Study search and characteristics of included patients

According to our electronic retrieval strategy, a total of 731 studies were identified. After removing duplicates, irrelevant articles, case reports, reviews, and meta-analyses, the full texts of the remaining studies were scrutinized. After excluding studies that did not provide the appropriate data, sixteen studies were included for final assessment and were deemed to be suitable for quantitative meta-analysis [9, 11,12,13,14,15,16,17,18,19,20,21,22,23,24,25]. The process of article selection is summarized in Fig. 1.

Fig. 1

Flow diagram of the study selection process

Study characteristics and publication bias analysis

Table 1 presents a detailed summary of the study characteristics. Among the sixteen included studies, fifteen were retrospective observational studies, and only one was a prospective cohort study. A total of 3685 patients were included, of which 1428 (38.7%) patients underwent uniport VATS, 296 (8%) patients underwent two-port VATS, and 1961 (53.2%) patients underwent three-port VATS. Seven articles used a propensity score or a matched pair method. All the eligible studies had NOS scores ranging from six to eight.

Table 1 Basic characteristics of the publications selected for meta-analysis

Using Comprehensive meta-analysis Version 3.0, we formed funnel plots to assess publication bias for this meta-analysis. Publication bias was evaluated for the blood loss and visual analogue pain score on POD 1. The results are summarized in Fig. 2; symmetrical funnel plots for perioperative blood loss showed that there was little publication bias in our meta-analysis. However, the funnel plot for visual analogue pain score on POD 1 did not show the satisfying symmetry.

Fig. 2

Funnel plots for results included in the meta-analysis: A Blood loss of operation, B Visual analogue pain score on postoperative day 1

Operative outcomesNumber of lymph nodes retrieved

Fourteen studies included data related to the number of lymph nodes retrieved, in which four studies compared two-port VATS with uniport VATS and twelve studies compared three-port VATS with uniport VATS. Thus, two studies involved comparisons of uniport VATS with two-port VATS and three-port VATS, respectively. A combined total of 3048 patients were included. The mean numbers of lymph nodes retrieved for the uniport, two-port, and three-port VATS groups were 18.12 ± 7.6, 18.47 ± 9.2, and 18.13 ± 7.04, respectively; there were no significant differences between the uniport VATS group and the other two VATS groups (uniport vs. three-port: SMD = 0.021, 95% CI − 0.077, 0.118; uniport vs. two-port: SMD = 0.032, 95% CI − 0.243, 0.179). However, low heterogeneity was detected (uniport vs. three-port: p = 0.094, I-squared: 40%; uniport vs. two-port: p = 0.156, I-squared: 43%). The results of this analysis are given in Fig. 3A.

Fig. 3

Forest plots for results: A Number of lymph nodes retrieved, B Operative time, C Conversion rate, D Morbidity rate

Operative time

Sixteen studies included data related to operative time, in which four studies compared two-port VATS with uniport VATS and fourteen studies compared three-port VATS with uniport VATS. Thus, two studies involved comparisons of uniport VATS with two-port VATS and three-port VATS, respectively. A combined total of 3685 patients were included. The mean operative times for the uniport, two-port, and three-port VATS groups were 154.18 ± 37.9 min, 168.58 ± 48.5 min, and 148.84 ± 45.6 min, respectively. A forest plot suggested that compared with multiport VATS groups, regardless of two or three ports, uniport VATS was not associated with an increased operative time (uniport vs. three-port: SMD = 0.160, 95% CI − 0.159, 0.480; uniport vs. two-port: SMD = − 0.043, 95% CI − 0.304, 0.218). High heterogeneity was detected (uniport vs. three-port: p < 0.001, I-squared: 94%; uniport vs. two-port: p = 0.046, I-squared: 63%). The results of this analysis are summarized in Fig. 3B.

Conversion rate

Seven studies included data related to the rate of conversion to open thoracotomy or the need for additional ports, in which two studies compared two-port VATS with uniport VATS and six studies compared three-port VATS with uniport VATS. Thus, only one study involved comparisons of uniport VATS with two-port VATS and three-port VATS, respectively. A combined total of 1695 patients were included. The mean conversion rates for the uniport, two-port, and three-port VATS groups were 5.6%, 3.5%, and 4.0%, respectively. There were no significant differences when comparing the uniport VATS group with the other two VATS groups (uniport vs. three-port: SMD = 0.019, 95% CI − 0.332, 0.370; uniport vs. two-port: SMD = 0.001, 95% CI − 0.220, 0.221). High heterogeneity was detected in the uniport versus three-port analysis (uniport vs. three-port: p < 0.001, I-squared: 87%; uniport vs. two-port: p = 0.858, I-squared: < 0.001%). The results of this analysis are summarized in Fig. 3C.

