Reduced heart rate variability predicts severe intraventricular haemorrhage in extremely preterm infants

Intraventricular haemorrhage (IVH) is a serious complication of extreme prematurity. Reduced heart rate variability (HRV) has been linked to severe IVH (sIVH). This study aimed to identify HRV metrics best predictive of sIVH.

This prospective cohort study used ECG traces of preterm infants (≤ 28 weeks) collected in the first week of life. The earliest five 2 h noise-free ECG segments were processed in MATLAB, extracting eight metrics of HRV spanning the time, frequency and non-linear domains: SDRRI, HF, LF, VLF, SD1, SD2, alpha1 and alpha2. Group differences were tested using Wilcoxon rank-sum test. sIVH classification performance was assessed using Receiver Operating Characteristic Area Under the Curve (AUC).

Seven out of 48 infants developed sIVH (full cohort), of which five developed sIVH post-ECG recording (late sIVH cohort). SD1, SD2, HF, and SDRRI were reduced in the full cohort. All metrics but alpha2 were reduced in the late sIVH cohort. Non-linear metric SD1 demonstrated highest classifier performance (AUC = 0.97 in late cohort, 95% CI 0.89-1.00), achieving 100% sensitivity and 90.2% specificity. Frequency metric HF achieved second highest performance (AUC = 0.95, 95% CI 0.85-1.00).

Abnormalities in SD1 and HF are most predictive of sIVH in preterm infants.

Severe intraventricular haemorrhage (sIVH) may be predicted with high sensitivity and specificity using measures of heart rate variability (HRV). A decrease in levels across HRV metrics derived using time, frequency and non-linear domains was associated with sIVH. HRV metrics representing the parasympathetic tone were the most useful in the prediction of sIVH. SD1, a short-term variability metric derived using the Poincaré analysis, was found to be most predictive of sIVH. Bedside HRV monitoring may allow early identification and intervention to prevent sIVH.