Literature retrieval results
Figure 1 shows the screening and selection processes. The primary literature search initially identified 91 studies. After 28 duplicates were deleted, 63 potential eligible studies were screened on titles and abstracts. Of these studies, 55 were excluded because they were not able to meet the inclusion criteria. Subsequently, the remained 8 studies were assessed for full-text eligibility. Finally, 7 studies satisfied the inclusion criteria(6, 13-18).
Research characteristics and quality assessment
The characteristics of included studies were summarized in Table 1. In total, 846 Eligible patients were included, including 426 patients in perineural dexamethasone group and 420 patients in intravenous dexamethasone group. The population ranged from 47 to 59.2 years old. The type, concentration, and dose of local anesthetics used for nerve block varied from studies. The studies also varied in the dose of dexamethasone used, ranged from 1 mg to 10 mg, and the most commonly used dose was 4mg. One study compared low-dose (4mg) dexamethasone with high-dose (8mg) dexamethasone administered by either perineural or intravenous routes, and data for both doses were included in our study (6). All included studies were considered as high quality.
Duration of analgesia: This variable was reported in five studies, with pooled results showing that perineural dexamethasone significantly prolonged analgesia compared to intravenous dexamethasone (WMD: 1.699 hours; 95% CI: 0.0014, 3.384; I2: 64.3%; P: 0.048, random model) (Figure 2a).
Duration of sensory block: Two included studies investigated this outcome, and the result showed no significant difference between perineural and intravenous dexamethasone (WMD: 264.579 minutes; 95% CI: -490.767, 1019.924; I2: 86.7%, P: 0.492, random model) (Figure 2b).
Duration of motor block: This variable was assessed in two studies. Overall, there was no significant difference between the two groups (WMD: 110.48 minutes; 95% CI: -284.379, 505.339; I2: 74.2%, P: 0.583, random model) (Figure 2c).
Postoperative pain at 12 hours: In view of this variable, the pooled result showed perineural dexamethasone was associated with a significant lower postoperative pain score at 12 hours (three studies, WMD: -0.652; 95% CI: -1.129, -0.176; I2: 43.1%, P: 0.007, fixed model) (Figure 3a).
Postoperative pain at 24 hours: This outcome was reported in three studies, the pooled analysis demonstrated no significant difference between the groups (WMD: -0. 309; 95% CI: -1.275, 0.658; I2: 72%, P: 0.531, random model) (Figure 3b).
Opioid consumption within 24 hours: Two studies measured this outcome. In this analysis, no significant difference was observed between the two groups (SMD: -0.019; 95% CI: -0.244, 0.206; I2: 0%, P: 0.866, fixed model) (Figure 3c).
Adverse events: In view of the incidence of adverse events, the pooled results showed no statistical difference for either outcome. Dyspnea (three studies, RR: 0.936; 95% CI: 0.27, 3.24; I2: 0%, P: 0.916, fixed model), sleep disturbance (two studies, RR: 1.028; 95% CI: 0.526, 2.01; I2: 28.7%, P: 0.935, fixed model), and postoperative glucose change (two studies, WMD: -1.5; 95% CI: -4.723, 1.722; I2: 0%, P: 0.361, fixed model).