Factors influencing rapid progress in child health in post-conflict Liberia: A mixed methods country case study on progress in child survival, 2000-2013

Objectives Only 12 countries in the WHO’s African region met Millennium Development Goal 4 (MDG 4) to reduce under-five mortality by two-thirds by 2015. Given the variability across the African region, a four-country mixed methods study was undertaken to examine barriers and facilitators of child survival prior to 2015. Liberia was selected for an in-depth case study due to its success in reducing under-five mortality by 73% and thus successfully meeting MDG 4. Liberia’s success was particularly notable given the civil war that ended in 2003. We examined some factors contributing to their reductions in under-five mortality. Design A case study mixed methods approach drawing on data from quantitative indicators, national documents and qualitative interviews was used to describe factors that enabled Liberia to rebuild their maternal, neonatal and child health (MNCH) programmes and reduce under-five mortality following the country’s civil war. Setting The interviews were conducted in Monrovia (Montserrado County) and the areas in and around Gbarnga, Liberia (Bong County, North Central region). Participants Key informant interviews were conducted with Ministry of Health officials, donor organisations, community-based organisations involved in MNCH and healthcare workers. Focus group discussions were conducted with women who have experience accessing MNCH services. Results Three prominent factors contributed to the reduction in under-five mortality: national prioritisation of MNCH after the civil war; implementation of integrated packages of services that expanded access to key interventions and promoted intersectoral collaborations; and use of outreach campaigns, community health workers and trained traditional midwives to expand access to care and improve referrals. Conclusions Although Liberia experiences continued challenges related to limited resources, Liberia’s effective strategies and rapid progress may provide insights for reducing under-five mortality in other post-conflict settings. Our case study used country-level indicator data for the years closest to 2000, 2005 and 2013, a review of national policies and strategies issued between 2007 and 2013, following the civil conflict, and key informant interviews and focus groups with community women conducted in 2013. Prior to 2000, country-level data for core MNCH indicators monitored by Countdown to 2015 were not reliably available from many African countries, including Liberia. Most of Liberia’s indicator data reported here were obtained from the World Bank Data Catalog,39 a repository of national, regional and global indicator data compiled from officially-recognised sources, including national Demographic and Health Surveys (DHS) and other national surveys. Data for indicators not readily available from the World Bank Data Catalog were obtained from the 2007 and 2013 Liberian DHS.40 41 We included indicator data most closely corresponding to the beginning and end of the study period to enable description of trends during the period. No DHS was conducted in Liberia between 1986 and 2007 due to the civil war, resulting in substantial missing data for the time period around 2000. We therefore also included 2007 DHS data to better visualise changes over time. Estimates were not always available for exact years 2000, 2007 and 2013, but we used data that were available within a 1 to 2 year window (see figure 2). Changes in child survival indicator coverage in Liberia, 2000, 2007 and 2013*. *Estimates were not always available for years 2000, 2007 and 2013, in which case the nearest estimate between 1999 and 2000, 2005 and 2007 or 2012 and 2013 was used; data were not available for the six indicators showing an asterisk (*) during the 2000 time period. †Among all births, both inside and outside a health facility. ‡Children 12–23 months old who have received BCG, measles and three doses each of DPT and polio vaccine (excluding polio vaccine given at birth). §Children under-five receiving oral rehydration and continued feeding. Source: World Development Indicators Data Catalog from the World Bank (http://datacatalog.worldbank.org; accessed August 2015) and Liberia Demographic and Health Surveys. ANC, antenatal care; ARI, acute respiratory infection; DPT, diphtheria, pertussis and tetanus. An information abstraction guide based on relevant global strategies related to child survival42–47 was developed to guide the document procurement and review process according to the following eight content areas: (1) healthcare system (including leadership, structure, human resources for health, access and utilisation, monitoring and evaluation and accountability), (2) national health strategies and policies (and regulations and laws, when applicable), (3) MNCH interventions, (4) clinical standards and guidelines, (5) commodities and essential medicines, (6) financial flows and resources, (7) effective partnerships and (8) other contextual factors (eg, conflict, political environment, sanitation and hygiene, nutrition and food security, education and human rights). When reviewing documents for information pertaining to the eight content areas, answers to the four overarching questions presented in table 1 were sought from each