Benzodiazepine-associated delirium in critically ill adults

Irene J. Zaal, John W. Devlin, Marijn Hazelbag, Peter M. C. Klein Klouwenberg, Arendina W. van der Kooi, David S. Y. Ong, Olaf L. Cremer, Rolf H. Groenwold, Arjen J. C. Slooter

The association between benzodiazepine use and delirium risk in the ICU remains unclear. Prior investigations have failed to account for disease severity prior to delirium onset, competing events that may preclude delirium detection, other important delirium risk factors, and an adequate number of patients receiving continuous midazolam. The aim of this study was to address these limitations and evaluate the association between benzodiazepine exposure and ICU delirium occurrence. In a cohort of consecutive critically ill adults, daily mental status was classified as either awake without delirium, delirium, or coma. In a first-order Markov model, multinomial logistic regression analysis was used, which considered five possible outcomes the next day (i.e., awake without delirium, delirium, coma, ICU discharge, and death) and 16 delirium-related covariables, to quantify the association between benzodiazepine use and delirium occurrence the following day. Among 1112 patients, 9867 daily transitions occurred. Benzodiazepine administration in an awake patient without delirium was associated with increased risk of delirium the next day [OR 1.04 (per 5 mg of midazolam equivalent administered) 95 % CI 1.02-1.05). When the method of benzodiazepine administration was incorporated in the model, the odds of transitioning to delirium was higher with benzodiazepines given continuously (OR 1.04, 95 % CI 1.03-1.06) compared to benzodiazepines given intermittently (OR 0.97, 95 % CI 0.88-1.05). After addressing potential methodological limitations of prior studies, we confirm that benzodiazepine administration increases the risk for delirium in critically ill adults but this association seems to be limited to continuous infusion use only.