Guideline-Discordant Neuro-prognostication as an Indicator of Self-Fulfilling Prophecy Bias in Post-Cardiac Arrest Care

Abstract

Abstract Background Higher rates of ‘early’ withdrawal of life supportive therapy (WLST), often defined as a timeframe less than 72 hours after cardiac resuscitation, have been interpreted as indicating therapeutic pessimism and lower post-resuscitation quality of care. However, this approach overlooks guideline-congruent criteria and patient specific factors that justify early WLST, and conversely, ignores the possibility of guideline discordant WLST beyond 72 hours. In this retrospective study of post-cardiac arrest patients who underwent WLST due to predicted poor neurological outcome (WLST-N), we adjudicated guideline discordance at the individual patient level to better quantify the potential extent of self-fulfilling prophecy bias. Methods Out-of-hospital cardiac arrest survivors spanning over a decade at a single institution were identified from two databases – Group 1: Dallas-Fort Worth Resuscitation Outcomes Consortium Cardiac Arrest registry (Epistry) (2011–2015); Group 2: Get With The Guidelines® resuscitation registry (2016–2022). Electronic medical records were manually reviewed to identify patients who died from WLST-N. We assessed the concordance of objective clinical data with consensus recommendations from the American Academy of Neurology 2006 practice parameters for Group 1, and European Resuscitation Council 2014 guidelines for Group 2. Results Guideline-discordant neuro-prognostication was found in 9% of patients (13 of 139) who survived until Intensive Care Unit admission and 25% (13 of 52) of WLST-N cases. Early WLST-N at < 72 hours occurred in 35% (18 of 52) of WLST-N cases, but half of these prognostic decisions were adjudicated as appropriately predictive of poor outcome (Cerebral Performance Category 3–5). Conclusion The potential extent of self-fulfilling prophecy bias due to therapeutic pessimism may be much lower than suspected based on published rates of ‘early’ WLST-N.