Oral or Poster Presentation
Concurrent Session 1D - Neonatal Clinical Trials
Introduction: The diagnosis of neonatal sepsis is clinically challenging. We hypothesize that whole blood immune response transcriptomic markers may accurately discriminate bacterial from non-bacterial sepsis in a low/middle-income setting (LMIC).
Methods: Prospective cohort of infants <3 months evaluated for sepsis at the Kamuzu Central Hospital in Lilongwe, Malawi. De-identified standardized hospitalization and vital sign data was collected, verified and sent electronically to a REDCap database in Vancouver, Canada. Blood was collected for RNA-sequencing. Sepsis was categorized by the following symptoms/signs at presentation: impression of not feeding well according to parent/caregiver, ill-looking by physician assessment or severe recessions, lethargy, convulsions or abdominal distension on physical exam. Informed consent was obtained and our study was approved by the UBC and Malawi IRBs.
Results: We enrolled 314 infants over 15 months, collected >53,000 data points (demographics, clinical evaluation and laboratory tests) and 535 RNA samples. Median GA and BW were 38 weeks and 3000 grams. Median age at presentation was 4 days. 63 blood cultures were positive (20%). Infants were inborn (43%), presented from another facility (53%) or from home (3.8%). Mortality was 13% vs 9.6% for infants with positive versus negative blood cultures. Of 290/314 cases meeting clinical sepsis criteria, 59 had positive blood cultures. 88% of infants received antibiotics 5 days. Infants with positive blood cultures had lower admission weights (2984 vs 2617 grams, p<0.05) compared to infants with negative cultures. Infants who died received a shorter duration (2 vs 6 days, p<0.0001) of antibiotics due to early in-hospital mortality.
Conclusion: RNA sequencing of blood samples is ongoing. This cohort shows high rates of positive blood cultures and antibiotic use compared to North America. This study will be the first to validate the incidence of bacterial sepsis in a LMIC using blood transcriptomics, possibly revealing opportunities to safely optimize antibiotic utilization.