Endoscopic sphincterotomy for adults with biliary sphincter of Oddi dysfunction

Abstract

Background

The sphincter of Oddi comprises a muscular complex encircling the distal part of the common bile duct and the pancreatic duct regulating the outflow from these ducts. Sphincter of Oddi dysfunction refers to the abnormal opening and closing of the muscular valve, which impairs the circulation of bile and pancreatic juices.

Objectives

To evaluate the benefits and harms of any type of endoscopic sphincterotomy compared with a placebo drug, sham operation, or any pharmaceutical treatment, administered orally or endoscopically, alone or in combination, or a different type of endoscopic sphincterotomy in adults with biliary sphincter of Oddi dysfunction.

Search methods

We used extensive Cochrane search methods. The latest search date was 16 May 2023.

Selection criteria

We included randomised clinical trials assessing any type of endoscopic sphincterotomy versus placebo drug, sham operation, or any pharmaceutical treatment, alone or in combination, or a different type of endoscopic sphincterotomy in adults diagnosed with sphincter of Oddi dysfunction, irrespective of year, language of publication, format, or outcomes reported.

Data collection and analysis

We used standard Cochrane methods and Review Manager to prepare the review. Our primary outcomes were: proportion of participants without successful treatment; proportion of participants with one or more serious adverse events; and health‐related quality of life. Our secondary outcomes were: all‐cause mortality; proportion of participants with one or more non‐serious adverse events; length of hospital stay; and proportion of participants without improvement in liver function tests. We used the outcome data at the longest follow‐up and the random‐effects model for our primary analyses. We assessed the risk of bias of the included trials using RoB 2 and the certainty of evidence using GRADE. We planned to present the results of time‐to‐event outcomes as hazard ratios (HR). We presented dichotomous outcomes as risk ratios (RR) and continuous outcomes as mean difference (MD) with their 95% confidence intervals (CI).

Main results

We included four randomised clinical trials, including 433 participants. Trials were published between 1989 and 2015.

The trial participants had sphincter of Oddi dysfunction. Two trials were conducted in the USA, one in Australia, and one in Japan. One was a multicentre trial conducted in seven US centres, and the remaining three were single‐centre trials. One trial used a two‐stage randomisation, resulting in two comparisons. The number of participants in the four trials ranged from 47 to 214 (median 86), with a median age of 45 years, and the mean proportion of males was 49%. The follow‐up duration ranged from one year to four years after the end of treatment. All trials assessed one or more outcomes of interest to our review. The trials provided data for the comparisons and outcomes below, in conformity with our review protocol. The certainty of evidence for all the outcomes was very low.

Endoscopic sphincterotomy versus sham

Endoscopic sphincterotomy versus sham may have little to no effect on treatment success (RR 1.05, 95% CI 0.66 to 1.66; 3 trials, 340 participants; follow‐up range 1 to 4 years); serious adverse events (RR 0.71, 95% CI 0.34 to 1.46; 1 trial, 214 participants; follow‐up 1 year), health‐related quality of life (Physical scale) (MD −1.00, 95% CI −3.84 to 1.84; 1 trial, 214 participants; follow‐up 1 year), health‐related quality of life (Mental scale) (MD −1.00, 95% CI −4.16 to 2.16; 1 trial, 214 participants; follow‐up 1 year), and no improvement in liver function test (RR 0.89, 95% CI 0.35 to 2.26; 1 trial, 47 participants; follow‐up 1 year), but the evidence is very uncertain.

Endoscopic sphincterotomy versus endoscopic papillary balloon dilation

Endoscopic sphincterotomy versus endoscopic papillary balloon dilationmay have little to no effect on serious adverse events (RR 0.34, 95% CI 0.04 to 3.15; 1 trial, 91 participants; follow‐up 1 year), but the evidence is very uncertain.

Endoscopic sphincterotomy versus dual endoscopic sphincterotomy

Endoscopic sphincterotomy versus dual endoscopic sphincterotomy may have little to no effect on treatment success (RR 0.65, 95% CI 0.32 to 1.31; 1 trial, 99 participants; follow‐up 1 year), but the evidence is very uncertain.

Funding

One trial did not provide any information on sponsorship; one trial was funded by a foundation (the National Institutes of Diabetes and Digestive and Kidney Diseases, NIDDK), and two trials seemed to be funded by the local health institutes or universities where the investigators worked.

We did not identify any ongoing randomised clinical trials.

Authors' conclusions

Based on very low‐certainty evidence from the trials included in this review, we do not know if endoscopic sphincterotomy versus sham or versus dual endoscopic sphincterotomy increases, reduces, or makes no difference to the number of people with treatment success; if endoscopic sphincterotomy versus sham or versus endoscopic papillary balloon dilation increases, reduces, or makes no difference to serious adverse events; or if endoscopic sphincterotomy versus sham improves, worsens, or makes no difference to health‐related quality of life and liver function tests in adults with biliary sphincter of Oddi dysfunction.

Evidence on the effect of endoscopic sphincterotomy compared with sham, endoscopic papillary balloon dilation,or dual endoscopic sphincterotomyon all‐cause mortality, non‐serious adverse events, and length of hospital stay is lacking.

We found no trials comparing endoscopic sphincterotomy versus a placebo drug or versus any other pharmaceutical treatment, alone or in combination.

All four trials were underpowered and lacked trial data on clinically important outcomes. We lack randomised clinical trials assessing clinically and patient‐relevant outcomes to demonstrate the effects of endoscopic sphincterotomy in adults with biliary sphincter of Oddi dysfunction.

Plain language summary

What are the effects of endoscopic sphincterotomy for adults with biliary sphincter of Oddi dysfunction?

Key messages

– We do not know if endoscopic sphincterotomy compared with sham operation or dual endoscopic sphincterotomy benefits adults with biliary sphincter of Oddi dysfunction, or changes serious or non‐serious unwanted effects, quality of life, and liver function.

– We have no data comparing endoscopic sphincterotomy with sham operation, endoscopic papillary balloon dilation,or dual endoscopic sphincterotomyon deaths, non‐serious unwanted effects, and hospital stay.

– No trial compared endoscopic sphincterotomy with placebo medicine or another medicine, alone or combined.

– We lack randomised clinical studies assessing relevant outcomes in larger numbers of participants.

What is biliary sphincter of Oddi dysfunction?

The sphincter of Oddi is a muscular valve around the base of the ducts from the gallbladder and pancreas. The valve is usually closed (contracted), but it relaxes when eating to allow bile and pancreatic juices into the small intestine helping digest food.

Biliary sphincter of Oddi dysfunction is a condition where the sphincter cannot contract and relax normally, blocking bile flow and resulting in pain.

How is biliary sphincter of Oddi dysfunction treated?

There are many treatments, including medication, endoscopic sphincterotomy (cutting of the sphincter/muscle of Oddi using an endoscope), or surgery (biliary‐enteric drainage).

What is endoscopic sphincterotomy treatment?

Sphincterotomy is a procedure that cuts the sphincter at the end of the bile duct and pancreatic duct to open into the small intestine. Endoscopic means this is done using an endoscope, which is a flexible tube with a camera and light that is inserted from the mouth until it reaches the sphincter. This procedure is used to remove gallstones (small stones containing cholesterol that form in the gallbladder or duct) or any other blockages.

What did we want to find out?

We wanted to know if endoscopic sphincterotomy decreased the number of people with unsuccessful treatment; caused any unwanted effects (for example, death and other serious and non‐serious unwanted effects); and had an impact on quality of life, length of hospital stay, and liver function (measured using blood tests to diagnose and monitor liver disease or damage). To do this, we looked for randomised clinical studies that compared endoscopic sphincterotomy versus a placebo medicine (fake medicine with no treatment effect); a sham operation (fake surgical operation); any treatment with medicines, administered by mouth or through the endoscope, alone or in combination; or a different type of endoscopic sphincterotomy. Randomised clinical studies allocate participants to two or more groups, by chance, compare the effects of the treatments (for example, surgical procedure compared with no treatment, a placebo medicine, or with an existing treatment).

What did we do?

We searched medical databases for randomised clinical trials that fulfilled our predefined criteria. We summarised the results and assessed the quality of the evidence.

What did we find?

We found four trials with 433 participants with sphincter of Oddi dysfunction. All trials had limitations in design, conduct, and reporting of outcomes. The largest trial included 214 participants, and the smallest included 47 participants. Two trials were conducted in the USA, one in Australia, and one in Japan. The trials lasted one to four years. One trial included two comparisons. Only one trial explicitly provided information on sponsorship (funded by the National Institutes of Diabetes and Digestive and Kidney Diseases (NIDDK)). Two trials seemed to be funded by local health centres or universities where the investigators worked.

The trials compared endoscopic sphincterotomy against sham operation (three trials), endoscopic papillary balloon dilation (an alternative method of endoscopic sphincterotomy that uses a balloon to widen a narrowed part of the sphincter) (one trial), or dual endoscopic sphincterotomy (endoscopic sphincterotomy for biliary and pancreatic sphincters) (one trial). No trials compared endoscopic sphincterotomy against a placebo medicine or any treatment with medicines.

We performed only one meta‐analysis (a summary analysis needing data from at least two trials) on the 'number of people with unsuccessful treatment at one to four years after end of treatment', with data from three trials comparing endoscopic sphincterotomy versus sham operation. Because of the small number of randomised participants in the trials, and the small number of trials, we could not reach a conclusion on this outcome. Neither could we reach a conclusion regarding serious unwanted effects because only one trial provided data.

We found inconclusive results for the 'serious unwanted effects' outcome when comparing endoscopic sphincterotomy with endoscopic papillary balloon dilation, and for the 'number of people with unsuccessful treatment at one to four years after end of treatment' outcome when comparing endoscopic sphincterotomy with dual endoscopic sphincterotomy as there was only one trial in each comparison providing data for the two outcomes.

No trials reported death, length of hospital stay, or non‐serious unwanted effects.

What are the limitations of the evidence?

We are very uncertain of the results as participants may have known which treatment they received, there were problems with how the trials were run, and there were few trials and data.

How up to date is this evidence?

The evidence is up to date to May 2023.

Summary of findings

Summary of findings 1. Summary of findings table ‐ Endoscopic sphincterotomy compared with sham treatment for adults with biliary sphincter of Oddi dysfunction.

Endoscopic sphincterotomy compared with sham treatment for adults with biliary sphincter of Oddi dysfunction
Patient or population: adults with biliary sphincter of Oddi dysfunction Setting: hospital Intervention: endoscopic sphincterotomy Comparison: sham treatment
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with sham treatment	Risk with endoscopic sphincterotomy
Proportion of participants without successful treatment at longest follow‐up follow‐up: range 1 years to 4 years	547 per 1000	574 per 1000 (361 to 908)	RR 1.05 (0.66 to 1.66)	340 (3 RCTs)	⊕⊝⊝⊝ Very lowa,b,c	The included trials compared a similar experimental intervention (endoscopic sphincterotomy) with a similar control intervention (i.e. sham procedure) at the defined range of time points. 3 of 4 trials provided data for this analysis. We presented outcomes data analysed at the longest follow‐up, i.e. 1, 2, and 4 years (median 2 years).
Proportion of participants with ≥ 1 serious adverse events at 1 year	151 per 1000	107 per 1000 (51 to 220)	RR 0.71 (0.34 to 1.46)	214 (1 RCT)	⊕⊝⊝⊝ Very lowd	All 26 participants reported pancreatitis.
Quality of life measured with SF‐36 (Physical Component Summary) at 12 months		MD 1 lower (3.84 lower to 1.84 higher)	‐	214 (1 RCT)	⊕⊝⊝⊝ Very lowd	Higher score indicates a more favourable health state.
Quality of life measured with SF‐36 (Mental Component Summary) at 12 months		MD 1 lower (4.16 lower to 2.16 higher)	‐	214 (1 RCT)	⊕⊝⊝⊝ Very lowd	Higher score indicates a more favourable health state.
All‐cause mortality (or overall survival) ‐ not reported	‐	‐	‐	‐	‐	No trial reported this outcome.
Proportion of participants with ≥ 1 non‐serious adverse events ‐ not reported	‐	‐	‐	‐	‐	No trial reported this outcome.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_441764493269622060.

^a Downgraded one level for study limitations (some concerns or high risk of bias).
^b Downgraded one level for inconsistency (I² > 50%).
^c Downgraded two levels for very serious imprecision (CIs included appreciable benefit and harm, fewer than 400 participants).
^d Downgraded three levels for extremely serious imprecision (CIs included appreciable benefit and harm, based on a single trial, and fewer than 400 participants).

Summary of findings 2. Summary of findings table ‐ Endoscopic sphincterotomy compared with endoscopic papillary balloon dilation for adults with biliary sphincter of Oddi dysfunction.

Endoscopic sphincterotomy compared with endoscopic papillary balloon dilation for adults with biliary sphincter of Oddi dysfunction
Patient or population: adults with biliary sphincter of Oddi dysfunction Setting: hospital Intervention: endoscopic sphincterotomy Comparison: endoscopic papillary balloon dilation
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with endoscopic papillary balloon dilation	Risk with endoscopic sphincterotomy
Proportion of participants without successful treatment ‐ not reported	‐	‐	‐	‐	‐	The trial did not report this outcome.
Proportion of participants with ≥ 1 serious adverse events at 1 year	65 per 1000	22 per 1000 (3 to 205)	RR 0.34 (0.04 to 3.15)	91 (1 RCT)	⊕⊝⊝⊝ Very lowa,b	4 participants reported recurrent common bile duct stones.
Health‐related quality of life ‐ not reported	‐	‐	‐	‐	‐	The trial did not report this outcome.
All‐cause mortality (or overall survival) ‐ not reported	‐	‐	‐	‐	‐	The trial did not report this outcome.
Proportion of participants with ≥ 1 non‐serious adverse events ‐ not reported	‐	‐	‐	‐	‐	The trial did not report this outcome.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_441786773322584551.

^a Downgraded one level for study limitations (unclear risk of bias).
^b Downgraded three levels due to extremely serious imprecision (CIs included appreciable benefit and harm, based on a single trial, and fewer than 400 participants).

Summary of findings 3. Summary of findings table ‐ Biliary endoscopic sphincterotomy compared with dual endoscopic sphincterotomy for adults with biliary sphincter of Oddi dysfunction.

Biliary endoscopic sphincterotomy compared with dual endoscopic sphincterotomy for adults with biliary sphincter of Oddi dysfunction
Patient or population: adults with biliary sphincter of Oddi dysfunction Setting: hospital Intervention: biliary endoscopic sphincterotomy Comparison: dual endoscopic sphincterotomy
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with dual endoscopic sphincterotomy	Risk with biliary endoscopic sphincterotomy
Proportion of participants without successful treatment at 1 year	298 per 1000	194 per 1000 (95 to 390)	RR 0.65 (0.32 to 1.31)	99 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
Proportion of participants with ≥ 1 serious adverse events ‐ not reported	‐	‐	‐	‐	‐	The trial did not report this outcome.
Health‐related quality of life ‐ not reported	‐	‐	‐	‐	‐	The trial did not report this outcome.
All‐cause mortality (or overall survival) ‐ not reported	‐	‐	‐	‐	‐	The trial did not report this outcome.
Proportion of participants with ≥ 1 non‐serious adverse events ‐ not reported	‐	‐	‐	‐	‐	The trial did not report this outcome.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_441786783286153707.

^a Downgraded one level for study limitations (unclear risk of bias).
^b Downgraded three levels for extremely serious imprecision (CIs included appreciable harm and benefit, based on a single trial, and fewer than 400 participants).

Background

Description of the condition

The sphincter of Oddi comprises a muscular complex encircling the distal part of the common bile duct and the pancreatic duct regulating it. The sphincter of Oddi also prevents duodenal contents from entering the common bile duct and pancreatic duct (Della Libera 2007; Dowdy 1962). Sphincter of Oddi dysfunction – also known as papillary stenosis, sclerosing papillitis, biliary spasm, biliary dyskinesia, and postcholecystectomy syndrome (Geenen 1980) – refers to the abnormal opening and closing of the muscular valve, which impairs the circulation of bile and pancreatic juices. The condition manifests clinically with pain in the upper right quadrant of the abdomen, consistent with biliary or pancreatic ductal origin (Petersen 2004; Toouli 1989). People with sphincter of Oddi dysfunction may present with diarrhoea, fever, nausea, vomiting, loss of appetite, weight loss, and other symptoms (Durkalski 2010; Toouli 1999). Experts are still unsure what causes sphincter of Oddi dysfunction; it could be related to microlithiasis or inflammation of the first section of the small intestine. However, the condition has many risk factors such as agenesis of the gallbladder, preoperative cholelithiasis, lithotripsy, liver transplant, alcohol use, hypothyroidism, and irritable bowel syndrome (Kim 2022).

The true incidence of biliary sphincter of Oddi dysfunction remains unknown, as the diagnosis is based on excluding organic pathologies of the biliary tract (Behar 2006; Durkalski 2010). Biliary sphincter of Oddi dysfunction is frequently considered in the differential diagnosis of postcholecystectomy abdominal pain, particularly recurrent postcholecystectomy pain in women (Freeman 2007). When bile acid refluxes to the pancreatic tract, clinical symptoms of acute recurrent pancreatitis appear (Freeman 2007; Geenen 1980).

According to the Rome IV Criteria, a person diagnosed with functional biliary sphincter of Oddi disorder must have:

• symptoms of biliary pain;

• elevated liver enzymes or a dilated bile duct (but not both); and

• no bile duct stones or other structural abnormalities (Cotton 2016).

Supportive criteria are:

• normal amylase or lipase;

• abnormal manometry;

and

• hepatobiliary scintigraphy signs (Cotton 2016; Kyanam Kabir Baig 2016).

There are three types of biliary sphincter of Oddi dysfunction, characterised by the following key features (Cotton 2016; Kyanam Kabir Baig 2016).

• Biliary sphincter of Oddi dysfunction type I: pain and abnormal hepatic enzymes on two occasions and dilated bile duct.

• Biliary sphincter of Oddi dysfunction type II: pain and abnormal hepatic enzymes on two occasions, or dilated bile duct.

• Biliary sphincter of Oddi dysfunction type III: pain with normal hepatic enzymes and normal bile duct.

In the clinical context, fluctuating symptoms complicate the management of affected people (Cotton 2016).

There is no specific medical treatment for biliary sphincter of Oddi dysfunction. Pharmacological treatment includes analgesics (e.g. paracetamol, tramadol, or codeine phosphate), used alone or in combination with amitriptyline‐containing medications (Kalaitzakis 2010), antidepressants (e.g. amitriptyline or duloxetine), calcium channel blockers (e.g. nifedipine or nicardipine), nitrates, or ursodeoxycholic acid (Craig 2002; Kalaitzakis 2010; Pauls 2016). Many of these interventions have raised concerns over unacceptable rates of adverse effects, including peripheral oedema, or failure to reduce pain episodes (Bistritz 2006; Craig 2002; Vitton 2012). There is limited evidence on the effects of these drugs (Bistritz 2006; Craig 2002; Sgouros 2006), as only a few small trials have been conducted, with short follow‐up (between 12 weeks and 16 weeks of active treatment) (Bistritz 2006). Acupuncture is one non‐pharmacological alternative (Zhao 2017), and endoscopic treatments include botulinum toxin injection and endoscopic sphincterotomy (Cotton 2004; Pasricha 1994). Some researchers have suggested that intermittent botulinum toxin injection could be used in people with type III biliary sphincter of Oddi dysfunction (Hackert 2017; Menon 2020; Pauls 2016). The irreversible and invasive surgical practice of removing the biliary sphincter has raised ethical concerns, especially when the indicators for sphincterotomy are unexplained (e.g. recurring biliary pain, with or without abnormal liver enzymes or dilated bile duct).

