Sedative techniques for endoscopic retrograde cholangiopancreatography

Abstract

Background

Endoscopic retrograde cholangiopancreatography (ERCP) is an uncomfortable therapeutic procedure that cannot be performed without adequate sedation or general anaesthesia. A considerable number of ERCPs are performed annually in the UK (at least 48,000) and many more worldwide.

Objectives

The primary objective of our review was to evaluate and compare the efficacy and safety of sedative or anaesthetic techniques used to facilitate the procedure of ERCP in adult (age > 18 years) patients.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2010, Issue 8); MEDLINE (1950 to September 2011); EMBASE (1950 to September 2011); CINAHL, Web of Science and LILACS (all to September 2011). We searched for additional studies drawn from reference lists of retrieved trial materials and review articles and conference proceedings.

Selection criteria

We considered all randomized or quasi‐randomized controlled studies where the main procedures performed were ERCPs. The three interventions we searched for were (1) conscious sedation (using midazolam plus opioid) versus deep sedation (using propofol); (2) conscious sedation versus general anaesthesia; and (3) deep sedation versus general anaesthesia. We considered all studies regardless of which healthcare professional administered the sedation.

Data collection and analysis

We reviewed 124 papers and identified four randomized trials (with a total of 510 participants) that compared the use of conscious sedation using midazolam and meperidine with deep sedation using propofol in patients undergoing ERCP procedures. All sedation was administered by non‐anaesthetic personnel. Due to the clinical heterogeneity of the studies we decided to review the papers from a narrative perspective as opposed to a full meta‐analysis. Our primary outcome measures included mortality, major complications and inability to complete the procedure due to sedation‐related problems. Secondary outcomes encompassed sedation efficacy and recovery.

Main results

No immediate mortality was reported. There was no significant difference in serious cardio‐respiratory complications suffered by patients in either sedation group. Failure to complete the procedure due to sedation‐related problems was reported in one study. Three studies found faster and better recovery in patients receiving propofol for their ERCP procedures. Study protocols regarding use of supplemental oxygen, intravenous fluid administration and capnography monitoring varied considerably. The studies showed either moderate or high risk of bias.

Authors' conclusions

Results from individual studies suggested that patients have a better recovery profile after propofol sedation for ERCP procedures than after midazolam and meperidine sedation. As there was no difference between the two sedation techniques as regards safety, propofol sedation is probably preferred for patients undergoing ERCP procedures. However, in all of the studies that were identified only non‐anaesthesia personnel were involved in administering the sedation. It would be helpful if further research was conducted where anaesthesia personnel were involved in the administration of sedation for ERCP procedures. This would clarify the extent to which anaesthesia personnel should be involved in the administration of propofol sedation.

Plain language summary

What is the best sedative technique for patients undergoing endoscopic cholangiopancreatography (ERCP)?

Endoscopic retrograde cholangiopancreatography (ERCP) is an uncomfortable procedure that uses an endoscope and X rays to visualize problems with patients’ bile and pancreatic ducts. An endoscope is a medical device that can be used to examine the inside of a patient’s body and consists of a long, thin, flexible, fibreoptic tube with a light source and a video camera. A large number (in excess of several 100,000) of these procedures are performed annually on a worldwide basis. 

Patients have to be given medicines to make them adequately drowsy (sedated) or unconscious (anaesthetized) to tolerate the ERCP procedure. These medicines may be administered by anaesthetic or non‐anaesthetic healthcare personnel, and there is currently some debate as to who should administer these drugs to patients for ERCP procedures.

Two common ways of  providing sedation for ERCP are to administer combined intravenous midazolam (sedative agent) and meperidine (morphine‐like opioid) or the drug propofol (a sedative‐anaesthetic agent) to patients. This review evaluated and compared the safety and effectiveness of the sedative techniques in patients undergoing ERCP procedures.

From a worldwide literature search we identified only four randomized controlled papers appropriate for review, involving a total of 510 patients. These papers compared the use of midazolam and meperidine with propofol sedation techniques for patients undergoing ERCP procedures. All the sedation was administered by non‐anaesthetic healthcare personnel.

There was no significant difference between the sedation techniques as regards safety. There were no deaths in the trials and the number of major complications, such as lack of oxygen (hypoxaemia) and low blood pressure (hypotension), was comparable in both techniques. There was no difference in patient satisfaction in either group. However, the recovery of patients who received propofol was significantly better than for those who had received midazolam and meperidine for the procedure.

In conclusion, patients undergoing ERCP procedures under propofol sedation recover faster and better than those patients receiving midazolam and meperidine sedation. This would make propofol the preferred choice for these procedures as there was no difference in the safety of either technique. Further research should focus on the safety of the sedative techniques and involve anaesthesia personnel in the administration of the sedation.

Summary of findings

for the main comparison.

Propofol compared with midazolam and meperidine for ERCP (and related) procedures
Patient or population: Adults (over 18 years) with requirement for an ERCP procedure with diagnostic intent Settings: Typically acute hospital setting (two based in Germany, one in Thailand. Study from USA: setting not stated) Intervention: Deep sedation with propofol Comparison: Conscious sedation with midazolam plus meperidine
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Conscious sedation	Deep sedation
Primary outcome: Mortality	See comment	See comment	Not estimable	510 (4 studies)	⊕⊕⊕⊝ moderate (NB please see footnote)	0 participants died in the studies. Please see narrative for examination of 30 day follow‐up mortality (unrelated to sedation) findings.
Primary Outcome: Hypoxaemia	See Comment	See Comment	Not estimable	510 (4 studies)	⊕⊕⊕⊝ moderate (NB please see footnote)	No significant difference between groups.
Primary Outcome: Hypotension	See Comment	See Comment	Not estimable	510 (4 studies)	⊕⊕⊕⊝ moderate (NB please see footnote)	No significant difference between groups.
Primary outcome: Failure to complete procedure (due to sedation‐related problems)	See comment	See comment	Not estimable	360 (3 studies)	⊕⊕⊕⊝ moderate (NB please see footnote)	There were instances of failure to complete in 1 study (Kongkam 2008) In the deep sedation (propofol) arm 3 procedures were terminated before completion, with 2 due to agitation and 1 due to apnoea. In the conscious sedation (Midazolam/Meperidine) arm, 2 procedures were terminated before completion, with 1 due to agitation and 1 due to aspiration. With there being sedation‐related problems in both sedative‐arms of this study (and only this study), no firm conclusions can be drawn.
Secondary outcome: Recovery	See comment	See comment	Not estimable	359 (3 studies)	⊕⊕⊕⊝ moderate (NB please see footnote)	The recovery profile in patients who had received propofol was superior to that in patients receiving midazolam/meperidine in the three separate studies. Please see narrative of main text for full details.
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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 quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

