Design

This was a prospective cohort study design. Data were collected from a population-based sample, which was part of a larger research project: “The Impact of Violence on Reproductive Health in Tanzania and Vietnam”.

Sample (participants and settings)

All pregnant women, who attended antenatal care in Pasua and Majengo Health Clinics in Moshi, Kilimanjaro Region of Tanzania, were asked to participate.

Inclusion criteria were gestational age (GA) <30 weeks, determined by ultrasound.

Exclusion criteria were multiple pregnancies, plans to deliver outside of Moshi Municipality, pregnancies not resulting in live births and children or mothers who died during the study period.

Measures

Data for this study was collected via five structured interviews. The first interview was at inclusion. The second interview was at 34 weeks GA. The third took place immediately after birth, the fourth interview was at 40 days postpartum and the final at 2–3 years postpartum. All data were double-entered.

Sociodemographic and reproductive data (age, parity, marital status and education level) were collected at the first interview. At the second interview, the women were asked about exposure to IPV, smoking and alcohol use and tested for HIV. The sex of the child and whether it was born preterm (GA <37 weeks) was registered at the third interview. At the fourth interview we asked for signs of postpartum depression using Edinburgh Postnatal Depression Scale (EPDS) with a score of >13 as a cut-off for probable depression. The tool has been validated in Africa, but not in Tanzania.[27] We collected data on the duration of breastfeeding at the final interview.

Recruitment was carried out between March 1^stand December 31^st, 2014. All interviews were done in private settings by experienced research assistants. Questionnaires and translations were pilot tested. Pilots were not included in the cohort.

Intimate partner violence.

The exposure to IPV was assessed at the second interview using a Kiswahili version of the questionnaire used in the WHO Multi-Country Study on Women and Domestic Violence against Women.[28] The questionnaire has three dimensions, pertaining to different types of violence; emotional, physical and sexual (Fig 1). Participants were asked about exposure to IPV both at any given time point in the relationship and specifically during pregnancy. Women who were exposed to ongoing physical or sexual IPV were referred to a separate counselor connected to the project, who had the possibility of referring to specific support services if the participant gave consent.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Definition of the different types of intimate partner violence.

https://doi.org/10.1371/journal.pone.0217479.g001

IPV was categorized in 5 different groups, who where not mutually exclusive: “At least one type of IPV”, “Emotional IPV”, “Physical IPV”, “Sexual IPV” and “All three types of IPV”.

The reference group was comprised of the women who had no exposure to IPV during the current relationship. All analyses were made twice–Once for women exposed at any point in the relationship and once for women exposed specifically during pregnancy.

Statistical analyses

The distribution of cohort characteristics between exposed and un-exposed groups were compared using Pearson’s chi-square test. Fisher’s Exact test was used in the case of smoking, since less than 5 women smoked. Participants with missing values were excluded from the analysis.

Logistic regression was performed to estimate odds ratio (OR) with 95% confidence interval between IPV and premature termination of EB. We first estimated the crude OR and then adjusted for potential confounders, which were selected based on a priori assessment suggested by a directed acyclic graph (DAG),[29] including maternal age, education, HIV-status, alcohol use during pregnancy and parity(S1 Fig). As shown in the supporting information, we assume postpartum depression to be part of the pathway to reduced breastfeeding, and as such it is not adjusted for in the regression model. Instead it is addressed in the stratified analysis.

In order to determine if any groups were more vulnerable to the effects of IPV than others, we conducted a stratified analysis comparing women who were exposed to at least one kind of IPV at any point in their relationship to those who were not. We chose the variables maternal age, education, EPDS Score, HIV status and alcohol consumption as well as child sex for the stratification. The selection was made a priori based on clinical experience.

Statistical analyses were performed using STATA software package (version 15).

Ethics

Ethical approval was granted by Research Ethics Review Committee of Kilimanjaro Christian Medical University College of the Tumaini University, Makumira, Tanzania.

Written permission was granted by Moshi Municipality Executive Director for carrying out research at Majengo and Pasua Health Clinics.

Oral consent was collected from all participants and documented by the research assistants carrying out the interviews.

Main findings

To the best of our knowledge, this is the first longitudinal study to explore the association between IPV and EB in Africa and the first study to provide the 6 months’ time perspective. It provides strong evidence on the suspected adverse effects of IPV suggested by other studies.[13, 15, 18–21, 23, 25] Women who reported exposure to at least one type of IPV at any point in the relationship had more than 50% higher odds of terminating EB before the child was 6 months old. When exposed to all three types of IPV, women had twice the odds of premature termination of EB when exposed at any point in their relationship. When exposed to all three types of IPV during the index pregnancy the odds were tripled. As the odds increase with the severity of IPV, a dose-response relationship could be present. However, the current tools available to assess IPV do not allow for an appropriate analysis to ascertain that.