Description of the intervention

The management of biliary sphincter of Oddi dysfunction aims to normalise the functional motility of the sphincter (Corazziari 1999; Sherman 2002). Drainage of biliary and pancreatic secretions into the duodenum can be achieved through pharmacological treatment, endoscopic treatment (sphincterotomy), or surgical treatment (sphincteroplasty and biliary‐enteric drainage) (Tzovaras 1998).

Endoscopic sphincterotomy of the biliary sphincter of Oddi is used to treat different pathological conditions of the papilla of Vater, or to diagnose people with biliary diseases and choose the optimal therapeutic strategy (Bistritz 2006; Köksal 2018). It is a technically challenging invasive endoscopic procedure performed under visual and fluoroscopic guidance (Köksal 2018). Reported complications include postprocedure pancreatitis (in 18% of operations) (Freeman 2007), bleeding (1.7%) (Kwak 2020), infection (3.7%) (Kwak 2020), and perforation (less than 1%) (Freeman 2007; Kalaitzakis 2010).

One study with 38 participants reported similar morbidity and mortality in endoscopic treatment and surgical treatment (i.e. transduodenal sphincteroplasty) of sphincter of Oddi dysfunction (Schwartz 2019). In one narrative review, eight studies showed a morbidity rate of 13% to 39% in participants treated with endoscopic sphincterotomy, and another nine studies showed a morbidity rate of 7% to 35% in participants treated with surgical sphincteroplasty (Tzovaras 1998).

Potential adverse events should be taken into consideration when selecting people for endoscopic sphincterotomy. The choice of accessories and procedural techniques depends on the anatomy of individuals and the preference of endoscopists (Krutsri 2019; Neuhaus 2019). The application of technical guidelines is important, but many recommendations are based on a low level of evidence (Neuhaus 2019).

How the intervention might work

Biliary endoscopic sphincterotomy refers to the cutting of the biliary sphincter and intraduodenal segment of the common bile duct following selective cannulation, using a high‐frequency current applied with a sphincterotome knife inserted into the papilla (Köksal 2018). The procedure is used to relieve bile duct obstruction and to prevent frequent episodes of high bile duct pressure in all three types of biliary sphincter of Oddi dysfunction (Bistritz 2006).

Why it is important to do this review

The Cochrane review 'Sphincterotomy for biliary sphincter of Oddi dysfunction' by Craig and colleagues has been archived in the Cochrane Library because the last searches for trials were performed in 2000(Craig 2001). We identified one non‐Cochrane review that assessed endoscopic sphincterotomy for managing sphincter of Oddi dysfunction (Sgouros 2006). It included two randomised trials and eight non‐randomised studies, and the review authors concluded that endoscopic sphincterotomy was associated with long‐term symptomatic improvement in 69% (122/177) of people with manometrically confirmed type II sphincter of Oddi dysfunction (Sgouros 2006). We found no other systematic reviews.

We were aware of a new trial of EPISOD after 2006 (Evaluating Predictors and Interventions in SOD (sphincter of Oddi dysfunction); Cotton 2014), and we decided to perform a new systematic review with meta‐analyses, using current Cochrane methodology (Higgins 2022a), and including only randomised clinical trials that assessed the benefits and harms of sphincterotomy in people with sphincter of Oddi dysfunction.

Objectives

To evaluate the benefits and harms of any type of endoscopic sphincterotomy compared with a placebo drug, sham operation, or any pharmaceutical treatment, administered orally or endoscopically, alone or in combination, or a different type of endoscopic sphincterotomy in adults with biliary sphincter of Oddi dysfunction.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised clinical trials with a parallel‐group design that assessed the benefits and harms of sphincterotomy for adults with biliary sphincter of Oddi dysfunction. We planned to include trials reported as full‐text articles, those published as abstract only, and unpublished data. We included trials irrespective of whether they reported the outcomes specified in our review. We did not apply language or publication date restrictions.

We excluded pseudo‐randomised studies (i.e. quasi‐randomised studies) as the method of allocation to the study groups is not truly random.

Types of participants

We included adults (aged over 18 years) diagnosed with biliary sphincter of Oddi dysfunction according to the Rome IV diagnostic criteria or other criteria used by the trial authors.

We planned to include trials also with a subset of participants with biliary sphincter of Oddi dysfunction as long as data on these participants were reported separately, or if we could obtain disaggregated data from the trial authors.

Types of interventions

Experimental interventions

• Any type of endoscopic sphincterotomy

Control interventions

• Placebo drug

• Sham operation (i.e. sham intervention that omits the therapeutical step (Ciccozzi 2016))

• Any pharmaceutical treatment (nifedipine, trimebutine, or nitrates; injection of botulinum toxin; or others), administered orally or endoscopically

• Endoscopic sphincterotomy, different from that used in the experimental group

We considered co‐interventions if used equally in the experimental and control groups of a trial.

Types of outcome measures

We analysed the outcomes below, regardless of the follow‐up time points, using intention‐to‐treat (ITT) or per‐protocol analysis depending on the availability of data. The primary analysis on which we based our main conclusions included the outcome data at the longest follow‐up.

Primary outcomes

• Proportion of participants without successful treatment, where people with successful treatment:

  • had a Recurrent Abdominal Pain Intensity and Disability score (RAPID score)^a of fewer than six days of lost productivity at longest follow‐up after the endoscopic sphincterotomy procedure (Cotton 2014);

  • had not undergone a second intervention (Cotton 2014); and

  • did not require opioid treatment during months 10, 11, and 12, unless they needed it for a reason other than abdominal pain and used it for 14 days or less (Cotton 2014).

• Proportion of participants with one or more serious adverse events. We consider an event to be serious if it fulfilled the definition of serious adverse events of the International Conference on Harmonisation (ICH) Guidelines, that is, any event that:

  • led to death;

  • was life‐threatening;

  • required hospitalisation or prolongation of existing hospitalisation;

  • resulted in persistent or significant disability;

  • was a congenital anomaly or birth defect; or

  • was an important medical event that could have jeopardised the patient or could have required intervention to prevent any of the other outcomes listed above (ICH‐GCP 2016).

• Health‐related quality of life, measured with validated questionnaires such as the World Health Organization (WHO) quality of life questionnaire (WHOQOL), the European quality of life measure (EuroQol), or the 36‐item Short‐Form Health Survey (SF‐36).

^a RAPID is a reliable instrument for measuring disability resulting from abdominal pain in people with suspected sphincter of Oddi dysfunction.The RAPID score instrument is practical, and the high levels of concordance support its reliability for consistently measuring pain disability in people with suspected type III sphincter of Oddi dysfunction (Durkalski 2010).

Secondary outcomes

• All‐cause mortality (or overall survival if data for all‐cause mortality were missing)

• Proportion of participants with one or more non‐serious adverse events (see definition of serious adverse event in Primary outcomes)

• Length of hospital stay

• Proportion of participants without improvement in liver function tests (e.g. unchanged or increased activity of conjugated bilirubin, alanine aminotransferase, aspartate aminotransferase, or pancreatic enzymes (amylase and lipase))

When a trial publication did not report on one or more of the above outcomes, we checked the trial protocol, if available, to see whether the trial authors omitted any outcomes from the identified publications. If they did, we contacted them to enquire about the reason for omitting the outcomes. We did not exclude such trials but summarised our findings in a narrative format only.

Search methods for identification of studies

To minimise bias in our search results, our searches followed the guidance in Chapter 4 of the Cochrane Handbook for Systematic Reviews of Interventions (Lefebvre 2022a) and PRISMA‐S (Rethlefsen 2021).

Electronic searches

The Cochrane Hepato‐Biliary Group Information Specialist searched The Cochrane Hepato‐Biliary Group Controlled Trials Register via the Cochrane Register of Studies Web on 16 May 2023. We also searched the Cochrane Central Register of Controlled Trials (2023, Issue 5) in the Cochrane Library, MEDLINE ALL Ovid (1946 to 16 May 2023), Embase Ovid (1974 to 16 May 2023), LILACS (VHL Regional Portal; 1982 to 16 May 2023), Science Citation Index Expanded (1900 to 16 May 2023), and Conference Proceedings Citation Index‐Science (1990 to 16 May 2023). The latter two were searched simultaneously through the Web of Science platform.

Appendix 1 includes the search strategies with the date range of the searches.

Searching other resources

We searched the US Food and Drug Administration (FDA) website (www.fda.gov), the European Medicines Agency (EMA) website (www.ema.europa.eu/ema/), the WHO International Clinical Trial Registry Platform (www.who.int/ictrp), and ClinicalTrial.gov (clinicaltrials.gov/) for ongoing or unpublished trials. We also contacted relevant individuals and organisations for information about ongoing or unpublished trials.

We searched for relevant grey literature such as reports, dissertations, theses, and conference abstracts (e.g. in Google Scholar (scholar.google.com)).

We used the PubMed/MEDLINE 'similar articles search' tool on all included studies. We manually checked citations and reference lists of the included studies and any relevant systematic reviews we identified.

We contacted authors of identified trials for additional published or unpublished trials.

We searched these other resources in May 2023.

We used items from the PRISMA‐S checklist that were relevant to our review to ensure that we reported and documented our searches as advised (PRISMA-S Checklist; Rethlefsen 2021).

We examined papers for any retraction statements and errata as errata can reveal important limitations or even fatal flaws in included studies (Lefebvre 2022a; Lefebvre 2022b; Moylan 2016; Wager 2011). We checked for retracted studies at www.zotero.org/blog/retracted-item-notifications/ and retractiondatabase.org/RetractionSearch.aspx.

Data collection and analysis

We conducted the review according to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2022a). We used Review Manager to collect data and perform the analyses (RevMan 2023). We used Trial Sequential Analysis software to assess imprecision as a sensitivity analysis (Thorlund 2017; TSA 2021).

Selection of studies

Two review authors (CN and HHA) independently screened the titles and abstracts of the search results and coded the references as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. We retrieved the full‐text reports of all potentially eligible trials, and two review authors (CN and HHA) independently screened the studies for inclusion, recording the reasons for the exclusion of ineligible studies. We settled any disagreements through discussion. We identified and excluded duplicates and collated multiple reports of the same study so that each study, rather than each report, was the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Page 2021a; Page 2021b). We included trials regardless of whether they reported our outcomes of interest.

For screening of non‐English language publications, we, in the first instance, used Google Translate to assist eligibility assessment (translate.google.com). If needed, we sought translators through the Cochrane Hepato‐Biliary Group to assist with eligibility assessment and data extraction.

During the selection of trials, if we identified observational studies on the topic of our review (e.g. quasi‐randomised studies, cohort studies, or case reports) that had reported adverse events during the study period, we included the reported data in a narrative synthesis at the end of the results section but did not include them in a meta‐analysis. We did not specifically search for observational studies for inclusion in this review, which is a limitation. We are aware that by not searching for all observational studies on adverse events, we risk placing more weight on potential benefits than on potential harms, and overlooking uncommon and late adverse events (Storebø 2018).

Data extraction and management

We used a data collection form for study characteristics and outcome data, which was piloted on at least two trials in the review. We prepared a table with study characteristics, based on PICOT (participants, interventions, comparison, outcomes, and time), to explore and compare study elements across studies. This enables us to group trials for each of our comparisons. We then entered this information in the Characteristics of included studies table. We also recorded when outcome data were not reported in a usable way in the Characteristics of included studies table. We resolved disagreements by consensus, involving all review authors.

Two review authors (HN and CN) independently entered the following data into the Characteristics of included studies table​​​ in Review Manager (RevMan 2023).

• Methods: study design, study period, study centres and location, study setting, withdrawals/dropouts, and date of study.

• Contact for correspondence.

• Trial registration; study protocol published.

• Ethics committee approval.

• Participants: mean age, age range, sex, diagnostic methods, severity of condition, baseline liver function, inclusion and exclusion criteria.

• Interventions: experimental intervention, comparison, concomitant medications, and excluded medications.

• Outcomes: planned outcomes in trial protocol and reported outcomes in a trial.

• Length of follow‐up.

• Language of publication.

• Multiple publications identified and type of publications.

• Sample size calculation performed and reported; ITT or per‐protocol.

• Notes: funding for studies, notable conflicts of interest of trial authors, and any other information on correspondence with trial authors.

We resolved any disagreement through discussion. One review author (NHH) checked whether the data were entered correctly in Review Manager (RevMan 2023).

We recorded whether trials measured adverse events as the number of participants with an adverse event or the number of adverse events per participant. We also recorded occasions where multiple events in a participant have been incorrectly treated as independent events (Peryer 2022).

Assessment of risk of bias in included studies

Two review authors (CN and HN) independently assessed the risk of bias for each study using the RoB 2 tool (www.riskofbias.info/) (Higgins 2022b; Sterne 2019). We resolved any disagreements by consensus involving a third review author (HHA).

In our bias risk assessments, we used the ITT principle, which includes all randomised participants, irrespective of the interventions that participants actually received.

We considered the following five domains to assess the risk of bias in individually randomised trials (Higgins 2022b).

• Bias arising from the randomisation process.

• Bias due to deviations from intended interventions.

• Bias due to missing outcome data.

• Bias in measurement of an outcome.

• Bias in selection of the reported result.

Each domain used a series of signalling questions with the answers (yes, probably yes, no information, probably no, no) (Appendix 2), using the guidance of Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2022b). We determined the risk of bias in each domain (low risk, some concerns, and high risk). We included text alongside the judgements to provide supporting information for our decisions (see Risk of bias in included studies).

We assigned one of the three levels of judgement to each domain as indicated below.

• Low risk of bias: the trial was judged at low risk of bias for all domains for the outcome result.

• Some concerns: the trial was judged to raise some concerns in at least one domain for the outcome result, but was not at high risk of bias for any of the remaining domains.

• High risk of bias: the trial was judged at high risk of bias in at least one domain for the outcome result, or the study was judged to have some concerns for multiple domains in a way that substantially lowered confidence in the result.

The overall risk‐of‐bias judgement was the same as for the individual domains such as low risk of bias, some concerns, or high risk of bias. Judging a result to be at a particular level of risk of bias for an individual domain implied that the result had an overall risk of bias at least this severe. We used the RoB 2 Microsoft Excel tool to assess the risk of bias and produce an overall risk of bias assessment (available at: www.riskofbias.info/welcome/rob-2-0-tool/current-version-of-rob-2).

We summarised the risk of bias in the forest plots and text, based on ITT. The risk of bias assessments fed into one domain of the GRADE approach for assessing the certainty of a body of evidence (Schünemann 2022a; Schünemann 2022b). We focused on assessing risk of bias in the trial outcomes, presented below, at the longest follow‐up, as this contributed the most useful information to users of our review.

• Proportion of people without successful treatment.

• Proportion of people with one or more serious adverse events.

• Health‐related quality of life.

• All‐cause mortality (or overall survival).

• Proportion of people with one or more non‐serious adverse events.

• Length of hospital stay.

Measures of treatment effect

We reported the risk ratio (RR) and its 95% confidence intervals (CI) for dichotomous outcomes. We planned to report continuous outcomes using the mean difference (MD) and 95% CI if we could perform a meta‐analysis, using data from trials that used the same scale. We also planned to use standardised mean difference (SMD) and its 95% CI to report outcomes when different scales were used to measure the same outcome. We used the trial authors' definitions for overall survival. If time‐to‐event outcome data were provided as dichotomous data at a fixed time point (e.g. at least 12 months), we constructed a 2 × 2 table and expressed intervention effects as RR (Higgins 2022c). Otherwise, we planned to report the hazard ratio (HR) and its 95% CI.

Unit of analysis issues

The unit of analysis in randomised parallel‐group trials is the participant as randomised to the trial groups.

For dichotomous outcomes (e.g. presence/absence of a serious adverse event), we used participants as units of analysis, rather than events (i.e. the number of participants with a hospital admission rather than the number of admissions per participant). However, if a trial reported rate ratios, we planned to analyse these based on events rather than participants. Where a single trial reported multiple trial groups, we planned to include only the trial groups that were relevant for our comparisons. We recorded whether the trial measured serious adverse events as participants with any serious adverse events or the number of serious adverse events per participant. We also planned to record occasions where multiple events, in a participant, had been incorrectly treated as independent without considering the interdependence of the events. Where the number of events appeared to be equal to the number of participants, we planned to treat the events as the unit of analysis (Higgins 2022d).

Dealing with missing data

We contacted study authors to verify key study characteristics and obtain missing numerical outcome data (Geenen 1989; Takezawa 2004; Toouli 2000). When we did not receive a reply, and we believed the missing data could have introduced serious bias, we considered this in the GRADE rating for the affected outcomes.

We performed an ITT analysis whenever possible (Newell 1992). If there were missing standard deviations (SDs) for our continuous outcome (health‐related quality of life), we planned to contact the corresponding author to request any available data. If the data were not available, we planned to calculate the SDs from standard errors, CIs, t values, or P values (as appropriate) related to the differences between means in two groups, following the guidance described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2022c). When there was insufficient information to calculate the SDs, we planned to impute them. We planned to replace missing SDs for 'change from baseline' with those provided in other trials for the same outcome. If this approach was not applicable, assuming that correlation coefficients from the two intervention groups were similar, we might have imputed an SD of the change from baseline for the experimental intervention, following a formula presented in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2022). However, this approach was not necessary for this review as all trials reported the SDs.

We planned to perform a full ITT analysis by imputing missing participants' values in sensitivity analysis according to the following two extreme scenarios (Hollis 1999 [https://revman.cochrane.org/#/375420102716162345/dashboard/htmlView/1.40#REF‐Hollis‐1999]).

• Extreme‐case analysis favouring the experimental intervention ('best‐worst' case scenario): no dropouts/participants lost to follow‐up in the experimental arm, and all dropouts/participants lost to follow‐up in the control arm, were assumed to have experienced the outcome, with all randomised participants in the denominator.

• Extreme‐case analysis favouring the control intervention ('worst‐best' case scenario): all dropouts/participants lost to follow‐up in the experimental arm, and no dropouts/participants lost to follow‐up in the control arm, were assumed to have experienced the outcome, with all randomised participants in the denominator.

Assessment of heterogeneity

We planned to explore and describe clinical and methodological heterogeneity, considering the characteristics and the design features of the trial, participants, interventions, comparators, outcomes, treatment duration, and follow‐up. We provided a summary of the trial design, participants, interventions, comparisons, outcomes, time, funding, and the risk of bias assessments in the Included studies section.

