.jpeg.jpg "Ashley Bach, MD, MPH photo")
Matthew Kirschen, MD, PhD
Assistant Professor of Anesthesia and Critical Care Medicine, Pediatrics, and Neurology
Childrens Hospital of Philadelphia
Philadelphia, Pennsylvania, United States
Disclosure information not submitted.
France Fung, MD
Assistant Professor of Neurology
Children's Hospital of Philadelphia, United States
Disclosure information not submitted.
Nicholas Abend, MD MSCE
Associate Professor of Neurology, Pediatrics and Anesthesiology and Critical Care
Children's Hospital of Philadelphia, United States
Disclosure information not submitted.
Jimmy Huh, MD
Associate Professor of Anesthesia, Critical Care and Pediatrics
Children's Hospital of Philadelphia, United States
Disclosure information not submitted.
Shih-shan Lang, MD
Attending Neurosurgeon
Children's Hospital of Philadelphia, United States
Disclosure information not submitted.
Ian Yuan, MD
Assistant Professor of Anesthesia, Critical Care and Pediatrics
Children's Hospital of Philadelphia, United States
Disclosure information not submitted.
Kathryn Graham, BA, MLAS
Research Coordinator
The Children's Hospital of Philadelphia, United States
Disclosure information not submitted.
Jeffrey Berman, PhD
Research Assistant Professor of Radiology
Children's Hospital of Philadelphia, United States
Disclosure information not submitted.
Arastoo Vossough, MD PhD
Associate Professor of Radiology
Children's Hospital of Philadelphia, United States
Disclosure information not submitted.
Alexis Topjian, MD, MSCE, FCCM
Associate Professor of Anesthesiology, Critical Care, and Pediatrics
Children's Hospital of Philadelphia, United States
Disclosure information not submitted.
Title: Association of EEG and Brain MRI Findings with Outcomes After Pediatric Cardiac Arrest
Introduction:
EEG and brain MRI are each associated with outcomes after pediatric cardiac arrest (CA), but limited data exist regarding the advantage of multimodal prognostication. We hypothesized that the predictive accuracy of combined EEG and MRI findings for outcomes post-CA would be higher than that of EEG or MRI alone.
Methods:
Retrospective cohort study of pediatric CA with EEG initiated within 24 hours and MRI obtained within 7 days of return of spontaneous circulation (ROSC) from 2005-2019. Initial EEG background after ROSC was classified as normal, slow/disorganized, discontinuous/burst-suppression, or attenuated/featureless. Worse ischemia on MRI was defined as ≥10% of brain tissue voxels with apparent diffusion coefficient < 650x10-6 mm2/s. Outcomes were death and poor outcome (Pediatric Cerebral Performance Category score change ≥1 from baseline resulting in discharge score ≥3). EEG background was analyzed as a continuous variable. Chi-squared, logistic regression, and area under the receiver operating curve (AUROC) evaluated associations of EEG, MRI, and their combination with outcomes.
Results:
89 children were evaluated. EEG background was normal in 16(18%), slow/disorganized in 41(46%), discontinuous/burst-suppression in 12(14%), and attenuated/featureless in 20(22%). 32(36%) children had worse MRI ischemia. 28(31%) died and 58(65%) had poor outcome. Worsening EEG background was associated with death (OR 5.4 [2.8,10.5]; p< 0.001) and poor outcome (OR 4.4 [2.1,9.2]; p< 0.001). Worse MRI ischemia was associated with death (OR 26.6 [8.0,88.0]; p< 0.001) and poor outcome (OR 15.5 [3.4,71.2]; p< 0.001). The predictive accuracy of EEG background versus worse MRI ischemia did not differ (AUROC: death: EEG 0.87 vs MRI 0.84, p=0.58; poor outcome: EEG 0.79 vs MRI 0.73, p=0.17). Combination of EEG background and worse MRI ischemia had higher predictive accuracy for death and poor outcome than EEG (AUROC: death: 0.92 vs 0.87, p=0.032; poor outcome: 0.83 vs 0.79, p=0.066) or MRI (AUROC: death: 0.92 vs 0.84, p=0.018; poor outcome: 0.83 vs 0.73, p=0.001) alone.
Conclusions:
In this single center cohort, the combination of EEG background within 24 hours and MRI ischemia within 7 days of CA had higher predictive accuracy for outcomes than either modality alone, supporting the use of multimodal prognostication.