The SeLECT score is useful to predict post-stroke epilepsy

We read with interest Josef Finsterer's Comment1Finsterer J The SeLECT score is inappropriate to predict post-stroke epilepsy.Lancet Neurol. 2018; 17: 106-107Summary Full Text Full Text PDF PubMed Scopus (7) Google Scholar on our Article2Galovic M Döhler N Erdélyi-Canavese B et al.Prediction of late seizures after ischaemic stroke with a novel prognostic model (the SeLECT score): a multivariable prediction model development and validation study.Lancet Neurol. 2018; 17: 143-152Summary Full Text Full Text PDF PubMed Scopus (100) Google Scholar on the SeLECT score as a prognostic model for prediction of late seizures after ischaemic stroke. He does not take into account three major points about prognostic modelling3Foote C Woodward M Jardine MJ Scoring risk scores: considerations before incorporating clinical risk prediction tools into your practice.Am J Kidney Dis. 2017; 69: 555-557Summary Full Text Full Text PDF PubMed Scopus (7) Google Scholar and the differences from aetiological research.4Moons KGM Royston P Vergouwe Y Grobbee DE Altman DG Prognosis and prognostic research: what, why, and how?.BMJ. 2009; 338: b375Crossref PubMed Scopus (803) Google Scholar First, for a prognostic model to be clinically useful, the entry data have to be routinely and widely available and easily applicable in the clinical setting. Second, for a model to be relevant, it must provide outcome information satisfying the needs of the user at, or soon after, the event. For these purposes, data on whether other conditions or complications are present, what the seizure types and frequency would be, or how the epilepsy would be eventually treated are therefore extraneous. Last, for a model to be robust, it has to foretell outcomes with adequate accuracy and precision in different populations. The SeLECT score achieves these goals using only a few well defined parameters from a “plethora of possible factors”1Finsterer J The SeLECT score is inappropriate to predict post-stroke epilepsy.Lancet Neurol. 2018; 17: 106-107Summary Full Text Full Text PDF PubMed Scopus (7) Google Scholar and by being externally validated in three cohorts from different countries, minimising selection bias. In response to Finsterer's specific concerns, first, most of the issues he raised1Finsterer J The SeLECT score is inappropriate to predict post-stroke epilepsy.Lancet Neurol. 2018; 17: 106-107Summary Full Text Full Text PDF PubMed Scopus (7) Google Scholar were addressed either in our Article or in its online appendix.2Galovic M Döhler N Erdélyi-Canavese B et al.Prediction of late seizures after ischaemic stroke with a novel prognostic model (the SeLECT score): a multivariable prediction model development and validation study.Lancet Neurol. 2018; 17: 143-152Summary Full Text Full Text PDF PubMed Scopus (100) Google Scholar Data on seizure types, frequency of recurrent attacks, and antiepileptic drug treatment after early seizures are provided in table 1 of the Article.2Galovic M Döhler N Erdélyi-Canavese B et al.Prediction of late seizures after ischaemic stroke with a novel prognostic model (the SeLECT score): a multivariable prediction model development and validation study.Lancet Neurol. 2018; 17: 143-152Summary Full Text Full Text PDF PubMed Scopus (100) Google Scholar We are puzzled by his notion of “[u]nprecise inclusion and exclusion criteria”1Finsterer J The SeLECT score is inappropriate to predict post-stroke epilepsy.Lancet Neurol. 2018; 17: 106-107Summary Full Text Full Text PDF PubMed Scopus (7) Google Scholar because these are given in detail in the Methods section. This section also specifies the definition of white matter hyperintensities. We report that antiepileptic drugs administered for indications other than epilepsy (eg, for neuropathic pain or psychiatric conditions) did not influence our results because none of the stroke survivors included in the validation cohorts received such treatment (see Article2Galovic M Döhler N Erdélyi-Canavese B et al.Prediction of late seizures after ischaemic stroke with a novel prognostic model (the SeLECT score): a multivariable prediction model development and validation study.Lancet Neurol. 2018; 17: 143-152Summary Full Text Full Text PDF PubMed Scopus (100) Google Scholar appendix p 6). The appendix also describes that no association was found between a positive family history for epilepsy and the risk of late seizures after stroke (see Article2Galovic M Döhler N Erdélyi-Canavese B et al.Prediction of late seizures after ischaemic stroke with a novel prognostic model (the SeLECT score): a multivariable prediction model development and validation study.Lancet Neurol. 