Gains of Continuing Resuscitation in Refractory Out-of-hospital Cardiac Arrest: A Model-based Analysis to Identify Deaths Due to Intra-arrest Prognostication.

Objective: Prognostication bias, in which a clinician predicts a negative outcome and terminates resuscitation (TR) thereby ensuring a poor outcome, is a rarely identified limitation of out-of-hospital cardiac arrest (OHCA) research. We sought to estimate the number of deaths due to intra-arrest prognostication in a cohort of OHCA's, and use this data to estimate the incremental benefit of continuing resuscitation.

Methods: This study examined a cohort of consecutive non-traumatic EMS-treated OHCAs from a provincial ambulance service, between 2007 and 2011 inclusive. We used Cox and logistic regression modeling, adjusting for Utstein covariates, to estimate the probability of ROSC, survival, and favorable neurological outcomes as a function of resuscitation time, and applied these models to estimate the number of missed survivors in those who had TR (prior to 20, 30, or 40 minutes). We determined the time juncture at which (1) the likelihood of survival fell below 1%, and (2) the proportion of survivors who had achieved ROSC exceeded 99%.

Results: Of 5674 adult EMS-treated cases, 46% achieved ROSC, and 12% survived. The median time of TR was 27.0 minutes (IQR 19.0-35.0). Continuing resuscitation until 40 minutes yielded an estimated 17 additional survivors (95% CI 13-21), 10 (95% CI 7-13) with favorable neurological outcomes. The probability of survival of those in refractory arrest decreased below 1% at 28 minutes (95% CI 24-30 minutes). At 36 minutes (95% CI 34-38 minutes) >99% of survivors had achieved ROSC.

Conclusion: We identified possible deaths due to intra-arrest prognostication. Resuscitation should be continued for a minimum of 30 minutes in all patients, however for those with initial shockable rhythms 40 minutes appears to be warranted. Interventional trials and observational studies should standardize or adjust for duration of resuscitation prior to TR.