Neighborhood-level heterogeneity in childhood morbidity through generalized linear mixed models.

Objective: Childhood morbidities are crucial for improving long-term public health outcomes. This study aimed to examine the existence of child-specific and regional variation in childhood morbidity based on the cross-cutting study of the Performance Monitoring for Action Ethiopia community survey (PMA-ET), and its relationship to socioeconomic and demographic variables in families. Methods: We enrolled 2,581 children suffering from different illnesses from six regions of the country of the survey at 6 weeks postpartum. Generalized linear mixed models (GLMMs) with maximum likelihood estimation were used to assess children's comorbidity status, and the DHARMa package in R to provide readily interpretable scaled residuals and test functions for typical model misspecification problems for the fitted GLMMs. Results: GLMMs with two random intercept models show the presence of child morbidity variations. Cough, fever, and diarrhea were found to be the most frequent types of children's illnesses among the main illness categories that were recorded. Cooking fuel, wealth quartiles, mothers' marital status, mother age, parity, residence, mother's education status, and availability of electricity were significantly associated with children's morbidity. Conclusions: These data show that variations in children's comorbidity were associated with both regional and child-specific characteristics. Thus, general principles for designing policies and interventions are required to reduce child comorbidity.