on other pregnancy or delivery complications included excessive bleeding, convulsions, swelling of the hands, body, or face, fever, vaginal discharge, or “any other such symptom”.

Covariates considered in the analyses related to sociodemographics, parity, and study design. Sociodemographics assessed included maternal age (15 to 19, 20 to 24, 25 to 29, 30 or older), wealth (in quartiles), education (none, primary, secondary or higher), religion and caste identity (Scheduled Tribe/Scheduled Caste [SC/ST], Other Backward Caste [OBC], Muslim, or General Caste], and paternal education (none, primary, secondary or higher). The wealth index was constructed via principle component analysis of household characteristics and assets following the technique used in the NFHS [13]. Parity at survey was categorized as 1, 2, or 3 or more. The study design covariate simply accounted for Ananya treatment group (i.e., whether or not the respondent lived in an Ananya program district).

Given the study focus on IPV, we additionally considered gender equity variables as covariates. These included maternal age at marriage (under 18 vs. 18 or older), workforce participation in the past 12 months, personal bank account ownership, and ownership of a mobile phone, as they are factors related to autonomy that are significantly different for women and men in India [11]. For maternal complication outcomes, we also included covariates specific to the index child: sex of child, age of child (0 to 5 months, 6 to 11 months, 12 to 23 months), having four or more antenatal care visits, and having a skilled birth attendant at delivery.

Data analysis

Descriptive frequencies were calculated for all outcomes and covariates, cross-tabulated by lifetime IPV with Rao-Scott chi-square tests of independence. Univariate and multivariable logistic regression models were used to assess the associations between IPV on each health outcome. Although Ananya district, age of mother, and parity were forced into all adjusted models, backwards stepwise model building was conducted for each outcome using an inclusion cut point of p < 0.2. Institutional versus home delivery was also considered for inclusion, but 97% of those with a skilled birth attendant delivered in an institution and the variance inflation factors exceeded 6.5 in a collinearity check, so these variables were deemed overlapping. This approach was used to allow for a more parsimonious model, given the small cell sizes for some outcomes such as abortion. An additional model was conducted to test for an interaction between IPV and wealth. Final parsimonious models for all outcomes were then stratified by wealth index to provide insight into the intersectionality of IPV and poverty. All analyses adjusted for survey design and individual sampling weights and were conducted using Stata 14.2 SE; and statistical significance was evaluated at p < .05 unless otherwise indicated.


Description of the sample

Participants were aged 15–49 years (Mean = 25.3, Std Dev = 4.3). The majority (52.1%) had no formal education, and were from a socially marginalized group (as indicated by 26.5% SC/ST, 46.3% OBC, and 17.3% Muslim). [See Table 1.] For the index pregnancy, 30.1% were first children, and 53.2% were male. Receipt of four or more antenatal care visits was reported by only 20.0% of the sample, but 72.4% had a skilled birth attendant at delivery.

Table 1 Descriptive Statistics by Lifetime Experience of IPV (N = 13,803)

Almost half of participants (45.1%) reported IPV ever; 28.6% reported only physical IPV, while 2.3% of them had faced only sexual IPV and 14.3% had faced both physical and sexual IPV. [See Table 1.] A slightly smaller percentage (43.0%) had experienced IPV in the past 12 months, with 27.6% reporting only recent physical IPV, 2.7% reporting only recent sexual IPV only, and 12.7% reporting both. A history of miscarriage, stillbirth, and abortion was reported by 8.7, 4.6, and 1.3% of the sample, respectively. More than one in 10 women (10.7%) reported labor complications at last pregnancy, and 16.3% reported other complications during pregnancy or delivery.

Correlates of IPV

Chi-square tests document significant associations (P < .001) between IPV and each of the following outcomes: miscarriage, stillbirth, and other pregnancy or delivery complications [Table 1]. However, upon breaking down other pregnancy and delivery complications into its component types, it appeared that excessive bleeding, convulsions, and swelling may have been the drivers of this association. Though 4.6% of the sample reported bleeding, another 4.6% convulsions, 6.5% swelling, 6.2% fever, 2.9% vaginal discharge, and 1.4% another symptom, the women who had endured partner violence reported significantly higher proportions of these first three relative to those who had not (bleeding: 5.7% vs. 3.7%, P < .001; convulsions: 5.6% vs. 3.9% P = .02; swelling: 6.1% vs. 5.1%, P < .001). Moreover, 8.6% of the sample reported prolonged labor and 3.5% reported pre-term labor, but these did not differ by experience of IPV.

Chi-square analyses also indicated that women reporting IPV were significantly more likely to indicate greater sociodemographic vulnerability including low to no education, a husband with low to no education, and early marriage. (See Table 1.) IPV was also more likely among older women and those with higher parity, as well as those not reporting 4+ ANC visits or a skilled birth attendant (SBA) for the index pregnancy. In terms of economic indicators, IPV was associated with lesser wealth, lack of female work force participation in the last 12 months, and not having a bank account or phone, which were all significantly correlated with each other.

