Measuring transient return of spontaneous circulation and rearrest after in-hospital cardiac arrest: a prospective validation study.

Abstract

Cardiopulmonary resuscitation process metrics guide quality improvement efforts and are often obtained through automated data collection using defibrillator waveforms. Transient return of spontaneous circulation (ROSC) events are challenging to accurately capture and can distort process metrics. Prior studies have accounted for transient ROSC events using expert-defined annotation rules; however, these rules have not been prospectively validated. We aimed to validate existing approaches to transient ROSC identification. This prospective validation study used a convenience sample of in-hospital cardiac arrest cases in a large metropolitan hospital system from April 2025 to January 2026. Patients without retrievable defibrillator waveform data were excluded. Trained researchers attended arrests and documented observed times of transient ROSC (<20 min) and rearrest. A blinded researcher reviewed defibrillator waveforms and tested whether a previously used definition of ROSC (at least 60 s of a potentially perfusing ECG rhythm without chest compressions) accurately recognized transient ROSC compared to the directly observed standard. Results were reported using patient-level and event-level diagnostic results including sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). Of 48 patients, 34 (71%) were included. Of those 34, 13 (38%) experienced at least one directly observed transient ROSC event. At the patient-level, the 60-s waveform-derived definition had a sensitivity, specificity, PPV, and NPV of 100%. At the event level, there was a sensitivity of 86% (18/21) with two false positives (PPV of 18/20, 90%)). This validation study supports the use of a 60-s waveform-derived definition of transient ROSC with high accuracy compared to in-person identification.