A conventional assessment of Quality of the evidence (GRADE) was not possible due to the absence of statistical aggregation. It was not, for example, possible to estimate the level of heterogeneity among studies contributing to an outcome. The judgment of ⊕⊕⊕⊝ moderate is included to reflect the variability among the studies with respect to risk of bias of the studies

Background

Description of the condition

Endoscopic retrograde cholangiopancreatography (ERCP) is a complex gastrointestinal procedure that has been practised for over 30 years. Currently, as many as 48,000 ERCPs are performed annually in the UK alone (Williams 2007). The procedure was developed as a result of dramatic technical advances in flexible endoscopy. The technique involves imaging of the biliary tree and pancreatic duct following endoscopy and is used to aid the diagnosis of obstruction, for example by gallstones or cholangiocarcinoma. ERCP has a reported complication rate of between 5% and 10%, with a death rate of 0.1% to 1% (Cohen 2002; Freeman 2002; Vandervoort 2002). Complications include acute pancreatitis, haemorrhage and perforation. ERCP is an uncomfortable procedure that requires adequate patient sedation or general anaesthesia for successful completion.

Description of the intervention

Levels of sedation in current practice have been defined by the American Society of Anesthesiologists (ASA) House of Delegates (ASA 1999). The categories are minimal sedation, conscious sedation and deep sedation. Minimal sedation (anxiolysis) is a drug‐induced state in which patients respond normally to verbal commands. Although cognitive function and co‐ordination may be impaired, ventilatory and cardiovascular functions are unaffected. Conscious sedation, during which the patient has a purposeful response to verbal and tactile stimulation, uses drugs such as midazolam and meperidine and is attempted by many gastroenterologists for patients having ERCP procedures. However, unintended levels of deep sedation may frequently occur in patients undergoing conscious sedation (Patel 2005). Deep sedation, where a patient has a purposeful response following repeated or painful stimulation, may be associated with inadequate spontaneous ventilation and requires a level of care identical to general anaesthesia. Deep sedation is often performed by an anaesthetist. This degree of sedation is recognized as a separate entity to general anaesthesia in the United States but in the UK the distinction is not as clear (RCA 2002).

In addition to sedation, patients undergoing ERCP procedures often receive lignocaine spray to the throat before the procedure is commenced. This is performed to obtund the gag reflex and thereby facilitate the introduction of the endoscope and reduce the amount of sedative required. However, for those patients who are more than normally sensitive to the effects of sedation, or who have difficulty swallowing, the combined effects of sedation with oropharyngeal local anaesthesia may increase the risk of aspiration (NCEPOD 2004).

Many gastroenterologists prefer to perform ERCP procedures in the semiprone or prone position, although there is some debate as to whether position impacts on ease of performance of the procedure (Das 2008). This may create a conflict if anaesthesia personnel are involved in the sedation administration as the airway is more difficult to access in the prone position, and some anaesthesia specialists may be reluctant to administer deep sedation in this situation. Conversely, it has been argued that the prone position improves oxygenation and reduces the risk of aspiration (Das 2008).

How the intervention might work

Midazolam is a fast‐acting benzodiazepine with a short elimination half‐life and was first synthesized in 1976 by Fryer and Walser (Walser 1976). It has sedative, anxiolytic and amnesic properties. It is frequently used for conscious sedation, either on its own or in combination with meperidine. Meperidine (pethidine) is an effective opioid analgesic drug that was synthesized in 1937 by the Hoechst chemists Eisleb and Schaumann (Eisleb 1939). Meperidine is used with midazolam to facilitate gastrointestinal procedures such as colonoscopies and ERCPs.

Propofol is a more recent short‐acting intravenous anaesthetic agent which, in a dose range of 1.5 to 2.5 mg/kg, rapidly produces sleep followed by prompt recovery in four to six minutes (Sear 1985). Propofol is also used in intermittent boluses by anaesthetists or physicians to provide either conscious or deep sedation. It has a shorter half‐life than meperidine or midazolam and a quicker recovery time. There are some studies showing that the endoscopist rated the sedation and technical performance of ERCP more favourably when sedation with propofol was used instead of midazolam and meperidine (Jung 2000; Wehrmann 1999). A shorter ERCP duration has also been demonstrated when propofol was used for sedation (Chen 2005). Pain at the injection site is a frequent minor complication of propofol use while hypotension and respiratory depression are more serious risks. Deep sedation and potential respiratory embarrassment is more likely to occur with propofol than with midazolam so the personnel administering the drug should be skilled in airway and respiratory support.

Why it is important to do this review

The administration of propofol by non‐anaesthesia personnel is a controversial issue. Both the Canadian and American Society of Gastroenterologists have endorsed the administration of propofol by non‐anaesthetic personnel for non‐complicated endoscopic procedures in patients (Byrne 2008; Position Statement 2009). There is also a growing body of evidence that endoscopic procedures may be performed with minimal morbidity and mortality under propofol sedation (Deenadayalu 2008). However, the US Food and Drug Administration (FDA) has recently denied a petition by the American Society of Gastroenterologists to remove the stipulation that only anaesthesia trained personnel can administer propofol for sedation procedures (FDA 2010). Whether or not this ruling will be modified in the future is difficult to foresee but a key question will be whether the safety profile of anaesthesia personnel delivered propofol sedation is better than that of non‐anaesthesia personnel administered propofol sedation. Propofol may also be administered by a target‐controlled infusion system and this can provide effective sedation during ERCP. However, this technique usually has to be performed by an anaesthetist as continuous monitoring is required (Fanti 2004).

In summary, ERCP is an uncomfortable procedure that requires either adequate sedation or general anaesthesia. There are numerous options available and this systematic review aimed to identify the best method.

Objectives

The primary objective of this review was to evaluate and compare the efficacy and safety of sedative or anaesthetic techniques to facilitate the procedure of endoscopic retrograde cholangiopancreatography (ERCP) in adult patients (those over 18 years of age). The initial plan had been to identify studies where anaesthesiologists or anaesthetists were also involved in the administration of the sedation. However, as there were no relevant studies involving anaesthesia personnel, the review focuses on non‐anaesthesia personnel administered sedation.

Methods

Criteria for considering studies for this review

Types of studies

We included published randomized controlled and quasi‐randomized trials regardless of blinding or language of publication.

Types of participants

We included men and women aged over 18 years who were undergoing ERCP procedures.

Types of interventions

We initially planned three interventions (see 'Differences between protocol and review'). The types of interventions we examined were necessarily broadened from the protocol stage. The reason for this change was that none of the selected studies reported on sedation administered by anaesthesia personnel. The review considered all studies examining endoscopic retrograde cholangiopancreatography (ERCP) and related procedures (where ERCP comprised the majority of those procedures) performed under deep sedation using propofol versus conscious sedation using midazolam plus opioid, regardless of which healthcare professional administered the sedative.

