OBJECTIVE: To evaluate interrater agreement in categorizing treatment outcomes and drug responsiveness status according to the International League Against Epilepsy (ILAE) definition of drug-resistant epilepsy.
METHODS: A total of 1053 adults with focal epilepsy considered by the investigators to meet ILAE criteria for drug resistance were enrolled consecutively at 43 centers and followed up prospectively for 18-34 months. Treatment outcomes for all antiepileptic drugs (AEDs) used up to enrollment (retrospective assessment), and on an AED newly introduced at enrollment, were categorized by individual investigators and by 2 rotating members of a 16-member expert panel (EP) that reviewed the patient records independently. Interrater agreement was tested by Cohen's kappa (k) statistics and rated according to Landis and Koch's criteria.
RESULTS: Agreement between EP members in categorizing outcomes on the newly introduced AED was almost perfect (90.1%, k = 0.84, 95% confidence interval [CI] 0.80-0.87), whereas agreement between the EP and individual investigators was moderate (70.4%, k = 0.57, 95% CI 0.53-0.61). Similarly, categorization of outcomes on previously used AEDs was almost perfect between EP members (91.7%, k = 0.83, 95% CI 0.81-0.84) and moderate between the EP and investigators (68.2%, k = 0.50, 95% CI 0.48-0.52). Disagreement was related predominantly to outcomes considered to be treatment failures by the investigators but categorized as undetermined by the EP. Overall, 19% of patients classified as having drug-resistant epilepsy by the investigators were considered by the EP to have "undefined responsiveness."
SIGNIFICANCE: Interrater agreement in categorizing treatment outcomes according to ILAE criteria ranges from moderate to almost perfect. Nearly 1 in 5 patients considered by enrolling neurologists to be "drug-resistant" were classified by the EP as having "undefined responsiveness."