Maternal experience of intimate partner violence is associated with suboptimal breastfeeding practices in Malawi, Tanzania, and Zambia: insights from a DHS analysis

Background Despite widespread suboptimal breastfeeding practices and maternal experiences of intimate partner violence (IPV), the association of IPV and breastfeeding practices remains unclear. This study investigated the associations between maternal experience of physical, sexual, emotional violence, and controlling behaviors with suboptimal breastfeeding practices in Malawi, Tanzania, and Zambia. Methods Data included mother-infant dyads from the most recent Demographic and Health Surveys for Malawi (n = 1878), Tanzania (n = 3184), and Zambia (n = 3879). Intimate partner violence (physical, sexual, and emotional) was measured using the Revised Conflict Tactics Scale. Breastfeeding practices were early initiation of breastfeeding (within 1 h of birth), exclusive breastfeeding (in previous 24 h), and continued breastfeeding. Associations between experience of physical, sexual, or emotional violence or controlling behaviors and breastfeeding practices, as well as associations between the frequency of IPV and breastfeeding practices were assessed. Results Many Malawian (77.6%) and Zambian (67.7%) and just over half (53.6%) of Tanzanian mothers, initiated breastfeeding within 1 h of birth. Exclusive breastfeeding was 70.6% in Zambia and 60.1% in Malawi, while 55.0% of Tanzanian mothers breastfed exclusively. Continued breastfeeding to at least 1 year was high in Malawi 92.2%, Tanzania 93.4%, and Zambia 95.0%. Most mothers reported experiences of IPV in Tanzania 79.1%, Zambia 78.9%, and Malawi 73.9%. Mothers who experienced sexual IPV were significantly more likely to delay breastfeeding (Malawi [AOR 1.55 (1.14, 2.10)]; Tanzania [AOR 1.30 (1.04, 1.62)]; and Zambia [AOR 1.28 (1.06, 1.54)]). Sexual IPV in Malawi and Zambia was associated with greater odds of not exclusively breastfeeding (Malawi [AOR 1.90 (1.05, 3.45)]; Zambia [AOR 1.75 (1.15, 2.67)]). Tanzanian mothers who experienced IPV often or sometimes were two times more likely not to breastfeed at one-year post-delivery [AOR 2.23 (1.09,4.57)]. Conclusions In the three countries investigated maternal experience of IPV was associated with suboptimal breastfeeding practices. Policies and programs targeting improved breastfeeding practices should consider screening during antenatal and postnatal care for experience of violence and support initiatives to reduce IPV.


Background
Globally, undernutrition is linked to approximately 45% of all deaths among children under 5 years of age [1]. Optimal nutrition is critical to healthy growth and development, especially during the first 1000 days [2]. Breastmilk, the first food for infants, provides the nutritional needs for the first 6 months of life [1] and promotes optimal neurodevelopment while decreasing the incidence of infectious diseases and infant mortality [2]. Consequently, unless contraindicated, the World Health Organization (WHO) recommends that newborns be breastfed within the first hour of birth and that infants be exclusively breastfed (EBF) until they are 6 months of age. Additionally, the WHO recommends that breastfeeding be continued at least until the first birthday and up to 2 years or beyond [3].
In efforts to improve breastfeeding practices, researchers have sought to uncover the determinants of breastfeeding [9,10]. While the evidence is wide-ranging on factors associated with breastfeeding practices, psychosocial factors such as maternal experience of intimate partner violence (IPV) are not well-understood. Intimate partner violence refers to the behaviors within an intimate relationship (partner or spouse) that cause psychological or emotional, physical, or sexual harm [11]. Intimate partner violence also includes controlling behaviors, such as restricting access to financial resources, medical care, or limiting social interactions [11]. Reports show that maternal experience of IPV in Malawi, Zambia, and Tanzania is common [6][7][8]. In the 2007 DHS survey, 47% of Zambian women reported physical IPV and 41.8% reported that their husbands displayed three or more controlling behaviors [7]. In the 2010 DHS, 25% of Malawian women reported having experienced emotional IPV and 19% reported sexual IPV [8], while 39% of Tanzanian women reported physical IPV [6].
Despite the common occurrence of IPV, impacts on breastfeeding remain unclear. However, two contradicting theories, the deficit hypothesis [12] and the compensatory hypothesis [13] have been proposed to explain a probable link between IPV and breastfeeding practices. The deficit hypothesis suggests that mothers who experience violence may have physical, psychosocial, or emotional barriers to breastfeeding, and therefore, are less likely to breastfeed [12]. Contrarily, the compensatory hypothesis proposes that maternal experience of IPV may improve breastfeeding practices as a result of the mother's increased sensitivity to the needs of her infant [13].
Overall, most evidence across countries suggests that mothers who experience IPV are less likely to breastfeed optimally [14][15][16][17][18][19], which is consistent with the deficit hypothesis. However, one study that included Malawian, Tanzanian, and Zambian mothers reported that those who experienced IPV were more likely to have optimal breastfeeding practices [20]. Therefore, because previous findings in Malawi, Tanzania, and Zambia, unlike several other African countries [20], supported the less common compensatory hypothesis, further investigation of associations between maternal experience of IPV and breastfeeding in these three countries is needed.
Furthermore, most studies reported lifetime IPV [14][15][16][17][18][19][20] which may not capture differences between mothers who currently experience IPV as opposed to those who experienced IPV in the past. Women who experience either past or current IPV may both have negative effects [11], but their breastfeeding choices may differ. Perhaps women who experienced IPV in the past may have had time to receive resources or support to help mitigate the negative effects of IPV on breastfeeding, and thus align more with the compensatory hypothesis.
Additionally, data are lacking on impacts of controlling behaviors, despite these being a form of IPV [11]; most studies have focused on the associations between physical, sexual, or emotional IPV and breastfeeding [14][15][16][17][18][19][20]. Because controlling behaviors can make a mother less autonomous by isolating her from support systems or restricting her access to medical care, it is important to understand how this form of IPV may influence a mother's breastfeeding practices. Previous findings in India suggested that maternal autonomy was associated with a higher likelihood of breastfeeding [21]. Therefore, the purpose of this study was to investigate the associations between physical, sexual, and emotional violence, as well as controlling behaviors, with early initiation of breastfeeding, EBF, and continued breastfeeding in Malawi, Tanzania, and Zambia. Additionally, we analyzed associations between frequency of IPV with breastfeeding practices in each country. Besides advancing the understanding of IPV and breastfeeding, the results from this study may support interventions, programs, and policies that seek to improve breastfeeding practices.

