The reporting and conducting of this study were in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Statement (20). The query task for systematic reviews was to collect randomized controlled trials (RCTs) that compared the role of desflurane, sevoflurane, and propofol on recovery outcomes for obese patients undergoing general anesthesia was eligible in this study, and the publication language was not restricted. The potentially relevant articles were searched in PubMed, EmBase, and the Cochrane Library from January 2000 to December 2020, and the following search terms were used through Medical Subject Heading to text words: (“obese” OR “overweight” OR “bariatric surgery” OR “body mass index”) AND (“anesthesia” OR “anesthetic” OR “desflurane” OR “sevoflurane” OR “propofol” OR “total intravenous anesthesia” OR “general anesthesia”). Meta-analyses were screened out if they did not meet the inclusion criteria listed later, first at the title/abstract level, and later at the methods section level if necessary. We also reviewed the reference lists in retrieved studies for any further eligible studies.

The details of inclusion criteria were: (1) Participants: obese patients undergoing general anesthesia; (2) Intervention and control: any 2 of desflurane, sevoflurane, and propofol; (3) Outcome: time to eye opening, time to extubation, time to stating name, time for orientation to place, and time required for hand squeezing; and (4) Study design: the study had to have RCT design. The relevance of studies was assessed by reviewing the title and abstracts, and the full-text evaluations were evaluated to obtain whether potentially relevant trials reported an outcome of interest. The study selection was independently double-checked the inclusion and exclusion criteria by two reviewers, and the conflicts between reviewers were resolved by group discussion until a consensus was reached.

The following data items in each included study were independently extracted by two reviewers: first author’s name, publication year, region, sample size, mean age, male proportion, body mass index (BMI), anesthesia technique, surgical technique, and reported outcomes. Then the quality of each trial was assessed using the Jadad scale by the same two reviewers, which was based on randomization, blinding, allocation concealment, withdrawals and dropouts, and use of intention-to-treat analysis (21). Any disagreement between the two reviewers were settled by an additional reviewer referring to the full text of the articles.

The effectiveness of a treatment on postoperative recovery outcomes was assigned as a continuous variable, and the standard mean difference (SMD) with a 95% confidence interval (CI) was calculated before data pooling. Then the random-effects model was used to calculate pooled SMD and 95% CI for comparison of the postoperative recovery outcomes after using desflurane, sevoflurane, or propofol (22, 23). Heterogeneity across included trials was evaluated using the I² and Q statistic, and the significant heterogeneity was defined as I² > 50.0% or P < 0.10 (24, 25). Subgroup analyses were also conducted for the postoperative recovery outcomes of desflurane versus sevoflurane according to mean age, male proportion, BMI, and surgical technique, and the difference between subgroup analyses was assessed by using the interaction t-test, which assumed the distribution of effect estimates were normal (26). Publication bias for investigated outcomes was assessed by using the funnel plot, Egger, and Begg tests (27, 28). The P-value for pooled conclusions is 2-sided, and the inspection level was 0.05. All of the analyses in this study were conducted by using the software STATA (version 10.0; Stata Corporation, College Station, TX, USA).

The electronic searches yielded 1,641 articles, and the 1,187 articles were retained after duplicate articles were removed. A total of 1,094 studies were excluded by reviewing their titles and abstracts because they reported irrelevant. The remaining 93 potentially eligible studies were retrieved, and 82 studies were removed after detailed evaluations: patients not obese (n = 37), other interventions (n = 31), and lacking sufficient data (n = 14). Reviewing the reference lists of the remaining studies did not find any new eligible studies. Finally, 11 RCTs were selected for meta-analysis (9–19), and the details of the study selection process are shown in Figure 1.

The characteristics of included studies are summarized in Table 1, and 713 obese patients undergoing general anesthesia were involved. Six studies compared desflurane with sevoflurane, three trials compared desflurane with propofol, and the remaining two trials compared sevoflurane with propofol. The BMI for each trial ranged from 35.3 to 58.0 kg/m², and the sample size ranged from 23 to 183. Eight trials with high quality (three trials had five scores, five trials had four scores), and the remaining three trials with low quality (three trials had three scores).

