Exploring geographical variations in vaccination coverage in Nigeria: A Bayesian geo-additive Poisson regression model

Background: In developing countries, infant and childhood mortalities are related to childhood diseases. Low vaccination coverage increases the risks of child to various diseases. Nigeria with the least vaccination coverage in Africa needs to address this burden. This paper analyze geographical variations of vaccination coverage in Nigeria.

Methodology: according to WHO, children should complete the recommended vaccinations by 12 months of age. Analyses are based on 2003 Nigeria Demographic Health Survey dataset for children between age 12 and 60 months. A child is fully vaccinated by receiving BCG, measles, 3 doses of DPT and 3 doses of polio vaccine. An outcome variable of interest (say y) was created by summing the total number of vaccines received by each child. Hence, y is assumed to follow a Poisson distribution. Here the geo-additive modeling technique simultaneously estimates the fixed, nonlinear and spatial effects at a step. Inferences are based on Markov chain Monte Carlo technique.

Results: Children from younger (15-19years) and older (40-49years) mothers; and being males are significantly associated with low vaccination coverage. Residing in urban areas and from educated mothers are associated with high coverage. Significant spatial variations are evident: Jigawa, Taraba and Bayelsa states (low vaccination coverage) while Yobe, Gombe, FCT, Kwara and Osun states (high coverage).

Conclusion: Findings from this paper have permitted us the opportunity to discern states with low and high vaccination coverage. The approach is useful for policy makers in directing the scarce resources to states where they are crucially needed.