ORIGINAL ARTICLE: Clinical Endoscopy150The risk of post-ERCP pancreatitis and the protective effect of rectalindomethacin in cases of attempted but unsuccessful prophylacticpancreatic stent placementGASTNeel S. Choksi, MD,1 Evan L. Fogel, MD, MSc,2 Gregory A. Cote, MD, MS,2 Joseph Romagnuolo, MD, MS,3

Grace H. Elta, MD,1 James M. Scheiman, MD,1 Amitabh Chak, MD,4 Patrick Mosler, MD, PhD,5,6

Peter D. R. Higgins, MD, PhD, MSc,1 Sheryl J. Korsnes, MA,1 Suzette E. Schmidt, BSN, CCRP,2

Stuart Sherman, MD,2 Glen A. Lehman, MD,2 B. Joseph Elmunzer, MD,1 on behalf of theUnited States Cooperative for Outcomes Research in Endoscopy

Ann Arbor, Michigan, USABackground: It is believed, based on limited observational data, that an unsuccessful attempt to place a prophy-lactic pancreatic stent substantially increases the risk of post-ERCP pancreatitis (PEP).

Objective: To better understand the risk of PEP in patients with failed pancreatic stent placement (FPS) andthe impact of rectal indomethacin on this risk.

Design: Secondary analysis of randomized, controlled trial data.

Setting: University of Michigan and Indiana University.

Patients: A total of 577 clinical trial participants at elevated risk for PEP.

Interventions: Pancreatic stent placement.

Main Outcome Measurements: Within the placebo group, we compared PEP rates in patients with FPS,patients who underwent successful stent placement, and in those without a stent attempt. We also performeda regression analysis evaluating the association between FPS and PEP. To define the protective effect of indometh-acin, we repeated these analyses in the indomethacin group and in the full study cohort.

Results: The incidence of PEP among patients in the placebo group who experienced FPS was 34.7%, signifi-cantly exceeding rates in patients who underwent successful stent placement (16.4%) and in those without a stentattempt (12.1%). After we adjusted for known PEP risk factors, FPS was found to be independently associated withPEP. Among the indomethacin group and in the full cohort, FPS was not associated with a higher risk of PEP.

Limitations: Low event rate, FPS not prospectively captured.

Conclusion: FPS appears to confer an increased risk of PEP, which is attenuated by rectal indomethacin admin-istration. These findings highlight the importance of adequate training and proficiency before endoscopistsattempt pancreatic stent placement and the routine use of rectal indomethacin in high-risk ERCP cases. (Gastro-intest Endosc 2015;81:150-5.)(footnotes appear on last page of article)Prophylactic pancreatic stent placement (PSP) reducesthe risk of post-ERCP pancreatitis (PEP) by preventingpancreatic ductal hypertension and the consequentUse your mobile device to scan thisQR code and watch the author in-terview. Download a free QR codescanner by searching QR Scannerin your mobile devices app store.

ROINTESTINAL ENDOSCOPY Volume 81, No. 1 : 2015activation of trypsin that develop as a result oftransient procedure-induced obstruction of the pancre-atic orifice. Randomized, controlled trial (RCT) datademonstrate that PSP reduces the incidence of PEP byapproximately 70% and essentially eliminates the riskof severe pancreatitis.1,2 As such, PSP is widely regardedas an effective means of preventing PEP in high-riskcases, is commonly used in the United States,3 and isrecommended by the European Society of Gastrointes-tinal Endoscopy.4www.giejournal.org

Delta:1_given nameDelta:1_surnameDelta:1_given nameDelta:1_surnameDelta:1_given nameDelta:1_surnameDelta:1_given nameDelta:1_surnameDelta:1_given nameDelta:1_surnameDelta:1_given nameDelta:1_surnameDelta:1_given nameDelta:1_surnamehttp://www.giejournal.orgRicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

Choksi et al Rectal indomethacinAlthough PSP clearly reduces risk, it is associatedwith several important disadvantages. It can be technicallychallenging, time consuming, costly, and inconvenient forpatients.5-8 Perhaps the most important clinical disadvan-tage of PSP is the belief that attempting to place a pancre-atic stent with subsequent failure substantially increasesthe risk of PEP by inducing additional injury to the pancre-atic orifice and/or pancreatic duct without providingductal decompression.9 Because data evaluating the suc-cess rates and effectiveness of PSP in real-world practice,as opposed to expert centers, are not available, this phe-nomenon of attempted but failed pancreatic stent place-ment (FPS) may represent an underappreciated problemin ERCP practice.

