The diagnostic accuracy of video electroencephalography without event capture.

Knox A, Arya R, Horn PS, Holland K. The Diagnostic Accuracy of Video Electroencephalography Without Event Capture. Pediatr Neurol. 2018 Feb;79:8-13.

Abstract

OBJECTIVE:

The aim of this study was to quantify the accuracy of 24-hour video electroencephalography (vEEG) for the diagnosis of epilepsy when a patient's typical paroxysmal event was not captured (no-event vEEG).

METHODS:

We performed a retrospective chart review of all first-time 24 hour no-event vEEG studies at Cincinnati Children's Hospital Medical Center. Clinician diagnosis of epilepsy with a minimum of one year follow-up was used as the reference standard to calculate diagnostic accuracy. Sensitivity and specificity of routine EEG (rEEG) and vEEG were compared in patients with both studies, and factors affecting the accuracy of vEEG were explored with a multivariable analysis.

RESULTS:

No-event vEEG showed sensitivity of 0.54 (95% confidence interval [CI] 0.44 to 0.64) and specificity of 0.88 (95% CI 0.84 to 0.92) respectively, with a diagnostic odds ratio of 7.53 (95% CI 4.45 to 12.76). The sensitivity of vEEG was statistically superior to that of rEEG, whereas specificity was comparable. Age emerged as the only factor that affected the diagnostic accuracy of no-event vEEG.

CONCLUSION:

Even in the absence of a typical seizure or spell, video EEG is a useful test for predicting or excluding epilepsy, with diagnostic accuracy that is superior to rEEG and unaffected by the presence of a chronic neurological condition.

____________________________________________________________________

From the article

No-event vEEG studies are a common diagnostic dilemma encountered by neurologists. This study showed that even if an event in question is not captured, vEEG provides useful diagnostic information; individuals with epileptiform discharges on no-event vEEG are 7.5 times more likely to be diagnosed with epilepsy compared with those with a normal vEEG. No-event vEEG has greater diagnostic accuracy than rEEG, with significantly superior sensitivity and comparable specificity. It also has good positive and negative predictive value; in this cohort, those with normal no-event vEEG had only a 17% chance of subsequently being diagnosed with epilepsy.

The diagnostic accuracy of no-event vEEG did not differ substantially in patients with chronic neurological conditions, regardless of whether patients were stratified by history of developmental delay, abnormal neurological examination, or abnormal imaging. The only factor affecting diagnostic accuracy was patient age with improved diagnostic accuracy for older patients, consistent with previous studies showing better diagnostic accuracy of rEEG in adults.   The pathophysiological basis for this observation is unknown, but we speculate that the association between interictal epileptiform discharges and the diagnosis of epilepsy is less certain in children, given the higher prevalence of epileptiform EEG traits of unknown clinical significance such as central-temporal spikes as well as other maturational issues…

Results were likely affected by selection bias, as patients with electro-clinical syndromes often detected on rEEG (such as childhood absence epilepsy) were inadvertently excluded from our cohort. We believe this explains why the estimate of sensitivity in this study was lower than the pooled sensitivity reported in other studies (0.31 vs 0.58)...

In summary, vEEG is a useful clinical test to aid diagnosis of epilepsy in children. The data presented here fill an important knowledge gap and has practical implications for management of patients with events concerning for seizure. Consistent with other studies,  we found over 50% of vEEGs captured typical events. Because vEEG monitoring after rEEG was found to improve diagnostic accuracy even if an event is not captured, this study provides evidence for the pragmatic clinical practice of ordering vEEG monitoring rather than serial rEEGs  after an initial inconclusive rEEG. If no event is captured, the vEEG study is normal, and the clinical suspicion for epilepsy is low, epilepsy may be excluded with reasonable certainty. This strategy may help facilitate a more prompt and accurate diagnosis of epilepsy versus other paroxysmal disorders.