Health services for women, children and adolescents in conflict affected settings: Experience from North and South Kivu, Democratic Republic of Congo

Background: Insecurity has characterized the Eastern regions of the Democratic Republic of Congo for decades. Providing health services to sustain women’s and children’s health during protracted conflict is challenging. This mixed-methods case study aimed to describe how reproductive, maternal, newborn, child, adolescent health and nutrition (RMNCAH+N) services have been offered in North and South Kivu since 2000 and how successful they were. Methods: We conducted a case study using a desk review of publicly available literature, secondary analysis of survey and health information system data, and primary qualitative interviews. The qualitative component provides insights on factors shaping RMNCAH+N design and implementation. We conducted 49 interviews with government officials, humanitarian agency staff and facility-based healthcare providers, and focus group discussions with community health workers in four health zones (Minova, Walungu, Ruanguba, Mweso). We applied framework analysis to investigate key themes across informants. The quantitative component used secondary data from nationwide surveys and the national health facility information system to estimate coverage of RMNCAH+N interventions at provincial and sub-provincial level. The association between insecurity on service provision was examined with random effects generalized least square models using health facility data from South Kivu. Results: Coverage of selected preventive RMNCAH+N interventions seems high in North and South Kivu, often higher than the national level. Health facility data show a small negative association of insecurity and preventive service coverage within provinces. However, health outcomes are poorer in conflict-affected territories than in stable ones. The main challenges to service provisions identified by study respondents are the availability and retention of skilled personnel, the lack of basic materials and equipment as well as the insufficient financial resources to ensure health workers’ regular payment, medicaments’ availability and facilities’ running costs. Insecurity exacerbates pre-existing challenges, but do not seem to represent the main barrier to service provision in North and South Kivu. Conclusions: Provision of preventive schedulable RMNCAH+N services has continued during intermittent conflict in North and South Kivu. The prolonged effort by non-governmental organizations and UN agencies to respond to humanitarian needs was likely key in maintaining intervention coverage despite conflict. Health actors and communities appear to have adapted to changing levels and nature of insecurity and developed strategies to ensure preventive services are provided and accessed. However, emergency non-schedulable RMNCAH+N interventions do not appear to be readily accessible. Achieving the Sustainable Development Goals will require increased access to life-saving interventions, especially for newborn and pregnant women.

We used a case study design, modified from a standardized protocol agreed upon by the BRANCH consortium and adapted for each country context. In DRC, we conducted a mixed-methods case study combining a desk review of publicly available peer-reviewed and grey literature, primary qualitative and secondary quantitative data. The qualitative component aimed to understand how contextual factors shape RMNCAH+N decision making and service provision. The quantitative component focused on estimating coverage of RMNCAH+N interventions at provincial and sub-provincial level, and according to exposure to conflict. The case study focused on both North and South Kivu provinces in DRC, which have seen the start of the two Congolese wars and have undergone extensive violence since then. North Kivu has experienced higher intensity violence than South Kivu, both in terms of casualties and events. It was the location of 40% of all violent events occurring in DRC between 2012 and 2017 and suffered more casualties than all other DRC provinces until 2016 when violence erupted in Kasai. Few fatalities have occurred in South Kivu since 2012, despite numerous violent events throughout the years. Violence against civilians (33.2%) and battles with no change of territory (31.2%) were the most frequent forms of violent episodes in both provinces [9]. Variation in conflict-related events and deaths at the sub-provincial level is significant. In North Kivu, the territories of Rutshuru and Beni were the most affected in absolute numbers of casualties; Walikale becomes the most affected territory when considering population size, followed by Beni. In South Kivu, Uvira and Kalehe were the two most affected territories in absolute terms, and Uvira and Walungu in relative terms (see Additional file 1 for more details). (Note: from an administrative point of view, the DRC is organized in 26 provinces and 192 territories. The Health System follows a different structure: provinces are divided in health zones, and these in health areas.) The qualitative field work occurred in four health zones that were chosen for their history of conflict and insecurity (as per active armed clashes and population displacement) during the last 5 years together with accessibility. Selection was conducted with representatives of the Provincial Health Offices due to their knowledge of the provinces. In North Kivu, the research team visited the health zones of Mweso (Masisi territory) and Ruanguba (Rutshuru territory). Mweso experienced extensive violence that led to population displacement and attacks on health facilities, while Ruanguba was at the center of the March 23 Movement (M23) offensive in 2012–2013 [10]. In South Kivu, Minova (Kalehe territory) and Walungu (Walungu territory) were visited (Fig. 2); both were characterized by extensive conflict over land issues and customary power [11]. Map of North and South Kivu with qualitative case study sites (Mweso and Ruanguba in North Kivu; Minova and Walungu in South Kivu). (Adapted