3.1 Study Selection and Characteristics
A flow diagram of our systematic review is shown in Figure 1. The initial search identified 4439 potential studies. After a review of titles and abstracts, 1204 studies were rejected due to data duplication, irrelevant purpose, or comments. One hundred twenty-nine articles were searched for more detailed assessments and full-text reviews. Based on randomization and the sample size of the retrospective study, eight articles were finalized. A total of 12,718 subjects were included, including seven randomized trials and one retrospective study; the data were collected from 1995–2018.The baseline characteristics of the included studies are presented in Table 1.
3.2 Risk of bias and quality of evidence
The bias risk of the included studies was critically assessed using the Cochrane Collaboration risk of bias tool. Seven studies were randomized, and one retrospective study was a large sample study. All of the studies were unblinded and the remaining bias was low, thus the evidence was considered reliable. The bias assessment of each methodologic component from the eligible studies is shown in Figure 2.
(1) Recurrent acute pancreatitis: There was a total of 12,717 patients in 8 studies (4922 in the cholecystectomy group). Pancreatitis occurred in 41 patients (0.83%) after ES in the cholecystectomy group and 177 patients (2.27%) in the wait-and-see group (Figure 3). Recurrent pancreatitis was compared between the resection and gallbladder in situ group (RR, 0.38; 95% CI, 0.27–0.53; P < 0.00001; I2=0%).
The first subgroup analysis was based on a comparison of the high and low risk groups : low-risk group (RR, 1.26; 95% CI, 0.08–19.45; P = 0.87); high-risk group (RR, 0.34; 95% CI, 0.05–2.1; P = 0.24; I2 = 0%); and unclear-risk-group (RR, 0.37; 95% CI, 0.26–0.52; P < 0.00001; I2 = 0%). Another subgroup analysis revealed the following: randomized research group (RR, 0.56; 95% CI, 0.15–2.01; P < 0.37; I2 = 0%); and retrospective study group (RR, 0.37; 95% CI, 0.26–0.52; P < 0.00001).
There is not have I2 statistic in low-risk group and retrospective group, because two trials not have acute pancreatitis in both gallbladder in situ or cholecystectomy group and only trial in retrospective group. Randomized trials were compared with retrospective studies. The incidence of pancreatitis in the resection group was 0.45% and 0.83%, respectively. The incidence of pancreatitis in the gallbladder in situ group was 1.07% and 2.27%, respectively. Finally, the sensitivity analysis was performed using the exclusion method one-by-one. There was no significant difference in pancreatitis between small sample. The large sample study was different from the small sample statistical analysis because the incidence of pancreatitis was low and the difference in large samples was more stable and more apparent. Small sample results were unstable, which causes deviation in the statistical analysis due to individual cases.
(2) Cholangitis and recurrent jaundice (Figure 4): Among the 12,717 patients evaluated the rate of the incidence of cholangitis in the resection group and the gallbladder in situ group (RR, 0.31; 95% CI, 0.26–0.38; P < 0.00001; I2=0%).
The incidence of cholangitis and jaundice in the gallbladder group was less than the preserved gallbladder group. The first subgroup analysis: in low-risk group (RR, 0.56; 95% CI, 0.27–1.16; P = 0.12; I2= 1%), high-risk group (RR, 0.28; 95% CI, 0.09–0.81; P = 0.02; I2= 0%), unclear-risk-group (RR, 0.31; 95% CI, 0.25–0.37; P < 0.00001; I2 = 0%). There was no significant difference in the low-risk group; however, there was a difference between the high-risk and unidentified-risk-groups. Another subgroup analysis: Randomized experimental group (RR, 0.45; 95% CI, 0.25–0.8; P = 0.007; I2 = 0%), retrospective study group (RR, 0.31; 95% CI, 0.26–0.38; P < 0.00001). In the randomized and retrospective groups, the proportion of cholangitis and recurrent jaundice in the final cholecystectomy group was lower than the conservative group.
(3) Mortality (Figure 5): A comparison of mortality between the resection and conservative groups in the 8 studies (RR, 0.73; 95% CI, 0.52–1.02; P = 0.07; I2 = 14%) revealed that there was no significant difference in mortality between the two groups. Subgroup analysis based on risk grading: in the low-risk group (RR, 0.67; 95% CI, 0.36–1.25; P = 0.21; I2 = 0%), high-risk group (RR, 0.87; 95% CI, 0.37–2.03; P = 0.74; I2 = 67%), and unclear-risk-group (RR, 0.53; 95% CI, 0.24–1.18; P = 0.12; I2 = 0%).
All groups showed that there was no significant difference in mortality between the two groups. The high-risk group had higher heterogeneity and adopted a random effect pattern that was mainly caused by the Hammarstrom study using the excavation method. It may be because the sample size of this study was small, and the data have certain contingency.
(4) Biliary colic and cholecystitis (Figure 6): Cholecystitis and biliary colic were compared in the post-operative resection and wait-and-see groups (RR, 0.28; 95% CI, 0.24–0.32; P < 0.00001; I2 = 17%).
The comparison in the subgroup analysis showed the following: low-risk group (RR, 0.06; 95% CI, 0.01–0.23; P < 0.0001; I2 = 0%); high-risk group (RR, 0.11; 95% CI, 0.04–0.32; P < 0.0001; I2 = 0%); and unclear-risk-group (RR, 0.32; 95% CI, 0.27–0.36; P < 0.00001; I2 = 0%). Subgroup analysis based on research typing: Randomized research (RR, 0.08; 95% CI, 0.04–0.19; P < 0.00001; I2 = 0%) and retrospective study groups (RR, 0.30; 95% CI, 0.26–0.34; P < 0.00001).
In all analyses, the incidence of cholecystitis and biliary colic in the early resection of the gallbladder group was significantly lower than the gallbladder group, and the heterogeneity was lower in all subgroups.
3.4 Publishing bias
The funnel plot does not show substantial asymmetry (Figure 2).