The number of trials for time to eye opening when comparing desflurane with propofol, desflurane with sevoflurane, and sevoflurane with propofol were three, six, and two trials, respectively. We noted desflurane was associated with a shorter time to eye opening as compared with sevoflurane (SMD: −0.86; 95% CI, −1.43 to −0.28; P = 0.003; Figure 2A). However, desflurane (SMD: 0.05; 95% CI, −0.71 to 0.80; P = 0.906) or sevoflurane (SMD: 1.13; 95% CI, −0.00 to 2.27; P = 0.050) were not associated with the time to eye opening when compared with propofol (Figure 2A). There was significant heterogeneity in the comparisons of desflurane with propofol (I²= 86.6%; P = 0.001), desflurane with sevoflurane (I²= 79.7%; P < 0.001), and sevoflurane with propofol (I²= 89.3%; P < 0.001).

The number of trials for time to extubation when comparing desflurane with propofol, desflurane with sevoflurane, and sevoflurane with propofol were two, five, and one trial(s), respectively. We noted desflurane was associated with shorter time to extubation when compared with propofol (SMD: −1.13; 95% CI, −1.52 to −0.73; P < 0.001) or sevoflurane (SMD: −1.19; 95% CI, −2.15 to −0.22; P = 0.016). However, sevoflurane versus propofol show a longer time to extubation (SMD: 1.47; 95% CI, 1.03 to 1.91; P < 0.001) (Figure 2B). There was significant heterogeneity for the comparison of desflurane with sevoflurane (I²= 90.6; P < 0.001), while no evidence of heterogeneity for the comparisons of desflurane with propofol (I²= 0.0%; P = 0.485), and sevoflurane with propofol (I²= 0.0%; P = 0.866).

The number of trials for time to stating name when comparing desflurane with propofol, and desflurane with sevoflurane were one and four trial(s), respectively. We noted desflurane was associated with a shorter time to stating name when compared with propofol (SMD: −1.40; 95% CI, −2.32 to −0.48; P = 0.003) or sevoflurane (SMD: −2.09; 95% CI, −3.33 to −0.85; P = 0.001) (Figure 2C), and significant heterogeneity for the comparison of desflurane with sevoflurane was observed (I²= 90.3; P < 0.001).

Time for orientation to place when comparing desflurane with propofol, desflurane with sevoflurane, and sevoflurane with propofol each had one trial. We noted desflurane was associated with a longer time for orientation to place as compared with propofol (SMD: 0.65; 95% CI, 0.22 to 1.07; P = 0.003), while desflurane versus sevoflurane show shorter time for orientation to place (SMD: −0.88; 95% CI, −1.46 to −0.30; P = 0.003). However, there was no significant difference between sevoflurane and propofol for the time for orientation to place (SMD: 0.13; 95% CI, −0.49 to 0.75; P = 0.672) (Figure 2D). Finally, we noted desflurane was not associated with the time required for hand squeezing as compared with sevoflurane (SMD: −3.27; 95% CI, −8.04 to 1.51; P = 0.180; Figure 2E), and significant heterogeneity was observed across included trials (I²= 96.3; P < 0.001).

Subgroup analyses for the time to eye opening, time to extubation, and time to stating name when comparing desflurane with sevoflurane are shown in Table 2. We noted desflurane versus sevoflurane was associated with a shorter time to eye opening if the mean age of patients <40.0 years, proportion of men <50.0%, BMI ≥ 50.0 kg/m², or patients were treated with gastroplasty. Moreover, desflurane versus sevoflurane was associated with a shorter time to extubation when the proportion of men <50.0%, or patients were treated with gastroplasty. Finally, desflurane versus sevoflurane was associated with a shorter time to stating name in all subgroups.

The publication bias for time to eye opening, time to extubation, and time to stating name were also evaluated and presented in Supplementary Figures S1–S3. There was no significant publication bias for time to eye opening (P-value for Egger: 0.170; P value for Begg: 0.373), time to extubation (P-value for Egger: 0.417; P value for Begg: 0.917), and time to stating name (P-value for Egger: 0.073; P value for Begg: 0.086).