Predicting outcome of epilepsy surgery in clinical practice: Prediction models vs. clinical acumen. (March 2020)

• Record Type: Journal Article
• Authors: Fassin, Anne Katharina; Knake, Susanne; Strzelczyk, Adam; Josephson, Colin B.; Reif, Philipp S.; Haag, Anja; Carl, Barbara; Hermsen, Anke M.; Gorny, Iris; Möller, Leona; Pagenstecher, Axel; Nimsky, Christopher; Bauer, Sebastian; Sure, Ulrich; Menzler, Katja; Rosenow, Felix; Klein, Karl Martin
• Is Part Of: Seizure. Volume 76(2020)
• Page Start: 79
• Page End: 83
• Publication Date: 2020-03
• Subjects: Epilepsy -- Epilepsy surgery -- Presurgical evaluation -- Outcome prediction -- Prediction models; Epilepsy -- Periodicals; Epilepsy -- Periodicals; Seizures -- Periodicals; Épilepsie -- Périodiques; Electronic journals; Electronic journals; 616.853
• DOI: 10.1016/j.seizure.2020.01.016
• Languages: English
• ISSNs: 1059-1311

Abstract

Highlights: Preoperative estimates corresponded well to observed outcome after epilepsy surgery. Patient counselling about expected outcome in clinical practice was appropriate. The use of prediction models did not improve the accuracy of clinical predictions. Abstract: Purpose: Epilepsy surgery is an evidence-based treatment for drug-refractory focal epilepsy. We aimed to evaluate how well preoperative outcome estimates of epilepsy surgery in clinical practice correlated with postoperative outcome and to compare prediction by the clinical team with available scores (m-SFS, ESN). Method: Retrospective cohort study including patients with drug-refractory focal epilepsy who underwent resective epilepsy surgery at Epilepsy Center Hessen, Marburg, between 1998-2016. Patients were categorized into four groups based on their estimated chance of postoperative seizure freedom documented in preoperative medical records. Variables required for calculation of m-SFS and ESN were also extracted from presurgical medical records. Seizure outcome using Engel/ILAE classifications was extracted from postoperative medical records. Results: 148 patients were included and 98 had follow-up at 5 years. 69 (70%) had Engel I and 50 (51%) ILAE 1 outcome. Observed 5-year outcome for very good candidates was 20/22 (91%) Engel I and 14/22 (64%) ILAE 1, for good candidates 29/40 (73%) Engel I and 21/40 (53%) ILAE 1, for candidates with slightly reduced chance 11/18 (61%) Engel I and 9/18 (50%) ILAE 1 and for candidates with considerably reduced chance 1/5 (20%) Engel I and 1/5 (20%) ILAE 1.There were no significant differences in discrimination or overall performance between predictions by the clinical team, ESN and m-SFS. Conclusions: Preoperative outcome estimates corresponded well with observed outcome indicating adequate patient counseling.