Types of outcome measures

Primary outcomes

1. Mortality (all causes; due to sedation, anaesthesia or the ERCP procedure)

2. Major complications (e.g. hypoxaemia and hypotension due to sedation)

3. Failure to complete ERCP (due to sedation‐related problems)

Secondary outcomes

1. Sedation efficacy and recovery

Search methods for identification of studies

Electronic searches

We searched: the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2010, Issue 8); MEDLINE (Ovid) (1950 to September 2011); EMBASE (1950 to September 2011); CINAHL, Web of Science and LILACS (all to September 2011). We searched MEDLINE using the search terms detailed in Appendix 1 and CENTRAL and EMBASE using a similar strategy, as detailed in Appendix 2 and Appendix 3. We searched CINAHL, ISI Web of Science and LILACS using a similar strategy that is detailed in Appendix 4, Appendix 5 and Appendix 6. We searched all databases via the Ovid platform, with significant assistance from the Cochrane Anaesthesia Review Group (CARG).

Searching other resources

We searched for additional studies from:

1. the reference lists of retrieved trial materials and review articles;

2. conference abstracts and proceedings.

We did not impose a language restriction.

Data collection and analysis

Selection of studies

Three authors (DG, JN, PW) independently scanned the titles and abstracts identified by electronic and manual searching. One author (SP) then retrieved the potentially relevant studies for full evaluation.

Data extraction and management

Two authors (DG and SP) independently extracted data using a simple data extraction form to evaluate both risk of bias (using the Cochrane Handbook for Systematic Reviews of Interventions Figure 8.6a as our template (Higgins 2011)) and the interventions and outcomes.

Assessment of risk of bias in included studies

Two authors (DG and SP) independently assessed the studies of interest for risk of bias and methodological quality as described below. These were according to categories described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

We assessed studies for risk of bias using the following questions.

• Was the allocation sequence adequately generated?

• Was allocation adequately concealed?

• Was knowledge of the allocated interventions adequately prevented during the study?

• Were incomplete data adequately addressed?

• Are reports of the study free of suggestion of selective reporting bias?

• Free of detection bias?

• Free of attrition bias?

We graded each of these questions: high risk of bias, low risk of bias, or unclear for all included studies (Kongkam 2008; Riphaus 2005; Schilling 2009; Vargo 2002). An overall assessment was then made for each study:

A. low risk of bias (all of the criteria met);

B. unclear risk of bias (one or more criteria partly met);

C. high risk of bias (one or more criteria not met).

There were no disagreements on the assessments of the studies.

Measures of treatment effect

Primary measures of treatment effect were mortality, hypoxaemia, hypotension and failure to complete procedure due to a sedation‐related problem. Recovery, onset of sedation and patient co‐operation were examined secondarily.

Unit of analysis issues

There were no unit of analysis issues in all four studies, however in one (Schilling 2009) there were no reported values for recovery or onset of sedation. All studies were randomized trials.

Dealing with missing data

We did not identify any missing data.

Assessment of heterogeneity

Due to varying age groups and morbidity status of patients in the studies, we felt the degree of clinical heterogeneity was too high to be able to perform a meta‐analysis.

Assessment of reporting biases

We did not consider that any overall reporting biases were present, however there was one study which had a high risk of selective reporting bias (full details below).

Data synthesis

The degree of clinical heterogeneity was too high to be able to perform a meta‐analysis, so the review was written in a narrative style.

Subgroup analysis and investigation of heterogeneity

We did not perform subgroup analysis.

Sensitivity analysis

We did not perform sensitivity analysis.

Results

Description of studies

Results of the search

The studies included were randomized and conducted on patients undergoing endoscopic procedures that included ERCP procedures. The patients were allocated into two groups, one that received midazolam and meperidine and the other that received propofol by bolus injection or infusion.

Our search in September 2011 identified 124 potential papers for inclusion in the review (see Figure 1). DG, JN and PW reviewed these fully. As no studies fulfilled the initial selection criteria (due to the fact that there were no papers where anaesthesia personnel administered the sedation directly), these criteria were broadened. Following this we selected seven papers for further review. Three of these were excluded.

1.

Study flow diagram.

Included studies

Characteristics of included studies

We included four studies with 510 patients (Kongkam 2008; Riphaus 2005; Schilling 2009; Vargo 2002). These are further described in the 'Characteristics of included studies' table. Study sample sizes varied from 75 to 150 participants, and the studies involved non‐anaesthesia personnel administered sedation to patients undergoing endoscopic procedures including ERCP. Two of the trials were conducted in Germany, one in America and one in Thailand. Two of the papers (Kongkam 2008; Vargo 2002) describe blinding of assessors in the patient recovery period. Propofol is a rapidly acting agent with a short half life and is often given by intermittent bolus injection to patients undergoing ERCP procedures. Because of these features, we felt that it would have been difficult to adequately blind the endoscopist and procedural personnel to the type of sedative drug used. Two of the papers (Riphaus 2005; Vargo 2002) described gastroenterologist administered sedation with all drugs, one (Kongkam 2008) gastroenterologist directed non‐physician assistant administration of sedation, and one (Schilling 2009) physician controlled nurse administered sedation. In two of the papers ( Kongkam 2008; Vargo 2002) the participants were younger (aged 18 years and above) and of ASA 1 to 3 status, and in the other two (Riphaus 2005; Schilling 2009) the patients were older (over 80 years) and had a higher ASA status of 3 to 4. In three of the papers the propofol was administered by bolus injection and in one by initial bolus followed by infusion. All papers reported on the primary outcome measures of major complications, only one reported instances of failure due to sedation‐related problems (Kongkam 2008), and only three reported on all parts of the reviews secondary outcome, efficacy of sedation. The Schilling 2009 paper only reported on patient co‐operation.

Excluded studies

Characteristics of excluded studies

We excluded the Koshy 2000 and Agostoni 2007 papers because on closer inspection none of the procedures examined involved ERCP. Also, different drug combinations than those which we specified were used. The paper by Chen 2005 was excluded because the control group were given intramuscular diazepam and meperidine as opposed to intravenous midazolam and meperidine.

Risk of bias in included studies

We used the Cochrane Collaboration's risk of bias tables to assess the validity and quality of the included trials. The studies showed either moderate or high risk of bias primarily due to inadequate information on allocation concealment and blinding issues in the procedure room (see Figure 2, Figure 3 and Characteristics of included studies). Blinding of the operators to the drugs given immediately before the procedure would not have been feasible due to the reasons pointed out previously. Blinding of the participants and nurses who performed pre‐procedure and recovery checks was carried out. There were no incomplete data outcomes.