Our understanding of FPS and its implications is basedentirely on anecdotal evidence and a single observationalstudy comprising only 3 outcome events wherein a singleexpert endoscopist placed all the pancreatic stents.9

Our group recently published a large-scale, multicenterRCT showing that rectal indomethacin reduced the riskof PEP in high-risk patients, most of whom (O80%) hadundergone PSP.10 A subset of participants in this RCT,randomly assigned to receive rectal indomethacin or pla-cebo, experienced FPS.

By using these multicenter RCT data, we aimed tobetter understand the phenomenon of FPS by (1) deter-mining the incidence of PEP in high-risk ERCP casescomplicated by FPS, (2) assessing the prophylactic effectof indomethacin in patients with FPS, and (3) performingan exploratory evaluation of clinical and procedural factorsassociated with FPS.METHODS

This study was a secondary analysis of prospectivelycollected data obtained during a previously reported multi-center RCT of rectal indomethacin for preventing PEP.10

Briefly, after approval by the institutional review commit-tees of the involved centers, patients at elevated risk forPEP were randomly assigned to receive 100 mg of rectalindomethacin or placebo suppositories immediately afterERCP. All other procedural and clinical decisions were de-ferred to the treating physicians. Placement of a pancreaticduct stent was not mandated; however, O80% of patientsreceived a stent because the study enrolled patients atelevated risk for PEP. The primary endpoint of the studywas the development of PEP, defined as new onset ofpain in the upper abdomen, an elevation in pancreaticenzyme levels of R3 times the upper limit of the normalrange 24 hours after the procedure, and hospitalizationfor at least 2 nights.11 Patient demographics, risk factorsfor PEP, ERCP procedure details, and follow-up data wererecorded on standardized data collection forms by a mem-ber of the study team who was blinded to study groupassignment. All data were subsequently entered into awww.giejournal.orgWeb-based database and managed by an independentdata management service. The final database containedinformation on 602 study participants.

We included all RCT participants enrolled at 2 studycenters (Indiana University, University of Michigan),comprising 96% of total enrollment. To generate a studysample of FPS participants, we identified the subset of par-ticipants in the RCT who did not have a pancreatic stentplaced at the time of ERCP. Two investigators (N.C.,B.J.E.) who were unaware of study group assignment andoutcome then independently reviewed the procedure re-ports for each of these participants to adjudicate whetheran attempt at PSP was made by using a priorideterminedcriteria. A patient was deemed to have experienced FPSif 1 or more of the following criteria was explicitly presentin the procedure report: (1) PSP was unsuccessful despiteattempt; (2) pancreatic duct cannulation could not beachieved despite attempt; (3) there was failed cannulationof both the bile duct and pancreatic duct; (4) pancreaticduct cannulation was achieved, but the wire was lostbefore stent placement; and (5) pancreatic duct cannula-tion was achieved, but the ERCP was aborted beforecompletion of the procedure. In the context of a high-risk case, criteria 2 and 3 assumed that cannulation is thefirst and necessary element of successful PSP, and theinability to achieve cannulation represents a failure of allendoscopic objectives, including PSP. Criteria 4 and5 assumed that, during a high-risk case, placement of awire into the pancreas indicates that PSP is already under-way, and a prophylactic stent would have been placed hadaccess not been lost. In cases of pancreas divisum, the dor-sal pancreatic duct was considered the duct of interest. Pa-tients who did not have a pancreatic stent placed and didnot meet any of the 5 FPS criteria made up the no PSPattempt cohort.

