Abstracts

Rationale:
Triphasic waves (TW) are known markers of encephalopathy on EEG and are associated with a variety of metabolic, infectious, and drug related etiologies. In particular the cephalosporin antibiotic cefepime is thought to lead to an increased risk of triphasic waves. We tested these associations using a combined dataset of EEG reports, laboratory values, and medication records. 

Methods:
We leveraged a dataset consisting of all text reports of EEGs performed at the University of Chicago between September 2012 and April 2023 (n = 23692), and all laboratory tests and EKGs performed on and medications ordered for the same patients during that time. Multivariate and univariate logistic regressions in Matlab were performed to evaluate associations between cefepime administration, laboratory values or EKG metrics obtained within 2 days before or after the date of an EEG report, and the presence of TW on EEG reports.

Results:
A multivariate logistic regression with BMP and CBC values (19 different lab values, 7968 values for each lab) as predictors and the presence of TW on EEG as the dependent variable found significant associations between TW presence and anion gap, HCO3, chloride,  eGFR, glucose, hematocrit, and RDW. Additionally, 295 lab values were used to conduct univariate logistic regressions with the individual lab values as the dependent variable and TW presence data as the independent variable. After correcting p-values for multiple comparisons, 31 different labs had significant (p < 0.05) associations, primarily labs reflecting renal or hepatic function, though non-metabolic variables such as ventricular rate on EKG and absolute lymphocyte count were also significantly predictive. Finally, a univariate logistic regression between TW presence and if the patient had received cefepime for at least three continuous days prior to the EEG as the independent variable was also markedly significant (β = 1.45, p = 1.64 x 10