Barriers in the delivery of emergency obstetric and neonatal care in post-conflict Africa: Qualitative case studies of Burundi and Northern Uganda

Objectives Maternal and neonatal mortality and morbidity rates are particularly grim in conflict, postconflict and other crisis settings, a situation partly blamed on non-availability and/or poor quality of emergency obstetric and neonatal care (EmONC) services. The aim of this study was to explore the barriers to effective delivery of EmONC services in post-conflict Burundi and Northern Uganda, in order to provide policy makers and other relevant stakeholders context-relevant data on improving the delivery of these lifesaving services. Methods This was a qualitative comparative case study that used 42 face-to-face semi-structured indepth interviews and 4 focus group discussions for data collection. Participants were 32 local health providers and 37 staff of NGOs working in the area of maternal health. Data was analysed using the framework approach. Results The availability, quality and distribution of EmONC services were major challenges across the sites. The barriers in the delivery of quality EmONC services were categorised into two major themes; human resources-related challenges, and systemic and institutional failures. While some of the barriers were similar, others were unique to specific sites. The common barriers included shortage of qualified staff; lack of essential installations, supplies and medications; increasing workload, burn-out and turnover; and poor data collection and monitoring systems. Barriers unique to Northern Uganda were demoralised personnel and lack of recognition; poor referral system; inefficient drug supply system; staff absenteeism in rural areas; and poor coordination among key personnel. In Burundi, weak curriculum; poor harmonisation and coordination of training; and inefficient allocation of resources were the unique challenges. To improve the situation across the sites, efforts are ongoing to improve the training and recruitment of more staff; harmonise and strengthen the curriculum and training; increase the number of EmONC facilities; and improve staff supervision, monitoring and support. Conclusions Post-conflict health systems face different challenges in the delivery of EmONC services and as such require context-specific interventions to improve the delivery of these services. Ethics approval for the study was obtained from the Regional Committee for Medical and Health Research Ethics, South-East (Norway); le Comité National d’Ethique pour la Protection des êtres Humains Participant à la Recherche Biomédicale et Comportementale (Burundi); and Gulu University Institutional Review Committee (Uganda). We also received permission from local administrative and health authorities. All participants provided a written informed consent before participating in the study and their anonymity, privacy and confidentiality was respected. The study was conducted from June to September 2013 in three provinces in Burundi and a district in Northern Uganda. The highest administrative unit in Burundi is the province; with each province having a number of communes. On the other hand, Uganda is divided into four administrative regions; Central, Western, Eastern, and Northern, with the regions in turn divided into districts. In terms of size and population, a district in Uganda is similar to a province in Burundi. That is why we choose the second level administrative unit for our study site in Northern Uganda (district) and a first level administrative unit for our study site of Burundi (province). In Burundi the study was undertaken in the provinces of Bujumbura-Mairie, Bujumbura-Rural and Ngozi while in Northern Uganda our study site was the district of Gulu. The Gulu district is made up of 3 counties, 16 sub-counties, 70 parishes and 279 villages, with a population of 374,700 [34]. The 2008 census in Burundi [35] puts the population of the three provinces of Bujumbura-Mairie, Bujumbura-Rural and Ngozi at 497,166, 555,933 and 660,717 respectively. Study participants were recruited from among staff of non-governmental organizations (NGOs) and local health providers (LHPs) and only those knowledgeable of or experienced with EmONC-related activities were included in the study. These included frontline healthcare providers at health facilities; senior health administrators and decision makers; organisations involved in the provision of EmONC training, donation, and supply of essential EmONC medicines, equipment and other supplies; and organisations providing other forms of EmONC-related technical and material support within our study areas. The NGOs included local, national and international organizations working in the domain of maternal health, be it at the level of policy support or technical assistance, health system support and strengthening, or delivery of health services. We classified the NGOs into three main groups: NGO-Health providers (NGOs that also provide health services), NGO-Policy makers (mainly UN-based NGOs) and NGOs (non-UN-based NGOs that do not provide health services). The LHPs were drawn from clinics, health centres and hospitals, and included nurses, midwives and doctors working on maternal health issues in their institutions, mainly at the maternity, antenatal care, and obstetric and gynecological units in both public and private facilities. Others included senior administrators at ministries of health at the provincial, regional or district levels (LHP-Policy makers). This is a qualitative case study that used face-to-face semi-structured in-depth interviews (IDIs) and focus group discussions (FGDs) for data collection. Interviews and FGDs were conducted in the local language, French or English (where applicable) by the principal investigator (PCC) or trained local research assistants (RAs). All interviews and FGDs were guided by detailed ‘Interview and FGD guides’ that were developed in both the English and the French languages and piloted prior to the commencement of study. The complete ‘Interview and FGD guides’ have been reported elsewhere [36]. Interviews and FGDs with NGO staff and local health providers were held mainly at their places of work, and the lawn of some local hotels. All interviews in French and the local languages were undertaken by the trained local RAs while all the English interviews were undertaken by the principal investigator (PCC). Interviews and FGDs typically lasted from 50–130 minutes. The FGDs included between 5–8 participants. Interviews and FGDs were audio-recorded and field notes taken. Soft drinks, tea or coffee was provided to FGD participants during the discussion. We also provided transport reimbursement to FGD participants. A total of 42 IDIs and 4 FGDs were conducted across the study sites as shown on Table 1. This involved a total of 69 participants; 30 males and 39 females. In terms of their participant category, 32 were LHPs and 37 NGO staff. In each of the settings, RAs were recruited locally. A total of 6 RAs were recruited and trained across the study sites; 4 in Burundi and 2 in Northern Uganda. All RAs understood the local language(s) plus English and/or French and were educated up to the university level. In Burundi interviews and FGDs were mainly held in the French or Kirundi languages, while in Northern Uganda they were held in English. The RAs were also involved in the transcription of the audio files and translation of the transcripts into English (where applicable). All interviews and FGDs were transcribed and translated into English (where applicable). The English transcripts were then imported into the QRS Nvivo software (QSR International, 2012). The data was analysed using the framework method [37]. Three team members (including PCC and PB) open-coded the transcripts on Nvivo and Microsoft® Word (where the texts of interest are highlighted and the code first labelled using the ‘New Comment’ sub-menu under the ‘Review’ menu). Microsoft® Word was used for coding and analysis by one of the co-authors who did not have access to Nvivo. The codes were mainly descriptions and labels of specific concepts as the transcripts were read. Two team members reviewed the codes that were developed and the inter-coder reliability was high. Related codes were then collapsed into different categories, and the categories were subsequently grouped into specific themes. The themes were inductively and deductively developed, involving predetermined themes included in the interview and FGD guides and explicitly covered during the data collection and review, as well as those that emerged during the data review. There was therefore a constant interplay between data collection, analysis and theme development, with dominant themes that emerged in earlier interviews and FGDs being explored deeper in subsequent and later interviews and discussions. The theme development was jointly undertaken by three team members.