Cardiac Autonomic Function score: a novel risk stratification tool in the cardiac intensive care unit based on periodic repolarization dynamics and deceleration capacity of heart rate (LMU-eICU study)

Background:

Treatment capacities in intensive care units (ICU) are a critical resource in modern medicine. Current concepts of risk stratification are primarily based on evaluation of clinical and vital signs, as well as laboratory values. Here, we propose a novel ECG-based cardiac autonomic risk stratification score (CAF_ICU) based on periodic repolarization capacity (PRD), a repolarization biomarker of sympathetic overactivity and deceleration capacity of heart rate (DC), a parameter showing vagal imbalance.

Methods:
CAF_ICU was developed in a retrospective cohort of 355 patients, consecutively admitted to the cardiac ICU at the LMU-university hospital between February 1^st 2018 and November 30^th 2018. CAF_ICU was subsequently validated in a cohort of 702 patients admitted between December 1^st 2018 to April 30^th 2020 to the same cardiac ICU. All patients in both cohorts were in sinus rhythm.  PRD and DC were calculated using the ECG-signals acquired within the first night of the ICU-stay between 2:00 am and 2:30 am. The primary endpoint of the study was intrahospital mortality within 30 days after ICU admission. We used Cox-regression analysis to calculate the continuous hazard ratios for PRD and DC in the training cohort and we subsequently defined CAF_ICU for each patient, as the cumulative risk derived from both PRD and DC. Survival curves were generated using the Kaplan-Maier method and were dichotomized at the median value of CAF_ICU. The incremental prognostic value of CAF_ICU on top of the established Simplified Acute Physiology score 3 (SAPS3) score was tested using integrated discrimination improvement index (IDI), continuous net reclassification index (NRI) and AI-based decision trees. 

Results:

Thirty (8.5%) and 100 (14.2 %) patients reached the primary endpoint in the training and validation cohorts, respectively. CAF_ICU was significantly higher in non-survivors than survivors in both the training (2.58 ± 1.34 vs. 1.76 ± 0.97 units, p = 0.003) and validation cohorts (2.20 ± 1.05  vs. 1.70 ± 0.83 units, p < 0.001). Moreover, CAF_ICU was a strong predictor of the primary outcome in both the training (HR 12.94; 95% CI 3.08 – 54.32; p < 0.001; Figure 1A) and validation cohorts (HR 3.65; 95% CI 2.26 – 5.91; p < 0.001; Figure 1B). In the pooled cohort CAF_ICU significantly improved risk stratification based on SAPS3-score (increase in IDI 0.032; 95% CI 0.014 – 0.064; p < 0.001, and continuous NRI 0.332; 95% CI 0.243-0.403; p < 0.001). Figure 2 illustrates decision-tree based classification of patients into risk-groups based on SAPS3 and CAF_ICU scores. 

Conclusion:

ECG-based automatic cardiac autonomic risk stratification by means of PRD and DC is highly predictive of short-term mortality in the ICU and can be combined with the SAPS3-score to identify patients with increased risk for intrahospital mortality. 

Figure 1 Cumulative 30-days mortality rates in the training (A) and validation (B) cohorts stratified by cardiac autonomic function score (CAF_ICU) ≥ 1.5 units (red line) and < 1.5 units (blue line).

Figure 2 Decision-tree based classification of patients into risk-groups based on the Simplified Acute Physiology score 3 (SAPS3) and cardiac autonomic function score (CAF_ICU) scores. Risk 0: SAPS3-Score < 56; Risk 1 SAPS3-Score ≥ 56 and < 69 or SAPS3-Score ≥ 69 and CAF_ICU < 1.5; Risk 2: SAPS3-Score ≥ 69 and < 86 and CAF_ICU ≥ 1.5 and Risk 3: SAPS3-Score ≥ 86 and CAF_ICU ≥ 1.5.