To estimate the risk of PEP associated with FPS indepen-dent of the effect of indomethacin, we restricted the pri-mary analysis to participants in the placebo group ofthe RCT. Within this group, we compared the risk of PEPin patients with FPS, in patients who underwent successfulstent placement, and in patients who did not undergo anattempt at PSP. In addition, we performed an exploratorymultivariable logistic regression analysis evaluating theassociation between PEP (dependent variable) and FPS(main predictor variable) while adjusting for PEP risk intwo separate ways: (1) by including patients PEP riskscores as a covariate in one model and (2) by includingscores for patients propensities to receive a pancreaticstent in another model. The PEP risk score, which hasbeen described previously, was calculated for each partici-pant by assigning 1 point for each major RCT inclusion cri-terion and 0.5 point for each minor RCT inclusion criterion(Table 1).10 The propensity score predicts a participantsconditional probability of receiving a pancreatic stentbased on all other observed characteristics of that partici-pant, serving as a surrogate marker for PEP risk.Volume 81, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 151

http://www.giejournal.orgRicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

RicardoResaltado

TABLE 1. Major and minor study inclusion criteriaused to calculate post-ERCP pancreatitis risk score

Major criteria Minor criteria

Clinical suspicion ofsphincter of Oddidysfunction

Age!50 y and female

History of post-ERCPpancreatitis

History of recurrentpancreatitis

Pancreatic sphincterotomy R3 pancreatic injections,with at least 1 injection

to the tail

Precut (access)sphincterotomy

Pancreatic acinarization

O8 cannulation attempts Pancreatic brush cytology

Pneumatic dilation ofintact biliary sphincter

Ampullectomy

Rectal indomethacin Choksi et alTo assess the protective effect of indomethacin, theseanalyses were then repeated in the group of participantsassigned to receive indomethacin and in the full RCTcohort (placebo and indomethacin groups). To evaluatethe modifying effect of indomethacin on FPS, in theregression analysis of the full cohort we included an inter-action variable containing FPS and study groupassignment.

An exploratory analysis of clinical and procedural factorsassociated with FPS was conducted by performing sequen-tial univariate logistic regression analyses, with FPS as thedependent variable and the following predictor variables:age, sex, study center, suspicion of sphincter of Oddidysfunction, prior PEP, recurrent pancreatitis, pancreaticsphincterotomy, precut sphincterotomy, difficult cannula-tion, ampullectomy, pancreatic brush cytology, manometri-cally proven sphincter dysfunction, biliary sphincterotomy,bile duct stent placement, choledocholithiasis, pancreati-cobiliary malignancy, trainee involvement, and pancreaticacinarization. Any variable positively associated with FPS(P value ! .05) in the univariate analysis was includedin a multivariable logistic regression model with FPS asthe dependent variable. Seven predictor variables wereincluded in the final model: age, sex, study site, recurrentpancreatitis, difficult cannulation, choledocholithiasis, andtrainee involvement.

All statistical analyses were performed by using theSTATA 12 statistical package (StataCorp LP, College Sta-tion, Tex). The difference in the proportion of PEP be-tween groups was analyzed by using the Fisher exacttest, with a 2-tailed P value of! .05 indicating significance.Propensity score calculation was conducted by using thepsmatch2 command.152 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 1 : 2015RESULTS

A total of 577 RCT participants were included in thisanalysis. Forty-two participants experienced FPS (7.3%),which occurred in 10 of the 403 participants enrolled atIndiana University (2.4%) and 32 of the 164 participantsenrolled at the University of Michigan (19.5%). Seventy-seven of the 577 participants experienced PEP (13.3%);the PEP rate was 17.3% (51 of 294) in the placebo groupand 9.2% (26 of 283) in the indomethacin group.The risk of PEP associated with FPS in theplacebo group

A total of 294 RCT participants were randomizedto receive placebo. Of these, 23 (7.8%) experienced FPS.Eight of the 23 FPS cases (34.7%) resulted in PEPcompared with 39 of the 238 successful PSP cases(16.4%; P Z .04) and 4 of the 33 cases in which therewas no attempt at PSP (12.1%; P Z .02) (Fig. 1). Amongparticipants in the placebo group, a logistic regressionmodel adjusting for PEP risk showed that FPS was indepen-dently associated with PEP (odds ratio [OR] 3.24; 95% con-fidence interval [CI], 1.26-8.31 adjusted by PEP risk score;OR 3.15; 95% CI, 1.14-8.77 adjusted by pancreatic stentpropensity score).The protective effect of rectal indomethacin inpatients with FPS

Of the 283 patients who received indomethacin,19 (6.7%) experienced FPS. Only 1 (5.3%) of these 19 pa-tients with FPS developed PEP, compared with 23 of the239 patients who underwent successful stent placement(9.6%; P Z 1.0) and 6 of the 58 patients who did notundergo a PSP attempt (10.3%; P Z 1.0) (Fig. 1).

