In infants with congenital heart disease, decreases in heart rate variability parameters LF (AUC 0.8, p=0.013) and Alpha 1 (AUC 0.71, p<0.05) predicted cardiac arrest up to 3 and 9 hours prior.
Observational (n=47)
No
Do changes in heart rate variability (HRV) parameters predict impending cardiac arrest in infants with congenital heart disease?
Heart rate variability parameters, specifically LF and Alpha 1, can predict impending cardiac arrest up to 9 hours in advance in infants with congenital heart disease.
Effect estimate: AUC 0.8 for LF; AUC 0.71 for Alpha 1
p-value: p=0.013 for LF; p<0.05 for Alpha 1
Introduction: Infants with congenital heart disease (CHD) are at higher risk of cardiac arrest (CA) compared to other patients. Prediction of which patients may be in impending CA is based on auxiliary and often untimely data, resulting in late rescues or fatalities. Heart rate variability (HRV) analysis is a noninvasive, versatile tool by which changes in autonomic nervous system output can be identified quickly. Prior studies have demonstrated that HRV abnormalities can precede diagnosis of systemic illness (i.e. necrotizing enterocolitis). We aim to identify the HRV antecedents of cardiac arrest in infants with CHD. Methods: In this retrospective study, patients < 6 months old with CHD who had experienced CA between 2015-2020 at a single institution were analyzed. Continuous electrocardiogram (ECG) data was retrieved from the local physiological archive inclusive of the 72 hours prior to the arrest event. HRV was analyzed using Kubios. Variables of interest included common HRV parameters including LF, HF, LF/HF, and detrended fluctuation analysis (alpha 1 and alpha 2). A matched logistic regression analysis was conducted using the average HRV parameteres of each infant during two distinct time windows: a three-hour period surrounding the CA event (case) and a three-hour period prior to CA (test). The test window was shifted backward in three-hour increments until 72 hours prior to CA. Results: A total of 47 patients were analyzed. 57% (n =27) had single ventricle anatomy. 70% (n= 33) were repaired at time of CA. Logistic regression analysis revealed that in the three hours preceding CA, LF was the most significantly deranged HRV parameter (average decrease 1 point, p=0.013, AUC 0.8). Alpha 1, reflective of short-term variability, had an earlier significant decrease compared to baseline and predicted CA by up to 9 hours (average decrease 2.8 points, p < 0.05, AUC 0.71). This was true across all etiologies of the CA. Conclusions: In infants with CHD who experience CA, autonomic dysfunction can precede a CA by up to 9 hours, and more acute derangements may occur within 3 hours of CA. This window of alarm may allow clinicians to correct course to prevent CA in this high-risk group of patients.
Baylon et al. (Sun,) conducted a observational in Congenital heart disease with cardiac arrest (n=47). Heart rate variability (HRV) analysis vs. Baseline/prior time windows was evaluated on Prediction of cardiac arrest using HRV parameters (AUC 0.8 for LF; AUC 0.71 for Alpha 1, p=p=0.013 for LF; p<0.05 for Alpha 1). In infants with congenital heart disease, decreases in heart rate variability parameters LF (AUC 0.8, p=0.013) and Alpha 1 (AUC 0.71, p<0.05) predicted cardiac arrest up to 3 and 9 hours prior.