V29. Audit of routine EEGs performed at a German tertiary neurology hospital – Need for change in clinical practice?

Abstract

Listen

Translate article

Objective To evaluate the likelihood and kind of EEG pathology in the most common neurological conditions in which a routine EEG exam was performed at a neurological tertiary referral center to guide future clinical practice. Method Retrospective analysis of all EEGs ordered at the Department of Neurology and Neurooncology, University Hospital, Goethe-University Frankurt (Germany) in 2009; for the 70% most common main discharge ICD 10 diagnoses, the incidence of a normal or a pathological initial EEG report was assessed. Pathological reports were sub-classified according to reported generalized slow activity (“gen. slow”), focal slow activity (“focal slow”), and epileptiform activity with or without additional focal slow activity (“epil. act.”). Hence, a single EEG could fall into two categories. Continuous vs. intermittent pathology were not distinguished. Results 1491 (8% of all considered cases) of 1725 EEG reports were included in the analysis ( Figure , Table ). The Table indicates the nine main discharge diagnoses covering two thirds of all EEGs ranging from 19 to 323 EEG examinations per diagnosis and lists the proportion of pathologies by diagnosis ( Figure ) and the proportion of patients with a given diagnosis referred to EEG. Almost half of all EEGs were reported normal; in equal parts (20% each) of the remainder, generalized or focal slowing, or epileptiform activity were observed. The highest yield of pathological findings (>80%) was obtained in patients with intracranial mass or cerebral hemorrhage, the lowest (11%, all focal slow) in patients with multiple sclerosis. Patients in whom “epilepsy” or “intracranial mass” were diagnosed formed the largest cohort of patients referred to EEG and together with cerebral hemorrhage exhibited the highest incidence of epileptiform activity, while in only 6% of hemorrhage patients an EEG exam was ordered. Proportionally, given a certain diagnosis, an EEG was most frequently requested in patients eventually diagnosed with syncope, cerebral ischemia, or CNS inflammation. While 4/5 of “syncope EEGs” were normal, in each of the other two conditions, 20% of EEGs exhibited focal and another 20% generalized slow activity. Conclusion Whenever an EEG was deemed necessary, chances of a pathological finding were approximately 50% with conditions characterized by space occupying lesions taking the lead. This suggests that in such conditions EEG will likely confirm a clinical suspicion of relevant brain dysfunction likely prompting treatment modifications. In contrary, our data suggests that EEG requests could be more restricted in patients diagnosed with multiple sclerosis, Parkinson’s disease, syncope, or a psychiatric condition as 70–90% of such EEGs can be expected normal and only 2% of these to reveal epileptiform – and hence treatment relevant – activity. This is especially so, since a normal EEG is of low negative predictive value in general. In the 14% of epilepsy patients referred to EEG, focal EEG pathology was found in 42% encouraging the use of EEG when trying to confirm a focal epilepsy syndrome or even identify an irritative zone. A limitation of our study is that our data describes only those patients in whom an EEG was performed, and we do not know which circumstances triggered this investigation.