Pregnant women and health workers’ perspectives on perinatal mental health and intimate partner violence in rural Ethiopia: a qualitative interview study

Background: Mental health conditions are common during the perinatal period and associated with maternal, foetal, and neonatal morbidity and mortality. There is an established bidirectional relationship between mental health conditions and intimate partner violence (IPV), including during and after pregnancy. Mean lifetime prevalence of physical, sexual or emotional IPV exposure among women in rural Ethiopia is estimated to be 61% and may be even higher during the perinatal period. We aimed to explore the perspectives of women and antenatal care (ANC) health workers on the relationship between all types of IPV and perinatal mental health, to inform the adaptation of a psychological intervention for pregnant women experiencing IPV in rural Ethiopia. Methods: We conducted in-depth qualitative interviews with 16 pregnant women and 12 health workers in the Gurage zone of the Southern Nations, Nationalities and People’s Region of Ethiopia, between December 2018 and December 2019. We conducted thematic analysis of English-translated transcripts of audio-recorded Amharic-language interviews. Results: Participants contextualised IPV as the primary form of abusive treatment women experienced, connected by multiple pathways to emotional and bodily distress. Patriarchal norms explained how the actions of neighbours, family, community leaders, law enforcement, and government agents in response to IPV often reinforced women’s experiences of abuse. This created a sense of powerlessness, exacerbated by the tension between high cultural expectations of reciprocal generosity and severe deprivation. Women and health workers advocated a psychological intervention to address women’s powerlessness over the range of difficulties they faced in their daily lives. Conclusions: Women and health workers in rural Ethiopia perceive multiple, interconnected pathways between IPV and perinatal emotional difficulties. Contrary to expectations of sensitivity, women and health workers were comfortable discussing the impact of IPV on perinatal mental health, and supported the need for brief mental health interventions integrated into ANC. SNNPR is one of thirteen regions and chartered cities in Ethiopia. The district of Sodo, where this study was conducted, has a projected population of 161,097 [19]. Ethiopian primary care, including ANC, comprises satellite ‘health posts’ staffed by health extension workers (HEWs), health centres staffed by family doctors, nurses, midwives, and health officers, and primary hospitals. We conducted qualitative in-depth interviews with pregnant women and ANC health workers, to explore their perspectives and experiences of IPV and its relationship to mental health. We adopted the interpretivist view that participants and researchers make subjective interpretations of the social world through their lived experience [20]. We considered our research an inductive process, grounded in the data, albeit influenced by our theoretical orientations. We recruited health workers and pregnant women from health centres and health posts delivering ANC in Sodo and neighbouring Butajira, to triangulate the perspectives of different stakeholders [21]. We recruited a purposive sample to capture the views of women of diverse age, religion, and educational levels, and of health workers of diverse age, religion, specialty, qualifications, and years of experience. The research assistant (RA) asked health workers to identify pregnant women who had reported symptoms of emotional distress, or whom they suspected could be experiencing IPV. Due to the absence of a formal IPV screening system, health workers identified potentially eligible women under their care based on knowledge of their personal circumstances. Potentially-eligible women were also identified from participants in an ANC survey who had reported depressive symptoms on the locally-validated patient health questionnaire [22], and relationship difficulties on a five-item screening questionnaire [23]. We anticipated conducting a total of between 20 and 30 interviews, based on the literature in this field. We determined the point of theoretical saturation through discussion between the first author, last author, and the RA conducting interviews, once the RA noted that new perspectives were not being raised by participants beyond those already captured. The first and last author agreed the point of theoretical saturation, informed by the breadth of perspectives described in existing literature [24]. Eligible participants were required to be female, aged 16 years or over, pregnant, conversant in Amharic, able to understand the information (discuss the study with the RA and respond to questions about their interest in taking part), and giving informed consent. We did not formally screen women for IPV exposure, to keep recruitment as inclusive as possible, and avoid excluding women experiencing but not disclosing IPV. Previous qualitative research in the same setting which did not purposively sample women experiencing IPV found that it was frequently described by participants [16–18]. We therefore considered any otherwise eligible pregnant woman to meet inclusion criteria, if her ANC health worker suspected that she was experiencing IPV, or if she had endorsed relationship difficulties on the aforementioned ANC survey. Health workers were required to work in ANC, be conversant in Amharic, and give informed consent. Women were excluded from participation if they were acutely unwell, requiring emergency treatment, or the RA noted difficulty in them understanding, remembering, considering the advantages and disadvantages, or communicating their views