128 Nonlinear Dynamic Analysis of Heart Rate Variability Can Detect Sepsis in Severe Burns

Abstract

Abstract Introduction Sepsis after burn injury affects 50–84% of adult patients and ~55% of pediatric patients. The diagnosis of sepsis in patients with severe burns is complicated by the overlap of clinical signs of post-burn hypermetabolic response with those of sepsis. Earlier detection of sepsis could lead to improved outcomes and decreased morbidity and mortality. Nonlinear dynamic (ND) analysis of heart rate variability (HRV) can identify non-random relationships between different HRV parameters by using a complex oscillogram. ND can be obtained even when heart rate recordings are weak or poor in quality, which is common in the setting of burns. We hypothesized that ND could detect differences in HRV after severe burns. Methods All patients enrolled during the study period at our institution with >20% total body surface area (TBSA) burn were eligible for this study. High-definition heart rate monitoring was collected three times for each patient in five-minute intervals. The first collection occurred after patient ICU admission, and the third collection occurred 16–24 hours. Matlab was used to analyze heart rate using the following common ND methods: Correlation dimension (CD), detrended fluctuation (DF), largest Lyapunov exponent (LLE), approximate entropy (AE), fractal dimension (FD), Hurst exponent (HE), and recurrence quantification analysis (RQA). Additionally, several linear time-domain parameters were analyzed, including heart rate (R), RR interval (RR), SD1, and SD2. Results A total of six non-septic (NS) and four septic (S) patients were enrolled in the study. There were no significant differences between the two groups in age, sex, percent TBSA, or percent TBSA of third-degree burns. Two nonlinear parameters were significantly different between NS and S patients: AE and the density of recurrence points in RQA. AE was significantly lower in NS than S patients at all time points [(0.931±0.011 vs. 0.960±0.024, p=0.001); (0.923±0.019 vs. 0.959±0.025, p=0.008); (0.933±0.011 vs. 0.964±0.025, p=0.013)]. The density of recurrence points in RQA were significantly higher than in NS than S patients at all time points. [(90.4±40.2 vs. 41.4± 27.8, p=0.029), (102.2±39.4 vs. 40.1±22.9, p=0.008), (92.5±34.4 vs. 45.5±33.9, p=0.043)]. Conclusions The two ND parameters AE and RQA, can be used to distinguish between patients with and without sepsis in the setting of severe burns.