Literature search results

A total of 432 articles were retrieved through the database. Finally, by reading the titles and abstracts, 182 duplicate articles and 250 obviously irrelevant articles were excluded; see Fig. 1. After careful reading, a total of 19 articles [9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27] were finally included. The detailed information and quality evaluation of the characteristics of all included studies are summarized in Table 1. A total of 3517 patients with mediastinal tumors were included, of which 1742 patients underwent RATS treatment and 1775 patients underwent VATS treatment.

Fig. 1

Flow diagram of literature retrieval and screening

Table 1 Basic information about the included studiesQuality evaluation of the included literature

The NOS scores for the included cohort studies are shown in Table 1 and all studies were of high quality with NOS scores between 7 and 9.

Meta-analysis resultsComparison of tumor size

A total of 10 articles [10, 13, 14, 16, 18, 22,23,24,25,26] reported the tumor size of the two groups of patients, involving a total of 1346 patients. There was statistical heterogeneity between the studies (P < 0.00001, I2 = 83%, so a random-effects model was used for data analysis. The results [MD =  − 0.02, 95%CI (− 0.33, 0.30), P = 0.91] indicate that there is an insignificant difference in tumor size between the RATS group and the VATS group. The meta-analysis results are shown in Fig. 2.

Fig. 2

Meta-analysis forest plot of the tumor size

Comparison of operation time

A total of 15 articles [9,10,11,12,13,14, 16, 18, 19, 21, 23,24,25,26,27] reported the operation time of the two groups of patients, involving a total of 1475 patients. There was statistical heterogeneity between the studies (P < 0.00001, I2 = 87%, so a random-effects model was used for data analysis. The results [MD = 0.17, 95%CI (–7.61, 7.94), P = 0.97] indicate that there is a statistically insignificant difference in the operation time between the two surgical methods, suggesting that there is an insignificant difference in the comparison of operation time between the RATS group and the VATS group. The meta-analysis results are shown in Fig. 3.

Fig. 3

Meta-analysis forest plot of the operation time

Comparison of the incidence of intraoperative blood loss

A total of 10 articles [10, 12,13,14, 16, 19, 23,24,25,26] reported the intraoperative blood loss of the two groups of patients, involving a total of 746 patients. There was statistical heterogeneity between the studies (P = 0.005, I2 = 62%, so a random-effects model was used for data analysis. The results [MD = –5.20, 95%CI (− 9.28, − 1.12), P = 0.01] indicate that there is a statistically significant difference in intraoperative blood loss between the two surgical methods, suggesting that the RATS group had less intraoperative blood loss than the VATS group. The meta-analysis results are shown in Fig. 4.

Fig. 4

Meta-analysis forest of the intraoperative blood loss

Comparison of conversion to thoracotomy rate

A total of 6 articles [10, 12,13,14, 16, 19, 23,24,25,26] reported the thoracotomy rate of the two groups of patients, involving a total of 1020 patients. There was no statistical heterogeneity between the studies (P = 0.92 > 0.1, I2 = 0%, so a fixed-effects model was used for data analysis. The results [OR = 0.41, 95%CI (0.23, 0.72), P = 0.002] indicate that there is a statistically significant difference in the conversion to thoracotomy between the two surgical methods, suggesting that the incidence of conversion to thoracotomy in the RATS group was lower than that in the VATS group. The meta-analysis results are shown in Fig. 5.

Fig. 5

Meta-analysis forest plot of the conversion to thoracotomy rate

Comparison of incidence of the total postoperative complications

A total of 10 articles [12, 13, 16, 18, 21, 23,24,25,26,27] reported the total postoperative complications of the two groups of patients, involving a total of 1023 patients. There was no statistical heterogeneity between the studies (P = 0.88 > 0.1, I2 = 0%, so a fixed-effects model was used for data analysis. The results [OR = 0.57, 95%CI (0.34, 0.95), P = 0.03] indicate that there is a statistically significant difference in the incidence of postoperative complications between the two surgical methods, suggesting that the incidence of postoperative complications in the RATS group was lower than that in the VATS group. The meta-analysis results are shown in Fig. 6.

Fig. 6

Meta-analysis forest plot of incidence of the total postoperative complications

Comparison of the incidence of postoperative drainage time

A total of 9 articles [10, 11, 14, 16, 18, 19, 23, 26, 27] reported the postoperative drainage time of the two groups of patients, involving a total of 781 patients. There was statistical heterogeneity between the studies (P < 0.00001, I2 = 92%, so a random-effects model was used for data analysis. The results [MD =  − 0.72, 95%CI (− 1.13, − 0.32), P = 0.0004] indicate that there is a statistically significant difference in the postoperative drainage time between the two surgical methods, suggesting that the postoperative drainage time in the RATS group was shorter than that in the VATS group. The meta-analysis results are shown in Fig. 7.

Fig. 7

Meta-analysis forest plot of postoperative drainage time

Comparison of postoperative hospital stay

A total of 16 articles [10, 11, 13,14,15,16,17,18,19,20,21,22,23,24, 26, 27] reported the postoperative hospital stay of the two groups of patients, involving a total of 3211 patients. There was statistical heterogeneity between the studies (P < 0.00001, I2 = 82%, so a random-effects model was used for data analysis. The results [MD =  − 0.90, 95%CI (− 1.16, − 0.65), P < 0.001] indicate that there is a statistically significant difference in the postoperative hospital stay between the two surgical methods, suggesting that the postoperative hospital stay in the RATS group was shorter than that in the VATS group. The meta-analysis results are shown in Fig. 8.

Fig. 8

Meta-analysis forest plot of postoperative hospital stay

Comparison of hospitalization costs

A total of 4 articles [10, 21, 24, 25] reported the hospitalization costs of the two groups of patients, involving a total of 506 patients. There was significant statistical heterogeneity between the studies (P < 0.00001, I2 = 98%, so a random-effects model was used for data analysis. The results [MD = 2634.75, 95%CI (991.62, 4277.88), P = 0.002] indicate that there is a statistically significant difference in hospitalization costs between the two surgical methods, suggesting that the hospitalization costs of the RATS group were higher than those of the VATS group. The meta-analysis results are shown in Fig. 9.

Fig. 9

Meta-analysis forest plot of hospitalization costs

Sensitivity analysis

Sensitivity analysis was performed on each outcome indicator using the one-by-one deletion method. The results of the sensitivity analysis showed that the effect size results did not change after the above indicators were deleted one by one, indicating that the stability of the results statistically combined after the above indicators is very high.

Publication bias

By drawing funnel plots for each outcome indicator, it was found that all studies were evenly distributed on both sides of the funnel plot, indicating that the publication bias of this study is small. The funnel plot drawn for the incidence of total postoperative complications as an example is shown in Fig. 10.

Fig. 10