Elsevier

Resuscitation

Volume 165, August 2021, Pages 93-100
Resuscitation

Clinical paper
Shock decision algorithm for use during load distributing band cardiopulmonary resuscitation

https://doi.org/10.1016/j.resuscitation.2021.05.028Get rights and content

Abstract

Aim

Chest compressions delivered by a load distributing band (LDB) induce artefacts in the electrocardiogram. These artefacts alter shock decisions in defibrillators. The aim of this study was to demonstrate the first reliable shock decision algorithm during LDB compressions.

Methods

The study dataset comprised 5813 electrocardiogram segments from 896 cardiac arrest patients during LDB compressions. Electrocardiogram segments were annotated by consensus as shockable (1154, 303 patients) or nonshockable (4659, 841 patients). Segments during asystole were used to characterize the LDB artefact and to compare its characteristics to those of manual artefacts from other datasets. LDB artefacts were removed using adaptive filters. A machine learning algorithm was designed for the shock decision after filtering, and its performance was compared to that of a commercial defibrillator's algorithm.

Results

Median (90% confidence interval) compression frequencies were lower and more stable for the LDB than for the manual artefact, 80 min−1 (79.9–82.9) vs. 104.4 min−1 (48.5–114.0). The amplitude and waveform regularity (Pearson's correlation coefficient) were larger for the LDB artefact, with 5.5 mV (0.8–23.4) vs. 0.5 mV (0.1–2.2) (p < 0.001) and 0.99 (0.78–1.0) vs. 0.88 (0.55–0.98) (p < 0.001). The shock decision accuracy was significantly higher for the machine learning algorithm than for the defibrillator algorithm, with sensitivity/specificity pairs of 92.1/96.8% (machine learning) vs. 91.4/87.1% (defibrillator) (p < 0.001).

Conclusion

Compared to other cardiopulmonary resuscitation artefacts, removing the LDB artefact was challenging due to larger amplitudes and lower compression frequencies. The machine learning algorithm achieved clinically reliable shock decisions during LDB compressions.

Introduction

The cardiopulmonary resuscitation (CPR) guidelines stress the importance of high-quality CPR and early defibrillation for a successful outcome after out-of-hospital cardiac arrest (OHCA).1 Chest compressions during CPR introduce artefacts in the electrocardiogram (ECG) that impede a reliable rhythm analysis by the defibrillator.2 Therefore, the current practice mandates CPR interruption to analyse the ECG and provide a reliable shock decision. These “hands-off” intervals for rhythm analysis lead to intermittent lack of cerebral and myocardial blood flow, and may compromise the outcome of the patient.3, 4, 5, 6, 7 Methods allowing a reliable shock decision without interrupting CPR would therefore be of great value.

Shock decision algorithms during manual CPR rely on artifact suppression filters.8 These filters use additional signals like the thorax impedance9, 10, 11, 12, 13 or the compression depth9, 10, 11, 12 to adapt to the time-changing characteristics of manual chest compressions. However, shock decision algorithms in defibrillators were designed to analyse artefact-free ECGs, and thus show a degraded performance even after filtering the artefact.13, 12 To address this limitation, research has focused on machine learning (ML) algorithms that analyse the ECG after filtering the CPR artefact.14, 15, 16, 17 These algorithms learn the characteristics of the filtered ECG, including those of the filtering residuals, and have been shown to meet the American Heart Association's (AHA) sensitivity and specificity requirements for shock decision algorithms.

The use of mechanical compression devices is rising in the prehospital setting, although none of the more recent clinical trials have demonstrated improved survival.18, 19, 20 The benefits of mechanical CPR include guaranteed compression rate and depth, and the possibility for high-quality chest compressions even during transport and invasive procedures.21, 22, 23 At present there are two main mechanical chest compression technologies: piston-driven and load distributing band (LDB) devices. ECG analysis algorithms for a reliable shock decision during mechanical CPR have only been demonstrated for piston-driven devices.24, 25, 26, 27 These studies have shown that a fixed compression rate and depth results in a much more periodic artefact than that of manual CPR.25 Therefore, simpler adaptive filters that exploit the periodic nature of the artefact have been successfully used to remove mechanical artefacts without the need of additional reference signals. These adaptive filters in combination with ML-algorithms for the shock decision produced solutions compliant with the AHA requirements.26

This study demonstrates the first method for a reliable shock decision during LDB compressions using a combination of a CPR artifact filter and a ML-algorithm. The analysis also includes a characterization of the LDB artefact in the time and frequency domain, a preliminary step needed to properly design the CPR artefact suppression filters.

Section snippets

Data collection

The data used in this study were extracted from the randomized controlled Circulation Improving Resuscitation Care (CIRC)18 trial conducted between March 2009 and January 2011 by emergency services in the United States (three services) and Europe (two services). The CIRC trial was designed to compare the effectiveness of an LDB device (AutoPulse, ZOLL, Chelmsford, Massachusetts, USA) against manual CPR in terms of survival to hospital discharge.18 Data from 969 OHCA patients from one US

Results

Shockable rhythms comprised lethal ventricular arrhythmia, predominantly ventricular fibrillation VF (19.2%), but also VT (0.7%). Nonshockable rhythms included asystole (27.8%) and organized rhythms (52.3%). In total 5813 segments (from 896 patients) were extracted, whereof 1154 (from 303 patients) were shockable and 4659 (from 841 patients) nonshockable.

Discussion

The use of mechanical CPR devices has grown considerably in the last years through two main technologies, LDB and piston-driven devices. Mechanical CPR guarantees high-quality chest compressions when manual compressions cannot be delivered safely. Examples of such situations include ambulance transportation,32, 33 primary percutaneous coronary intervention,34, 35 as a bridge to extracorporeal CPR,36 and to facilitate uncontrolled organ donation after circulatory death.37

This study describes the

Conclusions

The first method for an automatic and clinically safe shock decision during LDB CPR was demonstrated, meeting AHA requirements. Filtering the LDB artefact was challenging because of the larger amplitudes and a ML-algorithm was required to provide a reliable shock decision on the filtered ECG.

The proposed solution together with solutions already available for piston-driven and manual chest compressions would cover rhythm analysis in every CPR scenario. This may open the possibility of bringing

Credit author statement

Iraia Isasi: Conceptualization, methodology, software, formal analysis, data curation, writing (original draft and review) and visualization.

Unai Irusta: Term, conceptualization, methodology, data curation, writing (review), funding and supervision.

Elisabete Aramendi: Term, conceptualization, methodology, data curation, writing (review), funding and supervision.

Jan-Age Olsen: Term, conceptualization, resources, data curation and writing (review).

Lars Wik: Term, conceptualization, resources,

Conflict of interest

Dr. Wik is member of the medical advisor board at Stryker and has patents licensed to Zoll and Stryker.

Acknowledgements

This work was supported by the Spanish Ministerio de Ciencia, Innovacion y Universidades through grant RTI2018-101475-BI00, jointly with the Fondo Europeo de Desarrollo Regional (FEDER), by the Basque Government through grant IT1229-19, and by the university of the Basque Country (UPV/EHU) under grant COLAB20/01.

References (38)

Cited by (5)

  • Application of Artificial Intelligence in Cardiopulmonary Resuscitation

    2023, Medical Journal of Peking Union Medical College Hospital
View full text