Session M1.3

Risk-Stratification following Acute Coronary Syndromes Using a Novel Electrocardiographic Technique to Measure Variability in Morphology

Z Syed*, BM Scirica, CM Stultz, JV Guttag

Massachusetts Institute of Technology
Cambridge, MA, USA

Information in the ECG signal can be used to noninvasively determine which patients are more likely to suffer from fatal events than others. In our work, we propose a new technique for analyzing surface electrocardiographic signals, which quantifies the extent to which subtle ECG morphology changes exist from beat to beat. Termed morphological variability (MV), this variable measures changes in the shape and timing characteristics of electrocardiographic events in sequential pairs of heart beats. In contrast to other techniques that focus on one specific aspect of the ECG signal, our approach integrates information from all waves and intervals in the beat, with the goal of providing a more global assessment of cardiac electrical performance.
We assessed the relationship among MV and death during the 3 month period after admission due to NSTE-ACS. MV was calculated in 400 patients using the first 24 hours of continuous ECG performed after admission for NSTE-ACS. For each patient, differences in morphology between every pair of consecutively occurring beats were measured using a variant of dynamic time-warping. This result was used to generate a morphologic distance (MD) time-series from the ECG signal. Finally the variability in the MD time-series was summarized using time-domain and frequency-domain measures similar to those used in heart rate variability analysis.
An increased risk of death was seen during a follow-up period of 90 days for patients with high MV (HR=5.96; p=0.002). The relationship between high MV and death could be observed even after adjusting for HRV based measures (adjusted HR=3.56; p=0.05). In the case of patients who died, MV also showed a linear association with the survival time, and a unit change in the MV measure was associated with an estimated 16 day decrease in survival time (p=0.015).
Our results suggest that patients with high MV are at considerably increased risk of death following NSTE-ACS. Morphologic variability may therefore offer a new noninvasive technique for identifying patients at high risk.

(Abstract Control Number: 75)