In a comparison of FPS within the entire cohort (indo-methacin and placebo groups), 8 of the 23 patients withFPS in the placebo group developed PEP (34.7%), whereasonly 1 of the 19 patients with FPS in the indomethacingroup developed PEP (5.3%) (P Z .027). Logistic regres-sion analysis of the full RCT cohort demonstrated thatFPS is independently associated with PEP only when aninteraction variable making up the FPS and study groupassignment is included in the model, further supporting in-domethacins protective effect on PEP in patients whoexperienced FPS.Factors associated with FPSThe exploratory risk factor analysis revealed that

increasing age (OR 1.03; 95% CI, 1.01-1.06), study site (Uni-versity of Michigan) (OR 6.87; 95% CI, 3.03-15.59), a historyof recurrent pancreatitis (OR 2.19; 95% CI, 1.06-4.54),and difficult cannulation (OR 3.79; 95% CI, 1.84-7.79) areindependently predictive of FPS.www.giejournal.org

http://www.giejournal.org

Figure 1. Risk of PEP among patients with FPS compared with no attempt and successful PSP in the placebo group (left panel ) and indomethacingroup (right panel ). PEP, post-ERCP pancreatitis; PSP, pancreatic stent placement; FPS, failed pancreatic stent placement.

Choksi et al Rectal indomethacinDISCUSSION

This secondary analysis of RCT data suggests that pa-tients who experienced attempted but failed PSP wereat significantly increased risk for PEP compared with pa-tients who underwent successful stent placement andthose in whom stent placement was not attempted. Pro-phylactic administration of rectal indomethacin appearsto significantly attenuate this risk. Exploratory risk factoranalysis suggests that a history of recurrent pancreatitisand difficult cannulation are clinical factors associatedwith FPS.

In this RCT, the PEP risk in patients who experiencedFPS but did not receive indomethacin was approximately35%. Although this risk is not as high as previously reportedin a small, non-controlled series,9 it is certainly higher thanthe rate of PEP observed among patients who underwentsuccessful stent placement in our RCT and in numerousother RCTs evaluating the protective effect of PSP,wherein the rate of PEP in patients who underwent success-ful stent placement ranged from 2% to 15%.6,7,12-15

The risk of PEP associated with FPS we observedalso appears to exceed the risk seen in the control armsof previously published RCTs involving PSP, wherein stentplacement was not attempted and pancreatitis occurred in13% to 29% of cases.6,7,12-15 In our RCT, the rate of PEP wasconsiderably (3-fold) higher in the FPS group than in theno attempt group, although this observation is confoundedby the fact that the decision to place a pancreatic stent inthis RCT was deferred to the discretion of the endoscopist.Patients in whom stent placement was not attempted werelikely to be at lower risk for PEP because they elicited lessconcern on the part of the endoscopist.16 Nevertheless, alogistic regression model adjusting for this confoundingwww.giejournal.orgalso suggested that FPS is an independent predictor ofPEP.

PSP can be a challenging intervention that is notalways technically feasible. Indeed, prior studies have re-ported FPS rates in the 5% to 10% range.6,7,15 Thesestudies were performed at referral medical centers by en-doscopists with expertise and experience in pancreaticendoscopy. Although broader data from academic andcommunity practices are not available, FPS is likely to bean underappreciated, and perhaps growing, problem inreal-world ERCP practice. In this context, the results ofthis study underscore the importance of (1) adequatetraining, experience, and procedure volume for those per-forming high-risk ERCP requiring PSP; (2) a responsibleapproach to referring high-risk cases that mandate PSPto tertiary-care centers; and (3) the routine use of rectalindomethacin in all high-risk cases. Some experts recom-mend rectal indomethacin for all patients undergoingERCP based on highly favorable cost and risk benefitratios as well as existing RCT data suggesting its efficacyacross the entire spectrum of PEP risk.17,18 However, giventhe potential for FPS and its consequences, the findingsof this study suggest that rectal nonsteroidal anti-inflammatory drug administration is particularly importantin high-risk cases, such as those that would have beenincluded in our RCT (Table 1).

