Study selection

A total of 707 potentially relevant studies were identified by the systematic literature search, of which 89 were duplicate studies. We screened 618 records based on the title and abstract. Among these, 604 records were excluded because of unrelated studies (n = 411), irrelevant comparator (n = 23), and not an RCT (n = 170). After reviewing the full text of eligible articles, one clinical protocol was excluded, and the full text of two studies were not available. Finally, 11 RCTs including 643 participants were selected for our meta-analysis [10,11,12,13,14,15,16,17, 19,20,21]. The flow diagram for the literature search, screening, and eligibility evaluation is shown in Fig. 1.

Fig. 1

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram of present study

Characteristics of Included Studies

The general characteristics and results of interest in the included studies are detailed in Table 1. The sample size of the included studies ranged from 21 to 200 patients. Thoracotomy was performed in 4 of 11 included trials [10, 13, 19, 20], while video-assisted thoracoscopic surgery (VATS) was carried out in six studies [11, 12, 14,15,16, 21], and only one study did not report the surgical approach [17]. The procedure was performed under general anesthesia (GA) in five trials [12,13,14, 16, 17], GA plus epidural anesthesia in three trials [10, 19, 20], GA plus paravertebral block in one trial [11], GA plus intercostal nerve block in one trial [21], and one study did not specify anesthesia methods [15]. In addition, 10 RCTs enrolled patients who were adults [10,11,12,13, 15,16,17, 19,20,21], whereas one RCT enrolled children aged 3–18 years [14].

Table 1 Characteristics of included studies

MP was administered intravenously in 10 trials [10,11,12,13,14,15,16,17, 19, 21], and epidural MP infusion was performed in one trial [20]. Furthermore, the time of MP administration was before surgery in six trials [10,11,12,13, 19, 21], during surgery in three trials [14, 16, 17], and both before and after surgery in two trials [15, 20]. The total volume of MP ranged from 40 to 120 mg in three trials, and the dose of MP ranged from 1 to 30 mg/kg in eight trials. All included RCTs used placebo saline as a control.

Risk of Bias

Regarding the risk assessment for the 11 included RCTs, 4 studies were rated as having a low risk of bias [11, 12, 14, 21], while 7 studies were deemed unclear risk [10, 13, 15,16,17, 19, 20]. The complete assessment for separate studies is provided in the supplementary material (eTable 1). The risk of bias summary and risk of bias graph can also be found in the supplementary materials (Fig. 2 and eFig. 1).

Fig. 2

Risk of bias graph: review authors’ judgements about each risk of bias item presented as percentages across all included studies. Green for low risk of bias, yellow for some concerns, and red for high risk of bias

Coprimary OutcomesRest Pain Scores at 24 h Postoperatively

Two studies (n = 248) utilized an 11-point scale, including a numeric rating scale (NRS) or VAS, to assess rest pain scores at 24 h postoperatively [12, 21]. The results indicated that the MP group made a significant difference in this outcome compared with the placebo group (WMD –0.50 cm; 99% CI −0.82 to −0.18; p < 0.0001; I2 = 0%) (Table 2 and Fig. 3). However, the difference was minimal and did not reach the MCID threshold (1.1 cm). The quality of evidence was “high” (eTable 2). No publication bias was detected (p = 1.00).

Fig. 3

A Band plot for weighted mean difference (WMD) of the change in rest pain scores from 24 to 48 h postoperatively between methylprednisolone versus control. B Band plot for weighted mean difference (WMD) of the change in dynamic pain scores from 24 to 48 h postoperatively between methylprednisolone versus control; Pooled estimates of the WMD for each time point are represented by a dark line and 95% confidence intervals are represented by the surrounding shaded region

Dynamic Pain Scores at 24 h Postoperatively

This outcome was also evaluated by two studies (n = 248) with an 11-point NRS or VAS [12, 21]. The dynamic pain scores at 24 h postoperatively were significantly reduced in the MP group compared with the placebo group (WMD –0.55 cm; 99% CI –0.97 to –0.13; p = 0.0007; I2 = 0%) (Table 2 and Fig. 3). Nevertheless, the improvement was not clinically meaningful based on MCID. The quality of evidence for this outcome was rated as “high” (eTable 2). Egger’s test showed that no publication bias existed (p = 0.71).

