Validation of the 2HELPS2B Score for Inpatient Seizure Risk Prediction
Abstract number :
2.018
Submission category :
3. Neurophysiology / 3B. ICU EEG
Year :
2019
Submission ID :
2421469
Source :
www.aesnet.org
Presentation date :
12/8/2019 4:04:48 PM
Published date :
Nov 25, 2019, 12:14 PM
Authors :
Thanujaa Subramaniam, University of Wisconsin, Madison; Aaron F. F. Struck, University of Wisconsin, Madison; Mohammad Tabaeizadeh, Massachusetts General Hospital, Harvard; Sarah Schmitt, Medical University of South Carolina; Andres Rodriguez, Emory Unive
Rationale: Seizure risk stratification is needed to boost inpatient seizure detection rates and to improve continuous EEG (cEEG) utilization. 2HELPS2B is a simple point system that addresses this clinical need, but needs validation. The objective of the study is to validate the 2HELPS2B score with an independent patient cohort and develop a practical guide for its implementation. Methods: This is a multicenter retrospective medical record review of clinical and EEG data on consecutive cEEG patients. 2HELPS2B was evaluated on the validation cohort using mean calibration error (CAL) and area under the ROC curve (AUC). Next, the first hour of cEEG was evaluated to assess potential for 2HELPS2B to be used in initial seizure risk stratification. Kaplan-Meier survival analysis was used to determine the required duration of EEG monitoring to achieve seizure risk <5% based on initial risk stratification. Results: Data from 2111 subjects were collected, with a median duration of EEG monitoring: 48H, median age: 51 years, and 47.3% female. CAL on the validation cohort was 4.0%, and was not statistically different from foundational cohort, p=0.13. AUC on the validation cohort was 0.804 and not statistically different from foundational cohort, p=0.29. 2HELPS2B calculated on the first hour of cEEG had CAL of 4.2% and was not statistically different from 2HELPS2B calculated over entire EEG, p=0.93. EEG duration to achieve seizure risk of <5% using initial 2HELPS2B: 2HELPS2B=0: 1H (only screening EEG), false negative rate (FNR) 3.11%; 2HELPS2B=1: 12H, FNR 3.96%; 2HELPS2B³2: 24H, FNR 3.07%. Conclusions: 2HELPS2B is a simple clinical tool to aid in seizure detection, clinical communication, and cEEG utilization in hospitalized patients. Funding: No funding
Neurophysiology