Published in:
01-01-2017 | Original Research Article
Anesthetics and Outcome in Status Epilepticus: A Matched Two-Center Cohort Study
Published in: CNS Drugs | Issue 1/2017
Login to get accessAbstract
Background
The use of anesthetics has been linked to poor outcome in patients with status epilepticus (SE). This association, however, may be confounded, as anesthetics are mostly administered in patients with more severe SE and critical illnesses.
Objective
To minimize treatment-selection bias, we assessed the association between continuously administered intravenous anesthetic drugs (IVADs) and outcome in SE patients by a matched two-center study design.
Methods
This cohort study was performed at the Johns Hopkins Bayview Medical Center, Baltimore, MD, USA and the University Hospital Basel, Basel, Switzerland. All consecutive adult SE patients from 2005 to 2013 were included. Odds ratios (ORs) for death and unfavorable outcome (Glasgow Outcome Score [GOS] 1–3) associated with administration of IVADs were calculated. To account for confounding by known outcome determinants (age, level of consciousness, worst seizure type, acute/fatal etiology, mechanical ventilation, and SE duration), propensity score matching and coarsened exact matching were performed in addition to multivariable regression models.
Results
Among 406 consecutive patients, 139 (34.2%) were treated with IVADs. Logistic regression analyses of the unmatched and matched cohorts revealed increased odds for death and unfavorable outcome in survivors who had received IVADs (unmatched: ORdeath = 3.13, 95% confidence interval [CI] 1.47–6.60 and ORGOS1–3 = 2.51, 95% CI 1.37–4.60; propensity score matched: ORdeath = 3.29, 95% CI 1.35–8.05 and ORGOS1–3 = 2.27, 95% CI 1.02–5.06; coarsened exact matched: ORdeath = 2.19, 95% CI 1.27–3.78 and ORGOS1–3 = 3.94, 95% CI 2.12–7.32).
Conclusion
The use of IVADs in SE is associated with death and unfavorable outcome in survivors independent of known confounders and using different statistical approaches. Randomized trials are needed to determine if these associations are biased by outcome predictors not yet identified and hence not accounted for in this study.