Effects of different sedation regimens on delirium in ICU patients

Delirium is common among ICU patients and worsens outcomes. The comparative effectiveness of sedatives remains uncertain.

We conducted a single-center, prospective cohort study using a prespecified 48-h landmark. Adults (≥18 years) with an expected ICU stay ≥24 h and continuous intravenous sedation ≥6 h were enrolled. The predominant sedative used during the first 48 h (dexmedetomidine, midazolam, or remimazolam) was recorded. Predominance was defined a priori as the agent with the greatest infusion time share, with ties broken first by standardized cumulative dose and then by earliest start time/longest continuous block. The primary outcome was incident delirium from hour 48 to ICU day 7 or discharge, as determined using standardized, sedation-gated CAM-ICU assessments. Secondary outcomes included time-in-target sedation, ventilator-free days, and prespecified hemodynamic adverse events. We used multinomial propensity-score overlap weighting and weighted log-link models to estimate risk ratios (RRs) with robust standard errors. A time-varying exposure analysis from ICU admission assessed robustness.

Of the 506 patients who were screened, 426 were included in the landmark cohort (dexmedetomidine = 162, midazolam = 192, and remimazolam = 72). Incident delirium occurred in 108 (25.4%) of the 426 patients. After overlap weighting, dexmedetomidine was associated with a lower risk of delirium versus midazolam (RR 0.70, 95% CI 0.52-0.92) and a similar, although not significant, pattern versus remimazolam (RR 0.74, 95% CI 0.52-1.03). Dexmedetomidine had a higher time-in-target sedation and more ventilator-free days, with more bradycardia, but no apparent increase in hypotension or hypertension. Mortality did not differ. The findings were directionally consistent in the time-varying analysis (HR 0.76, 95% CI 0.58-0.99).

In this prospective cohort, dexmedetomidine-predominant sedation was associated with a lower risk of post-landmark delirium and better sedation quality than midazolam, at the cost of more bradycardia and without differences in mortality. These findings warrant multicenter confirmation and randomized evaluation.