The SLANT Score Predicts Poor Neurologic Outcome in Comatose Survivors of Cardiac Arrest: An External Validation Using a Retrospective Cohort.

Abstract

Hypoxic brain injury is the leading cause of death in comatose patients following resuscitation from cardiac arrest. Neurological outcome can be difficult to prognosticate following resuscitation, and goals of care discussions are often informed by multiple prognostic tools. One tool that has shown promise is the SLANT score, which encompasses five metrics including initial nonshockable rhythm, leukocyte count after targeted temperature management, total adrenaline dose during resuscitation, lack of bystander cardiopulmonary resuscitation, and time to return of spontaneous circulation. This cohort study aimed to provide an external validation of this score by using a database of comatose cardiac arrest survivors from our institution. We retrospectively queried our database of cardiac arrest survivors, selecting for patients with coma, sustained return of spontaneous circulation, and use of targeted temperature management to have a comparable sample to the index study. We calculated SLANT scores for each patient and separated them into risk levels, both according to the original study and according to a Youden index analysis. The primary outcome was poor neurologic outcome (defined by a cerebral performance category score of 3 or greater at discharge), and the secondary outcome was in-hospital mortality. Univariable and multivariable analyses, as well as a receiver operator characteristic curve, were used to assess the SLANT score for independent predictability and diagnostic accuracy for poor outcomes. We demonstrate significant association between a SLANT group with increased risk and poor neurologic outcome on univariable (p = 0.005) and multivariable analysis (odds ratio 1.162, 95% confidence interval 1.003-1.346, p = 0.046). A receiver operating characteristic analysis indicates that SLANT scoring is a fair prognostic test for poor neurologic outcome (area under the curve 0.708, 95% confidence interval 0.536-0.879, p = 0.024). Among this cohort, the most frequent SLANT elements were initial nonshockable rhythm (84.5%) and total adrenaline dose ≥ 5mg (63.9%). There was no significant association between SLANT score and in-hospital mortality (p = 0.064). The SLANT score may independently predict poor neurologic outcome but not in-hospital mortality. Including the SLANT score as part of a multimodal approach may improve our ability to accurately prognosticate comatose survivors of cardiac arrest.