Unravelling the effects of neighbourhood contextual influences on childhood mortality and morbidity in Nigeria

Background: The burden of childhood stunting and mortality remains huge in developing countries and in particular in the Sub-Saharan Africa region to which Nigeria is located. Despite the body of evidence supporting an association between neighbourhood contextual influences and health outcomes, few studies have examined the relationship between neighbourhood-level risk factors and childhood undernutrition (stunting) and mortality independent of the individual-level risk factors in a single analytical framework in Nigeria. Most studies to date have focused on individual-level factors overlooking the contribution of neighbourhood or area level factors. Beyond the effect of neighbourhood contextual influences, a child’s health will be influenced by the state, region and national policies and programs that in turn will affect the proximate determinants of his or her survival and health status. Aims: We described the variation that existed between the states in Nigeria using league table, control chart and spatial clustering of childhood stunting (Study I) and examined the contribution of community contextual factors at predicting childhood stunting beyond individual-level factors (Study II). We further identified and examined the predictors of childhood mortality in Nigeria (Study III) and developed prognostic model predicting differences in childhood mortality in Nigeria communities (Study IV). We also quantified the contribution of neighbourhood socioeconomic disadvantage alongside individual-level socioeconomic status to childhood mortality in Nigeria using multilevel analysis (Study V). Methods: We used the Nigeria Demographic and Health Survey (DHS) dataset which comprised of 28,647 and 31,482 under-five children nested within 888 and 896 communities for the 2008 and 2013 surveys respectively from 37 states including the Federal Capital Territory. We used league table, control chart and geospatial analysis to describe variations in childhood stunting that existed between the states in Nigeria (Study I). In study II, we applied multivariable multilevel logistic regression analysis to describe the independent contribution of community contextual influences (factors) alongside the individual level factors on childhood stunting in Nigeria. We applied multivariable logistic regression analysis that included Receiver Operating Characteristics (ROC) Curve to construct a model that examined the factors associated with childhood mortality (Study III). In study IV, we used mixed multivariable Poisson regression analysis to develop a prognostic model predicting differences in childhood mortality in Nigeria communities. In Study V, we applied multivariable multilevel logistic regression analysis and considered three measures of individual socioeconomic status i.e. maternal educational attainment, household wealth status, and employment status of the mothers. At the neighbourhood (level 2) and state (level 3), we included poverty rate, unemployment rate, and illiteracy rate. Results: There were statistically significant variations in the odds of childhood stunting and mortality across the neighbourhoods (Study II, IV & V) and states (I) in Nigeria. This confirmed the evidence of community and state level contextual phenomenon influencing childhood survival and stunting. Children residing in socioeconomically disadvantaged neighbourhoods had higher odds of childhood morbidity and mortality compared to their counterparts living in more socioeconomically advantageous neighbourhoods (Study II, IV & V). The odds of childhood morbidity and mortality were associated with neighbourhood and state socioecological conditions even after adjusting for individual’s household socioecological conditions (Study II, III, IV & V). There was moderate positive correlation between neighbourhood and individual variations in childhood mortality and morbidity (Study II & V). The odds of childhood stunting and mortality were higher in children residing in rural areas (Study II, III, IV & V) and in settings with poor sanitation (Study III & IV). Other factors that increased the odds of childhood mortality included low level of maternal health seeking behaviour, not breastfed for >18 months, being from a polygamous family setting, large family and high birth order, non-usage of contraceptive by mother, and mother having first marriage during their teenage years(Study III). Good household wealth status, adequate birth interval, being a female child and having normal birth weight, increasing maternal educational attainment were all associated with odds of not suffering from childhood stunting and surviving beyond five years of age (Study II & V). Conclusions: By adopting several modelling approaches including the multilevel modelling, we added to the growing body of evidence the effects of the neighbourhood contextual influences on childhood stunting and survival in Nigeria. Our study revealed that individual i.e. children and parental factors; neighbourhood and socioecological environment were associated with childhood stunting and mortality. Efforts at reducing the burden of childhood stunting and mortality should be directed at establishment of poverty alleviation programmes, effective publicly funded health care delivery, promotion of hygienic environmental practices and health education more importantly at the neighbourhood level. Lastly, given the importance of socioecological factors at influencing the lifestyles of neighbourhoods and individuals, interventions targeting structural make up of these two entities are vital in order to meet the MDGs 1 and 4 regarding childhood stunting and mortality in Nigeria and in particular developing countries in general.