The results of the stratified analysis demonstrated that associations between IPV and EB are complex with several modifying factors. Knowing which subgroups of mothers are particularly vulnerable, enables health workers to target interventions in settings with limited resources.

While maternal education, alcohol use during pregnancy and signs of postpartum depression all influenced the outcome, the most prominent modifications were seen with mothers who are more than 30 years old and who gave birth to girls. This interesting finding may stem from several reasons. Boys might be better at breastfeeding, mothers might perceive them to be either more capable or more in need of exclusive breastfeeding or there might be societal factors deeming girls less desired by a violent partner, hence giving birth to a girl might escalate violence in the family, increasing the adverse effect.[32] That the odds of early weaning increase with maternal age, suggests more resilience in younger mothers. Health professionals should be aware of older mothers and mothers of girls and watch out for signs of abuse.

Strengths and limitations

The main strengths of this study were the longitudinal design and high follow-up rate. With more than 90% of participants completing all interviews, the risk of attrition bias is negligible, and we believe selection bias has been minimized, since all women who met the wide inclusion criteria, agreed to participate.

However, there are some limitations to the study. Firstly, inquiring about IPV might inspire the exposed mothers to react to the violence and adopt protective measures, thereby limiting the exposure causing an underestimation of the adverse effect. While IPV is a complex exposure to measure, our findings on IPV prevalence are in line with what we expected from previous studies.[33, 34]

Secondly, the WHO tool we used to assess IPV does not allow for a reliable assessment of continued IPV. Therefore, we do not know if the violence was ongoing for the participants at follow-up.

Thirdly, there is no failsafe way to measure EB. The two most common approaches are either the 24-hour recall method or the “since birth” method.[35, 36] Both come with the risk of over- and underreporting. We believe the “since birth” method provides the most reliable results. However, the time elapsed before the follow-up interview introduces an increased risk of information bias. The method of distant recall is not optimal. For practical and financial reasons, a close follow-up was not possible in this study.

There is a need to develop a validated tool to assess breastfeeding practices to increase the quality of future studies on the subject.

Interpretation

Our findings are consistent with the results from most studies on IPV and breastfeeding despite the different outcome measures used. Other studies examine either initiation, intention to breastfeed or EB in the past 24 hours or since birth, but not after the age of 6 months.[13, 15, 18–21, 23, 25] Our findings are in keeping with the patterns observed in previous research on breastfeeding outcomes and IPV. It also provides a significant addition to the current literature by virtue of the longitudinal 6 months perspective.

Our findings provide evidence for the deficit hypothesis,[37] which proposes that women exposed to IPV are less inclined to or less capable of breastfeeding their children. There is qualitative evidence that sexually abused women could attach negative sexual associations to their breasts and concurrently be less willing to breastfeed.[38] Women suffering from IPV are also likely to engage in negative coping behavior[39] and might be physically, psychologically and cognitively impaired.[40] They are also more likely to suffer from postpartum depression compared to their unexposed peers.[41–43]

The mechanism of breastfeeding contains both physiological and emotional elements which can be disrupted under stress. Animal studies have demonstrated how stress lowers lactogenesis by decreasing oxytocin levels.[11] Literature on the human stress response and lactation is sparse; however experimental studies suggest the same mechanisms might apply.[10, 12]

Although, the observed association between IPV and breastfeeding might reflect a causal link between the two factors, it is also possible that both IPV and premature termination of breastfeeding are independent proxy indicators for a dysfunctional family pattern. Besides the physiological stress response, we know self-efficacy and social support are paramount to breastfeeding and are likely to be impaired by IPV.[44] As described in another paper, mothers exposed to IPV in our cohort had significantly less social support than mothers without exposure.[45]

The only study with conflicting findings was Misch et al, who found IPV to have a positive effect on breastfeeding in Tanzanian and Zambian women, as mentioned in the introduction.[25] They proposed the effect could be due to compensatory behavior, where the mother used breastfeeding as a coping mechanism to deal with violence. This compensatory hypothesis was proposed by Levendosky et al. in 2003 in a study on the maternal relationship with pre-school children in families with domestic violence. The study did not address breastfeeding specifically.[46] The adverse effects of IPV stress the need for interventions to aid the mothers exposed to IPV. Violence is a convoluted cultural phenomenon making the exposure difficult to tackle. While a diversity of adverse effects of IPV by now are well documented, an effective intervention has yet to be developed, possibly because most attempts are aimed at victims instead of perpetrators.[47] Despite the lack of effective interventions several studies recommend screening for IPV during pregnancy.[16, 47] The salient reason for this is that awareness is the first step to leave a violent relationship. Furthermore, pregnant women are in a situation where they are uniquely motivated to seek help.[47]

Further research to find effective interventions against IPV, preferable aimed at the perpetrators, is vital. Right now, however, there is an acute need for assistance for women trapped in violent relationships. Safe exit possibilities are needed to ensure their continued ability to provide optimal care for their children.