Spatial modelling of the relationship between HIV prevalence and determinants of child mortality in Rwanda

Abstract

Background: Very high child survival gains and non-decreasing national HIV prevalence coincided in Rwanda between 2005 and 2010, making the expected relationships between corresponding indicators not obvious. Objective: To assess the factors through which HIV/AIDS epidemic affected child mortality in Rwanda and pathways by which its overall impact was circumvented during the studied period. Methods: HIV infection data for 6952 women and child mortality data for 8817 live births were extracted from the 2010 Rwanda Demographic and Health Survey (RDHS) data. HIV infection data were merged with women’s data from 2010 RDHS, while child mortality data were matched with district-level HIV prevalence rate data. Selected Bayesian geo-additive regression models were employed to conduct the intended assessments. Results: Socioeconomic and sexual behaviour-related factors played significant roles in HIV spread in women. Differences in child mortality were determined by district-level spatial factors, mother’s fertility, preceding birth interval, number of antenatal care visits, place of delivery, mother’s age, duration of breastfeeding, and household economic status. HIV prevalence was also a significant determinant of child mortality. However, it did not significantly affect statistical significance of the identified child mortality determinants. It only induced changes in the magnitude of the effects of household economic status, mother’s fertility, spatial factors and all nonlinear-effect determinants. No changes were detected in the effect of antenatal care visits and place of delivery. Conclusion: HIV prevalence had a significant effect on child mortality. It also influenced the effect of certain key child mortality determinants, most of which are not very important during infancy where most of child deaths occurred in Rwanda. Further child survival gains can be achieved by changing sexual behaviours. We also argue that the residual impact of HIV prevalence might have been significantly offset by an improved access to maternal and child health care services. (Word count: 298)