Abstracts

Rationale: Seizure outcome is central to the assessment of overall outcome following epilepsy surgery, vital to adequately informing patients. Outcome is commonly classified based upon modifications of the Engel (Classes I-IV) or International League Against Epilepsy (ILAE; Classes 1-6) systems. There is, however, a lack of data assessing inter-rater reliability for either or any comparison of them. In addition, there are few studies of outcome for resective surgery or vagal nerve stimulation (VNS) in children, with ongoing controversy about whether either are appropriate outcome measures in this age group. Methods: Using Cohen's kappa (k), we evaluated inter-rater reliability of 2 surgical trainees (not involved in decision for or conduct of surgery) for the Engel and ILAE systems when assessing the outcome of all patients with a minimum of 12 months follow-up after undergoing surgery for medically intractable seizures at the Wessex Neurological Centre between 2001 and June 2009.Results: 102 patients fulfilled the inclusion criteria (55 male; 51 children <18 years). Median age at surgery was 24.6 (range 1-62) years with median follow-up 29 (12-92) months. 75 patients underwent resective surgery (51 adults; 24 children): 35 with hippocampal sclerosis, 8 cortical dysplasia, 17 tumours, 5 cavernous haemangioma, 3 dual pathology (hippocampal sclerosis + ganglioglioma, temporal cortex sclerosis or subacute infarct), 1 tuberous sclerosis, 1 arterio-venous malformation; in 5 no histopathological diagnosis was made. 26 underwent VNS implants and 1 disconnection (all children). At last follow-up, Observer 1 (O1) classified 88% (n=67) and Observer 2 (O2) 87% (n=66) of patients as either Engel I or II (free from or rare disabling seizures), while 83% (O1; n=63) and 80% (O2; n=61) respectively were classified as seizure free. 73% (O1; n=56) and 68% (O2; n=52) were classified as ILAE 1 or 1a. The ILAE system had a very good degree of inter-rater reliability (k0.83; 95% confidence intervals, CI, 0.74, 0.91), the highest in our study. The Engel classification system similarly had good inter-rater reliability (k0.76; 95%CI 0.66, 0.85). For pediatric resective cases, the inter-rater reliability was very good for both classifications: ILAE k0.88; 95%CI 0.68, 1 and Engel k0.82; 95%CI 0.63, 1. When assessing VNS implantation outcome, the inter-rater reliability was moderate for both classifications: ILAE (k0.53; 95%CI 0, 1) and Engel (k0.50; 95%CI 0.21, 0.78).Conclusions: The current body of literature reporting seizure outcome based upon Engel s system is likely to have an acceptable degree of inter-rater agreement in children as well as adults, while it is even better for the ILAE system. Although most outcomes using the Engel system have been classified retrospectively, the ILAE system has been advocated for prospective studies. The availability here of data on preoperative seizure days also supports its uptake for retrospective use. Assessing seizure outcome following VNS implantation using either system may have an unacceptably low degree of inter-rater agreement; further exploration of other classifications is warranted.