Rescue Analgesic Consumption Within 24 h Postoperatively

Three studies (n = 195) reported rescue analgesic consumption within 24 h postoperatively [11, 12, 20]. There was a statistically significant but clinically nonsignificant difference in this outcome between the MP and the placebo groups (WMD –2.65 mg; 99% CI −4.63 to −0.66; p = 0.0006; I2 = 0%) (Table 2). Sensitivity analysis revealed the favorable robustness of the result by excluding trials of non-preoperative interventions, multiple administration, and non-thoracoscopic surgery (eTable 3) [20]. The quality of the results was assessed as “moderate” owing to the small sample size (eTable 2). The evaluation by Egger’s test indicated no publication bias for this outcome (p = 0.82).

Secondary OutcomesRest Pain Scores at 48 h Postoperatively

Rest pain scores at 48 h postoperatively were evaluated on an 11-point scale (VAS or NRS) by two studies (n = 262) [11, 21]. The pooled result showed no difference in this outcome between the MP and placebo groups (p = 0.07) (Table 2 and Fig. 3). According to the GRADE system, the quality of evidence was “high” (eTable 2). We did not find publication bias for this outcome (p = 0.60).

Dynamic Pain Scores at 48 h Postoperatively

Two studies (n = 262) examined dynamic pain scores at 48 h postoperatively by an 11-point scale (VAS or NRS) [11, 21]. Administration of MP in lung surgery did not significantly relieve dynamic pain at 48 h postoperatively compared with the placebo group (p = 0.05) (Table 2 and Fig. 3). The quality of evidence was “high” (eTable 2). The result did not exhibit publication bias (p = 0.41).

Rescue of Analgesic Consumption Within 48 h Postoperatively

Two studies (n = 120) reported this outcome [11, 20]. The combined results showed that MP did not significantly reduce rescue analgesic consumption within 48 h postoperatively compared with the placebo group (p = 0.59) (Table 2). The quality of evidence was “moderate” because of the small sample size (eTable 2). Egger’s test for publication bias was nonsignificant (p = 0.54).

Inflammatory IndicatorsIL-6 at 6 h

The serum IL-6 level at 6 h postoperatively was reported in two studies (n = 86) [14, 17]. There was a significant difference in the outcome between the MP and placebo groups (WMD –20.49 pg/mL; 95% CI −29.94 to −11.04; p < 0.0001; I2 = 0%) (Table 2). The evidence exhibited a “low” quality owing to allocation concealment, performance bias, detection bias, and small sample size (eTable 2). There was no publication bias (p = 0.46).

IL-6 at 24 h

Three studies (n = 254) detected IL-6 levels at 24 h postoperatively [10, 17, 21]. The results showed that MP did not decrease this indicator compared with the placebo group (p = 0.13) (Table 2). Sensitivity analysis excluding trials of non-preoperative interventions showed a significant difference for this outcome in the MP group (WMD −5.62 pg/mL; 95% CI −10.98 to −0.26; p = 0.04; I2 = 0%) (eTable 3) [17]. The quality of this evidence was “low” because of sequence generation, allocation concealment, performance bias, detection bias, and high heterogeneity (eTable 2). No publication bias was found for the result (p = 0.55).

IL-10 at 6 h

Two studies (n = 86) measured serum IL-10 levels at 6 h postoperatively [14, 17]. The MP group was not different from the placebo group for this indicator (p = 0.11) (Table 2). The quality of evidence was rated as “very low” because of allocation concealment, performance bias, detection bias, small sample size, and high heterogeneity (eTable 2). The outcome showed no publication bias (p = 0.23).

TNF-α at 6 h

This outcome was reported by two studies (n = 86) [14, 17]. The serum TNF-α level at 6 h postoperatively was not different between the MP group and the placebo group after lung surgery (p = 0.28) (Table 2). The quality of evidence was evaluated as “very low” owing to allocation concealment, performance bias, detection bias, high heterogeneity, and publication bias (eTable 2). Egger’s test showed publication bias (p < 0.001).

CRP at 24 h

Two studies (n = 373) detected CRP at 24 h [15, 21]. There was no significant difference in the level of CRP at 24 h postoperatively between the MP and placebo groups (p = 0.06) (Table 2). The quality of this evidence was “low” for sequence generation, allocation concealment, performance bias, detection bias, and high heterogeneity (eTable 2). There was no publication bias (p = 0.19).

Duration of hospitalization

Duration of hospitalization was reported in five studies (n = 565) [10,11,12, 15, 21]. Administration of MP did not significantly reduce the length of hospital stay compared with the placebo group (p = 0.38) (Table 2). Posthoc sensitivity analyses revealed that the result was credible and stable after excluding trials of non-preoperative interventions, multiple administration, and non-thoracoscopic surgery (eTable 3) [