Abstract 19906: Evaluation of Approximate Entropy as a CPR Quality Measure

Abstract

Introduction: The majority of standard cardiopulmonary resuscitation (CPR) quality metrics are based on univariate, linear measures of CPR process. Non-linear measures, such as approximate entropy (ApEn), which characterizes the disorder of a signal, have been utilized in other areas, such as heart rate analysis, to capture phenomena that elude standard linear measures. Objective: To provide a preliminary evaluation of the performance of ApEn as a CPR-quality metric. Methods: Case data for non-traumatic, EMS-treated out-of-hospital cardiac arrests were obtained from the Pittsburgh site of the Resuscitation Outcomes Consortium for the period spanning the years 2010 to 2014. Cases with at least 1 available defibrillator data file containing an accelerometer-derived chest compression (CC) depth channel were included. CC depth data traces were extracted with custom software created in MATLAB (2014b, The Mathworks, Nattick, MA). Individual CC for the entire length of each available file were located with a calibrated peak finding function and used to calculate individual CC depth (mm), rate (compressions/minute) and flow fraction time series, which were reported in averages over 1-minute epochs. Data from cases with multiple data files were merged to reflect the order of recordings. ApEn was calculated for each 1-minute epoch of depth and rate. Correlation coefficients were calculated to assess the relationship between ApEn, CC rate, CC depth and flow fraction overall and stratified by pre/post 5 minutes of CPR. Results: Of a total of 1,278 cases with defibrillator downloads, a subset of 1,266 cases had analyzable CC depth channels. A total of 1,875,209 individual CC were detected and analyzed, covering 18,719 minutes of CPR. Mean (SD) ApEn of CC rate was 0.92 (0.34) and of CC depth was 0.79 (0.23). ApEn of CC depth was weakly correlated with flow fraction (r = 0.22) and ApEn of CC rate (r = 0.27), but not CC rate or CC depth. Both before and after the 5-minute mark, ApEn correlated weakly with flow fraction (r = 0.14 vs. r = 0.24) and ApEn of CC rate ( r = 0.30 vs 0.26). Conclusion: ApEn was only weakly correlated with CC flow fraction, a common CPR quality measurement. Additionally, ApEn of rate and depth were not correlated with the absolute magnitude of CC rate or depth.