Selection of trials
After searching the database and comparing the present criteria, a total of 17 trials were included in the meta-analysis (16, 19, 20, 22-35). These trials are all randomized control trials. The selection process is present in Figure 1.
Basic characteristics of included studies
The characteristics of the included studies are shown in Figure 2. There are 12 studies written in English (16, 19, 20, 24-27, 29, 31-34), 5 studies written in Chinese (22, 23, 28, 30, 35). Trial sizes from 19 to 124 patients. A total of 461 and 463 patients were included in the dexmedetomidine group and control group, respectively. The type of surgeries was mainly thoracic surgery, others included gastrectomy (28, 31) and lower limb surgery (30, 32).
Risk of bias assessment
The risk of bias of the seventeen trials was assessed and summarized in Figure 3. Two trials didn’t give enough information about their randomization (19, 20). Six trials had the right method of their allocation concealment (16, 20, 27, 29, 31, 33). Blinding was used in thirteen trials (16, 19, 20, 22, 24-27, 29-31, 33, 34). Only one trial had attrition bias (24). All the seventeen trials reported all the end points in their method sections.
Results of meta-analysis
Meta-analysis of pH
Seven trials including 345 patients reported pH (20, 23, 24, 26-29). The pooled data showed no significant difference between the dexmedetomidine group and the placebo group (MD: 0.03; 95% CI: from 0.00 to 0.07, p=0.04). (Figure 4)
Meta-analysis of PaO2
Eight trials including 536 patients reported PaO2 (19, 20, 26-28, 33-35). The pooled data showed significant higher PaO2 in the dexmedetomidine group than the placebo group (MD: 10.96; 95% CI: from 0.77 to 21.15; p=0.04). (Figure 5)
Meta-analysis of PaCO2
PaCO2 was reported by eleven trials including 571 patients (16, 20, 22, 23, 26-29, 31, 33, 34). The pooled data showed significant difference between the dexmedetomidine group and the placebo group. The dexmedetomidine group had lower PaCO2 than the placebo group (MD: -0.88; 95% CI: from -1.66 to -0.11; p=0.002). (Figure 6)
Meta-analysis of PaO2/FiO2
PaO2/FiO2 was reported by ten trials including 512 patients (16, 22-25, 29-32, 35). The pooled data showed significant higher PaO2/FiO2 in the dexmedetomidine group than the placebo group (MD: 30.77; 95% CI: from 19.11 to 42.43; p<0.00001). (Figure 7)
Meta-analysis of respiratory index
Two trials including 140 patients reported the respiratory index (30, 32). The pooled data showed no significant difference between the dexmedetomidine group and the placebo group (MD: -0.07; 95% CI: from -0.31 to 0.18; p=0.60). (Figure 8)
Meta-analysis of time to extubation
The time to extubation was reported by two trials including 140 patients (25, 33). The pooled data showed no significant difference between the dexmedetomidine group and the placebo group (MD: 0.28; 95% CI: from -3.65 to 4.20; p=0.89). (Figure 9)
Meta-analysis of length of hospital stay
The length of hospital stay was reported by three trials including 206 patients (16, 22, 33). The pooled data showed significant difference between the dexmedetomidine group and the placebo group (MD: -1.19; 95% CI: from -2.21 to -0.16; p=0.02). (Figure 10)
Meta-analysis of events of pulmonary complications
Three trials including 140 patients reported the events of pulmonary complications (16, 22, 33). The pooled data showed significant difference between the dexmedetomidine group and the placebo group (RR: 0.28; 95% CI: from 0.09 to 0.82; p=0.02). (Figure 11)
Subgroup analysis
Dexmedetomidine group compared with control group administrated by I.V.
As shown in Figure 4, 5, 6, there was no significant difference in pH (MD: 0.04; 95% CI: from 0.00 to 0.07; p=0.06). There was no significant difference in PaO2 (MD: 4.99; 95% CI: from -2.49 to 12.46; p=0.19). In PaCO2, there was significant difference (MD: -1.87; 95% CI: from -2.87 to -0.88; p=0.0002).
Dexmedetomidine group compared with control group administrated by Non-I.V.
There were no significant differences in pH (MD: 0.03; 95% CI: from -0.08 to 0.14; p=0.60), PaO2 (MD: 21.91; 95% CI: from -3.14 to 46.97; p=0.09) and PaCO2 (MD: 0.65; 95% CI: from -0.59 to 1.88; p=0.30). Subgroup analysis was not performed for PaO2/FiO2 because of lacking of data. (Figure 4, 5, 6)