Regression analyses assessing associations between IPV with reproductive and maternal health outcomes

Adjusted logistic regression analysis revealed that women who had ever faced IPV in their relationships had higher odds of ever having had a miscarriage (AOR = 1.35, 95% CI = 1.11–1.65) or stillbirth (AOR = 1.36, 95% CI = 1.02–1.82). (See Table 2.) Women’s risk for ever having these outcomes increased with age, as expected. Women who married at age 18 or older (relative to married at < 18) had lower odds of miscarriage and stillbirth. Women with more educated husbands had higher odds of stillbirth, whereas those who owned a mobile phone had lower odds of stillbirth ever. No significant association between IPV and abortion was observed, but the wealthiest women (relative to the poorest) and employed women (relative to non-income generating women) had higher odds of ever having had abortion, whereas women who married at age 18 or older (relative to those married at < 18) and those of marginalized caste/religion (relative to general caste women) had lower odds of having received an abortion.

Table 2 Logistic Regression to Assess Associations between Lifetime Experience of IPV and Birth Outcomes

Multivariable analyses also documented significant associations between IPV and both labor complications at last birth (AOR = 1.27, 95% CI = 1.04–1.54), and other pregnancy or delivery complications at last birth (AOR = 1.68, 95% CI = 1.42–1.99). (See Table 3.) Notably, the bivariate association between IPV and labor complications was not significant, but became significant in adjusted analyses. Women had lower odds of labor complications if they were older and were higher parity, but women with more education, an educated husband, employment, a male child, and receipt of ANC and an SBA had higher odds of labor complications. In terms of complications during delivery, women who were older and those with a mobile phone had lower odds of this outcome, though again husband education, women’s employment, receipt of ANC and SBA, as well as a male index child were associated with higher odds of complications during delivery.

Table 3 Logistic Regression to Assess Associations between Lifetime Experience of IPV and Labor/Pregnancy Complications

Interaction effects of wealth index and IPV on reproductive and maternal health outcomes

To assess whether wealth moderated the effect of IPV on reproductive and maternal health outcomes, a wealth x IPV interaction term was included in the final parsimonious models described above, and wealth was restored to the model if it was otherwise dropped, to determine if the interaction affected our outcomes of interest. No interactions were observed at p < .15. Wealth stratified adjusted regression models were also conducted to assess whether observed associations between IPV and reproductive and maternal health outcomes differed by wealth quartile. Only miscarriage and stillbirth indicated differences in associations across wealth quartiles. (See Table 4.) IPV was not associated with miscarriage for poorest (Quartile 1) women (AOR = 0.98, 95% CI = 0.67–1.45), but significant IPV and miscarriage associations were seen with women in wealth quartiles 2 (AOR = 1.86, 95% CI = 1.19–2.92) and 3 (AOR = 1.55, 95% CI = 1.07–2.25). For stillbirth, a significant IPV effect was only seen for women in wealth Quartile 3 (AOR = 1.79, 95% CI = 1.04–3.08). Stratified analyses did not indicate meaningful differences in IPV effects on other outcomes by wealth quartile.

Table 4 Wealth-Stratified Logistic Regression Analyses Assessing Associations between IPV and Birth and Maternal Complication Outcomes


The findings from this study indicate that nearly half of women in Bihar, India (45%) reported physical and/or sexual IPV from their husbands, and most of these also reported IPV in the past year, which would be in the pregnancy or postpartum period for this representative sample of mothers. This prevalence is fairly comparable to that seen in recent state level data from Bihar, which reported that 43.2% of married women (15–49 years) had ever experienced spousal violence [11]. Further, those who experience such violence, are significantly more likely to report miscarriage, stillbirth, and abortion, as well as maternal health complications, compared to women who had not faced violence from their husbands. These findings highlight the importance of the health care sector in terms of screening for IPV and supporting women with IPV histories as they appear to experience a disproportionate burden of high risk pregnancy outcomes.

Consistent with findings from prior multi-country research not specific to India [1], this study demonstrates that exposure to IPV increases women’s odds of experiencing miscarriage, and extends this work by also documenting that there is a significant association between IPV and stillbirth. These findings, taken with prior research from India demonstrating associations between IPV and infant morbidity and mortality [18,19,20,21], suggest that IPV is an important factor in compromising fetal health. Similarly, current findings also document increased risk for maternal complications, results seen in prior research from India [9, 10]. Observed associations between IPV and reproductive and maternal health risks could be attributable to direct injury from IPV in pregnancy, though studies from India suggest that IPV declines with pregnancy [11, 22]. Rather it may be due to other forms of abuses that may continue in pregnancy (e.g., denial of rest or food) [10, 20, 23], or from the cumulative stress and mental health trauma caused by IPV, manifesting in physical health concerns [24, 25] and poor perinatal health outcomes such as preterm delivery, rupture of uterus, infection, placental abruption, fetal injury and death [26, 27]. Further research is needed to explore the mechanisms to explain these observed associations.

