Abstract: The Under-five mortality (U5M) rate is an important determinant of societal and national advancement- a key marker of wellbeing, value, and access. In spite of efforts to identify the predictors of U5M to reduce its high level in Nigeria, the problem remains a major cause for concern. This study estimated the potential role of unmeasured/unobserved factors at both family and community levels, using shared frailty models on the 2013 Nigeria Demographic and Health Survey (2013 NDHS) data. The Log-rank test was used to identify variables associated with U5M- hazard ratio estimates with P<0.05 were considered as statistically significant. Our findings suggested region, marital status, place of residence and place of delivery were significant determinants of U5M in both frailty models. We also found evidence of frailty effect on the risk, particularly at the community level- heterogeneity due to unmeasured/unobserved factors, which are generally ignored when we assess the risk using only observed variables. We, therefore, suggest that to achieve the sustainable development goals relating to child health in Nigeria, more significant efforts should be directed at identifying more determinants, such as to reduce the influence of unobserved factors and facilitate an extension of interventions to these factors.

Keywords:  Under-five mortality, Unobserved heterogeneity, Frailty model, Determinants, Nigeria.