Multivariable Clinical Trait Model Predicts Childhood Epilepsy Risk

Share this content:
Investigators retrospectively analyzed data of 451 children who visited an outpatient department for diagnostic workup associated with ≥1 paroxysmal event.
Investigators retrospectively analyzed data of 451 children who visited an outpatient department for diagnostic workup associated with ≥1 paroxysmal event.

A multivariable prediction model composed of several different clinical characteristic variables, including sex, medical history, age of first seizure, and event description, in combination with electroencephalogram readings, may be helpful to predict the risk for epilepsy in children, according to study results published in Pediatrics.

Clinical data from 451 children who visited an outpatient pediatric neurology department after ≥1 paroxysmal event were retrospectively analyzed. Only those patients with epilepsy or an unknown diagnosis who had ≥1-year follow-up data available were included in the study. An external cohort of 187 patients (ie, validation cohort) tested the validity of the multivariable logistic regression model. Presence or absence of epilepsy comprised the primary outcome measure.

According to tests on the external validation cohort, the model's ability to discriminate was considered excellent (area under the receiver operating characteristic curve, 0.86; 95% CI, 0.80-0.92). In addition, the model featured a positive predictive value of 0.93 (95% CI, 0.83-0.97) and a negative predictive value of 0.76 (95% CI, 0.70-0.80). The area under the receiver operating characteristic curve was 0.73 (95% CI, 0.58-0.87) for model discrimination in children who presented with an uncertain diagnosis after a good initial clinical workup. The use of only electroencephalogram predictor variables (area under the curve, 0.82; 95% CI, 0.77-0.88) was superior to the use of only clinical predictor variables (area under the curve, 0.67; 95% CI, 0.59-0.74) in a submodel analysis.

Limitations of the study include its retrospective nature and the reliance of self-reported data by children or their caregivers regarding their description of clinical events, a key characteristic featured in the model.

The investigators concluded by emphasizing the need for future research that evaluates "the added value of a decision model in clinical practice to manage expectations of patients and caregivers to prompt interventions."

Reference

van Diessen E, Lamberink HJ, Otte WM, et al. A prediction model to determine childhood epilepsy after 1 or more paroxysmal events. Pediatrics. 2018;142(6):e20180931.

You must be a registered member of Neurology Advisor to post a comment.

Sign Up for Free e-newsletters



CME Focus