Data source and sampling
Data were obtained from the most recent Demographic and Health Surveys (DHS) for the following neighboring African countries of Malawi (2015-2016), Tanzania (2015-2016), and Zambia (2013-2014). Each of the country's DHS used stratified, cluster sampling. While all women 15-49 years were eligible for interviews, only a random subsample were selected for the Revised Conflict Tactics Scale (CTS-2) [22] used to assess intimate partner violence. Our analysis for each country included women 15-49 years of age who had an infant aged between 0 and 24 months, who completed the CTS-2 scale and answered questions about breastfeeding. Figure 1 shows the original sample size for each country's mother-child datasets, the exclusion criteria and the final sample sizes used for analyses. Our conceptual framework outlining the pathways between maternal experience of IPV and breastfeeding is shown in Fig. 2.

Analyses
The three study outcomes were the WHO's first three Infant and Young Child Feeding (IYCF) indicators: 1) early initiation of breastfeeding 2) EBF, and 3) continued breastfeeding [3]. Early initiation of breastfeeding was measured as the proportion of children born in the last 24 months who were put to the breast within 1 h of birth. Exclusive breastfeeding was measured as the proportion of infants 0-5 months of age who were fed exclusively with breastmilk in the previous 24 h. Continued breastfeeding was measured as the proportion of children 12-15 months of age who were fed breastmilk in the previous 24 h.
Using the CTS2 [22], maternal experience of physical IPV included the following nine variables: ever been pushed, shook, had something thrown at you, slapped, punched with fist or hit with something, kicked/dragged, strangled/burnt, threatened with knife/gun/other weapon, or had arm twisted/hair pulled. Emotional IPV was assessed by three variables: reported experience of humiliation by husband/partner, threatened with harm, insulted or made to feel bad. Sexual IPV included: being physically forced into unwanted sex, forced into other unwanted sexual act, or physically forced to perform sexual act. In addition to the CTS2, the DHS questionnaire also included a series of questions about whether partners ever displayed controlling behaviors within their relationship. Controlling behavior was described as: whether the husband/partner was jealous if respondent talks with other men, whether he accuses the respondent of unfaithfulness, whether he forbids respondent to meet her friends, whether he tries to limit respondent's contact with family, whether he insists on knowing where respondent is, or whether he does not trust her with money.
Intimate partner violence exposure was measured by maternal reported experience of specific forms (physical, sexual, emotional, or controlling behaviors) and frequency (never, yes but not in the last 12 months, often/ sometimes). Not having experienced any form of IPV was coded 0 and having experienced a form of IPV was coded 1. Frequency of experience of any form of IPV was coded as: 0 = never, 1 = yes but not in the last 12 months, 2 = often/sometimes. Frequency of controlling behavior was not collected and therefore could not be analyzed.
Child, maternal, and paternal characteristics for each country were computed using descriptive statistics and weighted frequencies. The weighted proportions of children meeting breastfeeding indicators and percentage of mothers who reported experience of IPV were calculated for each country. Bivariate logistic regressions were performed to determine the association between IPV exposures and each breastfeeding outcome. Variables that 1 Children older than 24 months were excluded as well as those children with missing data on age. 2 Women who were not selected for the Conflict Tactics Scale, the module that included questions about intimate partner violence, were excluded. 3 Any mother-child dyads missing information about breastfeeding were excluded. reached a p -value < 0.05 in the bivariate analyses were analyzed using multivariate logistic regression. No variables reached a p -value between 0.10 and 0.05 in the bivariate analyses. Multivariate analyses were adjusted for strata, cluster, and for covariates including the child's sex, maternal age, occupation, education, literacy, exposure to media, type of delivery (Caesarean section or vaginal), delivery location, birth spacing, antenatal visits, household wealth index, and urban or rural residence [23]. Multicollinearity of explanatory variables was tested using the Variance inflation factor (VIF) and the tolerance test and results were within normal limits. Variables with a p -value < 0.05 were considered significant in the multivariate analyses. SAS, v. 9.4, was used for statistical analyses.

