Software Credentialed Access
Epileptiform activity and outcomes in toxic-metabolic encephalopathy
Paul M. Chen , Saskia S. Schuurmans Stekhoven , Hiba Haider , Jin Jing , Wendong Ge , Eric Rosenthal , M. Brandon Westover , Sahar F. Zafar
Published: July 10, 2026. Version: 1.0.0
When using this resource, please cite:
(show more options)
Chen, P. M., Schuurmans Stekhoven, S. S., Haider, H., Jing, J., Ge, W., Rosenthal, E., Westover, M. B., & Zafar, S. F. (2026). Epileptiform activity and outcomes in toxic-metabolic encephalopathy (version 1.0.0). Brain Data Science Platform. https://doi.org/10.60508/mfcz-ay53.
Abstract
Objective. To determine whether ictal-interictal continuum (IIC) EEG patterns and seizures are associated with outcomes in patients with toxic-metabolic encephalopathy (TME).
Methods. Retrospective analysis of continuous-EEG recordings and records from 121 Massachusetts General Hospital inpatients (2012-2017) with TME (excluding cardiac arrest/anoxic injury). IIC/seizure burden was related to modified Rankin Scale (mRS) outcomes and mortality with univariate and multivariable regression.
Results. IIC patterns were present in 100 of 121 patients (82.6%). Patients with IIC patterns were more often female and had higher APACHE-II severity scores. The project releases the de-identified analysis data and reproduction code; the committed code reproduces the paper's Table 1 exactly.
Conclusion. IIC patterns are common in TME and track with illness severity. This project provides the de-identified data and analysis code for reproducibility.
Background
Ictal-interictal continuum (IIC) EEG patterns are common in critically ill patients, but their prognostic significance in toxic-metabolic encephalopathy (TME) is unclear. This project releases the de-identified data and code examining IIC patterns and outcomes in TME.Software Description
One de-identified analysis table (TME_IIC_coding_deid.csv; 121 patients, surrogate SID; IIC/seizure coding, clinical covariates, mRS/mortality outcomes; no PHI) plus a Python reproduction script that regenerates the paper's Table 1 exactly. Raw medication dumps and MRN-bearing files are excluded.Technical Implementation
Retrospective analysis of cEEG from 121 MGH inpatients (2012-2017) with TME, excluding anoxic injury. IIC/seizure presence and burden were coded and related to modified Rankin Scale and mortality using univariate and multivariable regression. IIC presence follows the paper's minimum-GRDA definition.Installation and Requirements
Python 3.9+ with pandas, numpy, scipy, statsmodels (pip install -r requirements.txt). Run python reproduce.py. See REPRODUCE.md and DATA_SOURCE.md.Usage Notes
reproduce.py regenerates Table 1 (N=121; IIC 100/82.6%; female 66; APACHE 21 vs 16; age 64.0+-15.95) matching the paper exactly, and fits the IIC-burden vs mortality model. The original R analysis script was not preserved; this is a clean re-implementation.Release Notes
First public release: de-identified data + reproduction code (reproduces Table 1 exactly).Ethics
De-identified data collected under IRB approval at Massachusetts General Hospital; MRNs removed, no dates or names released.Acknowledgements
MGH Department of Neurology.Conflicts of Interest
See the associated publication (Crit Care Explor 2023;5:e0913).Access
Access Policy:
Only credentialed users who sign the DUA can access the files.
License (for files):
BDSP Credentialed Health Data License 1.5.0
Data Use Agreement:
BDSP Credentialed Health Data Use Agreement
Required training:
Discovery
Corresponding Author
Files
- be a credentialed user
- sign the data use agreement for the project