Determinants of Delayed Recovery of Consciousness After Analgosedation Discontinuation in the ICU: Insights From Patients With COVID-19 Hypoxemic Respiratory Failure

OBJECTIVES: Prolonged disorders of consciousness are common in critically ill patients receiving mechanical ventilation and are often attributed to prolonged sedative exposure in the setting of decreased drug clearance and/or reduced metabolism. Here, in a large sample of critically ill COVID patients obtained over a short period, we tested the assumption that prolonged unconsciousness following benzodiazepine and/or propofol sedation can be attributed to residual exposure. Further, we examine associations between clinical variables on time to recovery of consciousness (RoC) as a framework for the broader critically ill population. DESIGN: Retrospective cohort study. SETTING: Massachusetts General Hospital, Weill Cornell Medicine, Columbia University Irving Medical Center. PATIENTS: Seven hundred eighty-four patients with COVID-19 critical illness in Spring-Summer 2020. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We estimated the latest expected RoC (LERoC) using models of sedation exposure that account for sedative-specific pharmacokinetics and critical illness. Our primary exposure variable was a time-weighted dose of analgosedative agents at benzodiazepine and/or propofol cessation; our primary outcome was time to RoC. We estimated relative risks for late RoC (i.e., after LERoC) via multinomial logistic regression. Among individuals with late RoC, we fit a multivariate subdistribution hazard model for time to RoC. Seventy-three percent of patients had RoC before hospital discharge, yet 34% of patients achieving RoC did not do so within pharmacologically plausible sedative elimination times. Patients with late RoC were older and exhibited hypoxemia and acute kidney injury. Patients with dexmedetomidine as an adjunct sedative had a disproportionately larger incidence of early recovery. Patients with early vs. late RoC did not have significantly different discharge dispositions. CONCLUSIONS: In our cohort, the time to RoC was commonly prolonged beyond that expected from sedation exposures alone. These data may aid clinicians and families with expectations of RoC and warrant investigation of alternative determinants of delayed RoC in this population.