A Retrospective Cross Sectional Study of the Effectiveness of a Project in Improving Infant Health in Bwindi, South Western Uganda

Introduction: Low-cost community-based interventions to improve infant health potentially offer an exciting means of progressing toward the Sustainable Development Goals (SDGs). However, the feasibility of such interventions in low-income settings remains unclear. Bwindi Community Hospital (BCH), Uganda implemented a 3-year nurse-led community project to address child-health issues. Nurses supported Community Health Volunteers (CHVs) and visited mothers pre- and/or postnatally to assess and educate mothers and infants. CHVs gathered data and gave basic advice on health and hygiene to mothers. We hypothesized that increased interventions by nurses and CHVs and increased contact with households, would improve health and reduce infant mortality. Methods: This was a retrospective cohort study analyzing routine data of all children born between January 2015 and December 2016. There were three interventions: antenatal nurse visit, postnatal nurse visit and CHV participation. Children received different numbers of interventions. We defined four diverse outcomes: facility-based delivery, immunization completeness, nutritional status, and infant mortality. Odds ratios, adjusted odds ratios, and multivariate logistic regression were used to assess associations between interventions and outcomes. Results: Of the 4,442 children born in 2015 and 2016, 91% were visited by a nurse (81% antenatally and 10% postnatally); 7% lived in villages with a high participating CHV. Households receiving a postnatal visit were more likely to complete immunization (aOR: 1.55, p = 0.016) and have the infant survive (aOR: 1.90, p = 0.05). Children from a hard-to-reach village (no road access) were less likely to be delivered in a health facility (aOR: 0.55, p < 0.001) and less likely to survive in their first year (aOR: 0.69, p = 0.03). Having two or more interventions was associated with a child having all four positive outcomes (aOR 0.78, p = 0.03). Lack of baseline data, a control area, or integrated assessment data limited more detailed evaluation. Conclusion: Visits to mothers after birth, by a nurse to educate and identify child illness, were associated with lower infant mortality and improved infant health as measured by completion of immunizations. Community health interventions could potentially have a greater impact if focused on hard-to-reach areas. Building evaluation into all project designs, whether local or internationally funded, would enable greater learning, and hence better use of resources.