Neonatal care and community-level treatment seeking for possible severe bacterial infection (PSBI) in Amhara, Ethiopia

Background: In Ethiopia, neonatal mortality accounts for approximately 54% of under-five deaths with the majority of these deaths driven by infections. Possible Severe Bacterial Infection (PSBI) in neonates is a syndromic diagnosis that non-clinical health care providers use to identify and treat newborns with signs of sepsis. In low- and middle-income countries, referral to a hospital may not be feasible due to transportation, distance or finances. Growing evidence suggests health extension workers (HEWs) can identify and manage PSBI at the community level when referral to a hospital is not possible. However, community-based PSBI care strategies have not been widely scaled-up. This study aims to understand general determinants of household-level care as well as household care seeking and decision-making strategies for neonatal PSBI symptoms. Methods: We conducted eleven focus group discussions (FGDs) to explore illness recognition and care seeking intentions from four rural kebeles in Amhara, Ethiopia. FGDs were conducted among mothers, fathers and households with recruitment stratified among households that have had a newborn with at least one symptom of PSBI (Symptomatic Group), and households that have had a newborn regardless of the child’s health status (Community Group). Data were thematically analyzed using MAXQDA software. Results: Mothers were described as primary caretakers of the newborn and were often appreciated for making decisions for treatment, even when the father was not present. Type of care accessed was often dependent on conceptualization of the illness as simple or complex. When symptoms were not relieved with clinical care, or treatments at facilities were perceived as ineffective, alternative methods were sought. Most participants identified the health center as a reliable facility. While designed to be the first point of access for primary care, health posts were not mentioned as locations where families seek clinical treatment. Conclusions: This study describes socio-contextual drivers for PSBI treatment at the community level. Future programming should consider the role community members have in planning interventions to increase demand for neonatal care at primary facilities. Encouragement of health post utilization could further allow for heightened accessibility-acceptability of a simplified PSBI regimen. Two woredas (districts) within a 300-km radius of Bahir Dar (capital city of Amhara) were selected for inclusion in this study. Both woredas comprise a majority of rural residents and were selected based on data from the 2007 Population and Housing Census of Ethiopia, Statistical Report for the Amhara region. Indicators such as population size, number of households and type of settlement (urban versus rural) were considered during site selection (Table 1). Number of households varied between woredas. Woreda A comprised 6405 households with 767 households in Woreda B. Four rural kebeles were purposively selected with the goal of representing various influencers for PSBI decision makers in rural communities. Characteristics of woreda sites Source: Central Statistical Agency – Ethiopia. The 2007 Population and Housing Census of Ethiopia: Statistical Report for Amhara Region. Addis Ababa, Ethiopia; 2012 aHouseholds refer to housing units, per the Central Statistical Agency – Ethiopia definition Qualitative research methods were employed to identify care seeking determinants for neonatal possible severe bacterial infection (PSBI) in rural Amhara. The methodology for this study was informed by existing studies analyzing determinants of health-seeking behavior (such as illness perception and characterization) in low- and middle-income contexts [20]. Data were collected through focus group discussions (FGDs). The discussion guide was initially prepared in English and translated into Amharic. FGDs were led by one moderator and one note taker, with observations documented by the principal investigator to document setting, behavior and contextual information for analysis. All discussions were conducted in the Amharic language and interviews were backtranslated into English. FGDs were stratified amongst mothers, fathers and household units. The purpose of collecting information from FGDs with mothers, fathers and household units was to understand household responsibilities and community-level decision making for neonates with signs of PSBI. Mothers in rural Ethiopia are typically primary caretakers of the child, although approval from fathers is often needed to follow-through on decisions requiring financial resources. Additionally, fathers are typically regarded as household decision-makers. For this reason, participant groups were separated by parental roles. Focus groups of household units, including all members of the household concurrently, were included to understand responses that may be influenced by power dynamics i.e. how maternal responses were given in the presence of the father. Household units typically comprised mothers, fathers and peripheral family members such as grandmothers and siblings of the newborn. Prior to accessing kebele sites, the research team (one moderator, one notetaker and the investigator) met with health center directors, health center supervisors and health extension workers (HEWs) at their respective locations to describe the study’s purpose. HEWs were approached and functioned as a recruiting mechanism, sampling participants from the catchment area of health centers. HEWs were asked to purposively identify households with newborns from health center records, utilizing the defined inclusion criteria, and targeted eight to ten participants per kebele. Discussions with the selected participants were scheduled over the phone or through approaching households by foot. Participants were selected if they had a newborn in the