from [1]) The investigators conducted a desk review of relevant grey and peer reviewed literature. Iterative searches were conducted in both Pubmed and Embase databases by combining key and MESH terms related to RMNCAH service delivery and North and South Kivu. We limited the search to the period 2000 to 2017. We did full text review of 71 results and of those 23 were relevant to our study. We also searched the grey literature for reports about health service provision in North and South Kivu using humanitarian websites such as Reliefweb, Humanitarian Response, Global Health and Nutrition Cluster. An additional 60 reports were reviewed that provided additional contextual information about the health system in DRC. Qualitative data originated from individual or group interviews with representatives of private and public health service providers currently working in North and/or South Kivu (see details in Table 1). Fifty-one in-depth interviews (IDI) and four focus group discussions (FGDs), with a total of 84 respondents were conducted (Table 2). Additional information on qualitative and quantitative data used in the case study Informants for qualitative data Individual and group in-depth interviews were conducted with: i) Ministry of Health officials from the Provincial Office in charge of RMNCAH+N programs, as well as the Chief Medical Officers responsible for the selected health zones; ii) Staff of United Nations agencies and of national and international non-governmental organizations including senior program managers, technical leads and other positions responsible for RMNCAH+N program planning, implementation and coordination; and. iii) Healthcare providers including clinicians in charge of RMNCAH+N services, chief nurses and community health workers. Participation was voluntary. Oral informed consent was obtained from all participants. Participants needed to be 18 years of age or older and working in the position for more than 30 days. Quantitative data sources Following data sources were used: i) The Armed Conflict Location and Event Data (ACLED) [12] provided data on type and number of conflict events, number of fatalities, location and date. This is a reliable conflict data source and one of the leading data sets in conflict epidemiology [12–14]. ii) The 2001 and 2010 Multiple Indicator Cluster Surveys and 2007 and 2013–2014 Demographic and Health Surveys (DHS) reports [6, 15, 16] provided data on key interventions coverage indicators in the RMNCH+N continuum of care at national and provincial levels. Coverage is expressed as the proportion of people who benefited from a certain service among the target population. iii) National Health Facility Information System as available in the District Health Information Software 2 (DHIS2). In South Kivu, health facility data existed for the period 2012–2017; in North Kivu only for the period 2015–2017. Facility data analyses were restricted to selected RMNCH indicators for which data availability allowed for comparisons and trends assessment (i.e. first visit of Antenatal Care (ANC1); fourth visit of antenatal care (ANC4), third dose diphtheria -pertussis -tetanus vaccine (DPT3), measles immunization and Caesarian section rate for South Kivu, and ANC1, ANC4, DPT1, DPT3, assisted deliveries, caesarean section, maternal mortality for North Kivu). Health facility data were merged with conflict events at territory level using Microsoft Excel to allow the assessment of the effects of conflict on RMNCAH indicators. iv) The 2017–2018 Service Provision Assessment report [14] provided information about the proportion of health facilities providing RMNCAH+N services. v) Population estimates from the provincial health divisions: the estimates are irregularly updated based on health-related activities such as distribution of insecticide-treated bed nets and were used to estimate intervention coverage based on the health facility data. Participants to in-depth interviews and Focus Group Discussions in the qualitative study component, North and South Kivu, DRC Notes: aDPS Division Provinciale de Santé (Provincial Health District), MCZ Médecin Chef de Zone (Chief Medical Officer of the health zone) In each health zone, one hospital, one health center and one health post were visited. The chief nurse was interviewed in each facility. Two referral hospitals (one per province) were included. In each of them, we interviewed the head of the maternity department and the chief midwife. One FGD per health zone was conducted with community health workers (CHWs). Relevant United Nations (UN) agencies, non-governmental organizations (NGOs) as well as national faith based organizations implementing RMNCAH+N programs were invited to participate to the study. We aimed to include five NGOs per province to ensure a variety of opinions was captured. A four-day training was conducted in Bukavu to secure a common understanding of the study objectives and data collection tools. One interview guide for each targeted group was developed and tested in a health center in Bukavu that was not included in the sample. The interview guides for FGD and facility health workers were translated into Swahili. Data collection took place over 8 weeks from August to September 2018 and was typically completed over 3 days per health zone. Interviews in Bukavu and Goma were scheduled between field visits according to the availability of participants. Multiple sources of secondary quantitative data were used to investigate coverage of RMNCAH+N interventions between 2000 and 2017. All were publicly available or were requested from the Ministry of Health (MoH). These included the 2001 and 2010 Multiple Indicator Cluster Surveys (MICS), 2007 and 2013–2014 Demographic and Health Surveys (DHS) [6, 15, 16] as well as routine health facility data available via the online District Health Information System (DHIS2) (details in Table 1). The Armed Conflict Location and Event Data (ACLED) project provided data on conflicts events and fatalities [9, 12]. We defined conflict intensity by territory as the average number of conflict-related fatalities