3 Results
3.1 Study selection
The initial literature search yielded a total of 240 articles. Studies that were duplicated were excluded. Four articles were also excluded based on their titles and abstracts. The remaining 54 potentially eligible studies were retrieved for a careful review of the full texts. Among them, studies without a control group, review articles, studies with the same database, studies including non-laryngeal cancer patients, studies with unclear inclusion criteria and studies not written in English were excluded from the analysis. Finally, 7 articles were included in this review 10-16. A flow diagram describing the process involved in study identification and inclusion/exclusion is shown in Figure 1 . eTable 1 in the Supplement summarizes the literature search process and the keywords used.
3.2 Demographics
Table 1 lists the basic demographics of patients from the 7 articles, including two randomized studies and 5 nonrandomized studies. The pooled prevalences of preoperative tracheostomy and radiotherapy were both significantly higher in the SC group (P =0.04 andP =0.00, respectively). The bias assessment for each study is described in eTable 2 and eTable 3 in the Supplement .
3.3 Outcomes
3.3.1 Operative time Four of the included studies evaluated the operative time (total laryngectomy with/without neck dissection) of both procedures11,12,14,15. Two of these 4 studies included only cases of isolated total laryngectomy (without neck dissection)12,14. The pooled results of 4 studies showed that the operative time was lower in the SC group (MD, -63.2; 95% confidence interval [CI], -106.0 to -20.4), and a decrease in operative time by 63.2 minutes was observed in the SC group (Figure 2a)11,12,14,15. Subgroup analysis including the 2 studies without neck dissection also showed that the operative time was lower in the SC group (MD, -45.9; 95% CI, -64.2 to -27.6) with a decrease of 45.9 minutes12,14. (Figure 2b). 3.3.2 PCF All of the included studies recorded the incidence of postoperative PCF in both groups. In the present study, the incidence rates of PCF were 13.7% (28/204) and 27.2% (90/331) in the SC and MC groups, respectively. The pooled analysis showed that the PCF rate was lower in the SC group (OR = 0.38; 95% CI, 0.18-0.83; P = 0.016).(Figure 3). 3.3.3 Length of hospital stay Three of the included studies recorded the length of hospital stay in the two groups. The pooled analysis demonstrated that the length of hospital stay after TL was approximately 2.9 days shorter in the SC group than in the MC group (MD, -2.9; 95% CI, -5.6 to -0.1). (Figure 4). 3.3.4 Postoperative surgical site infection Two of the included studies recorded the incidence of postoperative surgical site infection in both groups. The pooled analysis showed no significant difference between the SC and MC groups regarding the rate of postoperative surgical site infection (OR = 0.41; 95% CI, 0.02-8.73; P = 0.565). (Figure 5).
3.4 Publication Bias
Funnel plots and the results of heterogeneity tests are presented ineFigure 1 in theSupplement . The plots of different parameters are generally symmetrical, suggesting no obvious publication bias. The results of the Egger intercept test also indicated no apparent publication bias.