Health facility delivery in sub-Saharan Africa: Successes, challenges, and implications for the 2030 development agenda

Background: Sub-Saharan Africa remains one of the regions with modest health outcomes; and evidenced by high maternal mortality ratios and under-5 mortality rates. There are complications that occur during and following pregnancy and childbirth that can contribute to maternal deaths; most of which are preventable or treatable. Evidence shows that early and regular attendance of antenatal care and delivery in a health facility under the supervision of trained personnel is associated with improved maternal health outcomes. The aim of this study is to assess changes in and determinants of health facility delivery using nationally representative surveys in sub-Saharan Africa. This study also seeks to present renewed evidence on the determinants of health facility delivery within the context of the Agenda for Sustainable Development to generate evidence-based decision making and enable deployment of targeted interventions to improve health facility delivery and maternal and child health outcomes. Methods: We used pooled data from 58 Demographic and Health Surveys (DHS) conducted between 1990 and 2015 in 29 sub-Saharan African countries. This yielded a total of 1.1 million births occurring in the 5 years preceding the surveys. Descriptive statistics were used to describe the counts and proportions of women who delivered by place of delivery and their background characteristics at the time of delivery. We used multilevel logistic regression model to estimate the magnitude of association in the form of odds ratios between place of delivery and the predictors. Results: Results show that births among women in the richest wealth quintile were 68% more likely to occur in health facilities than births among women in the lowest wealth quintile. Women with at least primary education were twice more likely to give birth in facilities than women with no formal education. Births from more recent surveys conducted since 2010 were 85% more likely to occur in facilities than births reported in earliest (1990s) surveys. Overall, the proportion of births occurring in facilities was 2% higher than would be expected; and varies by country and sub-Saharan African region. Conclusions: Proven interventions to increase health facility delivery should focus on addressing inequities associated with maternal education, women empowerment, increased access to health facilities as well as narrowing the gap between the rural and the urban areas. We further discuss these results within the agenda of leaving no one behind by 2030. We use data from Demographic and Health Surveys (DHS) conducted between 1990 and 2015 in 29 sub-Saharan African countries. The surveys are grouped into two: “earliest” surveys conducted since 1990 (before the onset of the MDG agenda) and “latest” or most recent surveys conducted since 2010 but before 2015, close to the MDG deadline of 2015. By implication, countries which had only one DHS during this period were not included in the analysis. The time interval between the earliest and most recent DHS data provides sufficient time to observe reasonable changes in health facility delivery between the period before the MDG agenda and the period close to the MDG deadline. A total of 24 surveys (from 12 countries) come from Western Africa; 8 surveys (from 4 countries) come from Middle Africa; 22 surveys (from 11 countries) come from Eastern Africa; and 4 surveys (from 2 countries) come from Southern Africa (Table 1). Time intervals between the earliest and latest surveys ranged between 5 and 23 years, averaging 15 years of observation. The pooled DHS data include 396,837 births from earliest surveys and 762,445 from latest surveys; yielding a total of 1.1 million births occurring in the 5 years preceding the surveys. The pooled data set was based on birth history files where each woman was asked for the date of birth (month and year) of each live-born child, the child’s sex, whether the child was still alive (and if the child had died) the age at death (in days for the first month, in months if the deaths occurred between 1 and 24 months, and in years thereafter). These data allowed child deaths to be located by time and by age. Countries and Demographic and Health Surveys included in the analysis for 29 sub-Saharan African countries Notes: aObservation time calculated based on the upper bound of the year. For example, the 2010–2011 year uses 2011 as the end point. Latest surveys defined as those from 2010 with the exception of Madagascar (2008–09) Source: [22] We performed statistical analysis using Stata version 14 [6]. We used descriptive statistics to describe the counts and proportions of women who delivered by place of delivery and their background characteristics at the time of delivery. The reference event for all analyses were most recent birth during the 5 years preceding the surveys. We consider the following predictors of place of delivery: wealth status ranking based on wealth quintiles; residence (urban/rural); mother’s characteristics (education, having at least one antenatal care (ANC) visit, age of mother at birth); community women’s education (none or at least primary education); birth order of child; and a dummy indicator for the survey round (earliest/latest). Place of delivery was coded as ‘1’ for children who were born in a health facility and ‘0’ for children who were delivered elsewhere (Table 2). The percent of missing data for the variables concerned ranges from 1.2 to 4.9% and these were excluded from the analyses. Variables used in the analysis of predictors of place of delivery among women with most recent births for 29 sub-Saharan African countries Note: “Ref.” – Reference category We used multilevel logistic regression model to estimate the magnitude of association in the form of odds ratios (ORs) between place of delivery and the predictors. In particular, multilevel models were constructed using the mixed effects modelling procedure where data have been collected in nested units. Sampling cluster was included in the model as nested random effects with country modelled as fixed effects. For the purposes of the analysis, we fit unadjusted regression models for each explanatory variable and then fit two additional models: Model 0 (empty model) excludes independent variables in order to decompose the total variance into its cluster and country components. Model 1 is the full model which includes all independent variables. The three-level multi-level model to estimate the cluster and country effects is written as follows, eq. (1): where πij is the probability that the ith woman of jth cluster in the kth country will deliver in a health facility; Xij is a set of variables for each ith woman of the jth cluster in the kth country. These covariates may be defined at the individual, community, or country level; β0 is the associated vector of standard regression parameter estimates; u0jk represents the random effect at the cluster level; and v0k is the random effect at the country level. The intercept or average probability of a woman delivering in a health facility is assumed to vary randomly across clusters and countries. Based on this approach, the fixed effects (measures of association) are presented as odds ratios (OR) alongside 95% confidence intervals (CI). We tested the goodness of fit of the multilevel model using the log likelihood ratio (LR) test. This approach led to estimation of unadjusted and adjusted ORs of the likelihood of health facility delivery. Independent variables were included if they were statistically significantly associated with the outcome variable with a cut-off p-value of < 0.05 in the multilevel logistic regression model. The adjusted ORs were an outcome of the multilevel logistic regression (i.e. eq. (1)) in estimating the net contribution of each covariate to the outcome variable, adjusting for other covariates in the model. An OR of 1 implied no difference whereas an OR > 1 implied the woman was more likely to deliver in a health facility; and an OR < 1 implied less likelihood of a woman delivering in a health facility. All statistical tests were set at 5% level of significance with associated 95% confidence intervals. The adjusted ORs of place of delivery from the multilevel logistic regression model for each country were used to conduct meta-analysis in Stata to develop a forest plot of the adjusted pooled effect (i.e. women who delivered the most recent child in a health facility compared to women who delivered elsewhere) across 29 countries. The pooled effect focuses on health facility delivery during latest survey rounds compared with earliest survey round. The pooled ORs with associated 95% confidence intervals (CI) were estimated using Mantel-Haenszel statistical methods. Heterogeneity among the surveys was assessed using I2 statistics, a measure of the proportion of total variability explained by heterogeneity rather than chance expressed as a percentage [7]. Roughly, an I2 of 0–40% represents no or little heterogeneity, 30–60% moderate heterogeneity, 50–90% substantial heterogeneity, and 75–100% considerable heterogeneity [8]. The meta-analysis applied random effects analytical model due to the considerable heterogeneity (> 75%) among the survey results. Observed likelihood of delivering in a health facility were compared with expected likelihood of health facility delivery which were obtained after adjusting for the risk factors in the regression model. Independent variables were subjected to multi-collinearity tests by performing correlations, variable inflation factor (VIF) and tolerance tests. The mean VIF was 1.43 whereas tolerance values were at least 0.5 [9]. The VIF between several variables that potentially had multicollinearity such as mother’s education, community women’s education, and wealth quintile were also at least 0.5; and these tests indicated no cause for concern for collinearity. We applied sample weights for descriptive analyses using the Stata svy command to account for undercounting and over counting due to the sample design of the survey [6].