Background With the introduction of therapeutic hypothermia (TH), the prediction of neurological outcomes in cardiac arrest (CA) survivors is challenging. Early, accurate determination of prognosis by emergency physicians is important to avoid unnecessarily prolonged critical care with a likely poor neurological outcome.

Methods This prospective observational study included patients with non-traumatic CA and return of spontaneous circulation (ROSC) between March 2009 and May 2012 at a tertiary academic hospital. Unconscious patients with ROSC were treated with mild TH (32°C–34°C) for 24 hours. Blood samples were collected for S-100B, neuron-specific enolase (NSE), C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) at 0, 24 and 48 hours post-ROSC. Neurological outcomes were evaluated at hospital discharge and dichotomised as good (cerebral performance category (CPC) 1 or 2) or poor (CPC 3, 4 or 5).

Results Of the 119 patients (68.1% male, 53±15.6 years old) who underwent TH, 46 patients had a good outcome (38.9%). Poor neurological outcomes were predicted using receiver operating characteristic analyses at cut-off values of 0.12 g/L for S-100B at 24 hours post-ROSC (sensitivity, 95.0%; specificity, 75.6%; area under the curve (AUC) 0.916; 95% CI of AUC: 0.846 to 0.961), 31.03 ng/mL for NSE at 48 hours post-ROSC (sensitivity, 83.9%; specificity, 96.9%; AUC 0.929; 95% CI of AUC: 0.836 to 0.979) and 11.2 mg/dL for CRP at 48 hours post-ROSC (sensitivity, 69.4%; specificity, 75.0%; AUC 0.731; 95% CI of AUC: 0.617 to 0.827). ESR was not significant.

Conclusions Among the biomarkers, S-100B at 24 hours and NSE at 48 hours post-ROSC were highly predictive of neurological outcomes in patients treated with TH after CA.