Thrasher Research Fund - Medical research grants to improve the lives of children

Project Details

E.W. "Al" Thrasher

Status: Funded - Closed

Smart discharges to improve post-discharge survival following admission for infection in newborns and young infants

J. Mark Ansermino, MBBCh, MSc, FRCPC

Summary

BACKGROUND: In many African countries, including Uganda, post-discharge mortality following in-hospital treatment of severe infectious illness is a neglected cause of mortality in young infants. GAP: Despite the burden of post-discharge mortality, scarce resources within affected health systems mean that the necessary follow-up of vulnerable children is usually not possible. If these health systems could target vulnerable children, then the scarce resources could be better focused towards these children. However, currently no strategies exist to predict post-discharge mortality in newborns and young infants. HYPOTHESIS: A parsimonious model to predict post-discharge mortality can be developed using simple, easy to collect clinical, social and demographic variables, collected at the time of hospital admission. METHODS: Children from birth to 6 months of age were eligible for enrollment if admitted with a proven or suspected infection to six hospital sites in Uganda. Enrolled children received routine in-hospital evaluation and care during their admission. Children were followed until 6 months post-discharge. The primary outcome was post-discharge mortality. A prediction model was developed and internally validated using elastic net regression and 10-fold cross validation. RESULTS: 2708 infants under 6 months of age were enrolled from six hospitals over approximately 24 months, until March 30, 2020. All subjects received follow-up for six months following discharge. In total 190 children died during admission (7.1%) and 208 (7.8%) died during the post-discharge period. Re-admission occurred in 390 infants (16%). Key admission risk factors for post-discharge mortality included abnormal tone (OR 3.52; 95%CI 2.45 – 4.99), prior admissions (OR 2.45; 95%CI 1.45-3.94), longer travel time (OR range: 2.10 – 4.14, based on distance category), moderate (OR 4.43 95%CI 2.93-6.57) and severe (OR 6.12 95%CI 4.32 – 8.62) malnutrition, hypoxia (OR 4.03 95%CI 2.62 – 6.13), decreased urine production (OR 2.01; 95% CI 1.46 – 2.74) and anemia (OR 2.38 95%CI 1.11-4.60). A model with an area under the ROC curve of 0.74 (95% CI 0.71-0.77) was developed using these variables. IMPACT: This prediction modeling research will lead to the integration of newborns and young infants into a scalable program of Smart Discharges, so those with significant post-discharge vulnerability are identified prior to discharge and targeted to receive life-saving interventions

Publications: