How reliable is ictal duration to differentiate psychogenic nonepileptic seizures from epileptic seizures?

We sought to investigate (1) differences in ictal duration between psychogenic nonepileptic seizures (PNES) and epileptic seizures (ES), (2) the odds of being PNES when seizures last ≥5min, and (3) the value of ictal duration as a diagnostic test to differentiate PNES from ES. We retrospectively reviewed video-EEG recordings and tabulated ictal durations of all PNES and ES. We estimated the mean ictal durations of PNES and ES using linear mixed models. The odds of being PNES when seizures last ≥5min were estimated using logistic regression. We used receiver operating characteristics (ROC) curves to study the overall diagnostic accuracy of ictal duration in differentiating PNES from ES. We studied 441 ES and 341 PNES recorded from 138 patients. The mean ictal duration of PNES (148.7s, 95% CI: 115.2-191.8) was significantly longer (p<0.001) than that of ES (47.7s, 95% CI: 37.6-60.6). The odds of being PNES was about 24 times higher (Odds ratio: 23.8, 95% CI: 7.9-71.3) when the ictal duration was ≥5min. The ROC curve yielded an area under the curve of 0.80 (95% CI 0.73-0.88). Youden's index identified 123.5s as the optimal threshold to diagnose PNES with 65% sensitivity and 93% specificity. Our results indicate that ictal duration is a useful test to raise suspicion of PNES. When a seizure lasts ≥5min, it is 24 times more likely to be PNES with the potential risk of misdiagnosis as status epilepticus.