Abstract 1769: Storms of Heart Rate Decelerations in Asymptomatic Infants Prior to Neonatal Sepsis
Background: Prior to clinical signs of illness, septic infants have reduced baseline heart rate (HR) variability and transient HR decelerations. While a predictive model based on statistical properties of these heart rate characteristics (HRC) is associated with impending illness, the analysis does not employ measures from discrete decelerations.
Objective: To test the hypothesis that heart rate deceleration frequency and characteristics improve the early detection of neonatal sepsis.
Methods: We developed a formalized wavelet-based template matching algorithm to detect decelerations in HR time series. Over 700,000 records of 4096-beat HR time series from 325 infants in the University of Virginia NICU were analyzed.
Results: There were 168 events of sepsis in 88 infants. The new deceleration metrics added independent information to our existing HRC analysis in predicting neonatal sepsis (p < 0.0001). Interestingly, some asymptomatic infants had storms of large decelerations unaccompanied by apnea at a rate of several per minute. Storms of frequent decelerations were highly predictive and diagnostic of sepsis, with up to 10 or more-fold increase in severe clinical illness. The Figure⇓ inset shows a 20-minute HR record from an asymptomatic infant 7 hours before the diagnosis of Pseudomonas sepsis. The plot shows the large fold-increase in risk of illness with increasing frequency of large decelerations.
Conclusion: This new wavelet-based heart rate deceleration analysis improves heart rate characteristics monitoring in predicting neonatal sepsis.