document. Overarching questions to explore the eight content areas as related to child survival during the review of national health policies and strategies, key informant interview and focus groups with community women MDG, Millenium Development Goal; MNCH, maternal, neonatal and child health. Policies and strategies pertaining to overall national health, MNCH and those from other sectors related to MNCH (eg, education, water and sanitation, and agriculture and nutrition) were obtained from the WHO African Region office, the WHO country focal points for Liberia and Liberia’s Ministry of Health (MOH; formerly Ministry of Health and Social Welfare). These documents were reviewed and any additional documents referenced and deemed important for the review (according to the abstraction guide) were obtained from WHO or MOH. The final list of reviewed documents can be found in online supplementary table S1. bmjopen-2018-021879supp001.pdf Each document was reviewed by two authors (MAB, CAH), and information was recorded and summarised according to the abstraction guide. To avoid biased interpretation of the information documented, the abstracted information was reported as it was stated in the original source, and efforts were made not to overstate or minimise the original information or add commentary not contained in the source. Because major differences in MNCH often exist between urban and rural areas, participants for the qualitative study were included from both urban and rural areas. The design of the parent study (consisting of four country case studies) used country DHS to compare region-specific under-five mortality rates and declines in mortality over the study period. Urban and rural sites for the qualitative study were to be selected from the region or county whose annual rate of reduction in under-five mortality most closely matched that of the nation as a whole. In the case of Liberia, the 1986 DHS only reported mortality for three counties, while the 2007 DHS reported mortality rates for Monrovia and six regions comprised of three counties each.41 48 As such, specific locations representative of Liberia’s progress as a nation could not be conclusively identified. Following discussions with the in-country primary investigator (SBK), Monrovia (Montserrado County) was selected as the urban location with focus groups conducted in the Paynesville and New Kru Town areas, and the areas in and around Gbarnga (Bong County, North Central region) were chosen as the rural location with focus groups conducted in Gbarnga and Totota. While we cannot ensure these locations experienced declines in under-five mortality similar to Liberia as a whole, the other three country case studies were also largely conducted in the capital (urban site) and a nearby rural region.35 36 38 Bong County was selected because it was reasonably accessible for conducting the study in a timely manner and was not markedly different from other areas of the country in terms of demographics and health infrastructure. Data were obtained from semistructured, key informant interviews with MOH officials (n=11 individuals interviewed), donor organisations (n=8), community-based organisations (CBO) involved in MNCH (n=14) and healthcare workers (HCWs) (n=14). Data were also obtained from four focus group discussions, two in Monrovia (n=16 total participants) and two in Bong County (n=21), with women who have experience accessing MNCH services. Interviews and focus groups were conducted between 30 October 2013 and 19 December 2013. All participants, whether key informants or focus group women, were eligible for the study if they met the following criteria: (1) being 18 years of age or older, (2) having adequate knowledge or experiences related to childhood survival specified for each participant group below, (3) speaking English or Liberian English and (4) being able to provide written informed consent. Specific inclusion criteria for each key informant group included the following: national or provincial-level officials working in government-level healthcare system administration, policy-making, programme development, leadership or any aspect of MNCH (MOH officials); directors, managers or other leaders of entities providing financial or other aid for MNCH services, or international or national organisations focusing on MNCH or having MNCH as one component of their mission (Donor organisations); directors, leaders or managers working for a CBO involved in or providing referrals to MNCH services; and professionally trained physicians, nurses, clinical officers or other health-related staff working in a health facility providing MNCH care (healthcare professionals). Similar numbers of participants from each key informant group were enrolled, and a range of ages, work experiences and positions/roles within each group was sought using department registers when available. Additionally, efforts were made to balance the number of urban and rural participants among the HCWs and CBO workers. Lists of potential key informants from each group were developed by the in-country research team with assistance, as needed, from the WHO National Professional Officer for Family Health and the MOH Deputy Programme Manager for the Expanded Programme on Immunisations. A letter signed by an official from the MOH was sent to each potential key informant participant informing