Chest compressions induce errors in end-tidal carbon dioxide measurement

Real-time measurement of end-tidal carbon dioxide (ETCO₂) is used as a non-invasive estimate of cardiac output and perfusion during cardiopulmonary resuscitation (CPR). However, capnograms are often distorted by chest compressions (CCs) and this may affect ETCO₂ measurement. The aim of the study was to quantify the effect of CC-artefact on the accuracy of ETCO₂ measurements obtained during out-of-hospital manual CPR.

We retrospectively analysed monitor-defibrillator recordings collected by two advanced life support agencies during out-of-hospital cardiac arrest. These two agencies, represented as A and B used different side-stream capnometers and monitor-defibrillators. One-minute capnogram segments were reviewed. Each ventilation within each segment was identified using the transthoracic impedance signal and the capnogram. ETCO₂ values per ventilation were manually annotated and compared to the corresponding capnometry values stored in the monitor-defibrillator. Ventilations were classified as distorted or non-distorted by CC-artefact.

A total of 407 1-min capnogram segments from 65 patients were analysed. Overall, 4095 ventilations were annotated, 2170 (32.4% distorted) and 1925 (31.8% distorted) for agency A and B, respectively. Median (IQR) unsigned error in ETCO₂ measurement increased from 1.5 (0.6-3.1)% for non-distorted to 5.5 (1.8-14.1)% for distorted ventilations; from 0.7 (0.3-1.2)% to 3.7 (1.0-9.9)% in agency A and from 2.3 (1.2-3.9)% to 8.3 (3.9-19.5)% in agency B (p < 0.001). Errors were higher than 10 mmHg in 9% and higher than 15 mmHg in 5% of the distorted ventilations.

CC-artefact causes ETCO₂ measurement errors in the two studied devices. This suggests that capnometer algorithms may need to be adapted to reliably perform in the presence of CC-artefact during CPR.