A multi-level statistical modeling of the determinants of neonatal mortality in Uganda

Abstract

Neonatal mortality remains a major public health challenge in Uganda despite improvements in maternal and newborn healthcare services. This study aimed to determine the maternal socioeconomic, neonatal, household environmental and community-level factors associated with neonatal mortality in Uganda using a multilevel logistic regression approach. The study employed a cross-sectional analytical design using secondary data from the 2022 Uganda Demographic and Health Survey (UDHS). A total of 18,178 live births nested within 680 community clusters were included in the analysis among which 118 neonatal deaths were recorded. Data were analyzed using Stata version 16. Descriptive statistics were used to summarize the study variables, chi-square tests examined bivariate associations and multilevel logistic regression models were fitted to account for the hierarchical structure of the data. The empty model revealed significant community-level clustering of neonatal mortality (ICC = 14.23%, LR test, p = 0.035) justifying the use of multilevel modelling. At the bivariate level, household wealth index (χ² = 11.203, p = 0.024) and place of residence (χ² = 6.429, p = 0.011) were significantly associated with neonatal mortality. In the intermediate multilevel model, neonates from households in the second wealth quintile had significantly higher odds of neonatal mortality than those from the poorest households (AOR = 2.159, p = 0.010). In the final multilevel model, after adjusting for maternal socioeconomic, neonatal, household environmental and community-level factors, household toilet facility remained the only statistically significant determinant of neonatal mortality with households having improved toilet facilities exhibiting higher odds of neonatal mortality than those with unimproved facilities (AOR = 5.480, p = 0.020). Although this association was contrary to expectations it should be interpreted cautiously because it may reflect residual confounding or unmeasured factors not captured in the dataset. The study concludes that neonatal mortality in Uganda is influenced by household socioeconomic conditions, household environmental characteristics and community-level factors. The findings highlight the importance of strengthening socioeconomic support, improving maternal and newborn healthcare services, addressing contextual community differences and conducting further research to better understand the unexpected association between household sanitation and neonatal mortality