Abstract 84: Hyperventilation Alert During Out-of-Hospital Cardiopulmonary Resuscitation Using an Automated Capnogram Analysis Algorithm

Abstract

Background Hyperventilation during cardiopulmonary resuscitation (CPR) inhibits venous blood return, decreases coronary perfusion pressure, and is associated with decreased survival rates. Accurate computation of ventilation rate during CPR by defibrillator-based algorithms is difficult due to artifact induced on impedance waveforms. Real time capnography may help detect inadvertent hyperventilation during resuscitation. Objective To evaluate an automated defibrillator-based capnogram (CO2) analysis method to alert for excessive ventilation rate in the presence of low end-tidal CO2 (EtCO2). Methods Recordings from 270 OHCA patients with capnogram waveforms were collected between 2008 and 2010 from two EMS agencies based in Oregon and Texas. The hyperventilation detection algorithm was added to a previously reported EtCO2 analysis program. The hyperventilation algorithm employs exponentially-weighted moving-average filtering (alpha=0.25) to compute respiration rate (RR) and EtCO2 trends from individual breath-breath intervals and breath EtCO2 measurements. A combination of RR and EtCO2 is used in the decision tree. The algorithm was tested using a threshold of RR > 12 bpm and EtCO2 < 20 mmHg. The added requirement of a low EtCO2 avoids false detection on post-resuscitation spontaneous respirations which are unlikely to have low EtCO2. Hyperventilation alerts generated by algorithm were manually validated by two reviewers who had access to CO2, ECG, compression, and impedance-based respiration waveforms. Results The automated program generated 414 hyperventilation alerts in 148 (55%) patients. A total of 24 alerts were identified as false positive due to artifacts in the capnogram waveform. No false negative alert was identified. This yields an algorithm sensitivity of 100% and positive-predictive value 94%. During validated hyperventilation, the average maximum ventilation rate was 25 ±11 bpm, with an average minimum EtCO2 of 6 ±4 mmHg. Conclusion Hyperventilation is common in OHCA resuscitation. An automated capnogram analysis method can accurately detect episodes of hyperventilation and has the potential to alert EMS personal to excessive ventilation rates during OHCA resuscitation.