Time-Domain, Frequency Domain and non-linear measurements in neonates' Heart Rate Variability with clinical sepsis

Sepsis, a critical bacterial infection of the bloodstream, is a serious cause of illness in neonatal period in both premature and at term newborns. It is important to look for parameters that can help earlier detection of sepsis in the newborn. Previous studies have shown that Heart Rate Variability is reduced when associated with sepsis and diminish the adaptive capacity of the individual, degrading the information transported by their signals. To test for the statistical significance in discriminating between healthy and sepsis diagnosed neonates we analyzed the Inter-Beat-Interval derived from 90 minutes electrocardiographic recordings obtained from 45 newborn, 17 with clinical diagnosis of sepsis and 28 healthy newborn as control. Statistically significant time-domain measures (p<;0.05) of the time series produced paradoxical results comparing sepsis with healthy subjects. Frequency-domain and Time-Frequency analysis show reduced low-frequency power and low/high frequency ratio (p<;0.05) in subjects with sepsis; conversely high-frequency power was significantly higher (p<;0.05) in the sepsis group. Non-linear Sample-Entropy measure showed significant difference between groups (p<;0.01) and lower values in subjects clinically diagnosed with sepsis suggesting lower Inter-Beat-Interval signal complexity.