Results
Child, maternal, and paternal characteristics Table 1 summarizes child, maternal, and paternal characteristics for Malawi, Tanzania, and Zambia. In all three countries, child stunting was high, in Malawi (33.8%), Tanzania (32.7%), and Zambia (36.1%). Incidence of underweight and wasting were lower but generally similar across countries. Child morbidity such as fever (36.2%) and diarrhea (36.6%) were particularly common among Malawian children.
For each country, primary education was the highest level achieved for more than half of the mothers (Malawi 68.3%; Tanzania 65.7%; Zambia 59.0%). In all three countries, the majority of mothers were working outside the home (Malawi 70.7%; Tanzania 80.7%; Zambia 58.5%). Across all three countries, the mean age for men was higher compared to women and more men had completed secondary education or higher. The most common occupation for men was agriculture in Malawi 39.5%; Tanzania 56.7%; Zambia 56.3%.
Many Tanzanian (62.4%) and Zambian (53.6%) women believed being beaten by their husbands was justified, but this was not a commonly reported belief among Malawian women (16.0%). More than half of Malawian (56.9%) and Zambian (53.5%) mothers reported being afraid of their husbands at least sometimes. Men rarely had been physically hurt by their wife or partner (Malawi 3.9%; Tanzania 1.8%; Zambia 6.3%), although it is important to note this question was self-reported by the wives/partners and the men were not asked directly.

Breastfeeding
Early initiation of breastfeeding and EBF varied across countries while continued breastfeeding practices were similar ( Table 2). While 77.6% of Malawian mothers initiated breastfeeding within 1 h of birth, this percentage was lower in Tanzania (53.6%) and Zambia (67.7%). Just over half (55.0%) of Tanzanian mothers exclusively breastfed but exclusive breastfeeding was higher in Malawi (60.1%) and Zambia (70.6%). Continued breastfeeding until the child's first birthday was high in all countries: Malawi 92.2%, Tanzania 93.4%, and Zambia 95.0%.

Intimate partner violence
Intimate partner violence was similar across countries ( Table 2). Many of the mothers in all three countries reported experiences of IPV in Tanzania (79.1%), Zambia (78.9%), and Malawi (73.9%). In all three countries, controlling behaviors were the most commonly reported IPV and sexual IPV was the least commonly reported. Among women who reported IPV, many experienced only one form. However, in both Malawi and Tanzania, 20.6% reported experiencing two forms of IPV, and in Zambia 11.8% experienced three forms of intimate partner violence. The frequency of the different types of IPV   Intimate partner violence and delayed initiation of breastfeeding    Table 5 Adjusted associations between maternal experience of intimate partner violence (IPV) and cessation of breastfeeding IPV often or sometimes were more likely not to have exclusively breastfeed, but this did not remain significant in the adjusted models. On the contrary, Tanzanian women who experienced emotional IPV more than 1 year prior to the survey were more likely to have exclusively breastfed [AOR 0.38 (0.18, 0.78), p < 0.01] compared to women who reported no experiences of intimate partner violence.