2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

3.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Randomization

Three of the studies (Riphaus 2005; Schilling 2009; Vargo 2002) had adequate sequence generation and one (Kongkam 2008) did not. Allocation concealment was adequate in one study (Riphaus 2005), unclear in two studies (Schilling 2009; Vargo 2002), and inadequate (high risk of bias) in one study (Kongkam 2008).

Blinding

Three of the papers showed an unclear risk of bias for blinding (Kongkam 2008; Riphaus 2005; Schilling 2009) and one Vargo 2002 did not attempt blinding in the procedure room (high risk of bias).

Incomplete outcome data

All four studies showed a low risk of attrition bias.

Selective reporting

Three studies showed an unclear risk of selective reporting (Riphaus 2005; Schilling 2009; Vargo 2002), with one at high risk (Kongkam 2008).

Other potential sources of bias

Three studies showed an unclear risk of selective reporting (Riphaus 2005; Schilling 2009, Vargo 2002), with one high (Kongkam 2008).

Effects of interventions

See: Table 1

The outcomes for the four trials are summarized in Table 2 and Table 3.

1. Primary outcomes.

Primary Outcome	Death	Hypoxaemia	Hypotension < 90 mm Hg	Failure to complete due to sedation problem
Vargo 2002	Not reported	No significant difference between groups	No significant difference between groups	No failures recorded in either group
Riphaus 2005	Not reported	No significant difference between groups	No significant difference between groups	No failures recorded in either group
Schilling 2009	Not reported	No significant difference between groups	No significant difference between groups	No failures recorded in either group
Kongkam 2008	Not reported	No significant difference between groups	No significant difference between groups	5 failures but no significant difference between groups

2. Secondary outcomes.

Secondary outcome	Recovery	Onset of sedation	Patient co‐operation
Vargo 2002	Propofol group recovered better and faster P < 0.001	Shorter in Propofol Group	No difference between the groups
Riphaus 2005	Propofol group recovered better and faster P < 0.01	Shorter in Propofol Group	Superior in propofol group
Schilling 2009	Not recorded	Not recorded	Superior in propofol group
Kongkam 2008	Propofol group recovered better and faster P < 0.001	No significant Difference between groups	No difference between the groups

Primary Outcomes

See Table 2 for the primary outcomes.

Mortality

No deaths were reported during sedation or immediately after. The Schilling 2009 trial reported three non‐sedation related deaths at 30 days after the procedure.

Major complications: hypoxaemia

There was no difference between the two sedation groups in the incidence of hypoxaemia during the procedure in any of the trials. However it was noted that in those trials that did not routinely administer oxygen for the procedure (Kongkam 2008; Vargo 2002) the rate of hypoxaemia was much higher than in the trials that did administer oxygen for the procedure (Riphaus 2005; Schilling 2009). Only two papers (Riphaus 2005; Schilling 2009) reported on the incidence of hypoxaemia in recovery. The Riphaus 2005 trial reported a higher incidence in those patients that had received midazolam and meperidine sedation, whereas the Schilling group reported a higher incidence in those that had received propofol.

Major complications: hypotension 

There was no significant difference in the incidence of serious hypotension between the two sedation groups in any of the trials. In two of the papers (Riphaus 2005; Schilling 2009) intravenous fluid appeared to have been administered during the procedure, whilst it was not clear whether patients in the other two papers (Kongkam 2008; Vargo 2002) received any intravenous fluid supplementation during this period.

Failure to complete procedure due to sedation‐related problem

Only one paper reported this, the Kongkam 2008 study. Five patients had to have the procedure terminated due to sedation‐related complications, three in the propofol group and two in the midazolam and sedation group. Reasons for termination included uncontrolled agitation, apnoea and aspiration.

Secondary Outcomes

See Table 3 for the secondary outcomes. 

Sedation efficacy and recovery: onset of effective sedation

The onset of effective sedation (as judged by time to oesophageal intubation) was reported as significantly shorter for propofol in the Riphaus 2005 and Vargo 2002 trials. The Kongkam 2008 trial found no significant difference between the two types of sedation, and the Schilling 2009 trial did not comment on this outcome.

Sedation efficacy and recovery: patient co‐operation and satisfaction

Patient co‐operation was reported as significantly better in the propofol group in the Riphaus 2005 and Schilling 2009 trials. The other two trials reported no difference in patient co‐operation in either sedation group. Three papers reported no difference in patient satisfaction between the two sedation groups, whilst the Schilling 2009 trial did not report on this.

Sedation efficacy and recovery: recovery

Three papers reported on this (Kongkam 2008; Riphaus 2005; Vargo 2002). All reported significantly improved recovery features in patients that had received propofol.  Recovery was quicker and better in all these papers. The Vargo 2002 trial also reported significantly improved recovery 24 hours after the procedure. The Schilling 2009 trial did not report on recovery in their patients.

Discussion

Our review benefits from a comprehensive search strategy and contributes to a controversial and current international clinical debate on sedation for ERCP procedures. The controversy hinges on the use of propofol (an anaesthetic induction agent) sedation by non‐anaesthetic personnel for endoscopy procedures. Propofol has had major media coverage recently (ASA Media Clip Report 2011).

Propofol has traditionally been regarded as an anaesthetic agent, and authorities such as the FDA recommend that it only be used by trained anaesthetic personnel. However major anaesthetic bodies from the USA and UK acknowledge that this may not be possible in all cases and that non‐anaesthesia personnel administered propofol is acceptable provided there is adequate training and support. Cost and local legal issues may prove a further confounding area in this dilemma.

We have identified a paucity of high‐level evidence regarding sedation techniques for ERCP procedures, which has limited our ability to draw meaningful conclusions regarding this subject (please see Table 1). All of the papers cited in this review used non‐anaesthesia personnel to administer the sedation, both propofol and otherwise. Therefore, from the available data we are unable to make any assessment regarding the safety of anaesthesia delivered sedation for ERCP procedures. This is significant in view of the FDA recommendations that propofol only be used by trained anaesthesia personnel.

Summary of main results

In summary, patients can safely undergo ERCP procedures after being sedated with either propofol or midazolam and meperidine. It seems clear that patients sedated with propofol for these procedures recovered quicker and better than patients who underwent the procedure after having received midazolam and meperidine. It is not clear whether one technique was safer than the other. Adherence to standard protocols during the procedure (relating to aspects such as monitoring and oxygen administration) should be routine practice. These protocols have been well described by major gastrointestinal bodies in the UK and USA (British Society of Gastroenterologists 2003; Position Statement 2009).