about the information provided. The RA provided eligible women and health workers with written and verbal information (uploaded to an open-access repository: Keynejad [25]) about the study, before inviting them to give written, informed consent. After reading the information sheet or having it read aloud (if unable to read), participants consented by signing or making a thumbprint authorised by a secondary school-educated, independent witness (if unable to write). The RA ensured that women understood the study by providing written and verbal information in simple language and checking their understanding. The RA was trained to identify verbal and non-verbal cues suggesting that a woman might not feel comfortable with or interested in participating, and to politely cease providing study information to such women. The information sheet explained that women’s responses would be anonymised but excerpts would be shared with others and used to develop an intervention. The voluntary nature of participation was emphasised throughout and could be withdrawn at any time before writing up. This study received ethical approval from King’s College London Psychiatry, Nursing and Midwifery Research Ethics Subcommittee (#HR-17/18–6063), and Addis Ababa University College of Health Sciences (#026/18/Psy). Interviews followed a standard operating procedure (SOP; see 26) addressing researcher conduct and contact with participants. The SOP provided guidance on how the RA should respond to distress, support participants experiencing IPV, and respond to any risks to children and adults identified during interviews. The SOP guided the RA to take precautions to prevent partners learning about a woman’s participation (for example, reading a leaflet she brought home), maintain confidentiality, seek consent to break confidentiality if required, and urgently discuss concerns with a nominated clinical contact person. The SOP included numbers for local services, to be provided if women felt safe to take written information home. Recruitment, consent, and interviewing were conducted by a female, master’s degree-qualified RA, in Amharic. We trained the RA using WHO clinical and policy guidelines on responding to IPV [26]. The RA received regular feedback on early interviews, on responding to verbal cues, and exploring unclear points. The RA kept field notes about interview interactions, which informed analysis. In-depth interviews were conducted in a private health centre room. Interviewers collected basic demographic information and followed topic guides, which were employed flexibly [25]. Topic guides addressed relationship problems and violence experienced by pregnant women, and perspectives on asking about mental health and IPV. Interviews also addressed participants’ perspectives on a perinatal mental health intervention, which are reported in a separate study [27]. Participants were reimbursed 100 Ethiopian birr (£2.68), which was considered appropriate locally [28]. All interviews were digitally audio-recorded, downloaded onto a computer, and password-protected, before being emailed to experienced transcribers. Transcribed, password-protected Amharic interviews were then emailed to translators, who emailed de-identified, translated interviews to the RA. Paper documents were stored in a locked cupboard. We followed the six phase approach to thematic analysis [29]. First author RK familiarised herself with the data by reading and re-reading transcripts alongside field notes, noting initial ideas. We then used NVIVO [30] to systematically code the dataset, generating initial codes and collating relevant quotations. Next, we reviewed initial codes, gathered them into themes, and reviewed themes in relation to constituent quotations. We refined themes, clarifying names and definitions through abstraction and interpretation. In the explanation phase, we referred back to our research questions while writing up. For triangulation through multiple analysis [21], the research team worked closely, for cross-cultural, collaborative ‘sense-making’ [31]. RK selected a transcript from a pregnant woman and one from a health worker, which were particularly rich in content. AM (interviewer), EF (trial coordinator), and TB (post-doctoral researcher) independently familiarised themselves with the data and generated initial codes [29]). RK reviewed these codes, to alter or elaborate initial codes. We remained reflexive of potential differences in the lived experiences of Ethiopian researchers and research participants, which might influence our interpretations. After engaging in initial explanation by sketching out an initial diagram displaying relationships between preliminary themes, RK shared her analysis with AM, EF, TB and CH (a UK researcher living permanently in Ethiopia), before online coding discussions, which informed iterative revisions. RK used reflexivity to identify personal biases, influenced by intersectional personal circumstances [32]. She used self-reflexivity of her position as an English-speaking, female researcher working in a high-income country where gender equality is advocated but not always achieved, to strive for ‘empathic neutrality’ in her interpretations [20]. We did not pursue respondent validation (‘member checking’) of our interpretations. Interviews explored sensitive subjects and there was potential for power imbalances between some participants. Respondent validation could enable more vocal respondents to challenge uncomfortable interpretations, which accurately represented lived realities [21]. We remained mindful that transcripts were English translations of conversations in Amharic. We recalled that meaning conveyed through spoken language incorporates cultural experiences and understandings which cannot always be translated [33].