We planned to assess statistical heterogeneity by visual inspection of forest plots, considering the direction and magnitude of effects, and the degree of overlap between CIs. We planned to evaluate statistical heterogeneity with the Chi² and I² statistics, using P < 0.10 as a cut‐off point for statistical heterogeneity, and interpret this as described by Deeks 2022.

• 0% to 40%: might not be important

• 30% to 60%: may represent moderate heterogeneity

• 50% to 90%: may represent substantial heterogeneity

• 75% to 100%: considerable heterogeneity

If we identified substantial heterogeneity (I² statistic greater than 50%), we planned to report it and explore the possible causes by prespecified subgroup analyses.

Assessment of reporting biases

We planned to assess publication biases, but this was not possible as there were fewer than 10 included studies (i.e. one trial per outcome and comparison) (Higgins 2022b).

Data synthesis

When a sufficient number of clinically similar trials were available, we performed meta‐analyses using the random‐effects model as our main analysis and the fixed‐effect model as a sensitivity analysis. We presented all results with 95% CIs. We entered data for analyses in Review Manager (RevMan 2023), according to the guidance provided in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2022).

We planned not to conduct meta‐analysis if there was a considerable unexplained heterogeneity, or if trials reported outcomes differently (e.g. impossible to calculate the same effect measure from the available statistics) as described in Chapter 12 (Table 12.1.a) of the Cochrane Handbook for Systematic Reviews of Interventions (McKenzie 2022). In such a setting, our plan was to summarise the main findings and results of the included trials in a narrative format.

Subgroup analysis and investigation of heterogeneity

In the event of substantial clinical, methodological, or statistical heterogeneity, we planned to identify the possible reasons for heterogeneity by evaluating individual trials and their subgroup characteristics. We also planned to carry out the following subgroup analyses if data permitted and regardless of the presence of heterogeneity.

• Trials at low risk of bias compared to trials with some concern and trials at high risk of bias together, as trials at some concerns or at high risk of bias may overestimate beneficial intervention effects or underestimate harmful intervention effects (Higgins 2022b).

• Trials without for‐profit funding compared with trials with for‐profit funding, as trials with for‐profit funding may overestimate beneficial intervention effects or underestimate harmful intervention effects (Lundh 2017).

• Men compared to women, as the incidence of biliary sphincter of Oddi dysfunction is higher amongst women (Bistritz 2006).

• Comparison of the three different types of biliary sphincter of Oddi dysfunction, as this may help to reveal which type is best addressed with sphincterotomy. This is also important given the controversy surrounding type III sphincter of Oddi dysfunction (Wilcox 2015).

We planned to use the Chi² test for subgroup comparisons.

Sensitivity analysis

We planned to carry out the following sensitivity analyses for the primary outcomes.

• Removing outlying studies, defined as studies with results that conflict with the rest of the studies (Deeks 2022).

• Excluding trials at overall high risk of bias.

• Conducting the analyses with the fixed‐effect model.

• Performing extreme‐case analyses (as reported in Dealing with missing data).

• Assessing imprecision with Trial Sequential Analysis.

Trial Sequential Analysis

We calculated the diversity‐adjusted required information size (DARIS) using the following parameters for dichotomous outcomes (Wetterslev 2009): proportion of events in the control group estimated from the included trials; anticipated intervention effect (relative risk reduction, RRR) of 15%; alpha of 2.5% because of the three primary outcomes, and beta of 10% (90% power) (Jakobsen 2014; Wetterslev 2017). For continuous outcomes, we planned to use a minimal relevant difference equal to SD/2; SD of the control group; alpha of 2.5% because of three primary outcomes; beta of 10% (90% power); and diversity of the meta‐analysis. We planned to add trials according to the year of publication and at any risk of bias. Based on the required information size, we planned to construct the trial sequential monitoring boundaries for benefits, harms, and futility using the Lan‐DeMets‐O'Brien‐Fleming alpha‐spending function. If the trial sequential monitoring boundary is crossed before the required information size is reached, a sufficient level of evidence is reached, results of the meta‐analysis can be considered conclusive if bias can be excluded, and no additional trials may be needed. Conversely, if the boundary is not crossed, the meta‐analysis is inconclusive, and more trials may be needed to detect or reject a certain intervention effect. When the cumulative Z‐curve crosses the futility boundaries, a sufficient level of evidence is reached that the two treatments do not differ by more than 15% (anticipated intervention effect used in information size estimation), and no additional trials may be needed. In all situations where no trial sequential monitoring boundaries are reached, further studies may be needed until the information size is reached, or until monitoring boundaries are crossed. In Trial Sequential Analysis where the cumulative Z‐value does not cross the monitoring boundaries for benefit, harm, or futility, the assessment of imprecision in GRADE (see below) is downgraded by two levels if the accrued number of participants is below 50% of the DARIS, and by one level if between 50% and 100% of DARIS. We did not downgrade for imprecision if the cumulative Z‐value reached or crossed benefit, harm, futility, or DARIS (TSA 2021). A more detailed description of the Trial Sequential Analysis method is available (see www.ctu.dk/tsa/; Thorlund 2017; TSA 2021).

Summary of findings and assessment of the certainty of the evidence

We created summary of findings tables for the following comparisons.

• Endoscopic sphincterotomy compared with sham treatment.

• Endoscopic sphincterotomy compared with endoscopic papillary balloon dilation.

• Biliary endoscopic sphincterotomy compared with dual endoscopic sphincterotomy.

We presented outcomes data analysed at the longest follow‐up (Table 1; Table 2; Table 3). We provided the time point for assessing the outcome, the range of the longest follow‐up, and the median when there was more than one trial in an analysis.

We used Review Manager (RevMan 2023), and assessed the certainty of evidence of the outcomes below in the summary of findings tables.

• proportion of people without successful treatment;

• proportion of people with one or more serious adverse events;

• health‐related quality of life;

• all‐cause mortality (or overall survival);

• proportion of people with one or more non‐serious adverse events.

Two review authors (HN and CN) independently conducted GRADE assessments using GRADEpro GDT. The two review authors resolved any discrepancy through discussion with a third review author (DK) until a consensus was reached. We used the five GRADE factors: risk of bias (the overall RoB 2 judgement), heterogeneity, imprecision, indirectness, and publication bias, to assess the certainty of the body of evidence, as the certainty of evidence relates to the trials that contribute data to the meta‐analyses for the prespecified outcomes. We justified all decisions to downgrade the certainty, using footnotes and comments whenever needed to help the reader understand our assessments. Regarding risk of bias, we used the overall judgement for an outcome result. 'Low' risk of bias indicates 'no limitation' (the certainty is not downgraded); 'some concerns' indicates either 'no limitation' or 'serious limitation' (the certainty is downgraded one level); and 'high' risk of bias indicates either 'serious limitation' or 'very serious limitation' (the certainty is downgraded two levels). We used the methods and recommendations described in Chapter 14 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2022a), and GRADEpro GDT.

The levels of evidence are defined as 'high', 'moderate', 'low', or 'very low' (Schünemann 2022a).

• High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.

• Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.

• Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.

• Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

We conducted the review according to our published protocol (Naing 2022). We reported any deviations from it in the Differences between protocol and review section.

Results

Description of studies

See Characteristics of included studies and Characteristics of excluded studies tables.

Results of the search

We identified 749 records from electronic database searches, conducted on 16 May 2023. We identified no additional references by handsearching the reference list of articles retrieved through these electronic database searches or by searching other resources. After removing 154 duplicates, we screened 595 references. Based on title, abstract, or both, we excluded 574 records. Amongst the remaining 21 full‐text records, we excluded 14 records with reasons for exclusion. We identified no ongoing trials. The remaining four trials (seven records) met our inclusion criteria.

The flow of records is given in the PRISMA flow diagram (Figure 1; Page 2021a; Page 2021b).

1.

PRISMA Diagram (Page 2021a; Page 2021b). Date of search 16 May 2023

Included studies

See Characteristics of included studies table for more details.

We included four randomised clinical trials of parallel‐group design (Cotton 2014; Geenen 1989; Takezawa 2004; Toouli 2000). The trials were published between 1989 and 2015, and were conducted in three countries: one in Australia (Toouli 2000), one in Japan (Takezawa 2004), and two in the USA (Cotton 2014; Geenen 1989). One was a multicentre trial conducted in seven centres in the USA (Cotton 2014), and the remaining three were single‐centre trials. In three trials, baseline demographic and clinical characteristics were comparable between the experimental and control groups (Cotton 2014; Geenen 1989; Takezawa 2004). One trial did not provide this information (Toouli 2000).

Participants

The four trials included 433 adults with biliary sphincter of Oddi dysfunction. The number of trial participants in the four trials ranged from 47 to 214 (median 86). About 22.5% of the participants were men, and the median age was 45 years (Table 4). All four trials included participants with various types of biliary sphincter of Oddi dysfunction. For instance, type III in Cotton 2014 (214 participants), and type II in Geenen 1989 (47 participants). Two trials did not provide this information (Takezawa 2004; Toouli 2000).

1. Description of the included studies.

Study	Country	Site	Participants	Male, n (%)	Median age (range), (years)	Clinical condition	Type of endoscopic sphincterotomy	Comparator	Funding
Cotton 2014	USA	7 centres	214	17 (7.9)	(19–64)	Type III postcholecyctectomy pain without significant abnormalities on imaging or laboratory studies	NA.	Sham treatment	NIDDDK
Geenen 1989	USA	Single	47	NA	NA	Type II	Traction‐type papillotome (Wilson Cook), USA	Sham treatment	Medical College of Wiscosina
Takezawa 2004	Japan	Single	91	62 (68.1)	69 (41–93)b	For removal of common bile duct stones	KD 22‐Q Olympus	Endoscopic papillary balloon dilation (BARD, USA and Wilson‐Cook, USA)	NA
Toouli 2000	Australia	Single	81c	8 (9.8)	45 (22–71)	Recurrent biliary‐type pain median of 2 years (range: 6 months to 30 years) after cholecystectomy.	Demling‐type papillotome	Sham treatment	Flinders Medical Centre, Adelaide, in collaboration with the Royal North Shore Hospital, Sydneya

^a Indicated as the study was approved/where the study was located.
^b Indicates data from the endoscopic sphincterotomy group.
^c Indicates recruited numbers (data analysis was available for 79).
n: number of participants; NA: not available/not reported; NIDDK: the National Institutes of Diabetes and Digestive and Kidney Diseases.

Three of the four trials included men and women (Cotton 2014; Takezawa 2004; Toouli 2000). In two trials, women predominated (Cotton 2014; Toouli 2000); in one trial, men predominated (Takezawa 2004). One trial did not provide this information (Geenen 1989) (Table 4).

Experimental intervention

The trials used equipment from different manufacturers or various forms of papillotomy to perform endoscopic sphincterotomy. The endoscopic sphincterotomy groups included 141 participants (Cotton 2014), 23 participants (Geenen 1989), 45 participants (Takezawa 2004), and 37 participants (Toouli 2000).

Control interventions

We identified the following three types of control interventions: sham treatment (73 participants (Cotton 2014); 24 participants (Geenen 1989); 42 participants (Toouli 2000)); endoscopic papillary balloon dilation (46 participants (Takezawa 2004)); and dual biliary and pancreatic sphincterotomy (47 participants (Cotton 2014)).

We found no trials using a placebo drug or any other pharmaceutical treatment.

Sham

This was a medical procedure, analogous to a placebo drug, where participants undergo surgical procedures that have the appearance of therapeutic interventions but during which the essential therapeutic manoeuvre is omitted (Tenery 2002). It is also known as 'placebo surgery'. For instance, participants in the control arm were prepared for surgery and had incisions of papilla (papillotomy) after which the 'noises' of sphincterotomy were created without proceeding to actual sphincterotomy (Toouli 2000).

Endoscopic papillary balloon dilation

This procedure involves passing a balloon over a preposition guide wire and being placed in the bile duct opening (Takezawa 2004).

Dual biliary and pancreatic sphincterotomy

This is a two‐procedure approach. People with biliary‐type pain and pancreatic sphincter hypertension are treated with pancreatic sphincterotomy in addition to biliary sphincterotomy (Cotton 2014).

Comparisons

The four trials made the following comparisons:

• endoscopic sphincterotomy versus sham (Cotton 2014; Geenen 1989; Toouli 2000);

• endoscopic sphincterotomy versus endoscopic papillary balloon dilation (Takezawa 2004);

• endoscopic sphincterotomy versus dual endoscopic sphincterotomy (Cotton 2014).

Outcomes

The four trials reported a limited number of outcomes. These included the primary outcomes of 'proportion of people without successful treatment' (i.e. treatment failure), 'health‐related quality of life', and 'serious adverse events' (i.e. pancreatitis, recurrent common bile duct stones), and the secondary outcome of 'proportion of participants without improvement in liver function tests'.

Three trials reported the proportion of people without successful treatment (i.e. treatment failure) (Cotton 2014; Geenen 1989; Toouli 2000), albeit with variation in measurement tools and time points for follow‐ups. Cotton and colleagues used the RAPID score for pain measurement (Cotton 2014), while the other two trials did not indicate a specific measurement tool for pain score. Two trials used a subjective assessment, such as 'good', 'fair', or 'no improvement' (Geenen 1989), and 'no change (worse)', 'improved symptoms', or 'asymptomatic' (no further episodes of typical abdominal pain) (Toouli 2000), using structured (standardised) questionnaire for recording the participants' symptoms after the intervention. The directions of validity of the RAPID score and the unspecified pain scores in the structured (standardised) questionnaire are likely comparable. Therefore, we could meta‐analyse the three trials.

Two trials reported serious adverse events (Cotton 2014; Takezawa 2004); however, they used different control interventions, and therefore, we did not meta‐analyse the data. Cotton 2014 reported serious adverse events such as proportion of postendoscopic retrograde cholangiopancreatography (post‐ERCP) pancreatitis. Post‐ERCP pancreatitis isdefined astypical pain associated with at least a three‐fold increase in serum amylase or lipase within seven days after ERCP with symptoms notable enough to require admission to hospital for treatment (or extension of an existing or planned admission) (Cotton 2014). Takezawa 2004 reported one serious adverse event (recurrent common bile duct stone). These two types of serious adverse events did not have the same level of clinical significance. For instance, while post‐ERCP pancreatitis was a potentially fatal illness, participants with recurring common bile duct stones required hospitalisation but were not commonly fatal.

One trial measured health‐related quality of life in two aspects, physical and mental, using the SF‐36 Physical and Mental Component Summaries (Cotton 2014).

No trials reported any other outcomes of interest.

Follow‐up

Follow‐up varied in the four trials: immediately after the procedure (Takezawa 2004); after the end of treatment (Cotton 2014; Takezawa 2004); at two years (Toouli 2000); and at one year and at a maximum of four years (Geenen 1989).

Dropouts

One trial reported that a total of 23/241 (9.5%) participants had missing or late RAPID scores and no record of narcotic use or re‐intervention: these were 15/141 (10.6%) in the experimental group and 8/73 (11%) in the control group (Cotton 2014).

One trial reported that a total of 7/47 (15%) participants were lost for the four‐year follow‐up: 5/23 (22%) in the experimental group and 2/24 (8%) in the control group (Geenen 1989). Four of these seven participants had relocation issues, and three refused to continue participation.

One trial reported that a total of 5/91 (5.8%) participants refused to undergo ERCP at one‐week follow‐up: 2/45 (4.4%) participants in the experimental group and 3/46 (6.5%) participants in the control group (Takezawa 2004).

One trial reported that two participants were excluded after randomisation: one participant was lost to follow‐up and one was withdrawn due to two episodes of manometric procedure‐related pancreatitis (Toouli 2000). However, the trial did not report to which group the two participants belonged.

Funding

One trial explicitly provided information about the funding sources;the National Institutes of Diabetes and Digestive and Kidney Diseases (NIDDK) (Cotton 2014). Two trials appeared to be funded by local health centres or universities where researchers worked (Geenen 1989; Toouli 2000). One trial did not provide any information on clinical trial support or sponsorships (Takezawa 2004).

Excluded studies

See Characteristics of excluded studies table.

We excluded 14 studies during the full‐text review of these publications since they failed to meet the eligibility criteria of our review. Four studies were about the efficacy of other experimental interventions (Fullarton 1992; Guelrud 1988; Khuroo 1992; Wehrmann 1996), three studies were narrative reviews (Craig 2002; Tang 2018; Thatcher 1987), four studies were retrospective (Ren 2021; Sugawa 2001) or prospective observational studies (Linder 2003; Neoptolemos 1988), one study was a controlled clinical study (Cicala 2002), one was a 'views' article (Cotton 2019), and one included participants with other conditions (Swan 2013).

None of the studies of possible interest in our review were retracted or withdrawn (Moylan 2016; Wager 2011).

Risk of bias in included studies

See Appendix 3, Table 5, and Figure 2.

2. Risk of bias assessment.

Study ID	Experimental	Comparator	Outcome	D1	D2	D3	D4	D5	Overall
Cotton 2014	Endoscopic sphincterotomy	Sham treatment	Quality of life (mental)	Low risk	Low risk	Low risk	Low risk	Low risk	Low risk
Cotton 2014	Endoscopic sphincterotomy	Sham treatment	Quality of life (physical)	Low risk	Low risk	Low risk	Low risk	Low risk	Low risk
Cotton 2014	Endoscopic sphincterotomy	Sham treatment	Proportion without treatment success	Low risk	Low risk	Low risk	Low risk	Low risk	Low risk
Geenen 1989	Endoscopic sphincterotomy	Sham treatment	Proportion without treatment success	Low risk	Some concerns	High risk	Low risk	Some concerns	High risk
Toouli 2000	Endoscopic sphincterotomy	Sham treatment	Proportion without treatment success	Some concerns	Some concerns	Low risk	Low risk	Some concerns	Some concerns
Cotton 2014	Endoscopic sphincterotomy	Sham treatment	Serious adverse events	Low risk	Low risk	Low risk	Low risk	Low risk	Low risk
Geenen 1989	Endoscopic sphincterotomy	Sham treatment	No improvement in liver function tests	Low risk	Some concerns	High risk	Low risk	Some concerns	High risk
Takezawa 2004	Endoscopic sphincterotomy	Endoscopic papillary balloon dilation	Serious adverse events	Low risk	Some concerns	Some concerns	Low risk	Some concerns	High risk
Cotton 2014	Endoscopic sphincterotomy	Dual endoscopic sphincterotomy	Proportion without treatment success	Some concerns	Some concerns	Low risk	Low risk	Low risk	Some concerns

D1: randomisation process.
D2: deviations from the intended interventions.
D3: missing outcome data.
D4: measurement of the outcome.
D5: selection of the reported result.

2.

Following our protocol, we could perform RoB 2 assessment at a specific time point (longest follow‐up) for:

• proportion of participants without successful treatment;

• serious adverse events;

• health‐related quality of life.

A risk of bias summary of the included trials, for each comparison analysis, is shown in Table 5 and Figure 2. It can also be visualised to the right of the forest plot of each outcome. Further details on how the RoB 2 tool was applied to each domain and for each trial outcome can be found in Appendix 3.

