Participants

Included studies will involve patients presenting with MBO and initially undergoing endoscopic drainage, with no age limitation. Both distal and hilum MBO will be included. There will be no restrictions on aetiology, which include, but are not limited to, pancreatic cancer, cholangiocarcinoma, gallbladder cancer, ampulla of vater cancer and metastasis. Patients with benign biliary diseases, and those with EUS-BD performing as a salvage procedure for failed ERCP will be excluded.

Interventions/comparison

The intervention comparisons are EUS-BD versus ERCP-BD. EUS-BD can be performed in several ways, choledochoduodenostomy, hepaticogastrostomy, antegrade procedure and rendezvous technique.15 All these methods of EUS-BD will be included except for rendezvous technique. Because the rendezvous approach is a cross-over technique using EUS to pass a guidewire via the papilla to perform an ERCP. There will be no restrictions on stent type (metal/plastic stent), dilation device (dilation catheter/cystotome/balloon), and whether patient has an indwelling duodenal stent or not.

Outcomes

There are two primary outcomes for this study: technical success (defined as successful stent placement as determined endoscopically or radiographically) and clinical success (defined as reduction of total serum bilirubin levels to less than half of the preoperative level within 4 weeks).10

There are six secondary outcomes: (1) adverse events: total, pancreatitis, cholangitis, cholecystitis, bleeding and bile peritonitis; (2) stent patency (HR) for interval from initial insertion to recurrence of obstruction); (3) stent dysfunction: stent occlusion, stent migration and tumour in/overgrowth; (4) reinterventions; (5) procedure duration and (6) overall survival (HR for death).

Study design

Only randomised controlled trials (RCTs) will be included. Unpublished trials and abstracts will be included if the methodology and data are accessible. We will only include studies that are presented in English language due to constraints in translational resources.

Exclusion criteria will be: (1) studies without a comparative arm of ERCP-BD; (2) observational studies, case reports, reviews, editorials and letters to editor; (3) duplicate studies, in vitro studies or animal studies and (4) no data on any of the primary or secondary outcomes.

Electronic searches

Two investigators (ZJ and YW) will independently search MEDLINE, EMBASE, Web of Science, the Cochrane Library, ClinicalTrials.gov and Google Scholar, for all entries through 31 October 2018. The search strategies will be decided on after a discussion among all reviewers. The primary search strategy will be used for PubMed MEDLINE (online supplementary appendix I). Modifications to the search strategy will be made for other databases. We will assess eligibility of the retrieved articles by title and abstract using predetermined inclusion criteria. If this information is insufficient for eligibility assessment, we will review the full article. If any up-to-date evidence is published during the review period, we will evaluate the eligibility of each study and consider its addition to the analysis.

Searching other resources

To further increase the robustness of the literature search, a manual recursive search of the reference sections of the retrieved articles, as well as the related articles option in PubMed, will be carried out to identify other potentially relevant articles.

Selection of studies

Decisions about study inclusion and exclusion will be made independently by two investigators (ZJ and YW). Disagreements will be resolved by consensus after a mutual discussion. The details of the study selection procedure are shown in a PRISMA flow chart. (figure 1)

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow diagram of the study selection process. ERCP, endoscopic retrograde cholangiopancreatography; EUS, endoscopic ultrasound.

Data extraction and management

Two investigators (YW and HL) will independently extract the appropriate data onto a data collection form (online supplementary appendix II). The following variables will be contained in the collection form: country and year of the study, study design, patient demographics and clinical characteristics, methods of EUS-BD, types of stents, technical success, clinical success, procedure duration, stent patency, stent dysfunction, reinterventions, adverse events, overall survival and follow-up information. When necessary data are not included in the published studies, the corresponding authors will be contacted for additional information. If there is no reply, we will analyse only the available data. If there are no data on any of the primary or secondary outcomes, those studies will be excluded from the meta-analyses.

Assessment of risk of bias in included studies

We will assign two independent investigators (YW and HL) to appraise methodological quality of the included trials with the Cochrane Collaboration’s tool for assessing risk of bias.16 The tool appraises existence of selection bias by assessing methods of randomisation and allocation concealment, performance and detection of biases by checking blinding of personnel and outcome assessment, and attrition and reporting bias by evaluating incomplete and selective data reporting. Each of the items is assigned a judgement of high, low or unclear risk.

Data synthesis

The HRs for time-to-event outcomes (stent patency and overall survival) will be calculated using the Excel sheet published by Tierney et al,17 based on Parmar et al’ s method of data extraction18 from Kaplan-Meier curves. Weighted mean differences (WMDs) will be calculated for continuous variables. Medians will be used if means are not available and SDs will be calculated or imputed when possible.19 Risk ratios (RRs) will be calculated for categorical variables. Owning to the assumption of inherently various study scenarios and study populations, a random effects model for all analyses will be assumed.13 Heterogeneity among studies will be assessed by calculating the I² statistics whereby I² <25% indicates no heterogeneity, 25%≤ I²<50% indicates mild heterogeneity, 50%≤ I²<75% indicates moderate heterogeneity and I² ≥75% indicates strong heterogeneity.20 We had planned that if sufficient studies (≥10) are included in the analysis of primary outcomes, we would construct funnel plots to evaluate publication bias,13 otherwise, Egger’s test will be applied.21 All statistical analyses will be performed using Review Manager V.5.3 (Cochrane Collaboration, 2014).

Subgroup analyses

In the case of possible strong heterogeneity, we will explore the possible sources using subgroup and meta-regression analyses. Subgroup analyses will be carried out based on geographical location, publication form, study design, location of biliary obstruction, indwelling duodenal stent, EUS-BD technique, stent type and definitions of adverse event. For those subgroups with only one study included, subgroup analyses will not be performed.

Sensitivity analysis

We will carry out a sensitivity analysis by systematically removing every study and checking the pooled results for the remaining studies to see if there is any significant change in test performance.