Use of and barriers to maternal health services in southeast Chad: Results of a population-based survey 2019

Objectives Chad reports the second highest maternal mortality worldwide. We conducted a survey in Sila region in southeast Chad to estimate the use of maternal health services (MHS) and to identify barriers to access MHS. Design Retrospective cross-sectional, population-based survey using two-stage cluster sampling methodology. The survey consisted of two strata, Koukou Angarana and Goz Beida district in Sila region. We conducted systematic random sampling proportional to population size to select settlements in each strata in the first sampling stage; and in the second stage we selected households in the settlements using random walk procedure. We calculated survey-design-weighted proportions with 95% CIs. We performed univariate analysis and multivariable logistic regression to identify impact factors associated with the use of MHS. Setting We interviewed women in selected households in Sila region in 2019. Participants Women at reproductive age, who have given birth in the previous 2 years and are living in Koukou Angarana and Goz Beida district. Primary outcomes Use of and access barriers to MHS including antenatal care (ANC), delivery care in a health facility (DC), postnatal care (PNC) and contraceptive methods. Results In total, 624 women participated. Median age was 28 years, 95.4% were illiterate and 95.7% married. Use of ANC, DC and PNC was reported by 57.6% (95% CI: 49.3% to 65.5%), 22.5% (95% CI: 15.7% to 31.1%) and 32.9% (95% CI: 25.8% to 40.9%), respectively. Use of MHS was lower in rural compared with urban settings. Having attended ANC increased the odds of using DC by 4.3 (1.5-12.2) and using PNC by 6.4 (3.7-11.1). Factors related to transport and to culture and belief were the most frequently stated access barriers to MHS. Conclusion In Sila region, use of MHS is low and does not meet WHO-defined standards regarding maternal health. Among all services, use of ANC was better than for other MHS. ANC usage is positively associated with the use of further life-saving MHS including DC and could be used as an entry point to the community. To increase use of MHS, interventions should include infrastructural improvements as well as community-based approaches to overcome access barriers related to culture and belief. In early 2019 (21 January 2019 to 07 February 2019) MSF in collaboration with the Chadian Ministry of Health performed a cross-sectional, retrospective, population-based survey using two-stage cluster sampling methodology in Koukou Angarana and Goz Beida district, located in the Chadian Sila region. The objectives of the survey were to estimate the use of MHS and to identify barriers to access MHS. The study’s two main research questions were: We defined following primary outcomes to answer the research questions: Koukou Angarana and Goz Beida are two districts in Sila region. In an exploratory field visit by MSF to Sila region in late 2018, the two districts were identified as the areas in Sila region with the biggest need for intervention and therefore chosen as the study area. According to population data from the district Ministry of Public Health in 2018, more than 80% of the population in Sila region lives in Koukou Angarana and Goz Beida district. The study population consisted of women at reproductive age, living in Koukou Angarana and Goz Beida district and who have given birth in the last 2 years. According to the 2017 Chadian national census projections, which are based on the Chadian general population census conducted in 2009, the proportion of women at reproductive age was estimated to be 21.8% of the population in Sila Region.18 Population figures for the settlement-level were obtained from the estimates produced by the district Ministry of Public Health (MPH) representative. These figures were derived from estimates produced during a bed net distribution and seasonal malaria prophylaxis campaign which took place in 2017. For the first stage of sampling, we selected settlements using probability proportional-to-size (PPS) sampling. For the PPS, a list of settlements with population figures at settlement-level was obtained from estimates produced by the district MPH representative. According to the provided estimates there were 330 settlements in Koukou Angarana district and 620 in Goz Beida district. In the second stage of sampling, we selected households in the settlements using random walk procedure as an adaptation of the WHO guidelines for cluster vaccination coverage surveys.19 We based the sample size calculation on the estimated prevalence of utilisation of ANC of 58.5% in Sila region9 aiming for a precision of 7.5%, assuming a design effect of 1.5 and non-response of 10%. For calculation of the sample size we used ENA software for SMART.20 According to this calculation a total of 315 women in each district (Goz Beida and Koukou Angarana) needed to be included in the sample. We selected 45 clusters in Goz Beida district and 45 clusters in Koukou Angarana district consisting of seven households each. We included one woman per household. Women were eligible for inclusion if they satisfied all of the following criteria: member of the selected households, being at least 15 years old, had given birth (live or stillbirth), living in Goz Beida or Koukou Angarana district in the previous 2 years and provided informed consent for herself and by the head of the household. If multiple women were eligible for inclusion in one household, we selected one woman per household at random. The survey questionnaire on maternal health consisted of 69 closed questions including demographic data and data regarding ANC, DC, PNC and CM as primary outcomes. The questionnaire was based on the main indicators identified as relevant and actionable by MSF. Additionally, within the questionnaire we used same questions as the latest national survey on demographics and health (DHS14/15) and as comparable studies within the region.16 21 22 The questionnaires were translated from French to Chadian Arabic and back translated to French to ensure consistency. Demographic data were collected on age, marital status, level of literacy, pregnancy status and live and stillbirths in the past 2 years and nutrition status of women being