We present details on the implications of assessments of risk of bias for each specific result in the Effects of interventions section.

Effects of interventions

See: Table 1; Table 2; Table 3

See Table 1, Table 2, and Table 3.

We could perform meta‐analyses only for the comparison 'Endoscopic sphincterotomy versus sham treatment' as the included trials compared a similar experimental intervention (endoscopic sphincterotomy) with a similar control intervention (i.e. sham procedure) at the defined range of time points. The other comparisons included only one trial each.

There were no trials comparing endoscopic sphincterotomy versus a placebo drug, or any other pharmacological treatment.

1. Endoscopic sphincterotomy versus sham

Three trials compared endoscopic sphincterotomy versus sham (Cotton 2014; Geenen 1989; Toouli 2000). We used the longest available follow‐up data per outcome as the primary time point.

Primary outcomes

1.1. Proportion of participants without successful treatment at the longest follow‐up (one to four years)

Three trials reported 132/201 (65.7%) participants had no treatment success in the experimental groups versus 76/139 (54.7%) participants in the control groups (Cotton 2014; Geenen 1989; Toouli 2000). We performed our analysis based on the ITT principle with data from Cotton 2014 and Geenen 1989, and per protocol from Toouli 2000. Endoscopic sphincterotomy compared with sham, at one to four years' follow‐up, may have little to no effect on treatment success (RR 1.05, 95% CI 0.66 to 1.66; 3 trials, 340 participants; very low‐certainty evidence; Table 1; Analysis 1.1).

1.1. Analysis.

Comparison 1: Endoscopic sphincterotomy versus sham operation, Outcome 1: Proportion of participants without treatment success at the longest follow‐up (1–4 years)

The follow‐up outcome data were for one year in Cotton 2014, two years in Toouli 2000, and four years in Geenen 1989.

The certainty of evidence was downgraded to very low because of study limitations (some concerns or high risk of bias); inconsistency (I² = 65%); and very serious imprecision (CIs included appreciable benefit and harm, fewer than 400 participants).

1.2. Proportion of participants with serious adverse events

One trial reported 15/141 (11%) participants had serious adverse events (i.e. post‐ERCP pancreatitis) in the experimental group versus 11/73 (15%) participants in the control group at one‐year follow‐up (Cotton 2014). Endoscopic sphincterotomy compared with 'sham treatment' may have little to no effect on serious adverse events (RR 0.71, 95% CI 0.34 to 1.46; 1 trial, 214 participants; very low‐certainty evidence; Table 1; Analysis 1.2).

1.2. Analysis.

Comparison 1: Endoscopic sphincterotomy versus sham operation, Outcome 2: Proportion of participants with serious adverse events (pancreatitis, at 1 year)

The certainty of evidence was downgraded to very low because of extremely serious imprecision (CIs included appreciable benefit and harm, based on one trial, and fewer than 400 participants).

1.3. Health‐related quality of life at one year

One trial provided data for physical and mental health‐related quality of life using the SF‐36 Physical and Mental Component Summaries (scale from 0 to 100 with 100 representing the best possible health state) (Cotton 2014). There were twice as many participants in the experimental group (141) than in the control group (73).

It is very uncertain whether there is a difference in the effect of endoscopic sphincterotomy compared with sham on physical health‐related quality of life (MD −1.00, 95% CI −3.84 to 1.84; 1 trial, 214 participants; very low‐certainty evidence; Table 1; Analysis 1.3).

1.3. Analysis.

Comparison 1: Endoscopic sphincterotomy versus sham operation, Outcome 3: Health‐related quality of life (measured with SF‐36 Physical scale at 1 year)

It is very uncertain whether endoscopic sphincterotomy compared with sham has a better effect on mental health‐related quality of life (MD −1.00, 95% CI −4.16 to 2.16; 1 trial, 214 participants; very low‐certainty evidence; Table 1; Analysis 1.4).

1.4. Analysis.

Comparison 1: Endoscopic sphincterotomy versus sham operation, Outcome 4: Health‐related quality of life (measured with SF‐36 Mental scale at 1 year)

The certainty of evidence was downgraded to very low because of extremely serious imprecision (CIs included appreciable benefit and harm, based on a single trial, and fewer than 400 participants).

Secondary outcomes

1.4. All‐cause mortality (or overall survival if data for all‐cause mortality were missing)

None of the three trials reported this outcome.

1.5. Proportion of participants with one or more non‐serious adverse events

None of the three trials reported this outcome.

1.6. Length of hospital stay

None of the three trials reported this outcome.

1.7. Proportion of participants without improvement in liver function tests

One trial found 6/23 (26%) participants had no improvement in liver function tests in the experimental group versus 7/24 (29%) in the control group (Geenen 1989). The result on whether there was a difference in effect between endoscopic sphincterotomy versus sham treatment at one‐year follow‐up was very uncertain (RR 0.89, 95% CI 0.35 to 2.26; 1 trial, 47 participants; Analysis 1.5).

1.5. Analysis.

Comparison 1: Endoscopic sphincterotomy versus sham operation, Outcome 5: Proportion of participants without improvement in liver function tests

2. Endoscopic sphincterotomy versus endoscopic papillary balloon dilation

One trial compared endoscopic sphincterotomy versus endoscopic papillary balloon dilation (Takezawa 2004).

Primary outcomes

2.1. Proportion of participants without successful treatment

The trial did not report this outcome.

2.2. Proportion of participants with serious adverse events

The trial provided data for this outcome at one‐year follow‐up, which was also the end of treatment. The trial found 1/45 (2%) participants had serious adverse events (recurrent common bile duct stones) in the experimental group versus 3/46 (6%) participants in the control group. The result is very uncertain (RR 0.34, 95% CI 0.04 to 3.15; 1 trial, 91 participants; very low‐certainty evidence; Table 2; Analysis 2.1).

2.1. Analysis.

Comparison 2: Endoscopic sphincterotomy versus endoscopic papillary balloon dilation, Outcome 1: Proportion of participants with serious adverse events (recurrent common bile duct stones, at 1 year)

The certainty of evidence was downgraded to very low because of study limitations (unclear risk of bias), and extremely serious imprecision (CIs included appreciable benefit and harm, based on a single trial, and fewer than 400 participants).

2.3. Health‐related quality of life

The trial did not report this outcome.

Secondary outcomes

2.4. All‐cause mortality (or overall survival if data for all‐cause mortality were missing)

The trial did not report this outcome.

2.5. Proportion of participants with one or more non‐serious adverse events

The trial did not report this outcome.

2.6. Length of hospital stay

The trial did not report this outcome.

2.7. Proportion of participants without improvement in liver function tests

The trial did not report this outcome.

3. Endoscopic sphincterotomy versus dual endoscopic sphincterotomy

Cotton 2014 also compared endoscopic sphincterotomy with dual endoscopic sphincterotomy in a second‐stage randomisation during which 99 participants with pancreatic sphincter hypertension were randomised to two groups; 52 participants were to undergo biliary sphincterotomy and 47 were to undergo dual sphincterotomy.

Primary outcomes

3.1. Proportion of participants without treatment success

One trial provided data for this outcome at one‐year follow‐up. Based on the ITT analysis, the trial reported 10/52 (19%) participants had no treatment success in the experimental group versus 14/47 (30%) participants in the control group (Cotton 2014). The result on whether there is a difference in effect between endoscopic sphincterotomy and dual endoscopic sphincterotomy on the proportion of people without treatment success is very uncertain (RR 0.65, 95% CI 0.32 to 1.31; 1 trial, 99 participants; very low‐certainty evidence; Table 3; Analysis 3.1).

3.1. Analysis.

Comparison 3: Biliary endoscopic sphincterotomy alone versus dual endoscopic sphincterotomy, Outcome 1: Proportion of participants without treatment success (at 1 year)

The certainty of evidence was downgraded to very low because of study limitations (risk of bias), and extremely serious imprecision (CIs included appreciable harm and benefit, based on a single trial, and fewer than 400 participants).

3.2. Proportion of participants with serious adverse events

The trial did not report this outcome.

3.3. Health‐related quality of life

The trial did not report this outcome.

Secondary outcomes

3.4. All‐cause mortality (or overall survival if data for all‐cause mortality were missing)

The trial did not report this outcome.

3.5. Proportion of participants with one or more non‐serious adverse events

The trial did not report this outcome.

3.6. Length of hospital stay

The trial did not report this outcome.

3.7. Proportion of participants without improvement in liver function tests

The trial did not report this outcome.

GRADE judgement for all outcomes and comparisons

We assessed the certainty of evidence as very low because of risk of bias, inconsistency, and imprecision (Table 1; Table 2; Table 3). Very low‐certainty evidence means that, irrespective of the effect estimate, the true effect can be substantially different from it.

Subgroup analysis and investigation of heterogeneity

We could not perform all planned subgroup analyses because of insufficient data.

Endoscopic sphincterotomy versus sham

The subgroup analysis we could perform was for risk of bias in the trials that reported the proportion of participants without successful treatment (Cotton 2014; Geenen 1989; Toouli 2000; Analysis 1.1). The statistical heterogeneity was substantial (I² = 65%). The subgroup analysis of the trial at low risk of bias showed a harmful effect of endoscopic sphincterotomy on treatment success (RR 1.23, 95% CI 1.01 to 1.49; Analysis 4.1). The subgroup analysis of the remaining two trials combined (i.e. with some concern and at high risk of bias) showed that endoscopic sphincterotomy may have little to no effect on treatment success (RR 0.94, 95% CI 0.41 to 2.18). Using P < 0.10 as a cut‐off point for statistical heterogeneity, there was no significant subgroup difference (test for subgroup differences: Chi² = 0.36, df = 1 (P = 0.55), I² = 0%; Analysis 4.1). However, the results of the subgroup analyses can, by no means, be considered informative and can be misleading, mainly because of insufficient trials per subgroup and the underlying very low‐certainty evidence. In such a situation, the subgroup analyses are unlikely to detect any true differences between subgroups.

4.1. Analysis.

Comparison 4: Subgroup analysis: endoscopic sphincterotomy versus sham, Outcome 1: Risk of bias. Proportion of participants without treatment success at the longest follow‐up (1–4 years)

Sensitivity analysis

In the comparison of endoscopic sphincterotomy versus sham for the proportion of participants without successful treatment, only one trial out of three was at low risk of bias and the same trial was an outlying result (i.e. that endoscopic sphincterotomy may result in an increase in the proportion of participants without successful treatment) (RR 0.65, 95% CI 0.36 to 1.19; Analysis 4.1.1; Cotton 2014).

Fixed‐effect model

The result produced with the fixed‐effect model on the proportion of participants without successful treatment (RR 1.14, 95% CI 0.94 to 1.39) is comparable to the result with the random effects‐model (RR 1.05, 95% CI 0.66 to 1.66; Analysis 1.1).

Best‐worst case scenario

We assumed that all participants who were lost to follow‐up in the experimental group had treatment success, and all those participants who were lost to follow‐up in the control group did not have treatment success (RR 0.91, 95% CI 0.62 to 1.33; Analysis 5.1). There was no difference in treatment failure between the interventions.

5.1. Analysis.

Comparison 5: Sensitivity analysis: proportion of participants without treatment success at the longest follow‐up (1–4 years), Outcome 1: Best‐worst case scenario analysis

Worst‐best case scenario

We assumed that all participants who were lost to follow‐up in the experimental group did not have treatment success, and that all those participants who were lost to follow‐up in the control group had treatment success (RR 1.29, 95% CI 0.77 to 2.17; Analysis 5.2). There was no difference in treatment failure between the interventions.

5.2. Analysis.

Comparison 5: Sensitivity analysis: proportion of participants without treatment success at the longest follow‐up (1–4 years), Outcome 2: Worst‐best case scenario analysis

Trial Sequential Analysis

1. Endoscopic sphincterotomy versus sham

Without treatment success at the longest follow‐up

Three trials provided data for the proportion of people without treatment success at one to four years after randomisation (Cotton 2014; Geenen 1989; Toouli 2000). The events proportion in the control group was 54.7% (76/139), an RRR of 15%, an alpha (type I error) of 2.5%, and a beta (type II error) of 10% (90% power). The boundary of the DARIS was ignored because of too little information. Figure not shown.

Health‐related quality of life one year after randomisation

We did not perform Trial Sequential Analysis for physical health‐related quality of life one yearafter randomisation because only one trial was included, which was insufficient to estimate the DARIS (Cotton 2014).

We did not perform Trial Sequential Analysis for mental health‐related quality of life one yearafter randomisation because only one trial was included, which was insufficient to estimate the DARIS (Cotton 2014).

Serious adverse events

We did not perform Trial Sequential Analysis for serious adverse events one yearafter randomisation because only one trial was included, which was insufficient to estimate the DARIS (Cotton 2014).

2. Endoscopic sphincterotomy versus endoscopic papillary balloon dilation

Serious adverse events

We did not perform Trial Sequential Analysis for serious adverse events one yearafter randomisation because only one trial was included, which was insufficient to estimate the DARIS (Takezawa 2004).

3. Endoscopic sphincterotomy versus dual endoscopic sphincterotomy

Without treatment success at one year

We did not perform Trial Sequential Analysis for the proportion of people without treatment success, one year after randomisation because only one trial was included, which was insufficient to estimate the DARIS (Cotton 2014).

Reporting bias

We could not assess reporting bias by creating a funnel plot for any of the comparisons because there were fewer than 10 trials per comparison.

Harms reported in quasi‐randomised studies or observational studies, identified while searching for randomised clinical trials

Brawman‐Mintzer and colleagues conducted a naturalistic follow‐up study of 72 participants from one included study (see reference under Cotton 2014). It reported that 5/72 (7%) participants had current depression (measured using Diagnostic and Statistical Manual of Mental Disorders‐IV depressive disorders), 6/72 (8%) participants had current anxiety (measured using Diagnostic and Statistical Manual of Mental Disorders‐IV anxiety disorders), and 14/72 (19%) participants had a positive physical or sexual abuse (or both) history.

Discussion

Summary of main results

We included four randomised clinical trials involving 433 adults with sphincter of Oddi dysfunction who received endoscopic sphincterotomy. Two of the trials included participants with type II (Geenen 1989) and type III sphincters of Oddi dysfunction (Cotton 2014). The remaining two trials did not provide specific information about the classification type (Takezawa 2004; Toouli 2000). Trials were conducted in Australia (Toouli 2000), Japan (Takezawa 2004), and the USA (Cotton 2014; Geenen 1989). Of these, one was a multicentre trial conducted in seven US centres (Cotton 2014). Trials received funding from the National Institutes of Diabetes and Digestive and Kidney Diseases (NIDDK) (Cotton 2014) (Table 4), and local health centres or universities (Geenen 1989; Toouli 2000); one trial provided no information on funding or sponsorship of clinical trials (Takezawa 2004).

Outcomes

Three trials compared endoscopic sphincterotomy with sham (Cotton 2014; Geenen 1989; Toouli 2000). We were able to perform a meta‐analysis for only one outcome in one comparison. Trial authors assessed and reported the outcome at a range of time points, using similar measures. For the remaining two outcomes (i.e. serious adverse events and quality of life), we were unable to perform a meta‐analysis because of the different control interventions, which were endoscopic papillary balloon dilation (Takezawa 2004) and dual endoscopic sphincterotomy (Cotton 2014). None of the included trials compared endoscopic sphincterotomy with a placebo drug or any other pharmaceutical treatment.

The evidence is very uncertain on the proportion of people without treatment success when comparing endoscopic sphincterotomy intervention with sham at one‐ to four‐year follow‐up. We are unsure why Cotton 2014 showed an outlying result on the outcome proportion of participants without treatment success at the longest follow‐up (range between one and four years) (Analysis 1.1). The meta‐analysed result showed substantial heterogeneity (I² = 65%). This could be associated with the risk of bias in two of the trials (i.e. one at high risk of bias (Geenen 1989) and one with some concerns for risk of bias (Toouli 2000)). Our subgroup analysis stratified according to risk of bias showed harm (i.e. more participants without treatment success) in the trial at low risk of bias (Cotton 2014), whereas the trial with some concern for risk of bias and the trial at high risk of bias showed that sphincterotomy may have little to no effect on treatment success (Geenen 1989; Toouli 2000). Both of these trials crossed the line of no effect. Both trials were smaller than Cotton 2014, with an inherent risk of random error. The heterogeneity may also be due to the severity of the disease as the trial participants in Cotton 2014 had sphincter of Oddi dysfunction type III, while in Geenen 1989, the participants had sphincter of Oddi dysfunction type II. As mentioned previously, Toouli 2000 did not report the sphincter of Oddi dysfunction type. We conducted sensitivity analyses for extreme cases (i.e. best‐worst and worst‐best‐case scenario analyses) on the outcome of the proportion of people without treatment success. Our sensitivity analysis as well as Trial Sequential Analysis revealed that there was insufficient information to reach firm conclusions.

There were no data on secondary outcomes such as all‐cause mortality, non‐serious adverse events, and length of hospital stays. Only one trial reported data on the proportion of participants without improvement in liver function tests (Geenen 1989).

The results of our review should be interpreted with great caution because only one of the four trials was at low risk of bias. Due to trial design and imprecision of results, the remaining three trials were at a high risk of bias or had some concerns of bias. The outcome data were subject to the risk of both type I (alpha) and type II errors (beta). As such, there is a high risk of both systematic and random error with the importance of 'play of chance' for the reliability of estimates based on sparse data (Keus 2010).

One naturalistic follow‐up study of the participants in Cotton 2014 by Brawman‐Mintzer and colleagues in 2014 included 72 participants (see reference under Cotton 2014). However, this follow‐up study did not report data on the outcomes relevant to our review.

Summary of ongoing studies

We found no ongoing trials. This may reflect that research in this area is not considered of interest.

Overall completeness and applicability of evidence

There are several issues about the applicability of the findings of this review in clinical practice to be considered.

The generalisability of the findings of this review was limited by the small number of trials identified with a small quantity of data. Only one was a phase III trial (Cotton 2014), and three were early‐phase trials (i.e. phase II trials). The phase III trial demonstrated more treatment failures with the experimental intervention than with a sham. It could be argued that sham procedures were somehow therapeutic, reflecting a potential confounding effect. However, it is highly unlikely that any such effect (i.e. pain relief following sham intervention) would last for one year (Cotton 2014).

There was a slight protocol deviation from the strict Rome III definition of biliary pain observed in Cotton 2014 as the trial included participants with some daily discomfort, in addition to episodes of pain. However, subgroup analyses revealed that the magnitude of this selection bias was negligible (Cotton 2014).

All trials were conducted in high‐income countries (Australia, Japan, the USA), which may limit the applicability of the findings to low‐income countries. These findings were based on analyses that involved only one small trial per comparison, which is a limitation of the evidence.

Moreover, three trials included in our review were randomised sham‐controlled trials.A shamproceduremight not be a true treatment representative in people with the target condition. Only one trial reported analysis of the impact on quality of life, which could be the most desirable benefit for people after receiving the treatment (Cotton 2014).