Intimate partner violence and cessation of breastfeeding prior to 1 year
Evidence signifying associations between IPV and continued breastfeeding until the child's first birthday were lacking ( Table 5). The only significant finding was that Tanzanian mothers who often or sometimes experienced emotional IPV were two times more likely not to continue breastfeeding for a full year [AOR 2.23 (2.09, 4.57), p < 0.05]. There were no significant associations between IPV and continued breastfeeding among Malawian and Zambian mothers.

Discussion
This study used nationally representative data from Malawi, Tanzania, and Zambia to investigate associations between physical, sexual, and emotional violence, as well as controlling behaviors, with early initiation of breastfeeding, EBF, and continued breastfeeding. Additionally, associations between frequency of IPV, whether often/sometimes or more than one-year prior, were analyzed for each breastfeeding indicator. The major finding was that maternal experience of various types of IPV was associated with suboptimal breastfeeding practices in all three countries.

Intimate partner violence and delayed initiation of breastfeeding
In all three countries, maternal experience of any IPV was negatively associated with early initiation of breastfeeding which is consistent with the deficit hypothesis. Furthermore, experiencing emotional IPV specifically was associated with delayed initiation of breastfeeding in Malawi and Tanzania, which conflicts with previous findings from DHS analyses that showed no significance between emotional IPV and early initiation of breastfeeding in these countries [20]. However, similar to our findings, maternal experience of emotional IPV was associated with breastfeeding avoidance and higher odds of delayed initiation of breastfeeding in both Spain and Kenya [20,23]. Furthermore, a population-based study including pooled results from 51 low-income and middle-income countries found maternal lifetime experience of emotional violence was associated with lower odds of early initiation of breastfeeding [14].
Because the few studies that have analyzed emotional or psychological IPV have been cross-sectional and cannot determine casual pathways, the mechanism of the association remains unclear. Experience of IPV has been associated with poor self-esteem [24] and both low selfefficacy and low confidence have been identified as barriers to optimal breastfeeding practices [25,26]. That delayed initiation of breastfeeding among Tanzanian mothers was associated both with those who experienced emotional IPV often or sometimes and those who had experienced emotional IPV in the past illustrates the potential long-term negative influence of emotional IPV on breastfeeding. However, the proposed link between experience of IPV, low self-efficacy, and delayed initiation is only one plausible pathway. Future studies may identify other ways by which experiencing emotional IPV leads to delayed initiation of breastfeeding.
Additionally, mothers in all three countries who experienced sexual IPV were more likely to delay breastfeeding. Results from Malawi and Tanzania [20] showed no significant association between IPV and delayed breastfeeding initiation. However, our results are consistent with findings in India [18] and Zambia [20]. The negative impact of experiencing sexual IPV on mothers may contribute to the barriers and challenges of early initiation of breastfeeding. One study has suggested that maternal experience of sexual IPV may lead mothers to develop negative associations with breastfeeding [27]. Furthermore, it has been hypothesized that any previous experience of sexual IPV may increase a mother's cortisol levels [12] which has been linked to delayed onset of milk production up to 3 or 4 days postpartum [28]. Due to the limited available data on sexual IPV and early initiation of breastfeeding, investigation of additional possible mechanisms of how experiencing sexual IPV negatively impacts early initiation of breastfeeding would be beneficial.
In Zambia, mothers whose partners had controlling behaviors were more likely to delay the initiation of breastfeeding. Perhaps the partner's controlling behaviors diminished women's empowerment, which has been identified as a key factor in optimal breastfeeding practices [21]. Because this study is the first to identify controlling behaviors as negatively associated with breastfeeding, future research should consider analyzing which specific controlling behaviors are most detrimental for breastfeeding practices. Furthermore, potential mediating factors should be identified to help women overcome the negative influence of controlling behaviors on early initiation of breastfeeding.
In addition to the negative association across countries between early initiation of breastfeeding and sexual or emotional IPV, and controlling behaviors, physical IPV was also negatively associated with early initiation of breastfeeding in Tanzanian women regardless of whether the IPV was recent or in the past. Even experiencing physical IPV more than 1 year prior to the study may negatively impact breastfeeding as IPV is known to have long-term negative effects [11]. Similar to sexual IPV, physical IPV can lead to increased maternal cortisol [28]. Additionally, a mother's body, including her mammary glands, may have sustained physical damage as a result of physical IPV [11] thus interfering with early initiation of breastfeeding.