Overall completeness and applicability of evidence

Although there is a dearth of randomized trials comparing propofol sedation with conventional conscious sedation, there are large cohort studies which deal with the issue of the safety of various regimes (Berzin 2011; Cote 2010; Williams 2007). One cannot deny the validity of these in weighing up the evidence as to the best method of sedation for ERCP procedures. For instance Berzin 2011 and Cote 2010 have demonstrated that the body mass index (BMI) and American Society of Anesthesiologists (ASA) class are associated with adverse events during endoscopic procedures. Identification of risk factors such as these may not always be possible in smaller randomized controlled trials.

Quality of the evidence

The four trials we reviewed were all randomized and reported comprehensively on one of our primary outcomes (major complications during the procedure). Mortality is a very rare occurrence and was only specifically addressed in one paper (Schilling 2009). Failure of the procedure due to sedation‐related problems was only reported in one trial (Kongkam 2008). Three of the four papers reported on our secondary outcome (Kongkam 2008; Riphaus 2005; Vargo 2002).

Two of the trials (Kongkam 2008; Vargo 2002) included relatively healthy (ASA 1 to 3 status) patients of a younger age group (aged 18 years and above) than the other trials (Riphaus 2005; Schilling 2009), which included patients over 80 years and ASA 3 to 4. Different protocols for sedation were noted between the papers. Only two papers (Kongkam 2008; Riphaus 2005) dealt exclusively with patients undergoing ERCP procedures, the other two trials (Schilling 2009; Vargo 2002) included ERCP procedures with other endoscopic procedures. Due to the clinical heterogeneity we felt that a meta‐analysis should not be undertaken.

Potential biases in the review process

We acknowledge that there are several other drug combinations that can be used in the field of sedation for ERCP procedures, such as balanced propofol sedation (Lee 2011) where a combination of propofol, midazolam and possibly fentanyl are used. We chose only to look at studies comparing propofol with midazolam and opioids because we felt that these were the standard and most widely used current techniques.

There may have been some limitation in assessing outcomes such as 'failure to complete procedure due to inadequate sedation' as the randomized controlled trials we identified had relatively small numbers of patients. The failure rate in this category of trial may differ from that of large observational studies of clinical practice. We chose to categorise hypoxaemia and hypotension as major complications, but there may be differing opinions as to whether this is justified. Many gastroenterologists may only regard these as major complications if they culminated in a clinically significant endpoint.

Agreements and disagreements with other studies or reviews

There were no other randomized trials addressing the topic under review. The large cohort studies referred to in the section Overall completeness and applicability of evidence, confirmed the safety of propofol sedation. They also identified additional patient risk factors that are important to consider in patients undergoing sedation for ERCP procedures.

Authors' conclusions

Implications for practice.

The improved recovery profile of patients undergoing ERCP procedures with propofol sedation versus midazolam and meperidine conscious sedation suggests that propofol may be the preferred choice of sedation for these procedures. However this has generated a debate as to whether only anaesthesia personnel should provide this form of sedation, or whether non‐anaesthesia personnel should also be able to administer propofol sedation to patients undergoing ERCP procedures.

Implications for research.

The lack of randomized studies addressing the issue of sedation for ERCP procedures, in particular those with anaesthesia involvement, is notable. Further appropriate trials with significant anaesthesia participation should focus on safety outcomes such as hypoxaemia and hypotension in patients undergoing sedation for ERCP procedures. This would help clarify the extent of anaesthesia involvement required in the administration of propofol for these procedures.

Appendices

Appendix 1. Search strategy for MEDLINE(R) (Ovid SP)

1 . exp pancreatitis/ or exp cholangiopancreatography, endoscopic retrograde/ or exp cholangiography/ or exp cholecystectomy/ or exp Cholecystectomy, Laparoscopic/ or exp Choledochostomy/ or exp Postcholecystectomy Syndrome/ or Endoscopy, Gastrointestinal/ or pancreatitis.ti,ab. or (pancrea* adj5 (acut* or biliar* or gallstone*)).tw. or (cholangio?pancreatograp* or cholangiograp*or ERCP or pancreatograph* or cholecystect* or choledochostom* or post?cholecystectom* or post? choledochoduodenostom* or post?choledochostom*).tw. or (endoscop* adj5 (retrograde or sphincterot*)).mp. or (laparoscop* adj5 cholecystect*).mp. or (endoscop* adj3 gastrointestinal adj3 surg*).mp.  
 2 . exp conscious‐sedation/ or exp midazolam/ or exp propofol/ or exp hypnotics‐and‐sedatives/ or exp anesthesia‐general/ or exp anesthesia‐inhalation/ or exp anesthetics‐general/ or exp anesthetics‐inhalation/ or exp anesthetics‐intravenous/ or exp anesthesia‐local/ or exp anesthetics‐local/ or exp Laryngeal Masks/ or Postoperative Complications/ or exp Respiratory Insufficiency/ or Pronation/ or Narcotics/ or Droperidol/ or Prone Position/ or an?esth*.ti,ab. or (dormicum or midazolam* or propofol* or pofol or dissopropylphenal or diprivan or disoprivan or disoprofol* or meperidine or pethidine or isonipecain or demerol or dolsin or dolosal or dolin or dolcontral or dolantin or dolargan or meperidine hydrochloride or l?dol or operidine or dehydrobenzperidol or dehirobenzperidol or droleptan or dehidrobenzperidol or inapsine).tw. or sedat*.mp. or (an?esth* adj5 (local or general)).tw. or ((respiratory or ventilator) adj3 (depression or failure)).tw. 

