From November 2000 to November 2022, randomized controlled trials on the treatment of malignant obstructive jaundice with ERCP or PTCD were searched in the EMBASE, PubMed, MEDLINE, and Cochrane databases using the same index terms “ERCP, PTCD, PTBD, MOJ; endoscopic retrograde cholangiopancreatography, percutaneous transhepatic cholangial drainage, malignant obstructive jaundice”. The included literature had to be randomized controlled trials. Retrospective controlled trials, unpublished literature, case reports, and reviews were also excluded. Two researchers reviewed all of the literature and abstracts according to the study’s requirements, excluding unqualified literature and reading the full text of any literature that could potentially be included to determine whether it met the inclusion criteria. All disagreements were resolved by discussion.

Data on the publication year, authors, number of subjects, methodological characteristics, and evaluation indices (technique success, clinical efficacy, and procedure-related complications) were extracted. The bias risk assessment tool provided by the Cochrane Library was used to assess the quality of randomized controlled trials by two researchers independently, including the method of random allocation and whether subjects and study implementers and measurement results were blinded. The tool also assesses whether the data are complete and selective reporting of research results and other possible sources of bias. A consensus was reached after discussion when a controversy arose. Otherwise, divergence was resolved by third parties.

The extracted data were statistically analyzed using the software package Rev Man 5.3. To compare outcomes, the odds ratio (OR) and mean difference (MD) were calculated as effect sizes for dichotomous and continuous variables, respectively, including their combined value and 95% confidence interval (95% CI). A χ2 test was conducted to examine the heterogeneity among the included studies using the inconsistency index (I2) statistic. Heterogeneity was identified as P>0.10, I2>50%, in which a random-effects model was used; otherwise, the fixed-effects model was used for homogeneity, and two-sided P<0.05 was considered statistically significant.

The search strategy identified 1432 articles, of which 154 duplicate articles were excluded, 1256 irrelevant articles were excluded after reading the titles and abstracts, and 22 articles remained initially. Full texts were assessed for eligibility (conference abstracts and full texts without original data for retrieval, duplicate published studies, letters, non-RCTs, retrospective analyses, and reviews were excluded). Finally, seven articles (7–13) were included in this study. Because of the immature technology recorded in the first RCT paper (12), there would have been significant heterogeneity if it was included, and the analysis would not truly reflect the efficacy and safety of the two procedures; consequently, that RCT was ultimately excluded. Figure 1 shows the literature search strategy and screening process, and the quality of the included studies is plotted in Figure 2 . The primary characteristics of the included studies are shown in Table 1 .

The overall technical success rate was reported in all six articles, and there was no heterogeneity among the outcomes; therefore, a statistical analysis was conducted using the fixed effect model. The results of the meta-analysis: Z=3.19, P=0.001, OR=0.31 (95% CI: 0.15-0.64). The difference was statistically significant, and the total success rate of surgery in the PTCD group was significantly higher than that in the ERCP group ( Figure 3 ).

The total clinical efficacy was reported in six studies, and heterogeneity was observed among the results of each study. The random-effects model was applied, and the results of the meta-analysis were as follows: Z=1.76, P=0.08, OR=0.46 (95% CI: 0.20-1.09), indicating that the difference was not statistically significant, and there was no significant difference in total clinical efficacy between the ERCP and PTCD groups ( Figure 4 ).

The incidence of overall procedure-related complications was described in six studies, and there was no heterogeneity among the results of each study. Statistical analysis was conducted using the fixed-effect model, and the results of the meta-analysis were as follows: Z=2.57, P=0.01, OR=0.55 (95% CI: 0.34-0.87), indicating that there was a significant difference in the total complication rate between the two groups, with the PTCD group having a higher overall complication incidence ( Figure 5 ).

The incidence of postoperative cholangitis was reported in all six articles, and there was heterogeneity among the results; thus, the random-effects model was used for statistical analysis. The results of the meta-analysis revealed Z=0.21, P=0.83, OR=0.87 (95% CI: 0.24-3.16), and there was no significant difference in the incidence of postoperative cholangitis between the ERCP and PTCD groups ( Figure 6 ).

Procedure-related pancreatitis was reported in all six articles, and there was no heterogeneity among the results; therefore, statistical analysis was conducted using the fixed-effect model. The results of the meta-analysis were as follows: Z=2.80, P=0.005, OR=5.29 (95% CI: 1.65-16.97). The difference was statistically significant, and the incidence of postoperative pancreatitis in the ERCP group was significantly higher than that in the PTCD group ( Figure 7 ).

There was no heterogeneity among the results; therefore, a fixed-effects model was used for statistical analysis. The results of meta-analysis: Z=1.90, P=0.26, OR=0.54 (95% CI: 0.19-1.58). The difference was statistically significant, and there was no significant difference in the postoperative bleeding rate between the ERCP and PTCD groups ( Figure 8 ).

Publication bias analysis based on a funnel plot of technique success. No publication bias was detected with the observed indicators ( Figure 9 ).