A major question surrounding PSP is the acceptableamount of time and effort that should be spent on theinsertion process in cases of difficult pancreatic access. Pre-sumably, there exists a point of diminishing return at whichthe risk of additional attempts outweighs the benefit ofstent placement, especially if insertion eventually provesunsuccessful. Because future clinical studies are unlikelyto answer this question in a methodologically rigorousVolume 81, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 153

http://www.giejournal.org

Rectal indomethacin Choksi et alfashion, endoscopists should be aware of this importantclinical balance and use their best judgment regardingacceptable stent placement efforts. Based on the resultsof this study, it may be reasonable to favor a strategy oflimited PSP efforts (and rectal indomethacin) when pancre-atic access proves challenging, especially in patients withprior recurrent pancreatitis or when a difficult biliary can-nulation has occurred.

In this RCT, we observed a significant difference in FPSrates between study sites. Because FPS was not defined apriori, the approach to PSP was not standardized, anddata on several factors that are potentially important inthe success or failure of stent placement are not available,the underlying causes of this difference cannot be definedconcretely. However, several potential explanations exist.First, systematic differences between centers in definingFPS or in the time and effort dedicated to PSP before thedecision to abort may have played a role. Additionally,technical differences between centers, including thechoice of endoscopic instruments, may have been impor-tant. In particular, some experts advocate the routine useof .018-inch guidewires for FPS to more safely and effec-tively negotiate potentially tortuous pancreatic ducts;others advocate catheter systems that allow endoscopistcontrol of the guidewire. Additional research is necessaryto elucidate whether these or other factors are importantin minimizing the FPS rate and in attenuating the risk ofPEP when FPS occurs.

Despite the findings of this study, PSP remains the stan-dard of care for preventing PEP in high-risk patients, basedon robust clinical trial data and existing guidelines. Howev-er, considering the implications of FPS as well as the otherpotential disadvantages of PSP, future research focused ondefining prophylactic strategies that allow more selectivePSP use would be of significant clinical value. For example,a previously published post hoc analysis of this RCT16 and anetwork meta-analysis19 have suggested that rectal indo-methacin mono-prevention could be equivalent to, if notsuperior to, the combination of indomethacin and PSPin high-risk cases. On this basis, a large-scale, multicenterRCT comparing rectal indomethacin alone versus combina-tion therapy is in the final planning phases and shouldbegin enrolling participants in early 2015. Furthermore,post-traumatic pancreatic orifice edema and stenosismay be mitigated by other pharmacologic agents suchas nitroglycerin20 and topically applied epinephrine.21

Additional studies are necessary to determine whetherthese agents can serve as a surrogate stent, providingductal decompression without the risks, costs, and incon-venience of PSP.

The results of this study should be considered in thecontext of 2 important limitations. First, our RCT datacollection form did not prospectively capture FPS. Thisendpoint was adjudicated post hoc by manual review ofprocedure reports by using the aforementioned criteriathat are based on clinical assumptions aiming to determine154 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 1 : 2015endoscopist intent. Although we believe these assump-tions are reasonable, FPS may have been under-captured(when stent attempt was not mentioned) or misclassified(if the description of the procedure was misinterpretedby the reviewer), potentially biasing study results. Second,the event rate in this study was small, reducing the preci-sion in our estimates of effect and leading to probableregression model overfitting and possible missed associa-tions because of low statistical power. Nevertheless, thisis the largest and most methodologically rigorous analysisof the phenomenon of FPS, and it provides important, clin-ically relevant data in this area. To shed additional light onPSP and the phenomenon of FPS, researchers in futureRCTs involving ERCP should prospectively collect data onthe endoscopic instruments (catheters, wires, and stents)used for PSP; underlying pancreatic duct anatomy; theduration, technical difficulty, effort, and personnel involvedin the attempt; and the reasons for failure.