Appendix 2. Search strategy for CENTRAL, The Cochrane Library

#1 MeSH descriptor Pancreatitis explode all trees  
 #2 MeSH descriptor Cholangiopancreatography, Endoscopic Retrograde explode all trees  
 #3 MeSH descriptor Cholangiography explode all trees  
 #4 MeSH descriptor Cholecystectomy explode all trees  
 #5 MeSH descriptor Cholecystectomy, Laparoscopic explode all trees  
 #6 MeSH descriptor Choledochostomy explode all trees  
 #7 MeSH descriptor Postcholecystectomy Syndrome explode all trees  
 #8 MeSH descriptor Endoscopy, Gastrointestinal explode all trees  
 #9 (pancreatitis or cholangio?pancreatograp* or cholangiograp*or ERCP or pancreatograph* or cholecystect* or choledochostom* or post?cholecystectom* or post?choledochoduodenostom* or post?choledochostom*):ti,ab  
 #10 (endoscop* near (retrograde or sphincterot*))  
 #11 (laparoscop* near cholecystect*)  
 #12 ((endoscop* near gastrointestinal) and surg*) or (pancrea* near (acut* or biliar* or gallstone*))  
 #13 (#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12)  
 #14 MeSH descriptor Conscious Sedation explode all trees  
 #15 MeSH descriptor Midazolam explode all trees  
 #16 MeSH descriptor Propofol explode all trees  
 #17 MeSH descriptor Hypnotics and Sedatives explode all trees  
 #18 MeSH descriptor Anesthesia, General explode all trees  
 #19 MeSH descriptor Anesthesia, Inhalation explode all trees  
 #20 MeSH descriptor Anesthetics, General explode all trees  
 #21 MeSH descriptor Anesthetics, Inhalation explode all trees  
 #22 MeSH descriptor Anesthetics, Intravenous explode all trees  
 #23 MeSH descriptor Anesthesia, Local explode all trees  
 #24 MeSH descriptor Anesthetics, Local explode all trees  
 #25 MeSH descriptor Laryngeal Masks explode all trees  
 #26 MeSH descriptor Postoperative Complications, this term only  
 #27 MeSH descriptor Respiratory Insufficiency explode all trees  
 #28 MeSH descriptor Pronation explode all trees  
 #29 MeSH descriptor Narcotics explode all trees  
 #30 MeSH descriptor Droperidol explode all trees  
 #31 MeSH descriptor Prone Position explode all trees  
 #32 (an?esth* or dormicum or midazolam* or propofol* or pofol or dissopropylphenal or diprivan or disoprivan or disoprofol* or meperidine or pethidine or isonipecain or demerol or dolsin or dolosal or dolin or dolcontral or dolantin or dolargan or meperidine hydrochloride or l?dol or operidine or dehydrobenzperidol or dehirobenzperidol or droleptan or dehidrobenzperidol or inapsine or sedat*):ti,ab  
 #33 an?esth* near (local or general)  
 #34 ((respiratory or ventilator) near (depression or failure))  
 #35 (#14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 OR #25 OR #26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32 OR #33 OR #34)  
 #36 (#13 AND #35) 

Appendix 3. Search strategy for EMBASE (Ovid SP)

1 . pancreatitis/ or endoscopic retrograde cholangiopancreatography/ or exp cholangiography/ or cholecystectomy/ or bile duct bypass/ or exp postcholecystectomy syndrome/ or gastrointestinal endoscopy/ or pancreatitis.ti,ab. or (pancrea* adj5 (acut* or biliar* or gallstone*)).ti,ab. or (cholangio?pancreatograp* or cholangiograp*or ERCP or pancreatograph* or cholecystect* or choledochostom* or post?cholecystectom* or post?choledochoduodenostom* or post?choledochostom*).ti,ab. or (endoscop* adj5 (retrograde or sphincterot*)).ti,ab. or (laparoscop* adj5 cholecystect*).ti,ab. or (endoscop* adj3 gastrointestinal adj3 surg*).ti,ab.  
 2 . exp conscious sedation/ or exp midazolam/ or exp propofol/ or exp general anesthesia/ or hypnotic sedative agent/ or exp inhalation anesthesia/ or anesthetic agent/ or inhalation anesthetic agent/ or exp intravenous anesthetic agent/ or local anesthesia/ or local anesthetic agent/ or laryngeal mask/ or postoperative complication/ or respiratory failure/ or body posture/ or narcotic agent/ or droperidol/ or an?esth*.ti,ab. or (dormicum or midazolam* or propofol* or pofol or dissopropylphenal or diprivan or disoprivan or disoprofol* or meperidine or pethidine or isonipecain or demerol or dolsin or dolosal or dolin or dolcontral or dolantin or dolargan or meperidine hydrochloride or l?dol or operidine or dehydrobenzperidol or dehirobenzperidol or droleptan or dehidrobenzperidol or inapsine).ti,ab. or sedat*.ti,ab. or (an?esth* adj5 (local or general)).ti,ab. or ((respiratory or ventilator) adj3 (depression or failure)).ti,ab.  
 3 . 1 and 2  
 4 . (placebo.sh. or controlled study.ab. or random*.ti,ab. or trial*.ti,ab. or ((singl* or doubl* or trebl* or tripl*) adj3 (blind* or mask*)).ti,ab.) not (animals not (humans and animals)).sh.  
 5 . 3 and 4 

Appendix 4. Search strategy for CINAHL (EBSCO host)

S1 (MH "Pancreatitis")  
 S2 (MH "Cholangiopancreatography, Endoscopic Retrograde")  
 S3 (MH "Cholangiography")  
 S4 (MH "Cholecystectomy") or (MM "Cholecystectomy, Laparoscopic")  
 S5 (MH "Choledochostomy")  
 S6 (MH "Endoscopy, Gastrointestinal")  
 S7 AB pancreatitis or TX ( cholangio?pancreatograp* or cholangiograp*or ERCP or pancreatograph* or cholecystect* or choledochostom* or post?cholecystectom* or post?choledochoduodenostom* or post?choledochostom* )  
 S8 AB pancrea* and TX ( acut* or biliar* or gallstone* )  
 S9 TX ( (endoscop* and (retrograde or sphincterot*)) ) or TX ( (laparoscop* and cholecystect*) ) or TX ( (endoscop* and gastrointestinal and surg*) )  
 S10 S1 or S2 or S3 or S4 or S5 or S6 or S7 or S8 or S9  
 S11 (MH "Conscious Sedation")  
 S12 (MH "Midazolam")  
 S13 (MM "Propofol")  
 S14 (MH "Hypnotics and Sedatives")  
 S15 (MH "Anesthesia, General") or (MH "Anesthetics, General")  
 S16 (MH "Anesthesia, Inhalation") or (MH "Anesthetics, Inhalation")  
 S17 (MH "Anesthesia, Intravenous") or (MH "Anesthetics, Intravenous")  
 S18 (MH "Anesthesia, Local") or (MH "Anesthetics, Local")  
 S19 (MH "Laryngeal Masks")  
 S20 (MH "Postoperative Complications")  
 S21 (MM "Respiratory Failure")  
 S22 (MH "Pronation") or (MH "Prone Position")  
 S23 (MH "Narcotics")  
 S24 (MH "Droperidol")  
 S25 AB ( an?esth* or sedat* ) or TX ( dormicum or midazolam* or propofol* or pofol or dissopropylphenal or diprivan or disoprivan or disoprofol* or meperidine or pethidine or isonipecain or demerol or dolsin or dolosal or dolin or dolcontral or dolantin or dolargan or meperidine hydrochloride or l?dol or operidine or dehydrobenzperidol or dehirobenzperidol or droleptan or dehidrobenzperidol or inapsine ) or TX ( ((respiratory or ventilator) and (depression or failure)) )  
 S26 S11 or S12 or S13 or S14 or S16 or S17 or S18 or S19 or S20 or S21 or S22 or S23 or S24 or S25  
 S27 S10 and S26 