divided by the average annual population between 2012 and 2017. As no standard threshold exists, we considered the distribution of deaths and classified territories with a conflict death rate over 20 per 100,000 persons per year as the most insecure (see Additional file 1). Using this definition on South Kivu, we classified as stable the territories of Idwji and Bukavu; as intermediate the territory of Kalehe1; and as conflict-affected the remaining territories of Shabunda, Fizi, Kabare, Mwenga, Uvira and Walungu. Of the seven territories of North Kivu, two (Beni and Walikale) were classified as insecure and the remaining five as stable. Key indicators in the RMNCAH continuum of care were extracted from MICS and DHS reports and used for trends analysis at national and provincial level. They included the percentage of all women, currently married women and sexually active unmarried women aged 15–49 currently using any modern family planning method; skilled birth attendance; antenatal care (ANC); three doses of diphtheria, pertussis, and tetanus vaccine (DPT); measles vaccination; percentage of children under age 5 with symptoms of acute respiratory infection for or whom advice or treatment was sought from a health facility or provide; and oral rehydration therapy for children with diarrhoea. Stunting and wasting prevalence data were also extracted from MICS and DHS reports. At health zone level, RMNCAH indicators with sufficient data were derived from the DHIS2, namely first visit of Antenatal Care (ANC1); fourth visit of antenatal care (ANC4), third dose of diphtheria -pertussis -tetanus vaccine (DPT3), measles immunization and Caesarian section rate for South Kivu, and ANC1, ANC4, DPT1, DPT3, assisted deliveries, caesarean section, maternal mortality for North Kivu. As denominators, we used estimated populations from the health zones available from DHIS, given that the last DRC population census dates back to 1984. For interventions programming, the DRC ministry of health update the populations estimates by applying a population growth rate of 3%. Occasionally, population estimates are based on local censuses conducted prior to mosquito nets distribution campaigns. In South Kivu, for instance, the last population count was performed in 2014. Qualitative data were collected in French or Swahili depending on the preference of the respondent. Audio recordings in both languages were transcribed in French. Data were managed and coded in NVivo [17]. A combined approach to codebook development was used, with predefined codes addressing the specific issues the study aimed to explore, completed by additional codes arising from unexpected participants’ experience. Two team members coded the transcripts after having tested and compared coding approaches to ensure harmonization. Framework analysis was used to explore qualitative data. A matrix output (with cases as row and codes as column) was developed to systematically summarize data and facilitate constant comparison within and across cases and topics [18]. With regard to quantitative data, we aggregated health facility data at territory level and merged them with the ACLED data, with Microsoft Excel, to enable analysis by conflict location. In fact, from an administrative point of view, the DRC is organized in 26 provinces and 192 territories, the latter being the lowest administrative unit in the ACLED. The Health System follows a different structure: provinces are divided in health zones (33 health zones in North Kivu, 34 in South Kivu, and 516 nationwide). DHIS2 data from South Kivu were checked for completeness and errors, as they included raw numbers aggregated at the health zone level. It was therefore possible to asses reporting completeness as well as denominators used in calculations; this was not feasible for DHIS2 data from North Kivu as only proportions as estimated by the MoH in North Kivu were available. The reporting completeness for South Kivu ranged between 75% (in 2013) and 96% in 2017. Since we had no good empirical basis for adjustments for incomplete report, we assumed that all non-reporting facilities provided zero services (see Additional file 3 for more details). The quantitative analysis comprised three steps: first, descriptive trends analysis was carried out to compare two coverage indicators (ANC1 and DPT3), one service provision indicator (caesarean section rate) and two institutional mortality indicators (maternal mortality and stillbirths), in conflict-affected versus more stable territories. The choice of these indicators was dependent on data availability in the DHIS2. Second, the Composite Coverage Index (CCI) [19] was calculated at the national and provincial level using DHS and MICS data for both North and South Kivu. The CCI is a weighted average coverage of eight preventive and curative interventions, namely oral rehydration therapy, acute respiratory infection, family planning, skilled birth attendance, antenatal care, measles vaccination and diphtheria, pertussis, and tetanus vaccine. The CCI gives equal weight to four stages of the RMNCAH continuum of care: family planning, maternal and newborn care, immunization, and case management of sick children, and has been shown to be a robust measure of RMNCH continuum of care with a strong association with health outcomes [20]. Third, the effect of insecurity on service provision in South Kivu was estimated by random-effects generalized least square regression modelling, with year as random effect to account for the temporal variability of the conflict intensity and numbers of violent episodes. Regression analysis was run on health facility (DHIS2) data aggregated at the territory level and linked with conflict events. We used Stata 15 [21] for this analysis and set the significance to p < 0.05. The paucity of data from North Kivu precluded the regression analysis for this province.

• altare, chiara
• malembaka, espoir bwenge
• tosha, maphie
• hook, christopher
• ba, hamady
• bikoro, stéphane muzindusi
• scognamiglio, thea
• tappis, hannah
• pfaffmann, jérôme
• bisimwa, ghislain
• boerma, ties
• spiegel, paul