In summary, the applicability of the evidence of this review to current practice in people with biliary sphincter of Oddi dysfunction is extremely limited, and we cannot generalise the findings of the review. All trial results should be interpreted with caution.

Quality of the evidence

We used the GRADE criteria to assess the certainty of the evidence, which is shown in Table 1; Table 2; and Table 3.

The GRADE assessment found very low‐certainty evidence for all outcomes in the three comparisons. This means that any estimate of the effect is uncertain. There are two main reasons for risk of bias, which have led to downgrading the certainty of evidence. First, two trials did not clearly describe randomisation or allocation concealment (Takezawa 2004; Toouli 2000). Despite our requests to the authors, we did not receive additional data to change judgements in the key areas. It is also worth noting that some other key sources of bias may exist, such as selective reporting and bias due to incomplete outcome data. The number of dropouts in Geenen 1989 was high. Three trials did not have prepublished protocols (Geenen 1989; Takezawa 2004; Toouli 2000). In addition, imprecision due to the small number of participants with few events as well as wide CIs in analyses reduced our confidence in the certainty of evidence of this review. Based on the findings of this review, endoscopic sphincterotomy may reduce or increase or have little to no effect on outcomes, but the evidence so far is very uncertain. Due to its high risk of bias, or some concerns for risk of bias, the results of these individual trials need to be interpreted with caution.

Potential biases in the review process

Potential biases for this review were limited due to comprehensive searches with no limit on publication status or language, with adherence to Cochrane methodology. Nonetheless, there are areas of concern in this review.

We did not specifically search for observational studies to include in this review, which may bias findings in favour of benefits of the interventions while downplaying harms and overlooking uncommon and late adverse events (Storebø 2018). A lack of reply from the primary authors upon request for missing information or more details of participant recruitment may have resulted in 'reporting bias'. We evaluated trials reporting on outcomes in our summary of findings using the Cochrane RoB 2 tool, which included rating of 'some concerns' on some domains due to insufficient information. Trials were excluded due to insufficient information, and we did not receive responses from the authors (see details in Dealing with missing data). As fraudulent studies are still rife, contacting study authors or the journals that have published the studies to obtain more information is advisable.

For the main outcome 'treatment failure', we could use data on ITT from two of the three trials as the remaining trial did not provide information on dropouts. ITT is the most appropriate analysis to assess the estimated effect of assignment to interventions in order to inform a health policy question about whether an intervention should be recommended in a particular health system (i.e. whether to use endoscopic sphincterotomy in this case), whereas the estimated effect of adhering to the intervention as specified in the trial protocol would be the most appropriate to inform a care decision by an individual patient (i.e. whether to use endoscopic sphincterotomy in this case) (Higgins 2022b). We performed a sensitivity analysis for the best‐worst and worst‐best scenario analyses to determine the impact on the effect of missing data or loss to follow‐up. This was only possible for one comparison. The effect estimates retained no difference in treatment failure between the interventions, showing the stability of estimates. The finding that endoscopic sphincterotomy is not a better treatment has major implications for clinical practice because it applies to many thousands of patients (Cotton 2014). However, this was based on a single trial with very low‐certainty evidence.

Agreements and disagreements with other studies or reviews

One meta‐analysis of 37 individual randomised clinical trials addressed the safety of experimental endoscopy compared with sham control procedures (Schulman 2017). Two randomised clinical trials with a focus on ERCP in sphincterotomy were eligible for our review (Cotton 2014; Toouli 2000). The review highlighted concerns over the use of sham procedures (Schulman 2017).

The earlier version of this Cochrane review assessed outcomes stratified by manometric pressure cut points (i.e. less than 40 mmHg or more than 40 mmHg) (Craig 2001). In this present review, we assessed the outcome regardless of manometric pressure cut points. This is because according to the American Society for Gastrointestinal Endoscopy Technology Committee, a basal sphincter pressure of 40 mmHg or greater is the manometric criterion used to diagnose sphincter of Oddi dysfunction (ASGE 2011). Moreover, the index value of 40 mmHg represents the mean plus three SDs (Geenen 1989). Hence, the participants in this review had biliary sphincter of Oddi dysfunction and a sphincter pressure of 40 mmHg or less. One trial identified for our review reported that there was no association between the primary outcome and manometric pressure cut points (Cotton 2014).

The earlier Cochrane review assessed surrogate outcomes such as symptomatic responses (Craig 2001), while in our review, we focused on an end‐outcome (i.e. treatment failure defined with validated score such as RAPID in Cotton 2014, and standardised (structured) questionnaire in Geenen 1989 and Toouli 2000), to provide clinically important information. We identified two new trials for our review (Cotton 2014; Takezawa 2004). Cotton 2014 contributed data on treatment failure. Moreover, our review included a comparison with another type of sphincterotomy (Takezawa 2004), which adds to a more comprehensive assessment.

One study reported postprocedure pancreatitis in 18% of operations (Freeman 2007). We found that 26/241 (12%) participants experienced postprocedure pancreatitis, and it was comparable between the two groups. As such, the use of a sham‐controlled trial in these individuals brings ethical concerns because complications can occur with even a sham operation. The choice of whether to perform endoscopic sphincter or sham procedures on people with sphincter of Oddi disorders represents clinically important issues.

Authors' conclusions

Implications for practice.

Based on very low‐certainty evidence in the trials included in this review, we do not know if endoscopic sphincterotomy versus sham or versus dual endoscopic sphincterotomy increases, reduces, or makes no difference to the number of people without treatment success; if endoscopic sphincterotomy versus sham or versus endoscopic papillary balloon dilation increases, reduces, or makes no difference to serious adverse events; or if endoscopic sphincterotomy versus sham improves, worsens, or makes no difference to health‐related quality of life and liver function tests in adults with biliary sphincter of Oddi dysfunction.

Evidence on the effect of endoscopic sphincterotomy compared with sham, endoscopic papillary balloon dilation,or dual endoscopic sphincterotomyon non‐serious adverse events, overall survival (all‐cause mortality), and length of hospital stay is lacking.

Our review shows that the impact of endoscopic sphincterotomy on serious adverse events needs further investigation, and this information should be used to inform patients considering endoscopic sphincterotomy.

The effect estimates of the proportion of participants without successful treatment based on three trials that compared endoscopic sphincterotomy with a sham procedure may have increased the samples for greater statistical power, but this was compromised by the trials' substantial heterogeneity. In the only trial at low risk of bias, endoscopic sphincterotomy seemed to increase the number of participants without treatment success. This observation raises concerns about the risk of endoscopic sphincterotomy intervention.

Our overall conclusions include mostly trials with 'some concerns' for risk of bias and high risk of bias. The trials were underpowered and lacked data on clinically important outcomes. Only one trial per comparison reported serious adverse events and quality of life, and meta‐analyses were not possible with these clinically important outcomes that are of interest to patients.

Overall, the evidence in this review is highly insufficient, and each result should be interpreted with caution. We lack well‐designed, large trials, assessing clinically and patient‐relevant outcomes to demonstrate the effects of endoscopic sphincterotomy in people with biliary sphincter of Oddi dysfunction.

Implications for research.

Much more evidence from trials at low risk of bias and of sufficient power is needed to conclude a decision for or against an intervention with endoscopic sphincterotomy in biliary sphincter of Oddi dysfunction.

Further research is indicated, including randomised multicentre trials in larger populations with sphincter of Oddi dysfunction. The methods of research should be robust and results should be reported transparently. Trialists should define and categorise participants according to the type of sphincter of Oddi dysfunction and, if possible, provide disaggregated data for different subgroups.

Therefore, we recommend including various types of sphincters of Oddi dysfunction to delineate the possible impact of endoscopic sphincterotomy on the evidence. Since there is no study on cost‐effectiveness, future research on this outcome may also be required.

Future trials, if and whenever planned, should be designed according to the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) (www.spirit-statement.org) statement and reported according to the CONSORT statement (www.goodreports.org/reporting-checklists/consort/).

History

Protocol first published: Issue 9, 2022

Risk of bias

Risk of bias for analysis 1.1 Proportion of participants without treatment success at the longest follow‐up (1–4 years).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Cotton 2014

Low risk of bias

No concerns about the randomisation process.

Low risk of bias

Unlikely to arise deviations.

Low risk of bias

All randomised participants were analysed (ITT).

Low risk of bias

Appropriate method of measuring the observer‐reported outcome.

Low risk of bias

Planned assessments and unlike to be selected.

Low risk of bias

The trial is judged to be at low risk of bias for all domains for this outcome.

Geenen 1989

Low risk of bias

Unlikely to have obvious concerns over randomisation.

Some concerns

Cannot be determined whether any deviations arose or not.

High risk of bias

14% of randomised participants were not analysed at 4‐year follow‐up.

Some concerns

Measured appropriately with no differences in the two groups and unlikely to influence the observer‐reported outcome.

Some concerns

Cannot be determined any selection o the reporte outcome.

High risk of bias

The trial is judged to be at high risk of bias due to a high risk of bias regarding 14% missingness of randomised participants at a 4‐year follow‐up.

Toouli 2000

Some concerns

Need more information on randomisation process.

Low risk of bias

Appropriately measured and unlikely to influence the observer reported outcome.

Some concerns

Cannot be determined whether deviations arose because of the trial context.

Low risk of bias

Outcome data were available for 97.5% randomised participants.

Some concerns

Cannot be determined whether there was selection o the reporte outcome.

Some concerns

The trial is judged to raise some concerns inn three domains for this result, but not to be at high risk of bias for any domains.

Risk of bias for analysis 1.2 Proportion of participants with serious adverse events (pancreatitis, at 1 year).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Cotton 2014

Low risk of bias

No concerns about the randomisation process.

Low risk of bias

Unlikely to arise deviations.

Low risk of bias

All randomised participants analysed (ITT).

Low risk of bias

Appropriate method of measuring the observer reported outcome.

Low risk of bias

Planned assessment and unlikely to be selected.

Low risk of bias

The trial is judged to be at low risk of bias for all domains for this outcome.

Risk of bias for analysis 1.3 Health‐related quality of life (measured with SF‐36 Physical scale at 1 year).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Cotton 2014

Low risk of bias

No concerns about the randomisation process.

Low risk of bias

Unlikely to arise deviations.

Low risk of bias

All randomised participants analysed (ITT).

Low risk of bias

Appropriate method of measuring the observer‐reported outcome.

Low risk of bias

Planned assessment and unlikely to be selected.

Low risk of bias

The trial is judged to be at low risk of bias for all domains for this outcome.

Risk of bias for analysis 1.4 Health‐related quality of life (measured with SF‐36 Mental scale at 1 year).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Cotton 2014

Low risk of bias

No concerns about the randomisation process.

Low risk of bias

Unlikely to arise deviations.

Low risk of bias

All randomised participants analysed (ITT).

Low risk of bias

Appropriate method of measuring the observer‐reported outcome.

Low risk of bias

Planned assessment and unlikely to be selected.

Low risk of bias

The trial is judged to be at low risk of bias for all domains for this outcome.

Risk of bias for analysis 1.5 Proportion of participants without improvement in liver function tests.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Geenen 1989

Low risk of bias

Unlikely to have obvious concerns over randomisation.

Some concerns

Can not be determined whether any deviations arose or not.

High risk of bias

14% of randomised participants were not analysed at 4‐year follow‐up.

Low risk of bias

Measured appropriately with no differences in the two groups and unlikely to influence the observer reported outcome.

Some concerns

Can not be determined any selection of the reported outcome.

High risk of bias

The trial is judged to be at high risk of bias due to a high risk of bias regarding 14% missingness of randomised participants at a 4‐year follow‐up.

Risk of bias for analysis 2.1 Proportion of participants with serious adverse events (recurrent common bile duct stones, at 1 year).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Takezawa 2004

Low risk of bias

No concerns about the randomisation process.

Some concerns

Can not be determined any deviations.

Some concerns

Many randomised participants were not analysed at 1‐year follow‐up.

Low risk of bias

Measured appropriately with no differences in the two groups and unlikely to influence the hard outcome.

Some concerns

Can not be determined any selection of the outcome.

Some concerns

The trial is judged to raise some concerns in three domains for this result, but not to be at high risk of bias for any domains.

Risk of bias for analysis 3.1 Proportion of participants without treatment success (at 1 year).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Cotton 2014

Low risk of bias

No concerns about the randomisation process.

Some concerns

Need more information.

Low risk of bias

All randomised participants analysed (ITT).

Low risk of bias

Appropriate method of measuring the observer reported outcome.

Low risk of bias

Planned assessment and unlikely to be selected.

Some concerns

The trial is judged to raise some concerns in two domains for this result, but not to be at high risk of bias for any domains.

Risk of bias for analysis 4.1 Risk of bias. Proportion of participants without treatment success at the longest follow‐up (1–4 years).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 4.1.1 Trials at low risk of bias

Cotton 2014

Low risk of bias

No concerns about the randomisation process.

Low risk of bias

Unlikely to arise deviations.

Low risk of bias

All randomised participants analysed (ITT).

Low risk of bias

Appropriate method of measuring the observer‐reported outcome.

Low risk of bias

Planned assessment and unlikely to be selected.

Low risk of bias

The trial is judged to be at low risk of bias for all domains for this outcome.

Subgroup 4.1.2 Trials at some concern and at high risk of bias

Geenen 1989

Low risk of bias

Unlikely to have obvious concerns over randomisation.

Some concerns

Cannot be determined whether any deviations arose or not.

High risk of bias

14% of randomised participants were not analysed at 4‐year follow‐up.

Low risk of bias

Measured appropriately with no differences in the two groups and unlikely to influence the observer reported outcome.

Some concerns

Cannot be determined any selection of the reported outcome.

High risk of bias

The trial is judged to be at high risk of bias due to a high risk of bias regarding 14% missingness of randomised participants at a 4‐year follow‐up.

Toouli 2000

Some concerns

Need more information on randomisation process.

Some concerns

Cannot be determined whether deviations arose because of the trial context.

Low risk of bias

Outcome data were available for 97.5% randomised participants.

Low risk of bias

Appropriately measured and unlikely to influence the observer reported outcome.

Some concerns

Cannot be determined whether there was selection of the reported outcome.

Some concerns

The trial is judged to raise some concerns in three domains for this result, but not to be at high risk of bias for any domains.

Risk of bias for analysis 5.1 Best‐worst case scenario analysis.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Cotton 2014

Low risk of bias

No concerns about the randomisation process.

Low risk of bias

Unlikely to arise deviations.

Low risk of bias

All randomised participants analysed (ITT).

Low risk of bias

Appropriate method of measuring the observer‐reported outcome.

Low risk of bias

Planned assessment and unlikely to be selected.

Low risk of bias

The trial is judged to be at low risk of bias for all domains for this outcome.

Geenen 1989

Low risk of bias

Unlikely to have obvious concerns over randomisation.

Some concerns

determined whether any deviations arose or not.

High risk of bias

14% of randomised participants were not analysed at 4‐year follow‐up.

Low risk of bias

Measured appropriately with no differences in the two groups and unlikely to influence the observer.

Some concerns

Can not be determined any selection o the reporte outcome.

High risk of bias

The trial is judged to be at high risk of bias due to a high risk of bias regarding 14% missingness of randomised participants at a 4‐year follow‐up.

Toouli 2000

Some concerns

Need more information on randomisation process.

Some concerns

Cannot be determined whether deviations arose because of the trial context.

Low risk of bias

Outcome data were available for 97.5% randomised participants.

Low risk of bias

Appropriately measured and unlikely to influence the observer reported outcome.

Some concerns

Cannot be determined whether there was selection o the reporte outcome.

Some concerns

The trial is judged to raise some concerns in three domains for this result, but not to be at high risk of bias for any domains.

Risk of bias for analysis 5.2 Worst‐best case scenario analysis.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Cotton 2014

Low risk of bias

No concerns about the randomisation process.

Low risk of bias

Unlikely to arise deviations.

Low risk of bias

All randomised participants analysed (ITT).

Low risk of bias

Appropriate method of measuring the observer‐reported outcome.

Low risk of bias

Planned assessments and unlike to be selected.

Low risk of bias

The trial is judged to be at low risk of bias for all domains for this outcome.

Geenen 1989

Low risk of bias

Unlikely to have obvious concerns over randomisation.

Some concerns

Cannot be determined whether any deviations arose or not.

High risk of bias

14% of randomised participants were not analysed at 4‐year follow‐up.

Low risk of bias

Measured appropriately with no differences in the two groups and unlikely to influence the observer‐reported outcome.

Some concerns

Cannot be determined any selection of the reported outcome.

High risk of bias

The trial is judged to be at high risk of bias due to a high risk of bias regarding 14% missingness of randomised participants at a 4‐year follow‐up.

Toouli 2000

Some concerns

Need more information on randomisation process.

Some concerns

Cannot be determined whether deviations arose because of the trial context.

Low risk of bias

Outcome data were available for 97.5% randomised participants.

Low risk of bias

Appropriately measured and unlikely to influence the observer reported outcome.

Some concerns

Cannot be determined whether there was selection o the reporte outcome.

Some concerns

The trial is judged to raise some concerns in three domains for this result, but not to be at high risk of bias for any domains.

