Abstracts

Rationale: Continuous EEG (cEEG) is the gold standard for detection of seizures in patients with cerebral insults. There is higher risk of seizures in the presence of altered mental status (AMS) or coma. However, there is no widely accepted standard practice for patient selection for cEEG. Application of cEEG is likely widely discrepant across providers and medical centers. This study evaluated the relationship between diagnosis and utilization of cEEG at a large, academic medical center.

Methods: This retrospective study was performed with approval by the IRB. Patients were included if they had an inpatient diagnosis with high risk of seizure and were admitted between July 2018 and December 2019. Patients were excluded if their code status was “allow a natural death.” Diagnoses (ICD codes) were divided into several categories: vascular, inflammatory, infectious, tumor, seizure, trauma and other, with each ICD code treated as a separate case. Cases were also stratified by altered mental status (AMS). AMS was determined using the Confusion Assessment Method (CAM) score, in which a positive score indicates AMS. AMS per admission was categorized as probable (3 positive CAM scores within 48 hours), possible (any positive CAM), or definitely not altered (only negative CAMs). The primary outcome variable was utilization of cEEG. Additional variables included age, race, gender, and mechanical ventilation. Chi-square tests were used to examine differences in the proportion of cEEG utilization across diagnostic categories and AMS strata, with Bonferroni correction for multiple comparisons. Individual predictors of cEEG were evaluated using logistic regression, applied to variables with significance in bivariate analysis.

Results: 3,521 patients met criteria (48.4% female, age mean 60.2 +/- 17.9 years), resulting in 4026 ICD code cases, of which 596 had cEEG (14.8%). Inflammatory diagnoses were associated with a higher rate of cEEG, whereas tumor was associated with a lower rate (ps < .05). In cases with not altered or probable AMS, inflammatory diagnoses were associated with higher rate of cEEG (ps < .05). There were no differences in use of cEEG when compared by AMS level (p=.169). Logistic regression indicated that the estimated odds of cEEG was greater with mechanical ventilation [odds ratio (OR)=5.7; 95% confidence interval (CI)=4.7, 6.8; p < .001], having an infectious (OR=2.1; 95% CI=1.2, 3.5; p=.008), inflammatory (OR=0.54, 95% CI=0.41, 0.71; p < .001), or seizure (OR=2.1, 95% CI=1.2, 3.5; p=.008) diagnosis. The estimated odds of receiving cEEG was less for a tumor diagnosis (OR=0.54, 95% CI=0.41, 0.71; p < .001).