Of the 263 studies identified in the search (see Fig. 1), we excluded 121 duplicates and further 74 citations after title and abstract screening. We assessed 78 full texts and included 23 RCTs in review [18-40]. There were 2,293 patients included in this study. Baseline characteristics of included trials are summarized in Table 1.
The selected studies were published between 2009 and 2020. The number of included participants from each study were ranged from 40 to 422. The mean age of participants ranged from 47 to 73 years. All participants met the sepsis criteria. Dexmedetomidine dose varied apparently among studies. The dosage of dexmedetomidine was similar across included trials with the most common regimen including a loading dose of 1 μg/kg·h, followed by a maintenance dose of 0.2 to 1 μg/kg·h. Eight of the included trials [21, 27, 28, 31, 33, 36, 37, 40] were judged to be at high RoB, while the remainder of the trials were judged either at low ROB or certain concerns. (see Fig. 2 and e-Fig. 1 for all RoB judgements). Table 2 and e-Table 1-7 depict the pooled outcomes with associated GRADE certainty of evidence.
Table 2 shows the summary of findings for all outcomes including the certainty of evidence. Pooled analysis outlined the dexmedetomidine group had a lower occurrence of mortality (RR 0.81; 95% Confidence Interval (CI) 0.71–0.93; risk difference (RD) 6% reduction; 95% CI 2% reduction to 10% reduction; P < 0.05;high certainty), with no significant heterogeneity (I2 = 27%, P = 0.14) (see Fig. 3). There was no obvious asymmetry in funnel plots by the visual inspection (see e-Fig. 2). The TSA results demonstrated that the information size needed for detecting an intervention effect, was 3398 patients. The cumulative Z curve crossed both, the conventional boundary for benefit and the trial sequential monitoring boundary for benefit (see Fig. 4), suggesting that current evidence is sufficient, while further studies are unlikely to change the current conclusion of the benefit with dexmedetomidine. A cumulative meta-analysis was conducted to assess changes over the time and sample size (see Fig. 5 and e-Fig. 3). A statistically significant reduction in mortality was observed in the studies performed from 2009 to 2019 (RR = 0.58 95% CI 0.45–0.75). After the addition of further recent RCTs[22, 24, 25, 32, 36], a significant reduction in 28-day mortality was still observed (RR 0.81; 95% CI 0.71–0.93).
While 15 studies [18-33, 35-40] included ICU stays in their evaluation index, the results indicated that the dexmedetomidine group could not reduce ICU stays in comparison with the other sedation groups (SMD: −0.26; 95% CI −0.70,0.18, p = 0.24, high certainty]) (see e-Fig. 4 ).
Duration of Mechanical Ventilation
There were eight studies [18, 20, 22, 24, 26, 29-31] explored the impact of dexmedetomidine on the duration of mechanical ventilation. We selected the fixed effect model since there was no heterogeneity in both subgroups (I2 = 0%). The meta-analysis did not find a reduction of mechanical ventilation time under dexmedetomidine use compared to other sedation (MD: −0.4; 95% CI -1.12,0.31, p= 0.27, high certainty) (see e-Fig. 5).
Duration of Ventilator-free Days
There were three studies [19, 20, 25] included the ventilator-free days as indicator, whereby the results indicated that the dexmedetomidine group could not increase the ventilator-free days in comparison with other sedation group (MD: 1.68; 95% CI −1.5,4.85, p = 0.3, very low certainty] )(see e-Fig. 6).
Levels of IL-6, TNF-α, Alanine Transaminase and Creatinine changes at 24 h
Random effect models were utilized in the four outcomes, whereby the results showed a significant lower levels of IL-6 and TNF-α at 24 h in the dexmedetomidine group in comparison with the groups of the other sedatives (SMD: -1.46; 95% CI -2.10, -0.83 , p＜0.05, low certainty; SMD: -1.20; 95% CI -1.78, -0.62 , p＜0.05, moderate certainty; see e-Fig. 7 and e-Fig. 8). However, random model analysis indicated that dexmedetomidine did not lead to a significant change in ALT and Cr at 24 h (Respectively: p = 0.17 and 0.30, low certainty; see e-Fig. 9).
Incidence of delirium
Three studies [20, 30, 31] explored the incidence of delirium for dexmedetomidine. In total, 50/162 (30.86%) patients in the dexmedetomidine group were reported as having experienced delirium, versus 57/165 (34.55%) patients in the control group. Following the meta-analysis, dexmedetomidine was not significantly associated with lower risk of delirium compared to control group: 327 patients; risk ratio 0.89; 95% CI 0.66 to 1.19, low certainty; P = 0.43(see e-Fig. 10). However, considering there were only three studies were included, this result need be interpreted prudently.
Incidence of Total Adverse Events
Eight studies [19, 20, 22, 24, 30, 32, 36, 39] included the incidence of adverse events, which was evidenced by 831 participants. The results demonstrated that there was no difference in the incidence of adverse events between the group of dexmedetomidine and propofol (OR = 1.06, 95% CI 0.50, 2.25, p = 0.87, moderate certainty; see e-Fig. 11a). In terms of arrhythmia and hypotension, the pooled odds ratios (ORs) were 2.36 (95% CI 1.15, 4.8; P = 0.02; I2 = 0%, high certainty; see e-Fig. 11b) and 0.82 (95% CI 0.22, 3.09; P = 0.76; I2 = 47%, low certainty; see e-Fig. 11c). According to the research findings, dexmedetomidine was significantly associated with higher risk of arrhythmia, but did not show significant differences in the incidence of hypotension in comparation with other sedative medications.
Subgroup analyses and sensitivity analyses
From the subgroup analyses of the primary outcome, we found that the APACHE II scores of patients in each study (> 20 or ≤ 20) and the risk of bias had no significant effect on all-cause mortality. (see e-Fig. 12 and e-Fig. 13). Among the majority of the studies, the heterogeneity results were not obviously altered after sequentially omitting each study, indicating that our results were statistically reliable (see e-Fig. 14).