Appendices

Appendix 1. Search strategies

Database	Date range	Search strategy
Cochrane Hepato‐Biliary Group Controlled Trials Register (searched via the Cochrane Register of Studies Web)	16 May 2023	(sphincterotom* or vater papillotom*) AND (((oddi* ADJ2 sphincter) AND (dysfunction OR dyskinesia OR stenosis OR constriction OR occlusion)) or SOD or BSOD or 'biliary dyskinesia' or 'papilla* stenosis' or 'biliary dyssynergia' or odditis or papillitis or 'post‐cholecystectomy pain' or 'upper quadrant pain')
Cochrane Central Register of Controlled Trials in the Cochrane Library	2023, Issue 5	#1 MeSH descriptor: [Sphincterotomy, Endoscopic] explode all trees #2 MeSH descriptor: [Sphincterotomy, Transduodenal] explode all trees #3 (sphincterotom* or vater papillotom*) #4 #1 or #2 or #3 #5 MeSH descriptor: [Biliary Dyskinesia] explode all trees #6 ((oddi* near/2 sphincter) AND (dysfunction OR dyskinesia OR stenosis OR constriction OR occlusion)) or (SOD or BSOD or biliary dyskinesia or papilla* stenosis or biliary dyssynergia or odditis or papillitis or post‐cholecystectomy pain or upper quadrant pain) #7 #5 or #6 #8 #3 and #7
MEDLINE ALL Ovid	1946 to 16 May 2023	1. exp sphincterotomy, endoscopic/ or exp sphincterotomy, transduodenal/ 2. (sphincterotom* or vater papillotom*).mp. [mp=title, book title, abstract, original title, name of substance word, subject heading word, floating sub‐heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 3. 1 or 2 4. exp Biliary Dyskinesia/ 5. (((oddi* adj2 sphincter) and (dysfunction or dyskinesia or stenosis or constriction or occlusion)) or (SOD or BSOD or biliary dyskinesia or papilla* stenosis or biliary dyssynergia or odditis or papillitis or post‐cholecystectomy pain or upper quadrant pain)).mp. [mp=title, book title, abstract, original title, name of substance word, subject heading word, floating sub‐heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 6. 4 or 5 7. 3 and 6 8. (randomized controlled trial or controlled clinical trial or retracted publication or retraction of publication).pt. 9. clinical trials as topic.sh. 10. (random* or placebo*).ab. or trial.ti. 11. 8 or 9 or 10 12. exp animals/ not humans.sh. 13. 11 not 12 14. 7 and 13
Embase Ovid	1974 to 16 May 2023	1. exp endoscopic sphincterotomy/ 2. exp vater papillotomy/ 3. (sphincterotom* or vater papillotom*).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 4. 1 or 2 or 3 5. exp Vater papilla stenosis/ 6. exp bile duct dyskinesia/ 7. (((oddi* adj2 sphincter) and (dysfunction or dyskinesia or stenosis or constriction or occlusion)) or (SOD or BSOD or biliary dyskinesia or papilla* stenosis or biliary dyssynergia or odditis or papillitis or post‐cholecystectomy pain or upper quadrant pain)).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 8. 5 or 6 or 7 9. 4 and 8 10. Randomized controlled trial/ or Controlled clinical study/ or randomization/ or intermethod comparison/ or double blind procedure/ or human experiment/ or retracted article/ 11. (random$ or placebo or parallel group$1 or crossover or cross over or assigned or allocated or volunteer or volunteers).ti,ab. 12. (compare or compared or comparison or trial).ti. 13. ((evaluated or evaluate or evaluating or assessed or assess) and (compare or compared or comparing or comparison)).ab. 14. (open adj label).ti,ab. 15. ((double or single or doubly or singly) adj (blind or blinded or blindly)).ti,ab. 16. ((assign$ or match or matched or allocation) adj5 (alternate or group$1 or intervention$1 or patient$1 or subject$1 or participant$1)).ti,ab. 17. (controlled adj7 (study or design or trial)).ti,ab. 18. (erratum or tombstone).pt. or yes.ne. 19. or/10‐18 20. (random$ adj sampl$ adj7 ('cross section$' or questionnaire$ or survey$ or database$1)).ti,ab. not (comparative study/ or controlled study/ or randomi?ed controlled.ti,ab. or randomly assigned.ti,ab.) 21. Cross‐sectional study/ not (randomized controlled trial/ or controlled clinical study/ or controlled study/ or randomi?ed controlled.ti,ab. or control group$1.ti,ab.) 22. (((case adj control$) and random$) not randomi?ed controlled).ti,ab. 23. (Systematic review not (trial or study)).ti. 24. (nonrandom$ not random$).ti,ab. 25. 'Random field$'.ti,ab. 26. (random cluster adj3 sampl$).ti,ab. 27. (review.ab. and review.pt.) not trial.ti. 28. 'we searched'.ab. and (review.ti. or review.pt.) 29. 'update review'.ab. 30. (databases adj4 searched).ab. 31. (rat or rats or mouse or mice or swine or porcine or murine or sheep or lambs or pigs or piglets or rabbit or rabbits or cat or cats or dog or dogs or cattle or bovine or monkey or monkeys or trout or marmoset$1).ti. and animal experiment/ 32. Animal experiment/ not (human experiment/ or human/) 33. or/20‐32 34. 19 not 33 35. 9 and 34
LILACS (VHL Regional Portal)	1982 to 16 May 2023	(mh:(sphincterotomy, endoscopic OR e04.210.120.850 OR e04.210.240.250.840 OR e04.502.250.250.250.840 OR e04.515.750.500 OR sphincterotomy, transduodenal OR e04.210.120.860 OR e04.515.750.750)) OR ((sphincterotom* OR vater papillotom*)) AND (mh:(biliary dyskinesia OR c06.130.120.250.098)) OR ((((oddi* AND sphincter) AND (dysfunction OR dyskinesia OR stenosis OR constriction OR occlusion)) OR (sod OR bsod OR biliary dyskinesia OR papilla* stenosis OR biliary dyssynergia OR odditis OR papillitis OR post‐cholecystectomy pain OR upper quadrant pain))) AND ( db:("LILACS"))
Science Citation Index EXPANDED (Web of Science)	1900 to 16 May 2023	#5 #3 AND #4 #4 TI=(random* or blind* or placebo* or meta‐analys* or trial*) OR TS=(random* or blind* or placebo* or meta‐analys*) #3 #2 AND #1 #2 TS=(((oddi* NEAR sphincter) AND (dysfunction OR dyskinesia OR stenosis OR constriction OR occlusion)) or (SOD or BSOD or biliary dyskinesia or papilla* stenosis or biliary dyssynergia or odditis or papillitis or post‐cholecystectomy pain or upper quadrant pain)) #1 TS=(sphincterotom* or vater papillotom*)
Conference Proceedings Citation Index – Science (Web of Science)	1990 to 16 May 2023	#5 #3 AND #4 #4 TI=(random* or blind* or placebo* or meta‐analys* or trial*) OR TS=(random* or blind* or placebo* or meta‐analys*) #3 #2 AND #1 #2 TS=(((oddi* NEAR sphincter) AND (dysfunction OR dyskinesia OR stenosis OR constriction OR occlusion)) or (SOD or BSOD or biliary dyskinesia or papilla* stenosis or biliary dyssynergia or odditis or papillitis or post‐cholecystectomy pain or upper quadrant pain)) #1 TS=(sphincterotom* or vater papillotom*)
Other sources: US Food and Drug Administration (FDA) website (www.fda.gov) European Medicines Agency (EMA) (www.ema.europa.eu/ema/) WHO International Clinical Trial Registry Platform (www.who.int/ictrp) ClinicalTrial.gov (clinicaltrials.gov/)	May 2023	Use of the following free terms alone or in combination, as appropriate. "endoscopic sphincterotomy" "sphincter of Oddi dysfunction" "biliary sphincter of Oddi dysfunction"

Appendix 2. Questions for assessing risk of bias

Bias domain	Issues addressed
Bias arising from the randomisation process	Whether: the allocation sequence was random; the allocation sequence was adequately concealed; baseline differences between intervention groups suggest a problem with the randomisation process.
Bias due to deviations from intended interventions	Whether: participants were aware of their assigned intervention during the trial; carers and people delivering the interventions were aware of participants' assigned intervention during the trial. Whether, when the review authors' interest is in the effect of assignment to intervention (see Section 8.2.2 of Higgins 2022b): (if applicable) deviations from the intended intervention arose because of the experimental context (i.e. do not reflect usual practice); and, if so, whether they were unbalanced between groups and likely to have affected the outcome; an appropriate analysis was used to estimate the effect of assignment to intervention; and, if not, whether there was potential for a substantial impact on the result. Whether, when the review authors' interest is in the effect of adhering to intervention (see Section 8.2.2 of Higgins 2022b): (if applicable) important non‐protocol interventions were balanced across intervention groups; (if applicable) failures in implementing the intervention could have affected the outcome; (if applicable) study participants adhered to the assigned intervention regimen; (if applicable) an appropriate analysis was used to estimate the effect of adhering to the intervention.
Bias due to missing outcome data	Whether: data for this outcome were available for all, or nearly all, participants randomised; (if applicable) there was evidence that the result was not biased by missing outcome data; (if applicable) missingness in the outcome was likely to depend on its true value (e.g. the proportions of missing outcome data, or reasons for missing outcome data, differ between intervention groups).
Bias in measurement of the outcome	Whether: the method of measuring the outcome was inappropriate; measurement or ascertainment of the outcome could have differed between intervention groups; outcome assessors were aware of the intervention received by study participants; (if applicable) assessment of the outcome was likely to have been influenced by knowledge of intervention received.
Bias in selection of the reported result	Whether: the trial was analysed in accordance with a prespecified plan that was finalised before unblinded outcome data were available for analysis; the numerical result being assessed is likely to have been selected, on the basis of the results, from multiple outcome measurements within the outcome domain; the numerical result being assessed is likely to have been selected, on the basis of the results, from multiple analyses of the data.

Appendix 3. Risk of bias assessment of the included outcomes (after consensus)