Intimate partner violence and non-exclusive breastfeeding
Associations between maternal experience of IPV and non-exclusive breastfeeding generally varied across countries. Similar to findings in other countries [14,17,18,29], children born to Zambian mothers who reported physical IPV were more likely to receive foods or fluids other than breastmilk before 6 months of age. Likewise, both Malawian and Zambian mothers who experienced sexual violence had higher odds of not exclusively breastfeeding.
Contrary to all other findings of this study, Tanzanian women who experienced emotional IPV more than 1 year prior to the survey were more likely to have exclusively breastfed. These results are consistent with the compensatory hypothesis. Since the reported IPV occurred more than 1 year prior to the study, it's possible that these mothers were able to receive help to overcome the possible negative effects IPV might have had on their breastfeeding practices. The results in Tanzania differed from Zambian mothers who, if they had experienced emotional violence often or sometimes, were more likely to give foods or other liquids to the infant before 6 months of age. These contradictory results between countries suggest there may be other mediating factors that are country specific, such as breastfeeding support for women or psychosocial counseling resources that mediate the influence of IPV on breastfeeding.
However, further investigation is warranted to better understand how prior experience of IPV led mothers in Tanzania to be more likely to exclusively breastfeed. Understanding Tanzanian mothers' intention and choice to exclusively breastfeed despite their experience of IPV could be beneficial in designing interventions that provide breastfeeding support for other mothers who experience intimate partner violence. Future studies may consider exploring the use of breastfeeding counselling or other culturally-specific social support interventions as ways to mitigate the negative influence of IPV on EBF in Malawi and Zambia.

Intimate partner violence and cessation of breastfeeding prior to 1 year
To the best of our knowledge, this study was the first to analyze the associations between IPV and continued breastfeeding. Intimate partner violence was associated with cessation of breastfeeding in Tanzania among mothers who experienced emotional IPV often or sometimes. These mothers were two times more likely to stop breastfeeding before the child's first birthday compared to mothers with no reported experience of intimate partner violence. Frequent experience of emotional IPV has been shown to be associated with poor self-esteem [24]. However, these findings were not consistent in Malawi and Zambia; thus, the association between maternal experience of IPV and continued breastfeeding remains unclear. In this context, however, it is important to consider that overall adherence to continued breastfeeding recommendations was high in all three countries.

Recommendations
National strategic plans are prioritizing improvements in breastfeeding in Malawi, Tanzania, and Zambia [30][31][32]. In support of these efforts, screening women for IPV during antenatal care and providing access to services may at least partially mitigate negative effects of IPV, and therefore may be imperative to support breastfeeding mothers. While the WHO (2016) recommends that screening for IPV be conducted during antenatal visits [33], the feasibility of this practice in each country must be addressed. In Tanzania, screening for IPV among women attending outpatient clinics has been initiated [34], but future studies may need to consider the feasibility of IPV screening during antenatal care in these countries.
Furthermore, public health activities that support prevention of sexual IPV, as well as adequate screening of sexual IPV during antenatal care, may allow mothers to receive support and guidance on early lactation management. In addition to screening for IPV during antenatal care, women also should be provided with appropriate physical exams to be aware of any possible bodily damage that may hinder her ability to breastfeed. Because breastfeeding counseling [35] and social support [36] have been effective facilitators of breastfeeding in other countries among women who experience of IPV, both of these interventions may mediate the association between maternal experience of IPV and optimal breastfeeding in Malawi, Tanzania, and Zambia.

Limitations
Despite its strengths, the study had limitations. The use of cross-sectional data does not provide insight on causality and EBF was only queried for the previous 24 h and therefore, may not be reflective of the infant's first 6 months of life. Additionally, the CTS-2 has not been validated in Malawi, Tanzania, and Zambia, and crosscultural reliability has not yet been established in these countries. Furthermore, some maternal experiences of IPV may have remained undisclosed, resulting in underestimation of IPV, while social desirability bias may have resulted in overestimation of breastfeeding adherence. Measures of emotional/psychological IPV have varied across studies which is a barrier to comparing study results. Lastly, measuring IPV alone does not account for other violent experiences a mother may have encountered which also may impact breastfeeding.

Conclusions
Early initiation of breastfeeding and exclusive breastfeeding practices remains below the WHO targets for 2030 [4] in Malawi, Tanzania, and Zambia. Each type of IPV was negatively associated with optimal breastfeeding practices in at least one country and at least two forms of IPV were associated with a higher likelihood of delayed breastfeeding in each country. Therefore, for interventions to be effective in improving breastfeeding practices, the potential negative impact of IPV on breastfeeding outcomes must be considered. Policies to implement IPV screening during antenatal and postnatal care and referral of mothers for culturally sensitive assistance and care may improve breastfeeding practices. The importance of breastfeeding remains a constant focus for public health programs in Africa. Therefore, both raising awareness on the negative impact of IPV on maternal, infant, and child health, as well as supporting programs that reduce IPV are crucial.