Distinct clinical clusters of paediatric patients with status epilepticus: Retrospective cohort study

Abstract Aim To characterize the clinical features, management, and outcomes of paediatric patients with status epilepticus, and to explore whether distinct clinical subgroups can be identified from clinical descriptions. Method This was an exploratory retrospective single‐centre cohort study of paediatric status epilepticus admissions to Switzerland's largest tertiary‐level paediatric hospital. We analysed 642 status epilepticus admissions from 467 patients (230 females; median age at first 4 years 11 months interquartile range 1 year 5 months–7 years 5 months). We applied descriptive statistics and machine‐learning approaches. A k ‐means clustering algorithm was used to identify distinct clinical subgroups, while least absolute shrinkage and selection operator regression tested whether clinical metrics could predict mortality. Results Age‐related differences in status epilepticus aetiology were observed: infants and younger children more often presented with acute symptomatic causes, whereas older children and adolescents were more likely to have pre‐existing epilepsy. Out‐of‐hospital treatment was associated with faster treatment initiation and better treatment response. Shorter status epilepticus onset to treatment latency correlated with higher response rates and reduced need for intensive care. Cluster analysis identified three clinical subgroups: (1) younger patients with acute status epilepticus associated with an infection, including febrile status epilepticus (‘febrile seizure’ cluster); (2) younger patients with acute status epilepticus and a more severe in‐hospital course (‘para‐infectious’ cluster); and (3) older patients with an established epilepsy diagnosis (‘known epilepsy’ cluster). Across the cohort, progressive epilepsy aetiology and a previous diagnosis of epilepsy were associated with increased mortality risk. Interpretation Paediatric status epilepticus comprises clinically distinct subgroups that are identifiable from routine clinical data. Such data‐driven clinical clustering may help refine risk stratification and inform clinical decision making in paediatric status epilepticus.