Time series analysis of non-invasive hemodynamic monitoring data in neonates with hypoxic-ischemic encephalopathy

Abstract

Background and aims: Hemodynamic instability is common in neonates with hypoxic-ischemic encephalopathy (HIE) undergoing therapeutic hypothermia (TH). Rewarming is a critical period and non-invasive circulatory monitoring may help guide cardiovascular supportive therapy. The aim of the study was to provide a comprehensive analysis of cardiac function parameters during TH and its relation to neurodevelopmental outcome.

Methods: In a prospective, observational study, 26 neonates with moderate-severe HIE were enrolled, born between 2016 and 2019. A hemodynamic monitor based on electrical velocimetry (ICON, Osypka Medical GmbH, Berlin, Germany) was used. Heart rate (HR), stroke volume (SV), cardiac output (CO) data were recorded continuously throughout TH and rewarming. Neurological outcome was assessed at 2 years of age using the Bayley Scales of Infant Development II. edition. Favorable outcome was defined as >70 points on both the psychomotor and mental scales. Time-series analysis was used and features of cardiac function were described to perform logistic regression modeling for outcome prediction.

Results: Fourteen (54%) patients had favorable and 12 (46%) had adverse outcome. Data collection started from median [IQR] of 11.8 [7.0; 24.3] hours (h) of life and lasted until 84.0. [81.8; 87.0] h. During TH, the mean HR of the favorable outcome group was significantly lower than that of the adverse outcome group (86 ± 13/min vs. 104 ± 18/min, p = 0.01). During rewarming HR increased similarly in both groups. SV was unaffected by rewarming, and showed a slowly increasing trend. SV of the favorable outcome group was significantly higher compared to the adverse outcome group (1.55 ± 0.23 ml/kg vs. 1.29 ± 0.30 ml/kg, p = 0.035). In line with this, CO was similar in both groups (136 ± 27 ml/kg/min vs. 134 ± 36 ml/kg/min), and a significant 25% increase in CO was observed during rewarming. Based on multiple regression modeling, HR during TH was independently associated with neurological outcome (p = 0.023).

Conclusion: Based on continuous hemodynamic monitoring, patients with adverse outcome have lower SV and higher HR to achieve similar CO to patients with favorable outcome during TH. HR during hypothermia is independently associated with the neurodevelopmental outcome.

Keywords: asphyxia; bioimpedance; electrical velocimetry (EV); hemodynamic monitoring; hypoxic ischaemic encehalopathy; neonates.

Figure 1

Flowchart showing the enrollment procedure of the study.

Figure 2

Vasoactive-inotropic (VIS) scores during therapeutic hypothermia in favorable and adverse neurodevelopmental outcome groups. Point 0 of the X axis marks the start of rewarming. Datapoints represent the median of individual patient scores with interquartile ranges. Each patient score was calculated by averaging hourly scores of the previous period. There was no statistical difference between the two outcome groups.

Figure 3

(A) smoothed curves of the heart rate time-series variable. For smoothing, LOWESS (locally-weighted scatterplot smoothing) was applied on every individual separately (thin lines), or on each group separately (thick lines). (B) smoothed curves of heart rate of the individual patients in different groups cut at the −24 h +6 h of therapeutic hypothermia. (C) Smoothed curves of the cardiac output time-series variable. (D) smoothed curves of cardiac output of the individual patients in different groups cut at the −24 h +6 h of therapeutic hypothermia. (E) Smoothed curves of the stroke volume time-series variable. (F) smoothed curves of stroke volume of the individual patients in different groups cut at the −24 h +6 h of therapeutic hypothermia.

Figure 4

Estimated individual heart rate regression lines and group-level averages of graphs.

Figure 5

Estimated individual cardiac output regression lines and group-level averages of graphs.

Figure 6

Estimated individual stroke volume regression lines and group-level averages of graphs.

Figure 7

Receiver operating charateristics curve of the regression model constructed for neurodevelopmental outcome with heart rate during hypothermia as predictor.