Appendix 5. Search strategy for ISI Web of Science

#1 TS=(pancreatitis or cholangio?pancreatograp* or cholangiograp*or ERCP or pancreatograph* or cholecystect* or choledochostom* or post?cholecystectom* or post?choledochoduodenostom* or post?choledochostom*) or TS=(endoscop* SAME (retrograde or sphincterot*)) or TS=(laparoscop* SAME cholecystect*) or TS=(endoscop* SAME gastrointestinal SAME surg*) or TS= (pancrea* adj5 (acut* or biliar* or gallstone*))  
 #2 TI=(an?esth* or sedat*) or TS=(dormicum or midazolam* or propofol* or pofol or dissopropylphenal or diprivan or disoprivan or disoprofol* or meperidine or pethidine or isonipecain or demerol or dolsin or dolosal or dolin or dolcontral or dolantin or dolargan or meperidine hydrochloride or l?dol or operidine or dehydrobenzperidol or dehirobenzperidol or droleptan or dehidrobenzperidol or inapsine) or TS=(an?esth* SAME (local or general)) or TS=((respiratory or ventilator) SAME (depression or failure))  
 #3 #2 AND #1 

Appendix 6. Search strategy for LILACS (BIREME)

("PANCREATITIS" or "CHOLANGIOPANCREATOGRAPHY" or "CHOLANGIOGRAPHY" or "CHOLECYSTECTOMY" or "CHOLEDOCHOSTOMY" or pancreatitis or cholangiopancreatograp$ or cholangiograp$ or ERCP or pancreatograph$ or cholecystect$ or choledochostom$ or postcholecystectom$ or postcholedochoduodenostom$ or postcholedochostom$ or (endoscop$ and (retrograde or sphincterot$)) or (laparoscop$ and cholecystect$) or (endoscop$ and gastrointestinal and surg$) or (pancreas$ and (acut$ or biliar$ or gallstone$))) and ("MIDAZOLAM" or "PROPOFOL" or "ANESTHESIA" or "RESPIRATORYINSUFFIENCY" or "NARCOTICS" or "PRONATION" or "DROPERIDOL" OR “an?esth$” or “sedat$” or dormicum or midazolam$ or propofol$ or pofol or dissopropylphenal or diprivan or disoprivan or disoprofol$ or meperidine or pethidine or isonipecain or demerol or dolsin or dolosal or dolin or dolcontral or dolantin or dolargan or meperidine hydrochloride or l?dol or operidine or dehydrobenzperidol or dehirobenzperidol or droleptan or dehidrobenzperidol or inapsineor or ((respiratory or ventilator) and (depression or failure)))

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Kongkam 2008.

Methods	Gastroenterologist directed administration of propofol versus midazolam/meperidine for ERCP ‐ a block randomized study, conducted over 18 months (December 2003 ‐ May 2005).
Participants	The study included patients of ASA 1‐3 of age range 18‐70 years.
Interventions	67 patients in each group were analysed after randomization to receive propofol by infusion versus midazolam and meperidine.
Outcomes	No mortality in either group. There were 5 instances of failure to complete the procedure, 3 in the propofol group (2 agitation, 1 apnoea) and 2 in the midazolam/meperidine (1 agitation, 1 aspiration). 21(31.3%) patients had an episode of oxygen saturation less than 90% in the midazolam/meperidine group versus 15 (22.4%) patients in the propofol group (P = 0.330). 6 patients (9%) had an episode of systolic blood pressure less than 90 mm Hg in the midazolam/meperidine group versus 6 (9%) of patients in the propofol group (P = 1.00). Recovery time was 34.25 minutes in the midazolam/meperidine group versus 17.24 minutes in the propofol group (P < 0.001).
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"The process of randomization was performed after ERCP scheduling. Patients were alternately assigned with one of the regimens by using block randomization."
Allocation concealment (selection bias)	High risk	Not detailed.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Due to the dramatically different pharmacokinetics of the drugs used for sedation, effectively blinding the treating physician to the drug regimes would not have been possible.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Although unclear from the text what the second reported (bracketed) figures are for mean decrease in SBP and o2 saturation.
Selective reporting (reporting bias)	High risk	The study declared that it was not double‐blinded and was primarily interested in the recovery/safety profile of the two sedation techniques. There were no safety profile data reported in the recovery period (e.g. O2 saturation rates, blood pressure).
Other bias	Unclear risk	Insufficient information. Unclear whether gastroenterologist directing the sedation was the same person performing endoscopy.
OVERALL	High risk	High risk of bias (one or more criteria not met).

Riphaus 2005.

Methods	Gastroenterologist administered propofol versus midazolam/meperidine for ERCP. A randomized by computer‐generated list study over a two year period 2000‐2002.
Participants	The study included 150 ASA3 and 4 patients aged above 80 years.
Interventions	75 patients in each group were analysed after randomization to receive propofol versus midazolam and meperidine.
Outcomes	No mortality in either group and all patients completed the studies. 7 (9%) of patients had an episode of oxygen saturation less than 90% in the midazolam/meperidine group versus 8 (11%) patients in the propofol group (ns). 4 patients (5.3%) had an episode of systolic blood pressure less than 90 mm Hg in the midazolam/meperidine group versus 6 (8%) of patients in the propofol group (ns). Recovery time was 31±8 minutes in the midazolam/meperidine group versus 22±7 minutes in the propofol group (P < 0.01).
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"All patients were randomly assigned by a computer‐generated list by an independent physician" Every patient undergoing ERCP was randomised only once (i.e. re‐examinations were not included)."
Allocation concealment (selection bias)	Low risk	"The investigators and the endoscopy nurses were blinded to the randomization group".
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Due to the dramatically different pharmacokinetics of the drugs used for sedation, effectively blinding the treating physician to the drug regimes would not have been possible.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete data reported.
Selective reporting (reporting bias)	Unclear risk	Repeat doses of midazolam ‐ timings and quantity not clear.
Other bias	Low risk	Detection bias: YES ‐ "The investigators and the endoscopy nurses were blinded to the randomization group"  Attrition bias: YES ‐ No attrition reported in either group.
OVERALL	Unclear risk	Unclear risk of bias (one or more criteria partly met).

Schilling 2009.

