Amplitude spectrum area is dependent on the electrocardiogram magnitude: evaluation of different normalization approaches

Abstract Objective. Amplitude Spectrum Area (AMSA) of the electrocardiogram (ECG) waveform during ventricular fibrillation (VF) has shown promise as a predictor of defibrillation success during cardiopulmonary resuscitation (CPR). However, AMSA relies on the magnitude of the ECG waveform, raising concerns about reproducibility across different settings that may introduce magnitude bias. This study aimed to evaluate different AMSA normalization approaches and their impact on removing bias while preserving predictive value. Approach. ECG were recorded in 118 piglets (1–2 months old) during a model of asphyxia-associated VF cardiac arrest and CPR. An initial subset (91/118) was recorded using one device (Device 1), and the remaining piglets were recorded in the second device (Device 2). Raw AMSA and three ECG magnitude metrics were estimated to assess magnitude-related bias between devices. Five AMSA normalization approaches were assessed for their ability to remove detected bias and to classify defibrillation success. Main results. Device 2 showed significantly lower ECG magnitude and raw AMSA compared to Device 1. CPR-based AMSA normalization approaches mitigated device-associated bias. Raw AMSA normalized by the average AMSA in the 1st minute of CPR (AMSA 1m-cpr ) exhibited the best sensitivity and specificity for classification of successful and unsuccessful defibrillation. While the optimal AMSA 1m-cpr thresholds for balanced sensitivity and specificity were consistent across both devices, the optimal raw AMSA thresholds varied between the two devices. The area under the receiver operating characteristic curve for AMSA 1m-cpr did not significantly differ from raw AMSA for both devices (Device 1: 0.74 vs. 0.88, P = 0.14; Device 2: 0.56 vs. 0.59, P = 0.81). Significance. Unlike raw AMSA, AMSA 1m-cpr demonstrated consistent results across different devices while maintaining predictive value for defibrillation success. This consistency has important implications for the widespread use of AMSA and the development of future guidelines on optimal AMSA thresholds for successful defibrillation.