Abstract Body:
Introduction: Traditional outcomes in cardiac arrest studies, including ROSC, survival and survival with favorable neurologic outcome, are limited in their ability to reflect the success or failure of resuscitation efforts. Additionally, the binary nature of these endpoints often requires large sample sizes for evaluation in trials. Alternatively, CPR duration is one of the strongest predictors of survival and neurologic outcome after IHCA and is also a direct result of the efficacy of intra-arrest interventions in their ability to expedite ROSC. Therefore, we aimed to explore a novel cardiac arrest outcome, Time to ROSC, which incorporates CPR duration as a continuous measure.

Methods: The Get With The Guidelines-Resuscitation database was used to estimate control rates of event outcomes and CPR duration in pediatric IHCA (p-IHCA) events with ≥1 to ≤120 minutes of CPR in children <18 years. Two approaches for Time to ROSC were developed using unique statistical methods for handling events without ROSC (ECPR, TOR). Method 1 was a time-to-event analysis using a Fine-Gray model in which ROSC was the event of interest and ECPR and TOR were treated as competing risks. Method 2 used CPR duration as a continuous variable in which events failing to achieve ROSC were assigned the maximum values of 120 (ECPR) or 121 (TOR). Events ending in ROSC were assigned the observed CPR duration. Sample size calculations were performed for a variety of treatment effect sizes for traditional binary endpoints as well as Time to ROSC Methods 1 and 2. Finally, these approaches were applied to a second dataset derived from a real-world p-IHCA trial (ICU-RESUS).

Results: Baseline control estimates and sample size calculations are shown in Table 1. Even when targeting modest effect sizes, the total sample size required for Time to ROSC Methods 1 and 2 remained <1000 patients. When applied to the ICU-RESUS trial, Time to ROSC Method 1 was significantly improved in the intervention group compared to control (HR of better outcome 1.21, p=0.012; Table 2 and Figure 1). Time to ROSC Method 2 was also significantly improved in the intervention group (median 6 [IQR 2, 120] vs 10 [IQR 3, 120]; p=0.029; Table 2).

Conclusion: Time to ROSC is an alternative cardiac arrest outcome that may be used over traditional endpoints. Further development of Time to ROSC using simulation studies across a variety of assumptions is underway.