Abstract 188: Neuroprognostication After Pediatric Cardiac Arrest

Abstract

Objective: Management decisions and parental counseling following pediatric cardiac arrest depend on the ability of physicians to make accurate and timely predictions regarding neurological recovery. This study examined the level of agreement in neuroprognostication among neurologists and among intensivists, and prediction accuracy in terms of time post cardiac arrest. Methods: Pediatric neurologists (N=10) and intensivists (N=9) each reviewed 18 cases of children successfully resuscitated from a cardiac arrest and managed in the pediatric ICU. Cases were sequentially presented in three sections: Day 1, Days 2-4, and Days 5-7 post arrest, with updated examinations, neurophysiologic data, and neuroimaging data as available. At each time point, physicians predicted outcome by Pediatric Cerebral Performance Category (PCPC). Predicted PCPC (p) vs. actual hospital discharge PCPC (a) outcomes were compared. Prediction accuracy was defined as Exact Accuracy (p-a = 0) and Close Accuracy (p-a = ±1). Descriptive statistics, Kappa coefficients (Κ) and generalized estimating equation were performed. Results: Agreement among neurologists improved over time (Day 1 Κ 0.27, Days 2-4 Κ 0.43, Days 5-7 Κ 0.62), as did agreement among intensivists (Day 1 Κ 0.30, Days 2-4 Κ 0.44, Days 5-7 Κ 0.57). For all physicians, prediction accuracy did not improve from Day 1 to Days 2-4, but did improve from Days 2-4 to Days 5-7 (p=0.001) [Exact Accuracy Days 2-4 and Days 5-7: 35% vs. 43%; Close Accuracy Days 2-4 and Days 5-7: 77% vs. 89%]. Prediction accuracy did not differ significantly between physician groups at any time point (p=NS). At Days 5-7, 16 of 19 physicians predicted PCPC = 5-6 at least once when actual PCPC ≤ 4. All physicians predicted PCPC ≤ 4 at least once when actual PCPC = 5-6. Conclusions: Inter-rater agreement among neurologists and among intensivists improved over time and converged to moderate-good levels at later time points. For all physicians, prediction accuracy improved over time; however, even at later time points, incorrect outcome predictions occurred. The ethical implications of these findings for clinical decision-making and allocation of health care resources after pediatric cardiac arrest are considerable.