In summary, attempted but failed pancreatic stent place-ment appears to confer an increased risk of PEP, whichis attenuated by prophylactic rectal indomethacin adminis-tration. These findings highlight the importance ofadequate training and proficiency for the endoscopistbefore PSP is attempted and routine use of rectal indo-methacin in high-risk ERCP cases. Given the risks and costsof PSP, identifying prophylactic strategies that minimizethe need for pancreatic stenting may be of substantial clin-ical and economic value.REFERENCES

1. Choudhary A, Bechtold ML, Arif M, et al. Pancreatic stents for prophy-laxis against post-ERCP pancreatitis: a meta-analysis and systematicreview. Gastrointest Endosc 2011;73:275-82.

2. Mazaki T, Masuda H, Takayama T. Prophylactic pancreatic stent place-ment and post-ERCP pancreatitis: a systematic review and meta-anal-ysis. Endoscopy 2010;42:842-53.

3. Brackbill S, Young S, Schoenfeld P, et al. A survey of physician practiceson prophylactic pancreatic stents. Gastrointest Endosc 2006;64:45-52.

4. Dumonceau JM, Andriulli A, Deviere J, et al. European Society ofGastrointestinal Endoscopy (ESGE) Guideline: prophylaxis of post-ERCP pancreatitis. Endoscopy 2010;42:503-15.

5. Das A, Singh P, Sivak MV Jr, et al. Pancreatic-stent placement for pre-vention of post-ERCP pancreatitis: a cost-effectiveness analysis. Gastro-intest Endosc 2007;65:960-8.

6. Tarnasky PR, Palesch YY, Cunningham JT, et al. Pancreatic stentingprevents pancreatitis after biliary sphincterotomy in patients withsphincter of Oddi dysfunction. Gastroenterology 1998;115:1518-24.

7. Fazel A, Quadri A, Catalano MF, et al. Does a pancreatic duct stentprevent post-ERCP pancreatitis? A prospective randomized study.Gastrointest Endosc 2003;57:291-4.

8. Zolotarevsky E, Fehmi SM, Anderson MA, et al. Prophylactic 5-Frpancreatic duct stents are superior to 3-Fr stents: a randomizedcontrolled trial. Endoscopy 2011;43:325-30.

9. Freeman ML, Overby C, Qi D. Pancreatic stent insertion: consequencesof failure and results of a modified technique to maximize success.Gastrointest Endosc 2004;59:8-14.

10. Elmunzer BJ, Scheiman JM, Lehman GA, et al. A randomized trial ofrectal indomethacin to prevent post-ERCP pancreatitis. N Engl J Med2012;366:1414-22.www.giejournal.org

http://refhub.elsevier.com/S0016-5107(14)01982-8/sref1http://refhub.elsevier.com/S0016-5107(14)01982-8/sref1http://refhub.elsevier.com/S0016-5107(14)01982-8/sref1http://refhub.elsevier.com/S0016-5107(14)01982-8/sref2http://refhub.elsevier.com/S0016-5107(14)01982-8/sref2http://refhub.elsevier.com/S0016-5107(14)01982-8/sref2http://refhub.elsevier.com/S0016-5107(14)01982-8/sref3http://refhub.elsevier.com/S0016-5107(14)01982-8/sref3http://refhub.elsevier.com/S0016-5107(14)01982-8/sref4http://refhub.elsevier.com/S0016-5107(14)01982-8/sref4http://refhub.elsevier.com/S0016-5107(14)01982-8/sref4http://refhub.elsevier.com/S0016-5107(14)01982-8/sref5http://refhub.elsevier.com/S0016-5107(14)01982-8/sref5http://refhub.elsevier.com/S0016-5107(14)01982-8/sref5http://refhub.elsevier.com/S0016-5107(14)01982-8/sref6http://refhub.elsevier.com/S0016-5107(14)01982-8/sref6http://refhub.elsevier.com/S0016-5107(14)01982-8/sref6http://refhub.elsevier.com/S0016-5107(14)01982-8/sref7http://refhub.elsevier.com/S0016-5107(14)01982-8/sref7http://refhub.elsevier.com/S0016-5107(14)01982-8/sref7http://refhub.elsevier.com/S0016-5107(14)01982-8/sref8http://refhub.elsevier.com/S0016-5107(14)01982-8/sref8http://refhub.elsevier.com/S0016-5107(14)01982-8/sref8http://refhub.elsevier.com/S0016-5107(14)01982-8/sref9http://refhub.elsevier.com/S0016-5107(14)01982-8/sref9http://refhub.elsevier.com/S0016-5107(14)01982-8/sref9http://refhub.elsevier.com/S0016-5107(14)01982-8/sref10http://refhub.elsevier.com/S0016-5107(14)01982-8/sref10http://refhub.elsevier.com/S0016-5107(14)01982-8/sref10http://www.giejournal.org

