Status of presurgical evaluation among patients with drug-resistant epilepsy identified with a computable electronic...

Drug-resistant epilepsy (DRE) is associated with increased injury risk, cognitive decline, psychiatric illness, and premature death. Epilepsy surgery can be curative among well-selected individuals but remains underutilized. This study sought to identify people living with DRE in electronic health record (EHR) data and determine factors associated with initiation of presurgical evaluation.

Using a computable phenotypic algorithm, we identified people with probable DRE and an encounter in our medical system's EHR between 4/1/2020 and 6/1/2022. We randomly sampled 200 people for manual chart abstraction by two independent reviewers. People with confirmed DRE were classified according to stage in the presurgical evaluation care pathway. Demographic and clinical variables were tested for association with initiation of presurgical evaluation.

The algorithm identified 3,027 people with probable DRE. Among 200 randomly sampled people, 87.5% (n = 175) had epilepsy, 42% (n = 84) had DRE, and 11.5% (n = 23) had epilepsy with undefined drug responsiveness. Among those with DRE, 57.1% (n = 48) had not initiated presurgical evaluation. Presurgical evaluation was associated with co-morbid mood disorder (OR = 3.88, 95% CI = 1.5-10.3, p = 0.007), shorter median time since last epilepsy-related visit (2.40 months, IQR: 0.72-6.60 vs 6.96 months, IQR: 3.00-21.84, p = 0.003) and 2nd to last epilepsy visit (8.76 months, IQR: 4.78-15.36 vs 13.08 months and IQR: 7.32-28.32, p = 0.008), and tracking by a surgical coordinator (OR = 46.00, 95% CI = 9.5-222.5, p < 0.001). Unknown MRI classification (OR = 0.04, 95% CI = 0.0-0.4, p = 0.001) and generalized seizures (OR = 0.04, 95% CI = 0.0-0.29, p < 0.001) were associated with lower odds of evaluation.

An EHR algorithm can identify people with DRE and undefined drug responsiveness with potentially modifiable gaps in care.