Sensitivity analysis is a crucial component of meta-analysis because it determines the overall credibility of the observed results. The results can be considered reliable if they remain consistent across sensitivity analyses. A meta-analysis of the remaining studies was conducted to assess the stability of the results. Individual investigations were eliminated item by item using a sensitivity analysis. After excluding each study and reintegrating the effect values, all were within the CI. There was no significant difference before removal (I² = 0), showing that the sensitivity of the included literature was low and that the results of this analysis were stable.

This meta-analysis favors PTCD over ERCP for achieving satisfactory technical success as initial treatment in patients with MOJ. Otherwise, the two treatments had the same effectiveness in biliary drainage. A study (19) reported that the ERCP failure rate is approximately 10%, and the reasons for failure include immature techniques, ambiguous identification of the duodenal papilla, anatomical variation, and severe biliary tract stricture or occlusion caused by malignant obstruction. In comparison, PTCD has a higher procedure success rate than ERCP and can be recommended as the first treatment or remedy after ERCP treatment failure. Clinical effectiveness refers to the improvement in jaundice due to biliary drainage. A comprehensive comparison showed that both treatment methods can effectively decompress malignant biliary obstruction and drain bile. There was no statistically significant difference between the two groups in the clinical efficacy of the procedure for malignant obstructive jaundice (P=0.08). A larger-scale study (20) found that patients with morbidities of high obstruction, biliary sepsis, and liver function with a lower Child‒Pugh classification would have poorer drainage effect, regardless of the difference in the patients’ age, sex, diagnosis, number of stents, obstruction, bile duct diameter, abdominal cavity effusion time, intrahepatic lesion, lymph node metastasis, and distant metastasis. Except for these factors, the reason for the same clinical efficacy in the PTCD group following a higher technique success rate could be explained by the fact that ERCP has a better effect on bile drainage. Internal bile drainage is more favorable for bile acid excretion (21). Oral administration of the lost bile from PTCD significantly shortened the time for total bilirubin to return to normal levels in the blood (22). In addition, the definition of clinical efficacy varied. For example, clinical effectiveness was defined as a 50% reduction in the serum total bilirubin level. In the study by Bao et al. (7), the time of decline was defined as less than two weeks, and in the study by Hany et al. (8), clinical effectiveness was defined as a 50% reduction in the serum total bilirubin level within ten days.

In this meta-analysis, ERCP was associated with fewer overall postprocedural adverse events and more procedure-related pancreatitis than PTCD, which is considered a prognostic factor in patients and a reference strategy in the management of MOJ. Mild complications affect the clinical efficacy in patients, while serious complications may cause disease progression or even lead to the death of patients. The mortality rates associated with ERCP and PTCD have been reported to be 0.1% and 2%, respectively (23, 24). In addition to the reasons for the operation itself, the experience of the operator and whether the operator has received systematic training are also correlated with the occurrence of postoperative complications (25). Short-term complications of ERCP and PTCD mainly include biliary infection, acute pancreatitis, hemorrhage biliary leakage, liver abscess, duodenal perforation, and pneumothorax, with an overall complication rate of 10% (26). In this study, there was a significant difference in the total incidence of postprocedural complications between the ERCP and PTCD groups (P=0.01), which differs from the results of another meta-analysis (27) published in 2017. There was an insignificant difference between the two groups, given that most of the included studies were retrospective. Postprocedural pancreatitis is a common complication of endoscopic retrograde cholangiopancreatography (ERCP). The incidence of ERCP-associated pancreatitis reported in the literature (28) is 2.1%-24.4%, and its high-risk factors include repeated intubation, incision of Oddi’s sphincter, and accidental insertion of the main pancreatic duct (29). Subgroup analysis of the included studies showed that the incidence of postoperative pancreatitis in the ERCP group was significantly higher than that in the PTCD group, and the difference was statistically significant. Both the PTCD and ERCP groups were prone to cholangitis, and biliary obstruction was a high-risk factor for cholangitis. In addition, blockage of the drainage stent, stent displacement, and poor drainage effects are common reasons. However, there was no significant difference in the incidence of postoperative cholangitis between the two groups in this study (P=0.83). It was (30, 31) reported that operative bleeding after ERCP and PTCD was 1.6% and 2-3%, respectively. In this study, there was no significant difference in the bleeding rate between the two groups (P=0.26), which was inconsistent with another meta-analysis (32) and may be related to the small sample size and the need for a large RCT sample.

This is the first meta-analysis to compare the efficacy and safety of ERCP with PTCD management of biliary obstruction based on definite RCTs. We systematically evaluated the short-term efficacy and safety of ERCP and PTCD for malignant obstructive jaundice in 6 RCT studies. However, there are still some shortcomings because the inherent limitations of the meta-analysis and the included studies may have weakened our analysis. We could not evaluate the long-term efficacy and safety because such data on 30-day mortality were only provided in one paper. In addition, due to such limitations, we could not analyze the efficacy and safety of different types of procedures. Additionally, there was heterogeneity in a few observation indicators in this study, attributed to the technical variance of operators in different institutions and long time spans, and our comparative analysis of specific complication rates and mortality was limited by the small sample size. Despite these limitations, we believe that our assessment is reliable for comparing the effectiveness and safety of the two methods.