Unique ID	111C	Study ID	111C	Assessor	CN
Ref or Label	Cotton 2014	Aim	Assignment to intervention (the 'ITT' effect)
Experimental	Endoscopic sphincterotomy	Comparator	Sham treatment	Source	Journal article(s); non‐commercial trial registry record (e.g. ClinicalTrials.gov record)
Outcome	Quality of life (mental)	Results	MD 1.00 (95% CI −4.16 to 2.16)	Weight	1
Domain	Signalling question			Response	Comments
Bias arising from the randomisation process	1.1 Was the allocation sequence random?			Y	Quote: "2:1 randomisation through a centralized web‐based system. A permuted block scheme with random block size" p.5
	1.2 Was the allocation sequence concealed until participants were enrolled and assigned to interventions?			PY
	1.3 Did baseline differences between intervention groups suggest a problem with the randomisation process?			N	2 treatment groups were well‐balanced with respect to baseline demographic and clinical characteristics, p.7 Table 4.
	Risk of bias judgement			Low	No concerns about the randomisation process.
Bias due to deviations from intended interventions	2.1 Were participants aware of their assigned intervention during the trial?			N	Participants, referring physicians, research co‐ordinators, p.5
	2.2 Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?			N
	2.3 If Y/PY/NI to 2.1 or 2.2: were there deviations from the intended intervention that arose because of the experimental context?			NA
	2.4 If Y/PY to 2.3: were these deviations likely to have affected the outcome?			NA
	2.5 If Y/PY/NI to 2.4: were these deviations from intended intervention balanced between groups?			NA
	2.6 Was an appropriate analysis used to estimate the effect of assignment to intervention?			Y	Analysed according to their assigned groups
	2.7 If N/PN/NI to 2.6: was there potential for a substantial impact (on the result) of the failure to analyse participants in the group to which they were randomised?			NA
	Risk of bias judgement			Low	Unlikely to cause deviations.
Bias due to missing outcome data	3.1 Were data for this outcome available for all, or nearly all, participants randomised?			Y	Outcome data were available for all randomised participants (Table 4)
	3.2 If N/PN/NI to 3.1: is there evidence that result was not biased by missing outcome data?			NA
	3.3 If N/PN to 3.2: could missingness in the outcome depend on its true value?			NA
	3.4 If Y/PY/NI to 3.3: is it likely that missingness in the outcome depended on its true value?			NA
	Risk of bias judgement			Low	All randomised participants analysed (ITT).
Bias in measurement of the outcome	4.1 Was the method of measuring the outcome inappropriate?			N	Used validated tool 'RAPID'
	4.2 Could measurement or ascertainment of the outcome have differed between intervention groups?			N	The same measurement tools were applied to both groups.
	4.3 Were outcome assessors aware of the intervention received by study participants?			N	Participants and study investigators were blinded.
	4.4 If Y/PY/NI to 4.3: could assessment of the outcome have been influenced by knowledge of intervention received?			NA
	4.5 If Y/PY/NI to 4.4: is it likely that assessment of the outcome was influenced by knowledge of intervention received?			NA
	Risk of bias judgement			Low	Appropriate method of measuring the observer‐reported outcome.
Bias in selection of the reported result	5.1 Were the data that produced this result analysed in accordance with a prespecified analysis plan that was finalised before unblinded outcome data were available for analysis?			Y	Planned assessment as stated in a trial registry. NCT00688662
	5.2 … multiple eligible outcome measurements (e.g. scales, definitions, time points) within the outcome domain?			PN	Unlikely to have many ways to measure
	5.3 … multiple eligible analyses of the data?			PN	Unlikely to have many ways to analyse.
	Risk of bias judgement			Low	Planned assessment and unlikely to be selected.
Overall bias	Risk of bias judgement			Low	The trial was judged to be at low risk of bias for all domains for this outcome.
Unique ID	111B	Study ID	111B	Assessor	CN
Ref or Label	Cotton 2014	Aim	Assignment to intervention (the 'ITT' effect)
Experimental	Endoscopic sphincterotomy	Comparator	Sham treatment	Source	Journal article(s); Non‐commercial trial registry record (e.g. ClinicalTrials.gov record)
Outcome	Quality of life (physical)	Results	MD 1.00 (95% CI −3.84 to 1.84)	Weight	1
Domain	Signalling question			Response	Comments
Bias arising from the randomisation process	1.1 Was the allocation sequence random?			Y	Quote: "2:1 randomisation through a centralized web‐based system. A permuted block scheme with random block size," p.5 "blinded to treatment allocation," p.5
	1.2 Was the allocation sequence concealed until participants were enrolled and assigned to interventions?			PY
	1.3 Did baseline differences between intervention groups suggest a problem with the randomisation process?			N	2 treatment groups were well‐balanced with respect to baseline demographic and clinical characteristics, p.7 Table 4.
	Risk of bias judgement			Low	No concerns about the randomisation process.
Bias due to deviations from intended interventions	2.1 Were participants aware of their assigned intervention during the trial?			N	Participants, referring physicians, research co‐ordinators, p.5
	2.2 Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?			N
	2.3 If Y/PY/NI to 2.1 or 2.2: were there deviations from the intended intervention that arose because of the experimental context?			NA
	2.4 If Y/PY to 2.3: were these deviations likely to have affected the outcome?			NA
	2.5 If Y/PY/NI to 2.4: were these deviations from intended intervention balanced between groups?			NA
	2.6 Was an appropriate analysis used to estimate the effect of assignment to intervention?			Y	Analysed according to their assigned groups.
	2.7 If N/PN/NI to 2.6: was there potential for a substantial impact (on the result) of the failure to analyse participants in the group to which they were randomized?			NA
	Risk of bias judgement			Low	Unlikely to cause deviations.
Bias due to missing outcome data	3.1 Were data for this outcome available for all, or nearly all, participants randomized?			Y	Outcome data were available for all randomised participants (Table 4)
	3.2 If N/PN/NI to 3.1: is there evidence that the result was not biased by missing outcome data?			NA
	3.3 If N/PN to 3.2: could missingness in the outcome depend on its true value?			NA
	3.4 If Y/PY/NI to 3.3: is it likely that missingness in the outcome depended on its true value?			NA
	Risk of bias judgement			Low	All randomised participants analysed (ITT).
Bias in measurement of the outcome	4.1 Was the method of measuring the outcome inappropriate?			N	Used validated tool 'RAPID'
	4.2 Could measurement or ascertainment of the outcome have differed between intervention groups?			N	The same measurement tools applied for both groups.
	4.3 Were outcome assessors aware of the intervention received by study participants?			N	Participants and study investigators were blinded.
	4.4 If Y/PY/NI to 4.3: could assessment of the outcome have been influenced by knowledge of intervention received?			NA
	4.5 If Y/PY/NI to 4.4: is it likely that assessment of the outcome was influenced by knowledge of intervention received?			NA
	Risk of bias judgement			Low	Appropriate method of measuring the observer‐reported outcome.
Bias in selection of the reported result	5.1 Were the data that produced this result analysed in accordance with a pre‐specified analysis plan that was finalised before unblinded outcome data were available for analysis?			Y	Planned assessment as stated in a trial registry. NCT00688662
	5.2 … multiple eligible outcome measurements (e.g. scales, definitions, time points) within the outcome domain?			PN	Unlikely to have many ways to measure
	5.3 … multiple eligible analyses of the data?			PN	Unlikely to have many ways to analyse
	Risk of bias judgement			Low	Planned assessment and unlikely to be selected.
Overall bias	Risk of bias judgement			Low	The trial is judged at low risk of bias for all domains for this outcome.
Unique ID	111A	Study ID	111A	Assessor	CN
Ref or Label	Cotton 2014	Aim	Assignment to intervention (the 'ITT' effect)
Experimental	Endoscopic sphincterotomy	Comparator	Sham treatment	Source	Journal article(s); non‐commercial trial registry record (e.g. ClinicalTrials.gov record)
Outcome	Proportion without treatment success	Results	RR 1.19 (95% CI 0.98 to 1.45)	Weight	1
Domain	Signalling question			Response	Comments
Bias arising from the randomisation process	1.1 Was the allocation sequence random?			Y	Quote: "2:1 randomization through a centralized web‐based system. A permuted block scheme with random block size," p.5
	1.2 Was the allocation sequence concealed until participants were enrolled and assigned to interventions?			PY
	1.3 Did baseline differences between intervention groups suggest a problem with the randomisation process?			N	2 treatment groups were well‐balanced with respect to baseline demographic and clinical characteristics, p.7 Table 4.
	Risk of bias judgement			Low	No concerns about the randomisation process.
Bias due to deviations from intended interventions	2.1 Were participants aware of their assigned intervention during the trial?			N	Participants, referring physicians, research co‐ordinators, p.5
	2.2 Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?			N
	2.3 If Y/PY/NI to 2.1 or 2.2: were there deviations from the intended intervention that arose because of the experimental context?			NA
	2.4 If Y/PY to 2.3: were these deviations likely to have affected the outcome?			NA
	2.5 If Y/PY/NI to 2.4: were these deviations from intended intervention balanced between groups?			NA
	2.6 Was an appropriate analysis used to estimate the effect of assignment to intervention?			Y	Analysed according to their assigned groups.
	2.7 If N/PN/NI to 2.6: was there potential for a substantial impact (on the result) of the failure to analyse participants in the group to which they were randomized?			NA
	Risk of bias judgement			Low	Unlikely to cause deviations.
Bias due to missing outcome data	3.1 Were data for this outcome available for all, or nearly all, participants randomized?			Y	Outcome data were available for all randomised participants (Table 4)
	3.2 If N/PN/NI to 3.1: is there evidence that the result was not biased by missing outcome data?			NA
	3.3 If N/PN to 3.2: could missingness in the outcome depend on its true value?			NA
	3.4 If Y/PY/NI to 3.3: is it likely that missingness in the outcome depended on its true value?			NA
	Risk of bias judgement			Low	All randomised participants analysed (ITT).
Bias in measurement of the outcome	4.1 Was the method of measuring the outcome inappropriate?			N	Used validated tool 'RAPID'
	4.2 Could measurement or ascertainment of the outcome have differed between intervention groups?			N	The same measurement tools applied for both groups.
	4.3 Were outcome assessors aware of the intervention received by study participants?			N	Participants and study investigators were blinded.
	4.4 If Y/PY/NI to 4.3: could assessment of the outcome have been influenced by knowledge of intervention received?			NA
	4.5 If Y/PY/NI to 4.4: is it likely that assessment of the outcome was influenced by knowledge of intervention received?			NA
	Risk of bias judgement			Low	Appropriate method of measuring the observer‐reported outcome.
Bias in selection of the reported result	5.1 Were the data that produced this result analysed in accordance with a pre‐specified analysis plan that was finalized before unblinded outcome data were available for analysis?			Y	Planned assessment as stated in a trial registry. NCT00688662
	5.2 … multiple eligible outcome measurements (e.g. scales, definitions, time points) within the outcome domain?			PN	Unlikely to have many ways to measure.
	5.3 … multiple eligible analyses of the data?			PN	Unlikely to have been many ways to analyse.
	Risk of bias judgement			Low	Planned assessment and unlikely to be selected.
Overall bias	Risk of bias judgement			Low	The trial is judged at low risk of bias for all domains for this outcome.
Unique ID	211A	Study ID	211A	Assessor	CN
Ref or Label	Geenen 1989	Aim	Assignment to intervention (the 'ITT' effect)
Experimental	Endoscopic sphincterotomy	Comparator	Sham treatment	Source	Journal article(s)
Outcome	Proportion without treatment success	Results	RR 0.50 (95% CI 0.24 to 1.05)	Weight	1
Domain	Signalling question			Response	Comments
Bias arising from the randomisation process	1.1 Was the allocation sequence random?			NI	Quote: "Patients were randomly assigned to undergo either group," p.83 No further details about randomisation method. Allocation concealment: sealed opaque envelopes
	1.2 Was the allocation sequence concealed until participants were enrolled and assigned to interventions?			PY
	1.3 Did baseline differences between intervention groups suggest a problem with the randomisation process?			NI	Baseline information not reported.
	Risk of bias judgement			Low	Unlikely to have obvious concerns over randomisation.
Bias due to deviations from intended interventions	2.1 Were participants aware of their assigned intervention during the trial?			N	Double‐blind clinical trial (participant, physician) Endoscopist was aware of the choice.
	2.2 Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?			PY
	2.3 If Y/PY/NI to 2.1 or 2.2: were there deviations from the intended intervention that arose because of the experimental context?			NI	Could not be determined.
	2.4 If Y/PY to 2.3: were these deviations likely to have affected the outcome?			NA
	2.5 If Y/PY/NI to 2.4: were these deviations from intended intervention balanced between groups?			NA
	2.6 Was an appropriate analysis used to estimate the effect of assignment to intervention?			Y	Appropriate analysis.
	2.7 If N/PN/NI to 2.6: was there potential for a substantial impact (on the result) of the failure to analyse participants in the group to which they were randomized?			NA
	Risk of bias judgement			Some concerns	Could not be determined whether any deviations arose or not.
Bias due to missing outcome data	3.1 Were data for this outcome available for all, or nearly all, participants randomized?			N	7/47 (15%) participants lost to follow‐up at 4 years.
	3.2 If N/PN/NI to 3.1: is there evidence that the result was not biased by missing outcome data?			PN	Likely to be biased.
	3.3 If N/PN to 3.2: could missingness in the outcome depend on its true value?			PY	4 participants declined to continue to participate; likely due to treatment effect.
	3.4 If Y/PY/NI to 3.3: is it likely that missingness in the outcome depended on its true value?			PY
	Risk of bias judgement			High	15% of randomised participants were not analysed at 4‐year follow‐up.
Bias in measurement of the outcome	4.1 Was the method of measuring the outcome inappropriate?			N	Appropriately measured the outcome. Standardised questionnaire, liver function tests, manometry, etc.
	4.2 Could measurement or ascertainment of the outcome have differed between intervention groups?			N	The same measurement in 2 groups.
	4.3 Were outcome assessors aware of the intervention received by study participants?			N	Quote: "patients were interviewed by independent observer who was unaware of the assigned treatment," p.83
	4.4 If Y/PY/NI to 4.3: could assessment of the outcome have been influenced by knowledge of intervention received?			NA
	4.5 If Y/PY/NI to 4.4: is it likely that assessment of the outcome was influenced by knowledge of intervention received?			NA
	Risk of bias judgement			Low	Measured appropriately with no differences in the 2 groups and unlikely to influence the observer‐reported outcome.
Bias in selection of the reported result	5.1 Were the data that produced this result analysed in accordance with a pre‐specified analysis plan that was finalised before unblinded outcome data were available for analysis?			NI	Cannot be determined – planned assessment.
	5.2 … multiple eligible outcome measurements (e.g. scales, definitions, time points) within the outcome domain?			PN	Unlikely to have other outcome measurements for the observer‐reported outcome.
	5.3 … multiple eligible analyses of the data?			PN	Only 1 possible way to analyse.
	Risk of bias judgement			Some concerns	Could not be determined if there was any selection of the reported outcome.
Overall bias	Risk of bias judgement			High	The trial was judged at high risk of bias due to a high risk of bias regarding 14% missingness of randomised participants at 4‐year follow‐up
Unique ID	311A	Study ID	311A	Assessor	CN
Ref or Label	Toouli 2000	Aim	Assignment to intervention (the 'ITT' effect)
Experimental	Endoscopic sphincterotomy	Comparator	Sham treatment	Source	Journal article(s)
Outcome	Proportion without treatment success	Results	RR 0.66 (95% CI 0.41 to 1.08)	Weight	1
Domain	Signalling question			Response	Comments
Bias arising from the randomisation process	1.1 Was the allocation sequence random?			PY	Quote: "patients were randomised by the draw of a card to undergo either endoscopic sphincterotomy or a sham sphincterotomy procedure," p.99
	1.2 Was the allocation sequence concealed until participants were enrolled and assigned to interventions?			NI
	1.3 Did baseline differences between intervention groups suggest a problem with the randomisation process?			NI	Could not be determined due to lack of information.
	Risk of bias judgement			Some concerns	Need more information on randomisation process.
Bias due to deviations from intended interventions	2.1 Were participants aware of their assigned intervention during the trial?			NI	Need more information about participants awareness. Quote: "a clinician who was not involved with the endoscopic therapy and was unaware of the results of manometry or randomisation," p.99
	2.2 Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?			PN
	2.3 If Y/PY/NI to 2.1 or 2.2: were there deviations from the intended intervention that arose because of the experimental context?			NA
	2.4 If Y/PY to 2.3: were these deviations likely to have affected the outcome?			NA
	2.5 If Y/PY/NI to 2.4: were these deviations from intended intervention balanced between groups?
	2.6 Was an appropriate analysis used to estimate the effect of assignment to intervention?			NA
	2.7 If N/PN/NI to 2.6: was there potential for a substantial impact (on the result) of the failure to analyse participants in the group to which they were randomized?			NA
	Risk of bias judgement			Some concerns	Could not be determined whether deviations arose because of the trial context.
Bias due to missing outcome data	3.1 Were data for this outcome available for all, or nearly all, participants randomized?			PY	2/81 (2.5%) randomised participants were not analysed.
	3.2 If N/PN/NI to 3.1: is there evidence that result was not biased by missing outcome data?			NA
	3.3 If N/PN to 3.2: could missingness in the outcome depend on its true value?			NA
	3.4 If Y/PY/NI to 3.3: is it likely that missingness in the outcome depended on its true value?			NA
	Risk of bias judgement			Low	Outcome data were available for 97.5% randomised participants.
Bias in measurement of the outcome	4.1 Was the method of measuring the outcome inappropriate?			N	Appropriate analysis
	4.2 Could measurement or ascertainment of the outcome have differed between intervention groups?			N	The same for the 2 groups.
	4.3 Were outcome assessors aware of the intervention received by study participants?			N	Quote: "2‐year follow up was done by a clinician who was not involved with the endoscopic therapy and was unaware of the results of manometry or randomisation," p.99
	4.4 If Y/PY/NI to 4.3: could assessment of the outcome have been influenced by knowledge of intervention received?			NA
	4.5 If Y/PY/NI to 4.4: is it likely that assessment of the outcome was influenced by knowledge of intervention received?			NA
	Risk of bias judgement			Low	Appropriately measured and unlikely to influence the observer‐reported outcome.
Bias in selection of the reported result	5.1 Were the data that produced this result analysed in accordance with a prespecified analysis plan that was finalised before unblinded outcome data were available for analysis?			NI	Could not be determined as trial registry was not found. Information from authors not received.
	5.2 … multiple eligible outcome measurements (e.g. scales, definitions, time points) within the outcome domain?			PN	No other way to measure the observer‐reported outcome.
	5.3 … multiple eligible analyses of the data?			PN	No other way to analyse.
	Risk of bias judgement			Some concerns	Could not be determined whether there was selection of the reported outcome.
Overall bias	Risk of bias judgement			Some concerns	The trial was judged to raise some concerns in 3 domains for this result, but not to be at high risk of bias for any domains.
Unique ID	111D	Study ID	111D	Assessor	CN
Ref or Label	Cotton 2014	Aim	Assignment to intervention (the 'ITT' effect)
Experimental	Endoscopic sphincterotomy	Comparator	Sham treatment	Source	Journal article(s); non‐commercial trial registry record (e.g. ClinicalTrials.gov record)
Outcome	Serious adverse events	Results	RR 0.71 (95% CI 0.34 to 1.46)	Weight	1
Domain	Signalling question			Response	Comments
Bias arising from the randomisation process	1.1 Was the allocation sequence random?			Y	Quote: "2:1 randomization through a centralized web‐based system. A permuted block scheme with random block size," p.5
	1.2 Was the allocation sequence concealed until participants were enrolled and assigned to interventions?			PY
	1.3 Did baseline differences between intervention groups suggest a problem with the randomisation process?			N	2 treatment groups were well‐balanced with respect to baseline demographic and clinical characteristics, p.7 Table 4
	Risk of bias judgement			Low	No concerns about the randomisation process.
Bias due to deviations from intended interventions	2.1 Were participants aware of their assigned intervention during the trial?			N	Participants, referring physicians, research co‐ordinators, p.5
	2.2 Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?			N
	2.3 If Y/PY/NI to 2.1 or 2.2: were there deviations from the intended intervention that arose because of the experimental context?			NA
	2.4 If Y/PY to 2.3: were these deviations likely to have affected the outcome?			NA
	2.5 If Y/PY/NI to 2.4: were these deviations from intended intervention balanced between groups?			NA
	2.6 Was an appropriate analysis used to estimate the effect of assignment to intervention?			Y	Analysed according to their assigned groups
	2.7 If N/PN/NI to 2.6: was there potential for a substantial impact (on the result) of the failure to analyse participants in the group to which they were randomized?			NA
	Risk of bias judgement			Low	Unlikely to cause deviations.
Bias due to missing outcome data	3.1 Were data for this outcome available for all, or nearly all, participants randomised?			Y	Outcome data were available for all randomised participants (Table 4)
	3.2 If N/PN/NI to 3.1: is there evidence that the result was not biased by missing outcome data?			NA
	3.3 If N/PN to 3.2: could missingness in the outcome depend on its true value?			NA
	3.4 If Y/PY/NI to 3.3: is it likely that missingness in the outcome depended on its true value?			NA
	Risk of bias judgement			Low	All randomised participants analysed (ITT).
Bias in measurement of the outcome	4.1 Was the method of measuring the outcome inappropriate?			N	Used validated tool 'RAPID'
	4.2 Could measurement or ascertainment of the outcome have differed between intervention groups?			N	The same measurement tools applied for both groups.
	4.3 Were outcome assessors aware of the intervention received by study participants?			N	Participants and study investigators were blinded.
	4.4 If Y/PY/NI to 4.3: could assessment of the outcome have been influenced by knowledge of intervention received?			NA
	4.5 If Y/PY/NI to 4.4: is it likely that assessment of the outcome was influenced by knowledge of intervention received?			NA
	Risk of bias judgement			Low	Appropriate method of measuring the observer‐reported outcome.
Bias in selection of the reported result	5.1 Were the data that produced this result analysed in accordance with a prespecified analysis plan that was finalized before unblinded outcome data were available for analysis?			Y	Planned assessment as stated in a trial registry.
	5.2 … multiple eligible outcome measurements (e.g. scales, definitions, time points) within the outcome domain?			PN	Unlikely to have many ways to measure.
	5.3 … multiple eligible analyses of the data?			PN	Unlikely to have many ways to analyse.
	Risk of bias judgement			Low	Planned assessment and unlikely to be selected.
Overall bias	Risk of bias judgement			Low	The trial was judged at low risk of bias for all domains for this outcome.
Unique ID	211B	Study ID	211B	Assessor	CN
Ref or Label	Geenen 1989	Aim	Assignment to intervention (the 'ITT' effect)
Experimental	Endoscopic sphincterotomy	Comparator	Sham treatment	Source	Journal article(s)
Outcome	No improvement in liver function tests	Results	RR 0.89 (95% CI 0.35 to 2.26)	Weight	1
Domain	Signalling question			Response	Comments
Bias arising from the randomisation process	1.1 Was the allocation sequence random?			NI	Quote: "Patients were randomly assigned to undergo either group". p.83 No further details about randomisation method. Allocation concealment: sealed opaque envelopes
	1.2 Was the allocation sequence concealed until participants were enrolled and assigned to interventions?			PY
	1.3 Did baseline differences between intervention groups suggest a problem with the randomisation process?			NI	Not mentioned details of the baseline information.
	Risk of bias judgement			Low	Unlikely to have obvious concerns over randomisation process.
Bias due to deviations from intended interventions	2.1 Were participants aware of their assigned intervention during the trial?			N	Double‐blind clinical trial (participant, physician) Endoscopist was aware of the choice.
	2.2 Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?			PY
	2.3 If Y/PY/NI to 2.1 or 2.2: were there deviations from the intended intervention that arose because of the experimental context?			NI	Could not be determined.
	2.4 If Y/PY to 2.3: were these deviations likely to have affected the outcome?			NA
	2.5 If Y/PY/NI to 2.4: were these deviations from intended intervention balanced between groups?			NA
	2.6 Was an appropriate analysis used to estimate the effect of assignment to intervention?			Y	Appropriate analysis.
	2.7 If N/PN/NI to 2.6: was there potential for a substantial impact (on the result) of the failure to analyse participants in the group to which they were randomized?			NA
	Risk of bias judgement			Some concerns	Could not be determined whether any deviations arose or not.
Bias due to missing outcome data	3.1 Were data for this outcome available for all, or nearly all, participants randomised?			N	7/47 (15%) participants were lost to follow‐up at 4 years.
	3.2 If N/PN/NI to 3.1: is there evidence that result was not biased by missing outcome data?			PN	Likely to be biased.
	3.3 If N/PN to 3.2: could missingness in the outcome depend on its true value?			PY	4 participants declined to continue to participate; likely due to treatment effect.
	3.4 If Y/PY/NI to 3.3: is it likely that missingness in the outcome depended on its true value?			PY
	Risk of bias judgement			High	15% of randomised participants were not analysed at 4‐year follow‐up.
Bias in measurement of the outcome	4.1 Was the method of measuring the outcome inappropriate?			N	Appropriately measured the outcome. Standardised questionnaire, liver function tests, manometry, etc.
	4.2 Could measurement or ascertainment of the outcome have differed between intervention groups?			N	The same measurement in 2 groups.
	4.3 Were outcome assessors aware of the intervention received by study participants?			N	Quote: "patients were interviewed by independent observer who was unaware of the assigned treatment," p.83
	4.4 If Y/PY/NI to 4.3: could assessment of the outcome have been influenced by knowledge of intervention received?			NA
	4.5 If Y/PY/NI to 4.4: is it likely that assessment of the outcome was influenced by knowledge of intervention received?			NA
	Risk of bias judgement			Low	Measured appropriately with no differences in the 2 groups and unlikely to influence the observer‐reported outcome
Bias in selection of the reported result	5.1 Were the data that produced this result analysed in accordance with a pre‐specified analysis plan that was finalized before unblinded outcome data were available for analysis?			NI	Could not be determined – planned assessment.
	5.2 … multiple eligible outcome measurements (e.g. scales, definitions, time points) within the outcome domain?			PN	Unlikely to have other outcome measurements for the observer‐reported outcome.
	5.3 … multiple eligible analyses of the data?			PN	Only 1 possible way to analyse.
	Risk of bias judgement			Some concerns	Could not be determined if there was any selection of the reported outcome.
Overall bias	Risk of bias judgement			High	The trial was judged at high risk of bias due to 14% missingness of randomised participants at 4‐year follow‐up.
Unique ID	411A	Study ID	411A	Assessor	CN
Ref or Label	Takezawa 2004	Aim	Assignment to intervention (the 'ITT' effect)
Experimental	Endoscopic sphincterotomy	Comparator	Endoscopic papillary balloon dilation	Source	Journal article(s)
Outcome	Serious adverse events	Results	RR 0.34 (95% CI 0.04 to 3.15)	Weight	1
Domain	Signalling question			Response	Comments
Bias arising from the randomisation process	1.1 Was the allocation sequence random?			Y	Quote: "Randomisation was done using sealed envelopes prepared by random number generation," p.632 1:1 ratio to either biliary or dual sphincterotomy, who had pancreatic sphincter hypertension.
	1.2 Was the allocation sequence concealed until participants were enrolled and assigned to interventions?			Y
	1.3 Did baseline differences between intervention groups suggest a problem with the randomisation process?			N	Comparable (Table 4) Quote: "There was no significant differences between the two groups (EPBD [endoscopic papillary balloon dilation] vs EST[endoscopic sphincterotomy])."
	Risk of bias judgement			Low	No concerns about the randomisation process.
Bias due to deviations from intended interventions	2.1 Were participants aware of their assigned intervention during the trial?			NI	Could not be determined as the required information was not described. Not found in a clinical registry. No information received from the corresponding author.
	2.2 Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?			NI
	2.3 If Y/PY/NI to 2.1 or 2.2: were there deviations from the intended intervention that arose because of the experimental context?			NI	Could not be determined.
	2.4 If Y/PY to 2.3: were these deviations likely to have affected the outcome?			NA
	2.5 If Y/PY/NI to 2.4: were these deviations from intended intervention balanced between groups?			NA
	2.6 Was an appropriate analysis used to estimate the effect of assignment to intervention?			Y	Appropriate analysis.
	2.7 If N/PN/NI to 2.6: was there potential for a substantial impact (on the result) of the failure to analyse participants in the group to which they were randomized?			NA
	Risk of bias judgement			Some concerns	Could not be determined any deviations.
Bias due to missing outcome data	3.1 Were data for this outcome available for all, or nearly all, participants randomised?			N	47 dropouts in both groups (51.6%). Figure 1 of publication, p.632 (24 in endoscopic sphincterotomy group and 23 in endoscopic papillary balloon dilation group)
	3.2 If N/PN/NI to 3.1: is there evidence that result was not biased by missing outcome data?			PN	Unlikely biased as data analysis was appropriate (Figure 1 of publication)
	3.3 If N/PN to 3.2: could missingness in the outcome depend on its true value?			PY	Non‐acceptance or untraceable (Figure 1 of publication and p.634) About the same reasons of dropouts in both groups.
	3.4 If Y/PY/NI to 3.3: is it likely that missingness in the outcome depended on its true value?			PN
	Risk of bias judgement			Some concerns	Many randomised participants were not analysed at 1‐year follow‐up.
Bias in measurement of the outcome	4.1 Was the method of measuring the outcome inappropriate?			N	Appropriately measured
	4.2 Could measurement or ascertainment of the outcome have differed between intervention groups?			N	The same measurement of the outcome in both groups.
	4.3 Were outcome assessors aware of the intervention received by study participants?			NI	Could not be determined.
	4.4 If Y/PY/NI to 4.3: could assessment of the outcome have been influenced by knowledge of intervention received?			PN	Unlikely to influence the hard outcome.
	4.5 If Y/PY/NI to 4.4: is it likely that assessment of the outcome was influenced by knowledge of intervention received?			NA
	Risk of bias judgement			Low	Measured appropriately with no differences in the 2 groups and unlikely to influence the hard outcome.
Bias in selection of the reported result	5.1 Were the data that produced this result analysed in accordance with a pre‐specified analysis plan that was finalised before unblinded outcome data were available for analysis?			N	Could not be determined. Not found trial registry. No reply from the author.
	5.2 … multiple eligible outcome measurements (e.g. scales, definitions, time points) within the outcome domain?			PN	Unlikely to have many ways to measure the outcome.
	5.3 … multiple eligible analyses of the data?			PN	Unlikely to have many ways to analyse.
	Risk of bias judgement			Some concerns	Selection of the outcome could not be determined.
Overall bias	Risk of bias judgement			High	The trial was judged to raise some concerns in 3 domains for this result, but not to be at high risk of bias for any domains.
Unique ID	111E	Study ID	111E	Assessor	CN
Ref or Label	Cotton 2014	Aim	Assignment to intervention (the 'ITT' effect)
Experimental	Endoscopic sphincterotomy (biliary)	Comparator	Dual endoscopic sphincterotomy	Source	Journal article(s); non‐commercial trial registry record (e.g. ClinicalTrials.gov record)
Outcome	Proportion without treatment success	Results	RR 0.66 (95% CI 0.32 to 1.34)	Weight	1
Domain	Signalling question			Response	Comments
Bias arising from the randomisation process	1.1 Was the allocation sequence random?			NI	Quote: "2:1 randomization through a centralized web‐based system. A permuted block scheme with random block size," p.5 But this subgroup was not mentioned.
	1.2 Was the allocation sequence concealed until participants were enrolled and assigned to interventions?			NI
	1.3 Did baseline differences between intervention groups suggest a problem with the randomisation process?			NI	Overall, 2 treatment groups were well‐balanced with respect to baseline demographic and clinical characteristics, p.7 But this subgroup was not mentioned.
	Risk of bias judgement			Some concerns	Need more information about randomisation process.
Bias due to deviations from intended interventions	2.1 Were participants aware of their assigned intervention during the trial?			NI	This subgroup was not mentioned.
	2.2 Were carers and people delivering the interventions aware of participants' assigned intervention during the trial?			NI
	2.3 If Y/PY/NI to 2.1 or 2.2: were there deviations from the intended intervention that arose because of the experimental context?			NI	This subgroup was not mentioned.
	2.4 If Y/PY to 2.3: were these deviations likely to have affected the outcome?			NA
	2.5 If Y/PY/NI to 2.4: were these deviations from intended intervention balanced between groups?			NA
	2.6 Was an appropriate analysis used to estimate the effect of assignment to intervention?			Y	Appropriate analysis.
	2.7 If N/PN/NI to 2.6: was there potential for a substantial impact (on the result) of the failure to analyse participants in the group to which they were randomized?			NA
	Risk of bias judgement			Some concerns	Need more information.
Bias due to missing outcome data	3.1 Were data for this outcome available for all, or nearly all, participants randomised?			PN	Outcome data were available for the selected participants (Table 5)
	3.2 If N/PN/NI to 3.1: is there evidence that result was not biased by missing outcome data?			PY	Unlikely biased.
	3.3 If N/PN to 3.2: could missingness in the outcome depend on its true value?			NA
	3.4 If Y/PY/NI to 3.3: is it likely that missingness in the outcome depended on its true value?			NA
	Risk of bias judgement			Low	Selected randomised participants were analysed for this outcome.
Bias in measurement of the outcome	4.1 Was the method of measuring the outcome inappropriate?			N	Appropriate method. Used 'RAPID' score.
	4.2 Could measurement or ascertainment of the outcome have differed between intervention groups?			N	The same measurement in both groups.
	4.3 Were outcome assessors aware of the intervention received by study participants?			NI	This subgroup was not mentioned.
	4.4 If Y/PY/NI to 4.3: could assessment of the outcome have been influenced by knowledge of intervention received?			PN	Unlikely influenced the observer‐reported outcome.
	4.5 If Y/PY/NI to 4.4: is it likely that assessment of the outcome was influenced by knowledge of intervention received?			NA
	Risk of bias judgement			Low	Appropriate method of measuring the observer‐reported outcome.
Bias in selection of the reported result	5.1 Were the data that produced this result analysed in accordance with a pre‐specified analysis plan that was finalised before unblinded outcome data were available for analysis?			Y	Planned assessment as stated in a trial registry. NCT00688662
	5.2 … multiple eligible outcome measurements (e.g. scales, definitions, time points) within the outcome domain?			PN	Unlikely to have many ways to measure.
	5.3 … multiple eligible analyses of the data?			PN	Unlikely to have been many ways to analyse.
	Risk of bias judgement			Low	Planned assessment and unlikely to be selected.
Overall bias	Risk of bias judgement			Some concerns	The trial was judged with some concerns in 2 domains for this result, but not to be at high risk of bias for any domains.

