Detection of poststroke epilepsy and accompanying anti-seizure treatment using Swedish national registers – a validation study
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Introduction Stroke is the most common cause of epilepsy, causing an estimated 14-21% of all epilepsy cases. Unfortunately, there is limited knowledge regarding the optimal treatment for post-stroke epilepsy (PSE) and there has been done few studies on PSE because of the difficulty in following elderly patients with multimorbidity over time; epileptic treatments can continue for years and many studies on the field has only evaluated 12 months. Swedish National Patient Register (NPR) can give unique possibilities in following large patient cohorts over a long term, but to know whether certain conclusions from register studies are accurate, it is important to validate the registers. Aim The goal of this study was to evaluate the accuracy of register-based criteria associated with post-stroke epilepsy in the Swedish NPR. Methods This was a retrospective validation study with a total study population of 177 deceased individuals who have been identified by The Swedish National Board of Health and Welfare that fulfiled register-based criteria of PSE. Medical records from these patients were examined thoroughly to see if stroke and subsequent epilepsy could be confirmed, and the collected data were filled into protocols. The data were analysed and positive predictive value (PPV) was used as a method to validate the outcome, calculated as a ratio between the number of patients who had likely or definite epilepsy after stroke to the total number of patients. Results From the 177 patients who were identified, a total of 133 (PPV: 75%, 95%CI 0.69-0.82) had epileptic seizures after stroke. 105 out of 133 epileptic patients were prescribed anti-seizure medication (ASM). Conclusions/implications The results of the study are informative for future register-based studies on PSE. Compared to another recent study validating epilepsy codes in several countries, this result can be seen as high: in our study, at least 75% of the identified patients actually had PSE. Considering the risk of misdiagnosis in epilepsy, register studies will always identify some patients who do not meet the diagnostic criteria for PSE. The accuracy of the register codes may be further increased by requiring that the patients had received an anti-seizure medication, or to only identify patients diagnosed with epileptic seizures; our study also included unspecified seizures which identified some patients who had experienced seizures unrelated to epilepsy.