To characterize a critically ill cohort with status epilepticus (SE) by the illness severity scoring systems SAPS II (Simplified Acute Physiology Score II), APACHE II (Acute Physiology and Chronic Health Evaluation II), and SOFA (Sequential Organ Failure Assessment), and to compare their performance with the STESS (Status Epilepticus Severity Score) for outcome prediction.
The prospective cohort study was carried out at the University Hospital Basel, a Swiss tertiary academic medical care center. Consecutive adult SE patients hospitalized in the intensive care units from 2011 to 2016 were included. Illness severity scores and additional clinical data were recorded. Logistic regression models using automated variable selection were applied to identify scores associated with no return to functional and neurological baseline and death. Measures of discrimination and calibration were assessed.
Among 184 patients, 33% returned to baseline. Median scores of the illness severity scores were within the lowest third of the possible scoring ranges, and all differed significantly between patients with and without return to baseline. The areas under the receiver operating curves for the prediction of no return to baseline and death ranged from 0.64 to 0.73, with the highest value for the STESS predicting no return to baseline. Measures of calibration revealed adequate model fit for all analyses. Among integral components of the scoring systems, only the Glasgow Coma Scale (GCS) differed significantly between patients with and without return to baseline. In multivariable analyses, decreasing GCS and increasing STESS had the strongest associations (odds ratio [OR] = 0.84, 95% confidence interval [CI] = 0.77‐0.93 and OR = 1.34, 95% CI = 1.05‐1.68, respectively) with no return to baseline independent of all other scoring systems, whereas the APACHE II revealed the strongest association with death (OR = 1.15, 95% CI = 1.06‐1.25).
Although complex illness severity scoring systems in SE patients facilitate benchmarking and comparisons with other severely ill patient cohorts, they offer no advantages over the STESS and GCS regarding prediction of no return to baseline.