2018; 17: 143-152Summary Full Text Full Text PDF PubMed Scopus (100) Google Scholar appendix p 11). Second, seizures due to hyponatraemia would not have influenced our results because we only considered spontaneous unprovoked seizures after stroke as late seizure and excluded those potentially provoked by hyponatraemia. Similarly, we mitigated the risk that seizures would be caused by a pre-existing brain insult and not by the index stroke itself by excluding participants with previous brain lesions or epileptogenic comorbidities. Third, only the German validation cohort (n=311) relied exclusively on telephonic follow-up. The other two validation cohorts (Austrian and Italian, n=858 combined) did regular face-to-face follow-up interviews with a neurologist—ie, the gold standard to diagnose seizures. The Swiss derivation cohort (n=1200) relied on a combined approach of screening participants with a validated telephonic questionnaire that, if answered positively, triggered a face-to-face neurological consultation. SeLECT performed well in all cohorts, including those with face-to-face follow-up, lending support to a good reliability and generalisability of this model. Fourth, we agree that distinguishing seizures from mimics can be difficult even for the experienced neurologist.5Crompton DE Berkovic SF The borderland of epilepsy: clinical and molecular features of phenomena that mimic epileptic seizures.Lancet Neurol. 2009; 8: 370-381Summary Full Text Full Text PDF PubMed Scopus (67) Google Scholar This limitation is inherent to most epilepsy studies, except perhaps those using continuous video-electroencephalographic monitoring to diagnose seizures. We did not, however, count indeterminate events as seizures in our study. Last, we agree with Finsterer that it would be interesting to have genetic and polysomnographic data for our participants. This would, however, require instrumental testing in a large set of participants (n=2369), which was beyond the scope of our study. Therefore, we believe Finsterer's critique is unsubstantiated. The strength of our prognostic model is that it can transform a person's complex individual characteristics into five easily ascertainable parameters that can be used to accurately predict the risk of post-stroke seizures. In other words, the SeLECT score is simple but not simplistic, and there is a robust rationale for the variables included in the final model (see Article2Galovic M Döhler N Erdélyi-Canavese B et al.Prediction of late seizures after ischaemic stroke with a novel prognostic model (the SeLECT score): a multivariable prediction model development and validation study.Lancet Neurol. 2018; 17: 143-152Summary Full Text Full Text PDF PubMed Scopus (100) Google Scholar appendix p 13). Triple external validation showed that this model can be used to predict with adequate accuracy late seizures after stroke. The SeLECT score fills an urgent gap for an evidence-based prognostic tool that allows better prediction of new-onset epilepsy following an ischaemic stroke. MRK reports grants and personal fees from UCB, and personal fees from Sage Therapeutics and Novartis, outside of the submitted work. JSD is supported by the UK National Institute for Health Research University College London Hospitals and University College London Biomedical Research Centre as a senior investigator. JWS reports grants and personal fees from Eisai and UCB; grants from GlaxoSmithKline, WHO, and Netherland Epilepsy Funds; and personal fees from Lundbeck and Teva, outside of the submitted work. JWS's current position is endowed by the Epilepsy Society; he is a member of the editorial board of The Lancet Neurology and receives research support from the Marvin Weil Epilepsy Research Fund. All other authors declare no competing interests. The SeLECT score is inappropriate to predict post-stroke epilepsyCauses of post-stroke epilepsy are manifold and not just restricted to the stroke itself (primary post-stroke epilepsy). These causes range from premorbid conditions already present before the stroke, to morbidity that develops after the stroke, either as a complication of, or independent of, the stroke. Seizures associated with a pre-existing disposition (eg, meningitis, head trauma, or encephalopathy caused by, for example, diabetes or hypertension) can be triggered by stroke or by co-pathologies developing after a stroke (secondary post-stroke epilepsy). Full-Text PDF Prediction of late seizures after ischaemic stroke with a novel prognostic model (the SeLECT score): a multivariable prediction model development and validation studyThis easily applied instrument was shown to be a good predictor of the risk of late seizures after stroke in three external validation cohorts and is freely available as a smartphone app. The SeLECT score has the potential to identify individuals at high risk of seizures and is a step towards more personalised medicine. It can inform the selection of an enriched population for antiepileptogenic treatment trials and will guide the recruitment for biomarker studies of epileptogenesis. Full-Text PDF

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.