Choksi et al Rectal indomethacin11. Cotton PB, Lehman G, Vennes J, et al. Endoscopic sphincterotomycomplications and their management: an attempt at consensus.Gastrointest Endosc 1991;37:383-93.

12. Sofuni A, Maguchi H, Itoi T, et al. Prophylaxis of post-endoscopicretrograde cholangiopancreatography pancreatitis by an endoscopicpancreatic spontaneous dislodgement stent. Clin GastroenterolHepatol 2007;5:1339-46.

13. Tsuchiya T, Itoi T, Sofuni A, et al. Temporary pancreatic stent to preventpost endoscopic retrograde cholangiopancreatography pancreatitis: apreliminary, single-center, randomized controlled trial. J HepatobiliaryPancreat Surg 2007;14:302-7.

14. Lee TH, Moon JH, Choi HJ, et al. Prophylactic temporary 3F pancreaticduct stent to prevent post-ERCP pancreatitis in patients with a difficultbiliary cannulation: a multicenter, prospective, randomized study.Gastrointest Endosc 2012;76:578-85.

15. Smithline A, Silverman W, Rogers D, et al. Effect of prophylactic mainpancreatic duct stenting on the incidence of biliary endoscopicsphincterotomy-induced pancreatitis in high-risk patients. GastrointestEndosc 1993;39:652-7.

16. Elmunzer BJ, Higgins PD, Saini SD, et al. Does rectal indomethacineliminate the need for prophylactic pancreatic stent placement in pa-tients undergoing high-risk ERCP? Post hoc efficacy and cost-benefitanalyses using prospective clinical trial data. Am J Gastroenterol2013;108:410-5.

17. Sethi S, Sethi N, Wadhwa V, et al. A meta-analysis on the role of rectaldiclofenac and indomethacin in the prevention of post-endoscopicretrograde cholangiopancreatography pancreatitis. Pancreas 2014;43:190-7.

18. Sun HL, Han B, Zhai HP, et al. Rectal NSAIDs for the prevention ofpost-ERCP pancreatitis: a meta-analysis of randomized controlled trials.Surgeon 2014;12:141-7.

19. Akbar A, Abu Dayyeh BK, Baron TH, et al. Rectal nonsteroidalanti-inflammatory drugs are superior to pancreatic duct stents inpreventing pancreatitis after endoscopic retrograde cholangiopan-www.giejournal.orgcreatography: a network meta-analysis. Clin Gastroenterol Hepatol2013;11:778-83.

20. Bai Y, Xu C, Yang X, et al. Glyceryl trinitrate for prevention of pancre-atitis after endoscopic retrograde cholangiopancreatography: ameta-analysis of randomized, double-blind, placebo-controlled trials.Endoscopy 2009;41:690-5.

21. Xu LH, Qian JB, Gu LG, et al. Prevention of post-endoscopic retrogradecholangiopancreatography pancreatitis by epinephrine sprayed onthe papilla. J Gastroenterol Hepatol 2011;26:1139-44.

Abbreviations: FPS, failed pancreatic stent placement; PEP, post-ERCPpancreatitis; PSP, pancreatic stent placement; RCT, randomized,controlled trial.

DISCLOSURE: All authors disclosed no financial relationships relevantto this article.

See CME section; p. 214.Copyright 2015 by the American Society for Gastrointestinal Endoscopy0016-5107/$36.00http://dx.doi.org/10.1016/j.gie.2014.07.033

Received April 3, 2014. Accepted July 11, 2014.

