Clinical and video-EEG features associated with drug resistance in childhood and juvenile absence epilepsy: a single-centre retrospective cohort study

Drug resistance occurs in a clinically important subgroup of children with childhood absence epilepsy (CAE) or juvenile absence epilepsy (JAE), but early risk stratification remains difficult. We investigated clinical and video-electroencephalography (video-EEG) features associated with drug resistance and assessed the internal validity of an exploratory three-variable model. In this single-centre retrospective cohort, we analysed 61 children with CAE or JAE treated at the Affiliated Hospital of Qingdao University between January 2022 and January 2025. Drug resistance was defined as failure of adequate trials of two tolerated and appropriately chosen antiseizure medication schedules. Candidate variables included epilepsy syndrome, pretreatment seizure burden and quantitative video-EEG measures. Exploratory logistic-regression models were developed, and discrimination was quantified using the area under the receiver operating characteristic curve (AUC). The final combined model was internally validated with 2,000 bootstrap resamples to estimate optimism-corrected AUC, calibration intercept and slope, and Brier score. Of 61 children, 41 had a favourable seizure outcome and 20 met the criteria for drug resistance. Drug-resistant patients had higher pretreatment seizure frequency and greater interictal sleep generalized spike-wave discharge (GSWD) density. In the adjusted exploratory model, JAE versus CAE (odds ratio [OR] 10.219, 95% confidence interval [CI] 1.499–69.657), log-transformed pretreatment seizure frequency (OR 2.002, 95% CI 1.105–3.628) and log-transformed interictal sleep GSWD density (OR 13.793, 95% CI 1.846-103.056) remained associated with drug resistance; the wide CIs indicated substantial uncertainty. The apparent AUC was 0.868. After correction for bootstrap optimism of 0.025, the AUC was 0.843. The optimism-corrected calibration intercept and slope were − 0.110 and 0.784, respectively, and the optimism-corrected Brier score was 0.143. In this single-centre cohort, JAE syndrome, higher pretreatment seizure frequency and greater interictal sleep GSWD density were associated with drug resistance. Internal validation showed overfitting; the model should therefore be considered hypothesis-generating until it has undergone independent validation.