Optimizing drug-resistant epilepsy identification in the Veterans Health Administration.

Area of Science:

• Neurology
• Health Informatics
• Clinical Research

Background:

• Accurate identification of drug-resistant epilepsy (DRE) is essential for patient care and research.
• Previous methods for defining DRE in administrative data using 2010 International League against Epilepsy (ILAE) criteria were complex.
• Developing reliable DRE identification methods is critical for improving patient outcomes.

Purpose of the Study:

• To develop and validate an algorithm for identifying DRE in administrative data.
• To improve the accuracy of DRE measurement in large patient cohorts.
• To refine clinical interventions for epilepsy management.

Main Methods:

• Retrospective analysis of national administrative data from the Veterans Health Administration (VHA).
• Chart reviews by epileptologists using 2010 ILAE criteria to confirm DRE.
• Logistic regression analysis to develop and optimize algorithms for DRE identification.

Main Results:

• An optimal algorithm identified DRE based on the number and duration of anti-seizure medications (ASMs) and intractable epilepsy ICD codes.
• The best algorithm achieved high accuracy (F1 score=0.726), sensitivity (0.74), specificity (0.81), and AUC (0.78).
• Factors associated with DRE included age, ASM use, EEG/MRI procedures, and intractable epilepsy codes.

Conclusions:

• The developed algorithm effectively identifies DRE in administrative data, similar to prior work but with a unique combination of factors.
• Fine-tuning algorithms for specific healthcare settings is important for optimal performance.
• Further validation in diverse cohorts is necessary to assess the algorithm's broader applicability.