Abstracts

Rationale: Long-term continuous video EEG monitoring in the Epilepsy Monitoring Unit (EMU), an essential clinical tool for the diagnosis of patients with spells, is a limited clinical resource. Knowledge of the predicting factors for length of stay (LOS) in the EMU may allow clinicians and administrators to more efficiently utilize bed space in the EMU. Methods: The records for all consecutive admissions to the EMU at the University of Colorado Hospital between December 1, 2010 and May 31, 2011 (n= 142) were retrospectively reviewed. Data on patient demographics, seizure types and frequencies, antiepileptic therapies, comorbidities, EMU LOS, time to first diagnostic event and event descriptions were abstracted and analyzed. Patients younger than 18 or older than 80 years, patients with intracranial electrodes, and patients admitted to the EMU for reasons other than spell capture and characterization were not included. Results: Data from 114 patients were ultimately included in the analysis. The mean LOS for all patients was 83.52 +/- 4.24 hours. The mean time to first diagnostic event in those patients who had events was 30.93 +/- 3.37 hours. Overall, patients with nonepileptic seizures had a statistically significantly shorter mean LOS than those patients with epileptic seizures or nondiagnostic studies. Patients with a normal EEG background had a statistically significant shorter mean LOS than patients with abnormal EEG backgrounds (slowing or epileptiform discharges). EMU LOS (in hours) was not significantly correlated with patient age, age at event onset, number of event types, or number of antiepileptic drugs at admission. EMU LOS (in hours) was modestly, but significantly, correlated with self-reported event frequency prior to admission (r= -0.350, p=0.0003). When analyzed separately, those patients that were diagnosed with nonepileptic seizures in the EMU (n= 43) also showed a statistically significant modest correlation between LOS and self-reported event frequency (r= -0.436, p= 0.0055), while the correlation was not significant for those patients diagnosed with epileptic seizures (n= 35). Conclusions: Pre-admission clinical features, such as self-reported event frequency, may predict EMU LOS. After confirmation with further prospective analyses, these factors could be used to schedule EMU admissions for maximum resource utilization.