Development of a model to predict electroencephalographic seizures in neonates with hypoxic ischemic encephalopathy treated with therapeutic hypothermia

Shavonne L Massey; Amanda G Sandoval Karamian; Mark P Fitzgerald; France W Fung; Abigail Abramson; Mandy K Salmon; Darshana Parikh; Nicholas S Abend

doi:10.1111/epi.18196

Development of a model to predict electroencephalographic seizures in neonates with hypoxic ischemic encephalopathy treated with therapeutic hypothermia

Epilepsia. 2024 Dec 16. doi: 10.1111/epi.18196. Online ahead of print.

Authors

Shavonne L Massey^{1

2}, Amanda G Sandoval Karamian^{3

4}, Mark P Fitzgerald^{1

2}, France W Fung^{1

2}, Abigail Abramson⁵, Mandy K Salmon⁶, Darshana Parikh¹, Nicholas S Abend^{1

2

7

8}

Affiliations

¹ Department of Pediatrics (Division of Neurology), Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA.
² Departments of Neurology and Pediatrics, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, USA.
³ Department of Pediatrics, Division of Neurology, University of Utah School of Medicine, Salt Lake City, Utah, USA.
⁴ Brain and Spine Center, Primary Children's Hospital, Salt Lake City, Utah, USA.
⁵ Jacobs School of Medicine and Biomedical Sciences, University at Buffalo, Buffalo, New York, USA.
⁶ Department of Otolaryngology, Oregon Health and Science University, Portland, Oregon, USA.
⁷ Department of Anesthesia and Critical Care, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, USA.
⁸ Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, USA.

PMID: 39680059
DOI: 10.1111/epi.18196

Abstract

Objective: Electroencephalographic seizures (ES) are common in neonates with hypoxic-ischemic encephalopathy (HIE), but identification with continuous electroencephalographic (EEG) monitoring (CEEG) is resource-intensive. We aimed to develop an ES prediction model.

Methods: Using a prospective observational study of 260 neonates with HIE undergoing CEEG, we identified clinical and EEG risk factors for ES, evaluated model performance with area under the receiver operating characteristic curve (AUROC), and calculated test characteristics emphasizing high sensitivity. We assessed ES incidence and timing in neonates subdivided by ES risk group (low, moderate, high) as determined by EEG risk factors.

Results: ES occurred in 32% (83/260) of neonates. Performing CEEG for only 24 h would fail to identify the 7% (17/260) of neonates with later onset ES (20% of all neonates experiencing ES). Identifying 90% or 95% of neonates with ES would require CEEG for 63 or 74 h, respectively. The optimal model included continuity and epileptiform discharges, both assessed in the initial 1 h of CEEG. It yielded an AUROC of .80, and at a cutoff that emphasized sensitivity, had sensitivity of 94%, specificity of 45%, positive predictive value of 44%, and negative predictive value of 95%. The model would avoid CEEG beyond 1 h in 32% (84/260) of neonates, but 6% (5/83) of neonates with ES would not have ES identified. ES incidence was significantly different (p < .01) across ES risk groups (6% low, 40% moderate, and 83% high). Only ~6 h of CEEG would identify all neonates with ES in the low-risk group, whereas 75 and 63 h of CEEG would be required to identify 95% of neonates with ES in the moderate-risk and high-risk groups, respectively.

Significance: Among neonates with HIE, a model employing two EEG variables from a 1-h screening EEG and stratifying neonates into low-, moderate-, and high-risk groups could enable evidence-based strategies for targeted CEEG use.

Keywords: continuous EEG monitoring; electroencephalogram; hypoxic–ischemic encephalopathy; neonatal; seizure; status epilepticus.

Abstract

Grants and funding