To investigate the impact of different sedation strategies on short-term outcomes in intensive care unit (ICU) patients receiving invasive mechanical ventilation (IMV), and to assess whether these effects differ in special populations such as the elderly, those with malignancy, or hepatic impairment.

This retrospective observational cohort study utilized data from 2 large critical care databases, MIMIC-Ⅳ (v2.2) and eICU-CRD (v2.0). Inclusion criteria were: age ≥18 years; endotracheal intubation and IMV within 48 h after ICU admission; ICU length of stay (LOS) ≥48 h; initiation of a single-agent sedation strategy (dexmedetomidine, propofol, or benzodiazepines) from 2 h before to 48 h after intubation. Patients receiving combined sedatives, undergoing re-intubation, or with single-drug sample sizes <100 were excluded. The primary outcome was time to extubation; secondary outcomes included in-hospital mortality, ICU LOS, incidence of delirium, and major adverse cardiovascular events (MACEs). Inverse probability of treatment weighting (IPTW) based on gradient boosting machine algorithms was employed to control confounding bias, followed by augmented inverse probability weighting (AIPW) doubly robust models to estimate average treatment effects (ATE), odds ratios (ORs), and 95% confidence intervals (CIs). Prespecified subgroup analyses were applied to examine interactions with age, malignancy, and hepatic status.

A total of 12561 patients were enrolled, comprising 1111 in the dexmedetomidine group, 8021 in the propofol group, and 3429 in the benzodiazepine group. After IPTW weighting, baseline variables achieved overall balance across groups. In the AIPW doubly robust models, compared with the dexmedetomidine group, the propofol group demonstrated prolonged time to extubation by 0.782 d (ATE=0.782 d, 95%CI: 0.769 to 0.796, P<0.001), and the benzodiazepine group by 1.791 d (ATE=1.791 d, 95%CI: 1.779 to 1.804, P<0.001); compared with benzodiazepines, the propofol group showed shortened extubation time by 1.009 d (ATE=−1.009 d, 95%CI: −1.018 to −1.000, P<0.001). For secondary outcomes, compared with dexmedetomidine, both propofol and benzodiazepines were associated with increased risk for in-hospital mortality (propofol: OR=1.905, 95%CI: 1.647 to 2.201, P<0.001; benzodiazepines: OR=2.768, 95%CI: 2.262 to 3.388, P<0.001) and for delirium (propofol: OR=1.905, 95%CI: 1.428 to 2.541, P<0.001; benzodiazepines: OR=2.382, 95%CI: 1.681 to 3.376, P<0.001). Additionally, ICU LOS was prolonged by 0.367 d and 1.012 d in the propofol and benzodiazepine groups, respectively (propofol: ATE=0.367 d, 95%CI: 0.351 to 0.382, P<0.001; benzodiazepines: ATE=1.012 d, 95%CI: 0.996 to 1.029, P<0.001), and the total incidence of MACEs increased by 6.527 and 15.199 events (propofol: ATE=6.527 events, 95%CI: 6.112 to 6.942, P<0.001; benzodiazepines: ATE=15.199 events, 95%CI: 14.727 to 15.671, P<0.001). Regarding ventilator-associated pneumonia (VAP), propofol was associated with a lower risk than dexmedetomidine (OR=0.722, 95%CI: 0.648 to 0.805, P<0.001), whereas no significant difference was observed between benzodiazepines and dexmedetomidine (OR=1.082, 95%CI: 0.934~1.252, P=1.000). Subgroup and interaction analyses revealed significant interactions between sedation strategy and malignancy (P-for-interaction=0.001) and moderate-to-severe liver disease (P-for-interaction=0.048). Among patients with malignancy, propofol showed a trend toward shorter time to extubation than dexmedetomidine (ATE=−1.326 d, 95%CI: -2.732 to -0.080, P=0.064); similarly, in patients with moderate-to-severe liver disease, propofol also demonstrated a trend toward shorter extubation time than dexmedetomidine (ATE=-1.232 d, 95%CI: −2.939 to -0.475, P=0.157), suggesting attenuated or even reversed benefits of dexmedetomidine in these special populations.

In ICU patients receiving invasive mechanical ventilation, dexmedetomidine, is associated with shorter time to extubation and lower risks of in-hospital mortality, delirium, and cardiovascular adverse events compared with propofol and benzodiazepines. However, these advantage of dexmedetomidine are diminished among patients with malignancy or moderate-to-severe liver disease.