them of the purpose of the study, risks and benefits of participation, and describing the interview process. These were followed up with a phone call or email to those interested. The final number of key informant interviews conducted was arrived at through a combination of approaches. Due to study logistics, we set a minimum number of 6 interviews to be conducted with both MOH and donor organisation representatives and a minimum of 12 interviews (half urban, half rural) to be conducted with both HCWs and CBO representatives. In an effort to achieve saturation, we prioritised diversity in the types of key informants we reached (online supplementary table S2). The in-country PI and research assistants monitored data collection and saturation. Women were recruited to participate in focus groups using informational flyers or advertisements posted in different health centres and surrounding communities. As with the key informants, a balance was sought in the level of education and participants with live and deceased children, as well as a diversity of experiences and opinions regarding access and utilisation of MNCH services. The number of focus groups was determined at the outset of the study and constrained by study logistics. Written informed consent was obtained from all enrolled participants. Community women (online supplementary table S3) were provided a small monetary compensation for their participation. Interview guides for key informants and discussion guides for focus groups with community women were developed, pilot tested through cognitive interviewing49 and revised as needed. The guides focused on experiences with MNCH services and barriers to and facilitators for improving child survival (table 1), pertaining to the eight content areas evaluated during the review of national health policies and strategies. Not all content areas were appropriate for each key informant group, but each topic was asked of at least two of the four groups. The content areas and overarching questions were developed to provide structure across the four country case studies of the parent study. However, they were intentionally broad to provide sufficient flexibility for participants within and across countries to discuss the issues most relevant to them. Focus group discussions with community women focused only on the healthcare system, MNCH interventions, medicines and contextual factors content areas. While participants could discuss the entire period from 2000 forward, most participants recalled more recent information and experiences. Prior to conducting interviews and focus group discussions, participants completed a brief survey to obtain basic demographic information, MNCH-related work experience (key informants only), socioeconomic information (focus group women only) and/or information on births and under-five deaths in the household (focus group women only). Key informant interviews were conducted in English by one research assistant using the appropriate interview guide and were audio-recorded. Key informants were encouraged to provide their perspectives openly and discuss a range of barriers and facilitators to child survival. Interviews typically lasted 60–90 min. The focus group discussions were conducted in Liberian English and were audio-recorded. Two Liberian research assistants (one male and one female) were present at each focus group to facilitate discussion and note-taking. Focus group participants were encouraged to provide their opinions openly, and research assistants were trained in techniques to promote open discussion. Focus groups typically lasted between 1½ and 2 hours. The researchers on this study included individuals with knowledge and experience of MNCH at the national and international levels and who had prior experience with health research in Liberia. Key research team members had prior experience with qualitative and quantitative research methods and research ethics. An in-person methods training was held to ensure high-quality data across sites. Ongoing remote training and troubleshooting was provided to the research team during the piloting and data collection stages of the study. To promote reflexivity, preliminary results were discussed at a workshop held after data collection and preliminary analysis was completed. Following completion of the interviews and focus groups, audio recordings were transcribed by the research assistants and field notes incorporated into the transcript. Transcripts were coded and analysed using the software Atlas.ti (Atlas.ti Scientific Software Development, Berlin, Germany).50 In keeping with a framework approach often used for qualitative, multidisciplinary health research,51–53 deductive themes were determined a priori based on our conceptual framework of overarching questions. Additional inductive themes were also identified on review of the transcripts. Deductive codes provided a useful way of comparing themes and concepts within and across countries. Text was coded and reviewed for patterns of consistency, variation, relationships between themes and exemplary cases or quotations.54 55 Patients were not involved in the design of this study. Results were disseminated to MOH and WHO representatives from Liberia, and a presentation and report detailing results were made available to these representatives to aid further dissemination to other stakeholders.