Contrary to prior research from India [7, 8] and elsewhere [1, 3,4,5], findings from this study do not document a significant association between IPV and abortion. The prevalence of abortion was very small in the current study (< 2%) and may have been too small to be able to detect the association; this self-reported abortion data may be an underestimate given stigma around the practice in India [28], particularly due to the laws related to sex-selected abortion. Given that the respondents were all women who had given birth in the preceding 23 months, the estimate is also not representative of the larger population of women in the state. Further research is needed to clarify these findings. Nonetheless, even with this small number of cases, analysis of covariates documents that economic indicators drive this outcome. Wealthiest and employed women are most likely to have had an abortion. This is likely because wealth can bring greater access to abortion services [29], particularly through increased women’s autonomy, mobility, and resource control [30, 31]. While women’s workforce participation in India is more likely among poorer women [13], such participation even among this poorer population suggests that pregnancies may be linked with perceptions of higher opportunity costs as a child can result in lost earnings. Similar findings on wealth, income generation and abortion have been documented in other studies from India [29, 32, 33].

Economic circumstances also affect observed associations between IPV and miscarriage and stillbirth, as seen in the wealth-stratified analyses of our study. Contrary to our hypothesis, the associations between IPV and miscarriage and stillbirth were not seen for the poorest women in our study. These findings suggest that potential effects of IPV on fetal health and survival may not be as influential as other factors related to deprivation, such as poor maternal nutritional status [34], lack of education, lack of skilled attendant at delivery, lack of consistent antenatal care, low birth weight, prematurity and previous stillbirths [35]. In contrast, the wealthiest women may be buffered somewhat from these potential IPV effects on fetal health, as indicated by insignificant or attenuated findings for this wealth stratum in this study. Notably, middle-income women were the population seen to be more affected by the associations of IPV with miscarriage and stillbirth. In the case of stillbirth, women’s asset ownership/access such as to a mobile phone was associated with lower risk, possibly because these women had facilitated access to health care, social support, or healthy behaviors in pregnancy via such means. Assets such as a bank account or a mobile phone were also associated with lower risk for IPV in this sample, findings seen in prior research as well [36, 37].

While study findings document the potential value of asset ownership/control as a means of helping women reduce their perinatal health risks, and show that women living with IPV are less likely to have these assets, women’s income generation is associated with both increased risk for IPV and increased risk for maternal health complications. These findings are consistent with other studies from India documenting greater vulnerability to IPV for working women and suggesting that income generation may not be a valid indicator of women’s economic empowerment in this context [38,39,40,41,42]. In contrast, research from India documents that women’s direct ownership or control over assets or household resources is associated with lower risk for IPV [38, 43], findings also seen in the current study; this suggests that these constructs are better indicators of women’s economic empowerment in India. Financial inclusion programs (e.g., microfinance/microloans [44, 45], bank accounts/digital accounts [38, 46]) and self-help groups supporting women’s engagement in these programs [44, 47, 48] may facilitate women’s asset ownership/control and help reduce IPV and related perinatal health outcomes beyond efforts that can be offered through the health sector.

Findings from this study should be considered in accordance with certain limitations. This study involved cross-sectional analysis of survey data from a midline evaluation of an intervention in a single state in India. Causality cannot be assumed, though use of a state-wide representative sample of mothers who recently have given birth (excluding stillbirths) does allow greater generalizability of findings to the state. We also adjusted for intervention group, to address the role of treatment as a confounder. Experience of IPV in this study is limited to self-reported physical and sexual violence, and does not include other manifestations like emotional and financial violence. These data also rely on participant self-reports of reproductive outcomes and IPV experiences, which could be subject to recall and social desirability bias. Social desirability bias could be more acute for sensitive items such as sexual IPV and abortion, and could lead to an under-reporting of these experiences. Future research should include longitudinal data to assess prospectively the impact of IPV on reproductive and maternal outcomes; data from medical records and across multiple states would also support assessment of validity and improve generalizability of findings.


Findings from this representative sample of mothers who recently gave birth in Bihar India document that almost half (43%) have experienced IPV in the past year, suggesting that such violence in the perinatal period is pervasive. Further, findings indicate that a history of IPV in their marital relationship is associated with increased maternal health risks, including miscarriage, stillbirth, and maternal complications. While observed associations were less likely to hold true for the richest and poorest women in terms of miscarriage and stillbirth, possibly due to the extremes of high resources (for the richest) and deprivation (for the poorest) being a more important contributors to fetal health risk than was IPV, IPV was associated with labor and delivery complications regardless of women’s income level. Notably, indicators of asset ownership were also related to lower risk for IPV and perinatal health concerns. These findings overall highlight the importance of maternal health interventions that can support women’s marital safety and address potential health concerns related to the violence may face, as a means of improving maternal and infant health outcomes. However, they also suggest the promise of economic interventions in conjunction with health approaches to produce better health outcomes from an empowerment perspective.