232 - Heart rate characteristics identify extremely low birth weight infants at increased risk for cerebral palsy

Publication Number: 3224.232

Lisa Letzkus, University of Virginia School of Medicine, Earlysville, VA, United States; Sherry Kausch, UVA Childrens, Charlottesville, VA, United States; Douglas Lake, University of Virginia, Charlottesville, VA, United States; Karen Fairchild, University of Virginia School of Medicine, Charlottesville, VA, United States

Background: Abnormal heart rate characteristics (HRC) can reflect central nervous system dysfunction and serve as physiologic markers to identify NICU patients at high risk for neurologic disabilities such as cerebral palsy (CP). The HRC (HeRO) Score predicts sepsis, with higher values indicating reduced HR variability (HRV) and superimposed HR decelerations, similar to the HR pattern in fetal distress. We previously reported that persistently high HRC scores are associated with severe IVH and adverse 2-year outcomes.

Objective: In the current work sought to identify neonatal HR patterns associated with later diagnosis of CP.

Design/Methods: We analyzed data from extremely low birth weight (ELBW, < 1000 g) infants enrolled in the HERO trial (NCT00307333) randomized to HRC score display or non-display. We included infants with 2-year assessment of Gross Motor Function Classification System (GMFCS) and at least 5d of HR data available for analysis. Two groups were identified: infants with GMFCS < 2 (“No/Mild CP”) and GMFCS ≥2 (“Moderate/Severe CP”). Hourly HRC Scores and 10-minute HRV and HR deceleration measures were summarized every 12h then averaged over the first 28 days from birth. We evaluated CP prediction of these cumulative HR metrics in two ways: positive predictive value (PPV) at the point where sensitivity=PPV and area under the receiver operator characteristics curve (AUC).

Results: Among 439 ELBW infants with data available, 405 had no/mild CP and 34 (7.7%) had moderate/severe CP (Table 1). The optimal CP PPV metric was cumulative low HRV measured on 10-min data segments as maximum minus mean HR (Fig.1) which had PPV=0.323 (11/34) compared to PPV=0.235 (8/34) for the HRC Score. In contrast the optimal CP AUC metrics were HRC Score (Fig.2) and HR decelerations measured on 10-min data segments as median minus mean HR (data not shown). HRC Score AUC was 0.790 compared to 0.757 for HRV. Both HRV and HRC Scores remained significant when gestational age was included in logistic regression models (respective AUC/p-value 0.823/0.005 and 0.840/0.001). Cross-validated AUC of a model combining the 3 HRC features was 0.820.

Conclusion(s): HR metrics can serve as early physiomarkers of CP risk with low HRV having the strongest clinical predictive value. Integration of physiologic analytics into NICU monitoring may enhance early identification and intervention for infants at risk of CP.

Table 1: Characteristics of Infants Based on CP Category
Slide1.jpeg

Figure 1:Heart rate variability first 28 days by group
Slide1.jpegAverage 28-day HRV is shown for infants with no/mild cerebral palsy (CP) versus those with moderate to severe CP. HRV is defined as the difference between maximum and mean heart rate over 10 minute data segments. The black dotted line indicates the decision threshold where the number of predicted positives equals the number of observed events, resulting in sensitivity equal to positive predictive value (PPV). The area under the receiver operating characteristic curve (AUC) is 0.757, with a positive predictive value (PPV) of 32.4%.

Figure 2. Heart rate characteristics (HRC) score during the first 28 days by group.
Slide1.jpegAverage 28-day HRC scores are shown for infants with no/mild cerebral palsy (CP) versus those with moderate to severe CP. The black dotted line indicates the decision threshold where predicted positives equal observed events, such that sensitivity equals positive predictive value (PPV). The area under the receiver operating characteristic curve (AUC) is 0.790, with a positive predictive value (PPV) of 23.5%.