household within the previous 2 years, were 18 years of age or older and had newborns with PSBI symptoms (symptomatic group, SG) or were residents of the target communities (community group, CG). CG data was provided to compare care methods and trajectories among participants that were not recruited based upon a pre-specified ailment or condition. Participants identified in the SG were chosen if they had a newborn in the household exhibiting one or more PSBI symptoms in the first 28 days of life, as documented in health center records. PSBI symptoms were defined per the 2015 WHO guidelines and included fast breathing, chest in-drawing, fever, hypothermia, no movement or movement only upon stimulation, poor feeding or no feeding, and/or convulsions [10]. Community group (CG) participants were purposively selected regardless of the newborn’s health status based on their residence in the target communities. HEWs selected participants that were accessible or approachable via foot or vehicle for recruitment. The sampling frame included ten discussions per woreda (Table 2). Final recruitment consisted of 11 FGDs due to participant unavailability in some woredas. Sampling strategy, per woreda * Numbers are reported per the number of groups and range of participants in each group Data were collected between July and August 2018. The moderator and note-taker were both experienced in qualitative data collection. The moderator, fluent in Amharic and English, was additionally trained as a clinical nurse. FGDs typically lasted 45–75 min. Focus group guides included questions on household decision-making, care-taking actions and responsibilities, illness causation and characterization, illness severity, decision-making power and methods or facilities for care (Additional files 1, 2 and 3). Although introduction questions specifically addressed the newborn period, some questions were generalized to extend responses towards infancy. Discussions with mothers and fathers were conducted at health post compounds. Focus groups with household members (including the mother, father and peripheral family members such as brothers, sisters or grandparents of the newborn) were conducted in participant households. Demographic information was collected after discussions to gather information on age, number of people residing in the household, income, occupation, and education level. For mothers, information on the number of live births and number of children was collected to determine child loss. Precision of qualitative instruments were improved after reviewing preliminary data from the first two focus groups. Using pre-testing and in-field revision, the research team identified opportunities to utilize additional probes to improve richness of data. The research team additionally decided to focus on recruiting SG participants to ensure understanding of PSBI-related illness recognition, characterization and care seeking behaviors. As a result, after three CG focus groups, data collection focused on the SG to ensure saturation on the experiences of SG mothers, fathers and households (Fig. 1). Recruitment outcome As a result of this recruitment strategy (Table ​(Table2),2), 14 participants met CG recruitment criteria and 42 participants met SG recruitment criteria (Fig. ​(Fig.1).1). Four FGDs with mothers (29 participants), four FGDs with fathers (13 participants) and three FGDs with households (9 participants) were across the West and East Gojam zones. Of all FGDs, five were conducted in Woreda A and six in Woreda B. All discussions were recorded utilizing audio recorders. After data collection ended, data were transcribed into English by one bilingual Amharic-English transcriber with ample qualitative research and transcription experience. To assure quality of translation, three transcripts were re-transcribed by the same transcriber to affirm data quality. Discussions amongst the study team were continued throughout analysis to increase clarity when statements were made within the local context of communities. Data were thematically analyzed utilizing MAXQDA software (version 18.1.1, Berlin, GA, 2018). Thematic analysis is an iterative analytical approach to identifying concepts (themes) in the transcriptions [21]. Analytic memos were first created to contextualize emerging patterns and concepts. Memos largely addressed three questions: 1) How do different family members play a role in newborn care and care decisions? 2) Where did participants go to seek treatment for the newborn? and 3) How were these methods or facilities accessed? A codebook was then developed inductively using concepts that emerged while reading transcripts. Three transcripts were independently coded by the principal investigator and one research assistant experienced in qualitative research. During discussions of each transcript, agreement was sought to increase intercoder reliability. A final codebook was then created using input from two investigators. Emerging themes were further identified through iterative analytic memos, transcript summaries as well as analysis of key code intersections. Key codes were extrapolated then compared across sampling methods, sites and types of family members. Frameworks were created and revised throughout analysis to conceptualize study findings. Coding, memos and interim analyses were then discussed across the team—including the lead investigator, senior authors, faculty and data collectors. During the consent process, participants were informed of the study’s purpose, procedure and implications. Verbal consent was obtained as most respondents did not read or write in Amharic. The Emory University Institutional Review Board (IRB) determined data were exempt from full human subjects research review (July 3, 2018). Activities were approved by the Amhara Public Health Institute (APHI) ethical review board (July 18, 2018). Letters of support were provided to West and East Gojam zonal health departments prior to data collection.