Pilot study on VF-waveform based algorithms for early detection of acute myocardial infarction during out-of-hospital cardiac arrest.

Introduction: On-scene detection of acute coronary occlusion (ACO) during ongoing ventricular fibrillation (VF) may facilitate patient-tailored triage and treatment during cardiac arrest. Experimental studies have demonstrated the diagnostic potential of the amplitude spectrum area (AMSA) of the VF-waveform to detect myocardial infarction (MI). In follow-up, we performed this clinical pilot study on VF-waveform based discriminative models to diagnose acute MI due to ACO in real-world VF-patients.

Methods: In our registry of VF-patients transported to a tertiary hospital (Nijmegen, The Netherlands), we studied patients with high-quality VF-registrations. We calculated VF-characteristics prior to the first shock, and first-to-second shock changes (Δ-characteristics). Primary aim was to assess the discriminative ability of the AMSA to detect patients with ACO. Secondarily, we investigated the discriminative value of adding ΔAMSA-measures using machine learning algorithms. Model performances were assessed using C-statistics.

Results: In total, there were 67 VF-patients with and 34 without an ACO, and baseline characteristics did not differ significantly. Based on the AMSA prior to the first defibrillation attempt, discrimination between ACO and non-ACO was possible, with a C-statistic of 0.66 (0.56-0.75). The discriminative model using AMSA + ΔAMSA yielded a C-statistic of 0.80 (0.69-0.88).

Conclusion: These clinical pilot data confirm previous experimental findings that early detection of MI using VF-waveform analysis seems feasible, and add insights on the diagnostic impact of accounting for first-to-second shock changes in VF-characteristics. Confirmative studies in larger cohorts and with a variety of VF-algorithms are warranted to further investigate the potential of this innovative approach.