Examination of infant mortality risk in Turkey with spatio-temporal Bayesian models

The infant mortality rate in Turkey declined from 13.9 deaths per 1,000 live births in 2009 to 9.3 deaths per 1,000 live births in 2017. This study explored the role of spatio-temporal Bayesian models in explaining this decline. Parametric, nonparametric spatio- temporal Bayesian models, and a Bayesian generalized linear model without space, time, and space-time interaction were applied using the Integrated Nested Laplace Approximation (INLA) method. Exceedance probabilities were used for detecting significant risk clusters. The unstructured spatial and structured temporal interaction random effect of the best-fitting spatio-temporal Bayesian model contributed more to explaining variation in the relative risk of infant mortality than the other random effects. From 2009 to 2017, in each year, significant risk clusters were consistently detected in the eastern and south-eastern Anatolia regions. An increase of 1,000 USD in the Gross Domestic Product (GDP) per capita reduced the relative risk of infant mortality by 2.8%. When determining the factors that may affect infant mortality in Turkey, it is also essential to consider the effects of space, time, and space-time interaction. In addition, decision-makers should consider the increase in GDP per capita as a factor in reducing infant mortality in Turkey by focusing on these significant risk clusters in the eastern and south-eastern Anatolia regions.