Menu
Menu
Menu
Menu
Background: Maternal and neonatal mortality remain high in southern Tanzania despite an increasing number of births occurring in health facilities. In search for reasons for the persistently high mortality rates, we explored illness recognition, decision-making and care-seeking for cases of maternal and neonatal illness and death. Methods: We conducted 48 in-depth interviews (16 participants who experienced maternal illnesses, 16 mothers whose newborns experienced illness, eight mothers whose newborns died, and eight family members of a household with a maternal death), and five focus group discussions with community leaders in two districts of Mtwara region. Thematic analysis was used for interpretation of findings. Results: Our data indicated relatively timely illness recognition and decision-making for maternal complications. In contrast, families reported difficulties interpreting newborn illnesses. Decisions on care-seeking involved both the mother and her partner or other family members. Delays in care-seeking were therefore also reported in absence of the husband, or at night. Primary-level facilities were first consulted. Most respondents had to consult more than one facility and described difficulties accessing and receiving appropriate care. Definitive treatment for maternal and newborn complications was largely only available in hospitals. Conclusions: Delays in reaching a facility that can provide appropriate care is influenced by multiple referrals from one facility to another. Referral and care-seeking advice should include direct care-seeking at hospitals in case of severe complications and primary facilities should facilitate prompt referral.
The study was conducted in two districts in southern Tanzania (Tandahimba and Newala districts) of Mtwara region in 2015, where the EQUIP study was previously implemented [19, 20]. Most inhabitants of these rural areas belong to the Makonde ethnic group, who primarily engage in subsistence farming. Although most people speak the language of their ethnic group, Kiswahili is widely spoken [21].The uptake of facility births here is high at over 80%, although the neonatal mortality rate is still above 30 per 1000 live births, which is higher than the national average; the national maternal mortality ratio is 556 deaths per 100,000 live births [4, 5]. As we were interested in recruiting participants who had experienced maternal or newborn morbidity and mortality, or family members of a woman who died after giving birth, different methods were employed to recruit respondents. These included using community health workers to help identify cases that were reported to them, snowball sampling methods once participants had been recruited [22], and referrals by community leaders. From these suggestions, we sampled 48 participants purposively [23, 24] from across eight different wards—a sub-district administrative structure—within the two study districts, with differing levels of uptake of facility care. The sample size was chosen to give a sufficient number of interviews to explore different contextual factors that may influence illness recognition and care-seeking in relation to both maternal and newborn illnesses and deaths [18]. Participants in FGDs were village leaders who were purposively selected by the field supervisor based on the inclusion of their wards in the study area. We conducted 48 group in-depth interviews (IDIs) with family members (16 families who had experienced maternal illness, eight families whose family member had died after giving birth, 16 families whose newborn had experienced illness, and eight families whose newborn had died) and five focus group discussions (FGDs) with community leaders [20]. Semi-structured interview guides were designed and pre-tested. All interviews were conducted in Swahili. A debriefing template was provided to the team by the field supervisor and filled in the same day. At the end of each day of fieldwork, the team held a debriefing session and sent a report to the study coordinator, who provided feedback to the team in the field within 12 h during the first week of data collection and within 48 h for the rest of the study period to improve probing and adaptation of interview guides as required. Please see Additional files 1, 2, 3, 4 and 5 for interview and FGD guides. All interviews were audio recorded and transcribed verbatim. After completion of data collection, all debriefing notes and IDI and FGD transcripts were imported into NVivo 9 [25] and analyzed by senior project staff with the data collection team. Analysis began with multiple readings of all data. A codebook was developed and used as a point of comparison and discussion between the team carrying out analysis until a high level of inter-rater agreement about codes could be reached. Thematic analysis was used to group codes together into higher-order sub-themes and overall themes, which we related to the three delays. These themes were reached through group consensus. Representative quotations were selected for each theme and are presented in the results section that follows. Given the sensitive nature of this data collection, a thorough informed consent process was undertaken with each participant, reiterating to them their right to conclude the interview at any point without consequence. Interviewers paid close attention to any distress that participants were under and were instructed to pause or end the interviews as necessary. Written informed consent was obtained from all participants prior to the start of the IDIs and FGD. All transcripts had identifying information removed, and the confidentiality of all participants was protected from the outset and throughout. Audio files and transcripts were stored in password-protected files on secured computers in locked offices, to which only team members had access. The study received ethics approvals from the internal review board of Ifakara Health Institute and the National Institute for Medical Research. Permission was also secured from the local authorities in the study area prior to making ethics applications and beginning any data collection.
