Abstracts

Rationale: There are insufficient data regarding the common problem of breakthrough seizures in the emergency department (ED) to guide standardized therapy or predict which patients will require admission.

Methods: Our prospective, observational protocol captured 58 variables for 117 patients, using variables and values defined in clinically relevant ways. A waiver of informed consent was obtained from our institutional IRB. Predictive variables (Table 1) were binary, ordinal or continuous. Preliminary univariate and multivariate analyses for admission have been completed and analysis of other outcomes are in progress. Data were summarised in terms of individual variables. Formal statistical analysis used linear logistic regression to access various factors associated with hospital admission and other outcomes.

Results: Age and gender were uniformly distributed (Table 1). Most patients were relatively well controlled at baseline.  Most seizures were brief and generalized. Few patients were severely disabled. Half were non-compliant with anti-epileptic drugs (AEDs). GCS varied widely. Treatment by emergency medical services (EMS) was relatively uncommon.  Lorazepam or midazolam were given in the ED to a third of the patients and an AED was given to two-thirds. Post-ictal symptoms were present in 2/3, usually altered consciousness.  Post-ictal duration evenly ranged from < 20 minutes to > 4 hours. Time in ED was > 4 hours for most patients.  Temperature was above normal in > 25% of patients.  Antibiotics were given when abnormal termperature was accompanied by elevated WBC. Metabolic disturbances were common but rarely significant. Head CT, performed in 59 patients, never demonstrated acute lesions.  Insurance status paralleled our hospital’s usual profile.
_x000D_ Most patients were admitted to the hospital and length of stay was > 4 days in more than a third of those admitted. Compliant patients were more likely to be admitted (Table 2). Risk factors for admission included age, GCS < 15, temperature > 37.3 °C, elevated lactate and WBC. Treatment with lorazepam or midazolam in the ED was a strong predictor of admission, but association was weaker with drugs given by EMS or with AEDs given in the ED. In preliminary multi-variate analyses, administration of lorazepam or midazolam remained a strong association with admission even after adjusting for age, GCS, non-compliance, lactate and WBC. Use of these anti-convulsants also was associated with admission to the ICU in univariate analyses._x000D_ Weaknesses of this data include its relatively small sample size and the fact that all patients presented to the ED during daytime hours; no patients presenting at night were included due to limited research coordinator availability. Strengths include the consistent clinically relevant data definitions and prospective acquisition._x000D_
Conclusions: This preliminary, prospective observational protocol provides a rich data set with which to design future randomized therapeutic trials and for developing statistical tools for predicting subsequent admissions. A larger data set would facilitate exploration of multi-variate relationships for testing hypotheses.

Funding: Albert Einstein Society, Einstein Health Network, Philadelphia, Pennsylvania