Abstract 20763: Automatic Detection of Ventilation Waveforms From the Bioimpedance Signal During Chest Compression Pauses

    loading  Checking for direct PDF access through Ovid

Abstract

Introduction: Because of limited evidence, the role of ventilation during cardiopulmonary resuscitation (CPR) is not well understood. Effective ventilation provided during 30:2 CPR has been linked to better prognosis in small studies, but definitive evidence is lacking.

Hypothesis: Prior to placement of an advanced airway, only the electrocardiogram and bioimpedance signals are recorded during CPR. This study proposes to create an automated algorithm that uses bioimpedance recordings to detect ventilation (lung inflation) during 30:2 CPR.

Methods: We analyzed data from the Dallas Fort-Worth site of the Resuscitation Outcomes Consortium Continuous vs. Interrupted Chest Compression Clinical Trial. Bioimpedance recordings from Physio-Control Lifepak 12 and 15 defibrillators were reviewed and analyzed from initial CPR until the time of advanced airway placement. A pause was defined as an interruption in chest compressions of 3 to 15 seconds.A ventilation waveform was defined as a single waveform with an amplitude >0.5 Ω and upstroke (inhalation) and down stroke (exhalation) intervals >0.5 s. For waveforms that overlap with chest compressions, the criteria were applied only to the exhalation or inhalation interval. Two independent clinicians reviewed and annotated the recordings using the defined criteria (inter-rater reliability score of 0.89). Those annotations were used as a gold standard for the automated algorithm. An automated electronic algorithm was developed in Matlab. We analyzed sensitivity (SE), specificity (SP), percentage of correctly detected pauses with and without ventilation waveforms, respectively, and the 95% confidence intervals.

Results: The electronic files of 560 patients with 7404 pauses were processed. Of these, 2375 pauses had detectable ventilation waveforms and 5029 pauses did not. The ventilation detection algorithm showed SE= 90.8% (89.5-91.9) and SP= 92.6% (91.9-93.3).

Conclusions: We created a novel, automated software program that can accurately discriminate chest compression pauses with and without ventilations based on the bioimpedance signal. This method can be used to evaluate ventilation during 30:2 CPR and allow ventilation metrics to be included in analyses for future resuscitation studies.

Related Topics

    loading  Loading Related Articles