Prevalence of delirium while on study drug. The top panel demonstrates that, among all patients, those sedated with dexmedetomidine (DEX) had a 70% lower likelihood of having delirium on any given day compared with patients sedated with lorazepam (LZ). Sepsis did not modify this relation (adjusted P for interaction = 0.94), meaning that dexmedetomidine reduced the risk of developing delirium whether patients had sepsis (lower panel) or not. * Number of patients assessed denotes the number of patients who were alive, in the ICU, and not comatose (Richmond Agitation-Sedation Scale (RASS)-3 or lighter) and are therefore assessable for delirium. Percentages of patients alive and without coma, but with delirium, are represented with black bars if on lorazepam and gray bars if on dexmedetomidine.