CI: confidence interval; ITT: intention to treat; MD: mean difference; N: no; NA: not available; NI: no information; PN: probably no; PY: probably yes; RAPID: Recurrent Abdominal Pain Intensity and Disability; RR: risk ratio; Y: yes.

Data and analyses

Comparison 1. Endoscopic sphincterotomy versus sham operation.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Proportion of participants without treatment success at the longest follow‐up (1–4 years)	3	340	Risk Ratio (IV, Random, 95% CI)	1.05 [0.66, 1.66]
1.2 Proportion of participants with serious adverse events (pancreatitis, at 1 year)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.3 Health‐related quality of life (measured with SF‐36 Physical scale at 1 year)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
1.4 Health‐related quality of life (measured with SF‐36 Mental scale at 1 year)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
1.5 Proportion of participants without improvement in liver function tests	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Comparison 2. Endoscopic sphincterotomy versus endoscopic papillary balloon dilation.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Proportion of participants with serious adverse events (recurrent common bile duct stones, at 1 year)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Comparison 3. Biliary endoscopic sphincterotomy alone versus dual endoscopic sphincterotomy.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Proportion of participants without treatment success (at 1 year)	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

Comparison 4. Subgroup analysis: endoscopic sphincterotomy versus sham.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Risk of bias. Proportion of participants without treatment success at the longest follow‐up (1–4 years)	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
4.1.1 Trials at low risk of bias	1	214	Risk Ratio (M‐H, Random, 95% CI)	1.23 [1.01, 1.49]
4.1.2 Trials at some concern and at high risk of bias	2	126	Risk Ratio (M‐H, Random, 95% CI)	0.94 [0.41, 2.18]

Comparison 5. Sensitivity analysis: proportion of participants without treatment success at the longest follow‐up (1–4 years).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Best‐worst case scenario analysis	3	340	Risk Ratio (M‐H, Random, 95% CI)	0.91 [0.62, 1.33]
5.2 Worst‐best case scenario analysis	3	340	Risk Ratio (M‐H, Random, 95% CI)	1.29 [0.77, 2.17]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Cotton 2014.

Study characteristics
Methods	EPISOD randomised clinical trial Multicentre, parallel design, prospective, sham‐controlled, randomised trial, phase III The trial used a 2‐stage randomisation Site: 7 centres in the USA Study period: 6 August 2008 to 23 March 2012
Participants	214 participants with pain after cholecystectomy without significant abnormalities on imaging or laboratory studies, and no prior sphincter treatment or pancreatitis Suspected sphincter of Oddi dysfunction type III based on Rome III criteria (indicated in a complementary study by Yaghoobi 2015; see reference under Cotton 2014). Cholecystectomy was performed on average 4 years before study enrollment (indicated in Brawman‐Mintzer 2014; see reference under Cotton 2014). Mean age (years): intervention group: 38 (SD11); control group: 39 (SD11) Sex: 17 (7%) men Inclusion criteria: aged 18–65 years had pain > 3 months after cholecystectomy no history of pancreatitis (or evidence for it on imaging or laboratory tests) no prior sphincter intervention Exclusion criteria: if within the last 6 months, levels of direct bilirubin, alkaline phosphatase, amylase, or lipase were > twice normal or if transaminases were > 3 times normal if narcotic analgesics were used daily in the past month if they were known to have pancreas divisum, prior surgical biliary diversion, abnormal endoscopic ultrasound (if done), significant psychiatric or other major medical problems, or were pregnant
Interventions	Total number of participants randomised Experimental: 141 Control: 73 Trial authors performed 2 per‐protocol analyses First analysis Experimental intervention: ERCP with sphincterotomy (141 participants) Sphincter of Oddi manometry was performed using a basal pressure > 40 mmHg in both leads, to define an abnormality in the biliary and pancreatic sphincters. Control intervention: sham treatment with no sphincterotomy (71 participants) Received small calibre, 3‐ to 5‐French diameter, pancreatic stents to reduce the risk of postprocedure pancreatitis; these passed spontaneously within 1–4 weeks. Second analysis Experimental intervention: biliary sphincterotomy alone (52 participants) Control intervention: dual (biliary and pancreatic) sphincterotomy (47 participants)
Outcomes	Relevant to this review Treatment success: defined as < 6 days of disability due to pain in the prior 90 days both at months 9 and 12 after randomisation, with no narcotic use and no further sphincter intervention (Treatment failure: if the 12‐month RAPID score was missing or collected outside the acceptable window (within 6 months of the expected 12‐month visit)) Safety outcome (ERCP‐related pancreatitis, procedure‐related bleeding, etc.) Quality of life (i.e. 12‐month SF‐36 Physical and Mental Component Summaries scores; HADS anxiety and depression scores) Resource utilisation Adverse events Not relevant to this review Pancreatic sphincter hypertension with biliary sphincterotomy Pancreatic sphincter hypertension with pancreatic and biliary sphincterotomy
Notes	Trial registration: Clinicaltrials.gov Identifier: NCT00688662 Funding: grant U01 DK074739 from the National Institutes of Diabetes and Digestive and Kidney Diseases (NIDDK). The study was funded by NIDDK as a co‐operative agreement. The NIDDK staff participated in the design of the study and monitored its progress with a steering committee and Data and Safety Monitoring Board (DSMB). The investigators were responsible for all elements of the trial including design, data collection, and analysis. The study was monitored by an independent DSMB appointed by NIDDK. Conflict of interest: disclosures Cotton PB: received a gift of endoscopic accessories (sphincterotomies) from Cook Medical to be used in the study, to prevent participants from being unblinded by receiving a bill, consulted for Olympus America, and received royalties from Cook Medical for devices not used in the study. Romagnuolo J: received lecture fees from Olympus and Cook Medical. Fogel E: received lecture fees from Olympus. Tarnasky P: received lecture fees from Boston Scientific. Aliperti G: received lecture fees from Oklahoma Endoscopy Society. Freeman M: obtained research grants from Boston Scientific and Cook Endoscopy. Jamidar P: received lecture fees from Olympus and Boston Scientific. Robuck P and Serrano J: employees of NIDDK during key parts of the study. Elta G: consulted for Olympus. Drossman D and Elta G: compensated by the EPISOD grant as consultants. All other authors had some of their work funded by the grant from NIDDK to perform this study and were reimbursed for expenses involved in travelling to investigator meetings. Sources obtained for risk of bias assessment Journal article with results of the trial Non‐commercial trial registry

Geenen 1989.

Study characteristics
Methods	Prospective, randomised, double‐blind clinical trial Allocation concealment: sealed opaque envelopes Blinding: double‐blind (participant, physician) The endoscopist was aware of the choice. Site: Medical College of Wisconsin, USA Study period: 1980–1983
Participants	Total number of participants randomised: 47 Experimental: 23 Control: 24 Trial authors did 2 analyses at 2 different time points At 1‐year follow‐up: 47 participants At 4‐year follow‐up: 40 participants (symptomatic people in the sham group were relocated to the experimental group) Mean age (years): 49 (SE 2) Sex: NA Inclusion criteria: adults with suspectedsphincter of Oddi dysfunction type II documented on ≥ 2 occasions unexplained pain resembling biliary pain and lasting for > 6 months after cholecystectomy either 1 or 2 of the following objective findings suggesting partial common duct obstruction common duct dilation to > 12 mm determined by ERCP delayed emptying of contrast medium during ERCP elevated liver function tests Exclusion criteria: people with pancreatic diseases, gastrointestinal diseases, or severe illness pregnant or aged < 18 years had undergone biliary tract operations other than cholecystectomy
Interventions	Experimental intervention: endoscopic sphincterotomy (23 participants) Sphincterotomy was performed using traction‐type papillotome (Wilson Cook) Control intervention: sham sphincterotomy (24 participants) Sham sphincterotomy was performed exactly as true sphincterotomy, except that the sphincterotome was positioned in the duodenal lumen Follow‐up: 3 months, and 1 and 4 years
Outcomes	Relevant to this review Improvement in pain scores 1 year after therapy Good, fair, or no clinical improvement Not relevant to this review None
Notes	Trial registration: not reported Funding: approved by the Human Research Studies Committee of Medical College of Wisconsin Conflict of interest: not reported Sources obtained for risk of bias assessment Journal article with results of the trial

Takezawa 2004.

Study characteristics
Methods	Prospective randomised clinical trial Site: Department of Gastroenterology Kitasato University East Hospital, Sagamihara, Japan Study period: May 1998 to May 2002
Participants	91 participants with bile stones Number included: 91; intervention group: 45; control group: 46 Median age (years): intervention group: 70 (range 40–90); control group: 69 (range 41–93) Sex: 62 (68%) men Inclusion criteria: written informed consent no history of previous papilloplasty no obstructive acute suppurative cholangitis no stone impaction at the sphincter of Oddi no acute or chronic pancreatitis no malignancy of the bile duct, liver, or pancreas no anomalous pancreaticobiliary ductal union, defined by Misra and Dwivedi (Misra 1990) no severe bleeding tendency Exclusion criteria: people with failure to collect bile before removal of common bile duct stones people with precut sphincter of Oddi
Interventions	Total number of participants randomised Experimental intervention: 104 participants Control intervention: 96 participants Trial authors performed a per‐protocol analysis Experimental intervention: endoscopic sphincterotomy (45 participants) Control intervention: endoscopic papillary balloon dilation (46 participants)
Outcomes	Relevant to this review Recurrent common bile duct stone Not relevant to this review Change in the levels of the 5 pancreatic enzymes Follow‐up: 1 year
Notes	Trial registry: no information. We contacted the corresponding author but received no reply. Funding: not reported Conflict of interest: "The review board for human research at our institute approved the study protocol." We e‐mailed the authors on 29 April 2023, but it did reach the author. Sources obtained for risk of bias assessment Journal article with results of the trial

Toouli 2000.

Study characteristics
Methods	Prospective randomised parallel sham‐controlled trial Clinician was blinded to the procedure. Site: Flinders Medical Centre, Adelaide, in collaboration with the Royal North Shore Hospital, Sydney Study period: May 1987 to October 1996
Participants	81 participants with recurrent biliary‐type pain after cholecystectomy 26 with sphincter of Oddi stenosis 21 with sphincter of Oddi dyskinesia 32 normal manometry 2 were excluded (1 lost to follow‐up, 1 withdrawn) Median age (years): 45 (range 22–71) Sex: 8 (9.8%) men Inclusion criteria: recurrent biliary‐type pain for ≥ 6 months in a participant who had undergone previous cholecystectomy plus ≥ 1 of the following: high liver function text (serum aminotransferases) dilated common bile duct (> 12 mm) positive morphine/neostigmine provocation Exclusion criteria: people with recurrent pancreatitis (abdominal pain with elevated serum amylase levels)
Interventions	Total number of participants randomised Experimental: 37 Control: 42 Trial authors performed per‐protocol analyses. Experimental intervention: endoscopic sphincterotomy (37 participants) Control intervention: sham procedure (42 participants)
Outcomes	Relevant to this review Assessment of pain at 24 months ('no change' (worse), 'improved symptoms', or 'asymptomatic' (no further episodes of typical abdominal pain)) Severe adverse events Not relevant to this review None
Notes	Trial registration: not reported Funding: not reported. Assumed as Flinders Medical Centre, Adelaide, and the Royal North Shore Hospital, Sydney Conflict of interest: not reported Sources obtained for risk of bias assessment Journal article with results of the trial

EPISOD: Evaluating Predictors and Interventions in Sphincter of Oddi Dysfunction; ERCP: endoscopic retrograde cholangiopancreatography; HAD: Hospital Anxiety and Depression Scale; NA: not available/not reported; SD: standard deviation; SE: standard error; SF‐36: 36‐item Short Form Health Survey; RAPID: Recurrent Abdominal Pain Intensity and Disability.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Cicala 2002	Not an RCT; a controlled clinical study
Cotton 2019	Not an RCT; a 'views' article
Craig 2002	A narrative review.
Fullarton 1992	An RCT that did not investigate sphincterotomy, compared 2 drugs.
Guelrud 1988	Drug efficacy trial (nifedipine vs placebo)
Khuroo 1992	Assessed drug therapy on sphincter of Oddi dysfunction
Linder 2003	Observational study
Neoptolemos 1988	Observational study
Ren 2021	Not an RCT; a retrospective observational study
Sugawa 2001	Not an RCT; a retrospective observational study
Swan 2013	A prospective trial that assessed participants with an intact papilla
Tang 2018	A review
Thatcher 1987	A record review
Wehrmann 1996	Efficacy of endoscopic botulinum toxin injection

RCT: randomised clinical trial.

Differences between protocol and review

We made the following changes from the protocol (Naing 2022). In the protocol, we had specified that whenever data for the same outcome were provided using different measurement techniques, we would perform separate analyses. For instance, a trial that was published in 2014 used the RAPID (Recurrent Abdominal Pain Intensity and Disability) instrument to measure treatment success (Cotton 2014), while earlier trials used other, unidentified instruments (Geenen 1989; Toouli 2000). Depending on how instruments have evolved over time, different instrument applications for evaluating treatment efficacy may be appropriate. In this review, we accepted both the RAPID and non‐RAPID instruments for measuring treatment outcomes as a result of the tool's evolution.

Contributions of authors

CN: collected data, assessed risk of bias, analysed and interpreted data, wrote the review with suggestions of team members.

HN: assisted with data synthesis, assessed risk of bias, rated certainty of the evidence, interpreted results, assisted with writing the review, commented on the review.

HHA: collected data, assisted with risk of bias assessment and data synthesis, assisted with writing the review, commented on the review.

CP: interpreted results, assisted with writing the review, commented on the review.

All authors approved the review for publication.

Text in this review may overlap with other Cochrane reviews. This is because protocols and reviews follow the Cochrane methodology.

Sources of support

Internal sources

• Newcastle University Medicine, Malaysia

  Internet access to literature through Newcastle University.

  HN performed this review mainly during free time and occasionally during office hours at this institution.

External sources

• The Cochrane Hepato‐Biliary Group Editorial Team, Denmark

  Provided guidance to authors and ran the peer review process.

Declarations of interest

CN: none

HN: Cochrane editor, not involved in the editorial process.

HHA: none

CP: Cochrane editor, not involved in the editorial process.

None of the authors received financial support for this review.

New