Morbidity rate

Thirteen studies included comparable data related to overall morbidity, in which four studies compared two-port VATS with uniport VATS and eleven studies compared three-port VATS with uniport VATS. Thus, two studies involved comparisons of uniport VATS with two-port VATS and three-port VATS, respectively. A combined total of 2705 patients were included. The mean morbidity rates for the uniport, two-port, and three-port VATS groups were 14.93%, 10.67%, and 18.70%, respectively. Uniport VATS was not associated with a higher rate of morbidity than two-port or three-port VATS (uniport vs. three-port: SMD = − 0.038, 95% CI − 0.127, 0.051; uniport vs. two-port: SMD = − 0.011, 95% CI − 0.169, 0.147). However, low heterogeneity was detected (uniport vs. three-port: p = 0.661, I-squared: < 0.001%; uniport vs. two-port: p = 0.402, I-squared: < 0.001%). The results of this analysis are summarized in Fig. 3D.

Perioperative blood loss

Fifteen studies included comparable data related to blood loss, in which three studies compared two-port VATS with uniport VATS and thirteen studies compared three-port VATS with uniport VATS. Thus, only one study involved comparisons of uniport VATS with two-port VATS and three-port VATS, respectively. A combined total of 3246 patients were included. As shown in Fig. 4, the uniport approach resulted in a significantly lower bleeding volume compared with two-port VATS (SMD = -0.417, 95% CI − 0.612, − 0.222). The uniport technique also seemed to have lower bleeding volume than three-port VATS, but no statistical distinction was observed (SMD = − 0.120, 95% CI − 0.252, 0.012). High heterogeneity was detected in the uniport versus three-port analysis (uniport vs. three-port: p = 0.003, I-squared: 60%; uniport vs. two-port: p = 0.435, I-squared: < 0.001%).

Fig. 4

Forest plot of perioperative blood loss for uniport VATS, two-port VATS, and three-port VATS groups

Duration of postoperative chest tube drainage

Fourteen studies reported data related to the duration of postoperative chest tube drainage, in which four studies compared two-port VATS with uniport VATS and twelve studies compared three-port VATS with uniport VATS. A combined total of 2757 patients were included. The results are summarized in Fig. 5A. The mean chest tube insertion times for the uniport, two-port, and three-port VATS groups were 4.19 ± 2.4, 4.60 ± 2.4, and 5.29 ± 2.7 days, respectively. There was a significant difference between the uniport and two-port VATS groups (SMD = − 0.224, 95% CI: − 0.417, − 0.030). The comparison between the uniport and three-port VATS groups also showed a significant difference (SMD = − 0.429, 95% CI: − 0.602, − 0.256). High heterogeneity was detected in the uniport versus three-port analysis (uniport vs. three-port: p < 0.001, I-squared: 72%; uniport vs. two-port: p = 0.221, I-squared: 32%).

Fig. 5

Forest plots for results: A Duration of postoperative chest tube drainage, B length of hospital stay

Length of hospital stay

Fifteen studies included comparable data related to length of hospital stay, in which three studies compared two-port VATS with uniport VATS and thirteen studies compared three-port VATS with uniport VATS. Thus, only one study involved comparisons of uniport VATS with two-port VATS and three-port VATS, respectively. A combined total of 2543 patients were included. The results of this analysis are shown in Fig. 5B. The mean lengths of hospital stay for the uniport, two-port, and three-port VATS groups were 7.25 ± 2.9, 8.04 ± 4.0, and 8.36 ± 3.2 days, respectively. There was a significant difference between the uniport and three-port VATS groups but not between the uniport and two-port VATS groups (uniport vs. three-port: SMD = -0.324, 95% CI − 0.489, − 0.158; uniport vs. two-port: SMD = − 0.170, 95% CI − 0.379, 0.039). High heterogeneity was detected in the uniport versus three-port analysis (uniport vs. three-port: p < 0.001, I-squared: 74%; uniport vs. two-port: p = 0.313, I-squared: 14%).

Visual analogue score of pain on postoperative day 1 and postoperative day 3

The results regarding visual analogue score of pain are shown in Fig. 6. Eight studies included data related to pain score, in which two studies compared two-port VATS with uniport VATS and seven studies compared three-port VATS with uniport VATS. Only one study involved comparisons of uniport VATS with two-port VATS and three-port VATS, respectively. A combined total of 1304 patients were included in the analysis of seven studies involving the pain scale on POD 1, and 1518 patients were included in the analysis of seven studies involving the pain scale on POD 3. High heterogeneity was detected regarding the pain scale on POD 1 (uniport vs. three-port: p < 0.001, I-squared: 93%; uniport vs. two-port: p < 0.001, I-squared: 96%). Forest plots showed that uniport VATS was significantly associated with lower pain on postoperative days 1 and 3 when compared with three-port VATS (VAS day 1: SMD = − 0.915, 95% CI − 1.408, − 0.422; VAS day 3: SMD = − 0.653, 95% CI − 1.149, − 0.156). This trend was also observed in the comparison between the uniport and two-port VATS groups (VAS day 1: SMD = − 1.124, 95% CI − 2.298, − 0.049; VAS day 3: SMD = − 0.942, 95% CI − 1.181, − 0.704).

Fig. 6

Forest plots for results: A Visual analogue score of pain on postoperative day 1, B Visual analogue score of pain on postoperative day 3