Menu
Menu
Menu
Menu
Background Globally, under-five mortality has declined significantly, but still remains a critical public health problem in sub-Saharan African countries such as Benin. Yet, there is no empirical information in the country using a nationally representative data to explain this phenomenon. The aim of this study was to examine how proximate and socio-economic factors are associated with mortality in under-five children in Benin. Methods We analysed data of 5977 under-five children using the 2017 to 2018 Benin Demographic and Health Surveys. Multivariable hierarchical logistic regression modelling technique was applied to investigate the factors associated with under-five mortality. The fit of the models were assessed using variance inflation factor and Pseudo R 2. Results were reported as adjusted odds ratios (aORs). All comparisons were considered to be statistically significant at p2 years of birth interval (aOR=1.52; 95% CI: 1.07 to 2.17). Among the proximate determinants, we found the probability of death to be higher in children whose mothers had no postnatal check-up (PNC) visits after delivery (aOR=1.79; 95% CI: 1.22 to 2.63), but there was no significant association between individual-level/household-level factors and under-five mortality. Conclusion This study has established that socio-economic and proximate factors are important determinants of under-five mortality in Benin. Our findings have shown the need to implement both socio-economic and proximate interventions, particularly those related to PNC visits when planning on under-five mortality. To achieve this, a comprehensive, long-term public health interventions, which consider the disparity in the access and utilisation of healthcare services in Benin are key.
Data from the 2017 to 2018 Benin Demographic and Health Surveys (BDHS) was used in this study. Specifically, the birth recode file, which contains data of all births, was used. Data of 5977 under-five children, which formed the unit of analysis in this study, were obtained by interviewing women who had given birth within 5 years to the survey. The BDHS used a multistage, stratified sampling design in selecting all eligible women and men for interviews from households that were considered as sampling units.23 This study adapted Mosley and Chen’s conceptual framework of child survival in developing countries (figure 1)22 as its conceptual framework. The framework helped in selecting variables available in the 2017 to 2018 BDHS datasets for the analyses. The adapted constructs of the conceptual framework are shown in figure 1. Conceptual framework of determinants of under-five mortality Source.22 PNC, postnatal check-up. The outcome variable was under-five mortality, defined as the death of a child within the first 5 years of life. We re-coded it into a binary variable as (0=No and 1=Yes). The explanatory variables considered in this study include community-level and household-level/individual-level socio-economic variables and proximate determinants variables. The community-level socio-economic variables include region and place of residence. The household-level and individual-level socio-economic variables were wealth index, mother’s ethnicity, mother’s religion, maternal education and occupation, and partner’s education and occupation. The proximate determinants include sex of child, birth size, birth rank and birth interval, mother’s age at childbirth, ANC visit, tobacco use, place of delivery, type of assistance during delivery and postnatal check-up (PNC) visits. The various categories for these determinants can be found in table 1. U5MR (per 1000 live births) and uOR by explanatory variables (n=5977, weighted) *p<0.05; **p<0.01; and ***p<0.001. U5MR, under-five mortality rate; uOR, unadjusted OR. Descriptive and multiple regression analyses were performed in this study. The first step of the analyses involved the use of frequency tabulations to describe the proportions of all the explanatory variables, followed by a distribution of under-five mortality per the explanatory variables, with their respective CIs. Then, we conducted a bivariate logistic regression analysis with each of the explanatory variables and the outcome variable (under-five mortality) to assess the link between all the potential determinants and deaths in under-five children without adjusting for the effect of other covariates. This was followed by a multicollinearity test on all the explanatory variables to determine if there was evidence of multicollinearity. Using the variance inflation factor (VIF), the multicollinearity test results indicated no collinearity among the explanatory variables (mean VIF=1.49, max VIF=1.66 and minimum=1.01). Next, we performed a multivariable hierarchical logistic regression analysis in three stages. First, community-level socio-economic determinants variables were fitted in the first model to assess their association with under-five mortality (Model I). This was followed by the inclusion of household-level/individual-level socio-economic variables (Model II). Proximate determinants variables were added in the final model to also examine their association with deaths in under-five children (Model III). Categories of the explanatory variables that had the lowest under-five mortality rates were used as reference categories. The goodness-of-fit of the logistic models was assessed using Pseudo R2. Data processing and analysis were performed using Stata, V.14.2 (Stata Corp, College Station, Texas, USA). We further applied sample weight (v005/1 000 000) to correct for oversampling and undersampling. The SVY command in Stata was used to account for the complex survey design and generalisability of the findings. We further used the ‘syncmrates’ command to calculate the under-five mortality rates using the synthetic cohort probability.15 Patients and the public were not involved in the design and conduct of this research.
N/A