Current affiliations: Division of Gastroenterology, University of MichiganMedical Center, Ann Arbor, Michigan (1), Division of Gastroenterology,Indiana University Medical Center, Indianapolis, Indiana (2), Division ofGastroenterology, Medical University of South Carolina, Charleston,South Carolina (3), Division of Gastroenterology, University HospitalsCase Medical Center, Cleveland, Ohio (4), Division of Digestive Diseasesand Nutrition, University of Kentucky Medical Center, Lexington,Kentucky, USA (5), Division of Gastroenterology and Hepatology,Kantonsspital Graubuenden, Chur, Switzerland (6).

Reprint requests: Neel Choksi, MD, University of Michigan Medical Center,3912 Taubman Center, Ann Arbor, MI 48109.Volume 81, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 155

http://refhub.elsevier.com/S0016-5107(14)01982-8/sref11http://refhub.elsevier.com/S0016-5107(14)01982-8/sref11http://refhub.elsevier.com/S0016-5107(14)01982-8/sref11http://refhub.elsevier.com/S0016-5107(14)01982-8/sref13http://refhub.elsevier.com/S0016-5107(14)01982-8/sref13http://refhub.elsevier.com/S0016-5107(14)01982-8/sref13http://refhub.elsevier.com/S0016-5107(14)01982-8/sref13http://refhub.elsevier.com/S0016-5107(14)01982-8/sref14http://refhub.elsevier.com/S0016-5107(14)01982-8/sref14http://refhub.elsevier.com/S0016-5107(14)01982-8/sref14http://refhub.elsevier.com/S0016-5107(14)01982-8/sref14http://refhub.elsevier.com/S0016-5107(14)01982-8/sref15http://refhub.elsevier.com/S0016-5107(14)01982-8/sref15http://refhub.elsevier.com/S0016-5107(14)01982-8/sref15http://refhub.elsevier.com/S0016-5107(14)01982-8/sref15http://refhub.elsevier.com/S0016-5107(14)01982-8/sref16http://refhub.elsevier.com/S0016-5107(14)01982-8/sref16http://refhub.elsevier.com/S0016-5107(14)01982-8/sref16http://refhub.elsevier.com/S0016-5107(14)01982-8/sref16http://refhub.elsevier.com/S0016-5107(14)01982-8/sref16ahttp://refhub.elsevier.com/S0016-5107(14)01982-8/sref16ahttp://refhub.elsevier.com/S0016-5107(14)01982-8/sref16ahttp://refhub.elsevier.com/S0016-5107(14)01982-8/sref16ahttp://refhub.elsevier.com/S0016-5107(14)01982-8/sref16ahttp://refhub.elsevier.com/S0016-5107(14)01982-8/sref17http://refhub.elsevier.com/S0016-5107(14)01982-8/sref17http://refhub.elsevier.com/S0016-5107(14)01982-8/sref17http://refhub.elsevier.com/S0016-5107(14)01982-8/sref17http://refhub.elsevier.com/S0016-5107(14)01982-8/sref19http://refhub.elsevier.com/S0016-5107(14)01982-8/sref19http://refhub.elsevier.com/S0016-5107(14)01982-8/sref19http://refhub.elsevier.com/S0016-5107(14)01982-8/sref20http://refhub.elsevier.com/S0016-5107(14)01982-8/sref20http://refhub.elsevier.com/S0016-5107(14)01982-8/sref20http://refhub.elsevier.com/S0016-5107(14)01982-8/sref20http://refhub.elsevier.com/S0016-5107(14)01982-8/sref20http://refhub.elsevier.com/S0016-5107(14)01982-8/sref21http://refhub.elsevier.com/S0016-5107(14)01982-8/sref21http://refhub.elsevier.com/S0016-5107(14)01982-8/sref21http://refhub.elsevier.com/S0016-5107(14)01982-8/sref21http://refhub.elsevier.com/S0016-5107(14)01982-8/sref22http://refhub.elsevier.com/S0016-5107(14)01982-8/sref22http://refhub.elsevier.com/S0016-5107(14)01982-8/sref22http://dx.doi.org/10.1016/j.gie.2014.07.033http://www.giejournal.org

The risk of post-ERCP pancreatitis and the protective effect of rectal indomethacin in cases of attempted but unsuccessful prophylactic pancreatic stent placementMethodsResultsThe risk of PEP associated with FPS in the placebo groupThe protective effect of rectal indomethacin in patients with FPSFactors associated with FPS

DiscussionReferences