Abstract 232: Sensitivity of Individual Data Modalities for Detecting Return of Spontaneous Circulation and Rearrest with Defibrillator Data

Abstract

Background Determining the presence of return of spontaneous circulation (ROSC) and re-arrest (RA) in out-of-hospital cardiac arrests offers a challenge for health care providers and researchers. A better understanding of how different data streams reflect patient vital status could improve patient care by reducing time to identification of ROSC and RA and by improving the efficiency of retrospective analyses. We hypothesized that some data streams in the absence of others may be insufficient in determining rhythm status and we sought to determine the extent of these insufficiencies. Methods Defibrillator data for non-traumatic cardiac arrests collected by the Resuscitation Outcomes Consortium’s (ROC) Pittsburgh site from 2006-2008 and 2011 were analyzed. Custom software was used to identify ROSC/RA status during the resuscitation and post-resuscitation phases of EMS treatment. The data streams included audio, force of compression (COMP), transthoracic impedance (IMP) and patient care reports (PCR). Official ROSC/RA status was established using all available channels. We then analyzed each channel for indicators of ROSC/RA at those time-points when ROSC/RA were noted in the initial analysis. Sensitivities and specificities for detecting ROSC/RA were calculated for each channel relative to the clinical benchmark, a combination of ECG +IMP signals in Stata 12. Results The ROSC sensitivities for each signal compared to our clinical benchmark were .890(audio), .912(ECG), .934(IMP), .868(COMP) and .813(PCR). ROSC/RA combined sensitivities were calculated to be .851(audio), .900(ECG), .933(IMP), .880(COMP) and .719(PCR). The RA sensitivities were calculated to be .767(audio), .860(ECG), .933(IMP), .833(COMP) and .430(PCR) while specificities were calculated to be 1(audio), 1(ECG), 1 (IMP), 1(COMP) and 0(PCR). Conclusion The ECG and IMP individual signals produced the highest sensitivity in detecting ROSC/RA when compared to our benchmark of ECG + IMP. When analyzing a case with just one signal, the sensitivity for detection of ROSC/RA may be grossly inadequate particularly with only PCR information. However, the most commonly available pairing of data streams in modern defibrillators, ECG and transthoracic impedance, is accurate.