Abstracts

To investigate the reasons an emergent electroencephalogram (EmEEG) is ordered in the Intensive Care Units (ICUs) compared to the hospital Ward, examine its usefulness and find predictive variables for its results. We retrospectively identified all EEGs ordered between December 1997 and March 2002 and performed within one hour from their request in our University hospital. We compared the tests ordered by the four hospital ICUs with those ordered by the Ward and developed predictive models for the results based on clinical variables. The ICUs ordered 129 (49.4%) of all EmEEGs during the study period and the Ward 132 (50.6%) of the tests. On the requisition forms, the test was ordered to rule out status epilepticus more frequently by the ICUs (68.2% vs 52.7%, chi square test p [lt] 0.01) and to rule out seizures by the Ward (28.2% vs 17.8%, p [lt] 0.05). The Neuro-ICU ordered the test more frequently to exclude non-convulsive status than the other ICUs (OR, 95% CI 16, 3.2-79, p [lt] 0.001). Compared to non-ICU, ICU patients with head trauma or post cardiopulmonary arrest had the tests more frequently ordered (3.2, 1.2-8.4 and 17, 4-74, p [lt] 0.01) and patients with stroke less (0.3, 0.12-0.6, p =0.001). The frequency of suspicious clinical activity (subtle muscle twitching or strange dystonic posturing) or recent tonic-clonic seizure when ordering the test did not differ between ICUs and Ward. EEG findings consistent with convulsive status epilepticus and generalized slowing were found more frequently in the ICU recordings (4.8, 1.3-17, p = 0.009 and 1.7, 1.1, 2.8, p = 0.03). Normal EEG, interictal epileptiform activity or focal non-epileptic slowing were more frequently present on the Ward recordings (3.3, 1.1-10, p = 0.02, 2, 1.1-3.3, p = 0.03 and 2.5, 1.1-5, p = 0.02, respectively). In at least 12.4% of ICU patients the test was expected to lead to an anti-epileptic medication change. In multivariate logistic regression models, cardiopulmonary arrest (3, 1.2-8, p [lt] 0.05) and age (1.03, 1.003-1.05, p [lt] 0.05) were predictive of any epileptic activity found on EmEEGs in ICU patients. In our university hospital, the ICUs order EmEEG more often than the Ward to exclude status epilepticus, although based on clinical signs the suspicion level may not be higher. The Neuro-ICU, particularly, orders the test more frequently than to other ICUs to exclude non-convulsive status epilepticus. Indeed, status epilepticus is confirmed more frequently in the ICUs than the Ward by the test. Cardiopulmonary arrest and increasing age are predictors of any epileptic activity on EmEEG in ICU patients.