Background: Acute Respiratory Infections (ARIs) account for more than 6% of the worldwide disease burden in children under the age of five, with the majority occurring in Sub-Saharan Africa. Rural children are more vulnerable to and disproportionately affected by ARIs. As a result, we examined the rural–urban disparity in the prevalence of ARI symptoms and associated factors among children under the age of five in Sub-Saharan Africa. Methods: We used the most recent Demographic and Health Survey (DHS) data from 36 countries in Sub-Saharan Africa. The study included 199,130 weighted samples in total. To identify variables associated with ARIs symptoms, a multilevel binary logistic regression model was fitted. The Adjusted Odds Ratio (AOR) with a 95% CI was used to determine the statistical significance and strength of the association. To explain the rural–urban disparity in ARI prevalence, a logit-based multivariate decomposition analysis was used. Results: Being female, ever breastfeeding, belonging to a poorer, better wealth status, and having better maternal educational status were significantly associated with lower odds of ARIs among under-five children. Whereas, small size or large size at birth, not taking vitamin A supplementation, being severely underweight, having diarrhea, didn’t have media exposure, never had the vaccination, being aged 36–47 months, and being aged 48–59 months were significantly associated with higher odds of ARIs among under-five children. The multivariate decomposition analysis revealed that the difference in characteristics (endowment) across residences explained 64.7% of the overall rural–urban difference in the prevalence of ARIs, while the difference in the effect of characteristics (change in coefficient) explained 35.3%. Conclusion: This study found that rural children were highly affected by ARIs in SSA. To reduce the excess ARIs in rural children, public health interventions aimed at impoverished households, home births, and unvaccinated and malnourished children are crucial.
This study used the Demographic and Health Surveys (DHSs) data of 36 sub-Saharan African countries conducted from 2005 to 2019, which was conducted using nationally representative samples to estimate core demographic and health indicators of the whole country. To recruit the samples, a multistage stratified cluster sampling technique was used, with Enumeration Areas (EAs) serving as primary sampling units and households serving as secondary sampling units [34]. The Kids Record dataset (KR file) was used for this study after we obtained an authorization letter from the measure DHS program for data access. The outcome variable was ARI symptoms among under-five children. The presence of ARIs is defined as children having a history of cough within two weeks accompanied by short, rapid breathing or difficulty of breathing and fever within two weeks preceding the survey. In DHS, mothers of under-five children were asked whether their children had a history of cough within two weeks preceding the survey. For children who had a cough, the mother was asked whether the child’s cough was accompanied by short, rapid breathing or difficulty of breathing and fever within two weeks preceding the survey. It was obtained from the question “did he/she breathe faster than usual with short, rapid breaths or have difficulty breathing in the 2 weeks preceding the survey?”. Then categorized as “Yes” if a child meets all the above-mentioned criteria and “No” otherwise [35]. The independent variables were categorized into child characteristics, mother characteristics, household characteristics, and contextual factors. Child characteristics include child age, sex of the child, breastfeeding, vitamin A supplementation, diarrhea in the last two weeks, ever had vaccinated, type of birth, child size at birth, and child nutritional status (stunting, wasting, and underweight); mothers-related characteristics include maternal age, media exposure, and maternal education, and household characteristics and contextual characteristics include household wealth status, residence, and country. To assess a child’s nutritional status DHS used anthropometric measures (height, age, and weight), height for age measures stunting, weight for height measured wasting, and weight for age measured for underweight. Stunting is defined as the height for age z-score less than 2 standard deviations below the median of the reference population, wasting is defined as the weight for height z-score less than 2 standard deviations below the median of the reference population, and underweight is defined as weight for age z-score less than 2 standard deviations below the median of reference population [36]. To adjust for the non-response and sampling design, the data were weighted using the primary sampling unit, strata, and weighting variable. STATA version 16 statistical software was used for analysis. Since the DHS data has a hierarchical nature, under-five children within the same cluster might share similar characteristics to children from different clusters. This could violate the assumptions of the traditional logistic regression model; these are the independence of observations and equal variance assumptions. Therefore, a multilevel binary logistic regression model was fitted to identify factors associated with ARIs using EAs as a random variable. The presence of the clustering effect was assessed using the Intra-class Correlation Coefficient (ICC) and Likelihood Ratio (LR) test. ICC quantifies the degree of heterogeneity of ARIs between clusters (the proportion of the total observed individual-level variation in ARIs that is attributable to cluster variations) Six models were fitted and model comparison was made using deviance as the models were nested models. Null model (empty model), Model I (residence), Model II (Model I + child characteristics), Model III (Model II + mothers characteristics), Model IV (Model III + household characteristics), Model V (Model IV + country-level characteristics (sub-Saharan African region)) were fitted and a model with lowest deviance value was chosen as the best-fitted model for the data. We identified the independent variables based on previous literature conducted on determinants of ARIs. Variables with a p-value less than 0.2 in the bi-variable analysis were considered for multivariable analysis. The Adjusted Odds Ratio (AOR) with a 95% Confidence Interval (CI) and p-value < 0.05 in the multivariable model were used to declare significant determinants of ARIs. Logit-based Multivariate Decomposition analysis was used to identify factors that contributed to the rural–urban difference in ARIs. The analysis was based on the logit link function which uses the output from the binary logistic regression model by dividing the difference in ARIs among under-five children into components. The overall rural–urban difference in the prevalence of ARIs among under-five children can be explained by the difference in composition between residences (i.e., differences in characteristics or endowment) and/or the difference in effects of the explanatory variables across residences (i.e., differences in coefficients). The mvdcmp STATA command was used to generate the overall and detailed multivariate decomposition analysis results [37]. Variables with a p-value < 0.2 in the bi-variable Logit-based multivariate decomposition analysis were considered for the multivariable Logit-based multivariate decomposition analysis. Finally, p-value < 0.05 and the corresponding coefficient (B) with a 95% confidence interval were used to declare significant factors that contributed to the rural–urban difference in ARIs. There was no need for ethical clearance as the researcher did not interact with respondents. The data used was obtained from the MEASURE DHS Program, and permission for data access was obtained from the measure DHS program through an online request from http://www.dhsprogram.com. The data used for this study were publicly available with no personal identifier. For details about the ethical considerations of the DHS, the program sees https://dhsprogram.com/methodology/Protecting-the-Privacy-of-DHS-Survey-Respondents.cfm.
N/A