pregnant or lactating at time of the study. For each type of service used (ANC, DC, PNC), we collected data on type of attendant (skilled or traditional) and payment of services. Additionally, service-specific data were collected. For ANC, we collected the number of ANC visits per pregnancy and if ANC visits included blood pressure measure, tetanus and malaria prophylaxis. For DC, we collected additional data about mode of transport and time to a health facility. Regarding CM, we collected data on knowledge about CM, use of CM and preferred method among those who had never used any CM. For any service not used, we asked about barriers to access care. For barriers to access DC, multiple answers were possible. We used the following definitions: Women of reproductive age: Women aged 15–49 years were defined as women of reproductive age. Literacy: Being able to read a phrase written on a piece of paper. Nutrition status in pregnant and lactating women: Severe acute malnutrition: mid-upper arm circumference (MUAC) of <185 mm; moderate acute malnutrition: MUAC 185–<230 mm. MHS: Includes ANC, DC and PNC. Use of ANC: At least one ANC visit during pregnancy. Complete ANC: At least one blood pressure check, receipt of malaria prophylaxis and tetanus vaccination during pregnancy. Use of DC: Assistance during birth by a skilled birth attendant in a health facility. Use of PNC: At least one visit within 42 days of birth. Urban and rural: Categorisation into urban or rural setting was based on administrative delineations. We defined and categorised reported barriers to access care retrospectively. We categorising access barriers into five groups according to previously used groups in low-income countries23 : (i) barriers related to transport, including distance to healthcare facility and security issues; (ii) barriers related to culture, and belief, including belief of not being sick enough to use service, not being the custom to use service, using medications from the market, preference for traditional practitioner, preference of birth at home, lack of trust in the services and not aware of the potential benefit of seeking care; (iii) barriers related to economic factors, including not having money for trip or service fee; (iv) barriers related to family support, including having no time or not time yet, no child care, no one to accompany or no permission; and (v) barriers related to quality of care including no staff at healthcare facility. Answers of not knowing a reason for not using MHS or not understanding the question were categorised as ‘others’. The categorisation is available in the online supplemental table 1. bmjopen-2021-048829supp001.pdf We conducted a 3-day training on ethics, survey techniques, sampling method, questionnaires, data collection followed by a 1-day pilot for surveyors including debriefing session afterwards. According to results of debriefing minor changes to the questionnaire were made: changes to answer options, correction of translation errors and improvement of skip patterns of the digital questionnaire. Data collection took place within the households by interviewing eligible and consenting women. Surveyors worked in teams of two and collected the data for the study with KoboCollect electronic survey software on tablets. We did not collect personal identifiable data. All data were stored password-protected by MSF Holland. All data will be archived at the MSF Headquarters in Amsterdam for a duration of 5 years after the survey. Primary outcomes of the study were use of and access barriers to ANC, DC, PNC and CM. In order to calculate correct population level summary statistics, we weighted all analysis for the population of Goz Beida and Koukou Angarana districts and additionally for the number of women in the household. Furthermore, all analyses were conducted accounting for the effect of clustering induced by the two-stage sampling design.24 Most of the variables were categorical variables grouped according to the answer possibilities given in the questionnaire. Continuous variables were age, number of live and stillbirths within the last 2 years and number of ANC visits. Those were grouped in the following way: age (<29 years; ≥29 years), number of ANC visits (<4 visits; ≥4 ANC visits), number of live births within last 2 years (no live birth within the past 2 years; 1 live birth within the past 2 years, >1 live births within the past 2 years), stillbirths within the last 2 years (no stillbirth within the past 2 years; >1 stillbirth within the past 2 years). All categorical variables were reported as proportions with 95% CIs. For continuous variables, such as age, we determined the mean, median, range and IQR. As part of the descriptive analysis we compared urban versus rural settings. For comparing categorical variables in the two groups we used χ2 tests with the Rao-Scott adjustment for the complex survey design.24 Results are presented as proportions, 95% CIs and p values. We conducted three multivariable analyses using logistic regression to understand factors associated with use of (1) ANC, (2) DC and (3) PNC. For all models we included as independent variables potential influencing factors such as maternal age (<29 years vs >29 years; age group split at median age), level of literacy (illiterate vs literate), setting of living (rural vs urban) and history of live births (0, 1, >1 live birth in the past 2 years) and stillbirths (no stillbirths, >1 stillbirth within the past 2 years). For the model regarding DC in a health facility we additionally included ‘having ever used ANC’ and for the model regarding PNC we additionally included ‘having ever used ANC’ and ‘DC in a health facility’. Results are presented as adjusted ORs, 95% CIs and p values. All statistical analyses were performed using the complex survey design commands (svy) in the Stata V.15 statistical package.25 Study consent was collected at different levels. The surveyors informed all village heads about the study in Chadian Arabic and asked for written consent before they proceeded to speak to households in the village. Within selected households, the surveyors informed the women about the study objectives and asked for written consent. If the head of the household was different from the woman, his or her written consent was also requested. No patient involved.