We read with interest the paper by Moore et al. [1] assessing the use of flumazenil for reversal of iatrogenic benzodiazepine-associated delirium in alcohol withdrawal. Alcohol withdrawal is associated with significant morbidity and requires extensive resource utilization for patient management. While we acknowledge that the treatment of patients with alcohol withdrawal can result in iatrogenic benzodiazepine overdose, this adverse effect can be limited by symptom-triggered administration of benzodiazepines, which is perhaps the most widely accepted and evidence-based approach to therapy [2]. The retrospective and uncontrolled nature of the data presented by the authors suggests substantial deviations from this approach and includes other departures from commonly accepted care such as the use of multiple benzodiazepines, neuroleptics, opioids, and a host of other drugs. Patient selection is unclear and the dosing and indications for benzodiazepines are unspecified. The “success” and “safety” of flumazenil is based on a chart review and relies on subjective interpretation, potentially leading to bias. Additionally, improvement after flumazenil was considered diagnostic of benzodiazepine-induced delirium despite the fact that some patients with hepatic encephalopathy may respond to flumazenil [3] and no information is provided about the subjects’ liver function. Finally, though paradoxical reactions to benzodiazepines may rarely occur, we feel it is unlikely to be the primary cause for hyperactive delirium several days after treatment initiation with benzodiazepines, as other causes, including ICU delirium, or persistent alcohol withdrawal are more likely to be etiological.

We commend the authors for considering a novel pharmacodynamic question involving the parallel occurrence of alcohol withdrawal and benzodiazepine delirium. We hope that if the authors continue to investigate this question in future studies, they will use a controlled and blinded protocol that is methodologically sound and addresses issues of subject consent in research.