Methods	Nurse administered (physician directed) propofol sedation versus midazolam/meperidine in high risk octogenarians for ERCP, EUS or double‐balloon endoscopy (DBE). A computer‐generated list randomized trial conducted over 15 months, from March 2006 ‐ June 2007.
Participants	151 patients over 80 years, ASA grade 3‐4.
Interventions	Randomized into 76 patients receiving propofol versus 75 receiving midazolam/meperidine.
Outcomes	No immediate mortality however follow up to 30 days revealed mortality in 1 propofol patient and 2 midazolam/meperidine patients from non‐sedation related events. There were 3 instances of failure to complete the procedure in the propofol arm and 2 in the midazolam/meperidine group. These were not related to sedation problems, but to inability to intubate the duodenum with the endoscope in all cases. 7 (9.3%) of patients had an episode of oxygen saturation less than 90% in the midazolam/meperidine group versus 9 (11.8%) patients in the propofol group (ns). 2 patients (2.6%) had an episode of systolic blood pressure less than 90 mm Hg in the midazolam/meperidine group versus 4 (5.2%) of patients in the propofol group (ns).
Notes	The study outcomes focused on safety aspects (e.g. cardiopulmonary events) and did not consider recovery outcomes in any detail. Neither the supervising nurse nor the trained physician initiating/supervising the sedation were involved in the endoscopic procedure. The physician remained in the room for the procedure and was able to manage any complications.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"All patients were randomly assigned to one of the three different interventions by a computer‐generated list by an independent physician". Presume this was computer‐generated list of random numbers. Also ‐ report states patients were randomly assigned to one of the three interventions? Presume this was meant to read one of the two sedation techniques.
Allocation concealment (selection bias)	Unclear risk	"All patients were randomly assigned to one of the three different interventions by a computer‐generated list by an independent physician". No further details supplied.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Due to the dramatically different pharmacokinetics of the drugs used for sedation, effectively blinding the treating physician to the drug regimes would not have been possible.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete data.
Selective reporting (reporting bias)	Unclear risk	There are instances where it seems data are not reported completely (e.g. "Patients sedated with propofol showed a significantly lower oxygen saturation rate during the recovery period (8% vs. 28%, P≤0.01)" No other data given. Also, the authors declare need for additional intervention in case of adverse events as a second outcome measure. This not reported completely: "All necessary intervention were managed by trained nurses" i.e. number of interventions/adverse events not detailed.
Other bias	Low risk	Aside from those mentioned in "free of selective reporting".
OVERALL	Unclear risk	Unclear risk of bias (one or more criteria partly met).

Vargo 2002.

Methods	Gastroenterologist administered propofol versus midazolam/meperidine for ERCP or EUS. A block‐randomized study (period not stated).
Participants	The paper studied patients over 18 years of age with an ASA of 1 to 3, and included EUS procedures also.
Interventions	38 patients received propofol and 37 midazolam and meperidine.
Outcomes	No mortality in either group and all patients completed the studies. 14 patients had an episode of oxygen saturation less than 90% in the midazolam/meperidine group versus 21 patients in the propofol group (P = 0.13). 6 patients had a decrease of systolic blood pressure greater than 25% from baseline in the midazolam/meperidine group versus 7 the propofol group (P = 0.96). Recovery time was 70.5±7.1 minutes in the midazolam/meperidine group versus 18.6±6.5 minutes in the propofol group (P<0.001).
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"During the study period, any patient undergoing ERCP or EUS was randomized only once (i.e. re‐examinations were not included). A 1:1 randomization was used, with the randomization blocks supplied by the biostatistics department of the Cleveland Clinic Foundation".
Allocation concealment (selection bias)	Unclear risk	"The subjects, the research nurse who conducted all pre‐and postprocedural patients assessment and the recovery room personnel were blinded to the type of medication used in the sedation. The randomization was conducted in the procedure room, hence the endoscopist and procedure room personnel were not blinded to the patient sedation regimen".
Blinding (performance bias and detection bias) All outcomes	High risk	Blinding was not attempted for procedure.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No incomplete data reported.
Selective reporting (reporting bias)	Unclear risk	Only rated sedation by "pt.co‐operation on a VAS scale".
Other bias	Low risk	Detection bias: Yes ‐ "The same research nurse who was blinded to the type of sedation used conducted all pre‐and postprocedural evaluations." Attrition bias: Yes ‐ 5/165 (3%) patients excluded due to failed duodenal intubation. 2/77 (3%) in the propofol group, 3/78 (4%) patients were excluded from the meperidine/midazolam group.
OVERALL	High risk	High risk of bias (one or more criteria not met).

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Agostoni 2007	RCT but only endoscopic ultrasound (EUS) procedures were examined
Chen 2005	RCT but diazepam and meperidine were given intramuscularly
Koshy 2000	Non‐randomized and did not include ERCP procedures

Differences between protocol and review

The scope of the review was necessarily broadened from the protocol stage due to the lack of appropriate papers identified. The original intention was to review sedative techniques for ERCP using the following comparisons.

1. Conscious sedation using midazolam plus opioid performed by gastroenterologist versus deep sedation using propofol and performed by anaesthesia personnel. 
 2. Conscious sedation using midazolam plus opioid performed by gastroenterologist versus general anaesthesia performed by anaesthesia personnel. 
 3. Deep sedation using propofol versus general anaesthesia, both performed by anaesthesia personnel.

The search was widened to consider all studies examining ERCP and related procedures (for example endoscopic ultrasound (EUS)) performed under conscious sedation using midazolam plus opioid versus deep sedation using propofol, regardless of which professional administered the sedative. We were only able to identify studies involving non‐anaesthesia personnel administered sedation, and so the review focused on these.

Contributions of authors

Conceiving the review: Davinder Garewal (DG), Steve Powell (SP), Jonas Nordmeyer (JN), Stephen J Milan (SM) 
 Co‐ordinating the review: SP 
 Undertaking manual searches: SP 
 Screening search results: JN, DG, Pallavi Waikar (PW) 
 Organizing retrieval of papers: SP 
 Screening retrieved papers against inclusion criteria: JN, DG, SM 
 Appraising quality of papers: JN, DG, SM, PW 
 Abstracting data from papers: JN, DG, SM, PW 
 Writing to authors of papers for additional information: JN, DG, SM 
 Providing additional data about papers: JN, DG, SM 
 Obtaining and screening data on unpublished studies: JN, DG, SM 
 Data management for the review: JN, DG, SP, SM 
 Entering data into Review Manager (RevMan 5.1): SP 
 RevMan statistical data: SP, SM 
 Other statistical analysis not using RevMan: SP, SM 
 Double entry of data: (data entered by person one: SP; data entered by person two: DG) 
 Interpretation of data: JN, DG, SP, SM 
 Statistical inferences: SM 
 Writing the review: DG, SP, PW 
 Guarantor for the review (one author): DG 
 Person responsible for reading and checking review before submission: DG

Sources of support

Internal sources

• St George's Healthcare NHS Trust, London, UK.

External sources

• No sources of support supplied

Declarations of interest

None known

New