Abstract 13019: Unique Changes in Electrocardiographic Metrics on Continuous Telemetry Prior to In-Hospital Cardiac Arrests Identified With Automated Algorithms
Introduction: Despite advances, survival to discharge for in-hospital cardiac arrests (IHCA) remains less than 25%. Patients at risk of IHCA are often on continuous ECG monitoring, but no automated methods exist to leverage this information.
Hypothesis: We hypothesize that ECG changes detectable with automated algorithms in the hours preceding IHCA may aid in predicting which patients are at risk of IHCA.
Methods: We conducted a retrospective study on 79 IHCA and 1743 age-matched control patients. Telemetry data was processed with SuperECG (Mortara Instruments, Milwaukee, WI) for PR, QRS duration (QRSd), ST, QTc, RR, and QRS amplitude (AMP) metrics. Median, discontinuity, tachyarrhythmia and pacer dependence filters were applied. For cases, we selected the final (C1) and preceding (C2) 3-hour blocks before IHCA. For controls, we randomly selected two consecutive 3-hour blocks (T1, T2). C1 was compared to C2 and to a pooled control group (TP), consisting of T1 and T2, to identify changes unique to the final 3 hours. C2 was compared to TP, and T1 to T2 for internal control. Single and dual metric comparisons were performed. Maximal false positive rate of 5% was used to define thresholds for metric changes. We used chi-square testing and adjusted for multiple comparisons.
Results: For C1 vs C2, we found higher percentages of significant QRSd prolongation (19.0% vs 3.8%, P=0.003) and AMP decrease (11.4% vs 1.3%, P=0.009). For C1 vs TP, we found higher percentages of QRSd prolongation (19.0% vs 4.6%, P<0.001), AMP decrease (11.4% vs 4.5%, P=0.004), and ST elevation (11.4% vs 3.0%, P<0.001). For dual metric comparisons, highest sensitivity was found with having both QRSd prolongation and ST elevation (C1 vs C2: 21.5% vs 3.8%, P<0.001 and C1 vs TP: 21.5% vs 4.7%, P<0.001). The sensitivity for meeting any single or dual metric comparison was 44.3% for C1 versus a specificity of 88.6% (P<0.001) for C2 and 86.5% (P<0.001) for TP. T1 vs T2 and C2 vs TP were statistically insignificant for all comparisons above.
Conclusions: Significant ECG changes, identified by automated methods, are seen in the hours prior to IHCA. With further refinement and investigation, algorithms to detect ECG changes can be used in conjunction with traditional methods to help identify patients at risk of IHCA.
Author Disclosures: A. Kuo: None. D.H. Do: None. Y. Daniel: None. Y. Bai: None. Q. Ding: None. D. Mortara: Ownership Interest; Significant; Mortara Instruments. X. Hu: None. N.G. Boyle: None.
- © 2016 by American Heart Association, Inc.