Patient Satisfaction After MRCP and ERCPKrishna Menon, M.D., F.R.C.P.(C.), Alan N. Barkun, M.D., F.R.C.P.(C.), M.Sc. (Epid.), F.A.C.P.,Joseph Romagnuolo, M.D., F.R.C.P.(C.), Gad Friedman, M.D., F.R.C.P.(C.),Shailesh N. Mehta, M.D., F.R.C.P.(C.), Caroline Reinhold, M.D., F.R.C.P.(C.), M.Sc. (Epid.), andPatrice M. Bret, M.D., F.R.C.P.(C.)Division of Gastroenterology and Department of Diagnostic Radiology, McGill University and the McGillUniversity Health Centre, Montreal; and Department of Radiology, University of Toronto, Toronto, Canada

OBJECTIVE: Magnetic resonance cholangiopancreatography(MRCP) is an accurate diagnostic test for detecting abnor-malities of the pancreaticobiliary system. Because it is non-invasive, MRCP appears to be more tolerable than ERCP,although this has not been studied. The purpose of this studyis to compare patient satisfaction after MRCP and ERCPperformed sequentially.

METHODS: We prospectively recruited 34 patients undergo-ing ERCP, for whom an MRCP was able to be performedbefore ERCP. Patient satisfaction was assessed by validatedquestionnaires using seven-point Likhert scales (individualratings and direct comparisons). The following dimensionswere explored: anxiety, pain, discomfort, tolerability (rela-tive to expectations), willingness to repeat the procedure,and overall preference.�2 and Student’st tests (paired andunpaired) were performed, and 95% CIs were provided.

RESULTS: Two patients (5.9%) were unable to undergoMRCP because of claustrophobia. The remaining 32 com-pleted both tests (94% same day) and all questionnaires.Average age was 56� 18 yr, and 66% were women. In 23patients, some degree of biliary obstruction was suspected;nine patients had pancreatitis. Patients reported a lowerdegree of pain (p � 0.001) and discomfort (p � 0.047) withMRCP, but MRCP was more difficult than they expected(p � 0.012). Patients were marginally more willing to repeatMRCP (ns,p � 0.09). On direct comparisons, patients weremore satisfied with MRCP regarding anxiety (p � 0.04) andpain (p � 0.001). Patients displayed a higher overall pref-erence for MRCP compared with ERCP (p � 0.01); how-ever, only 59% clearly preferred MRCP over ERCP. Themost common problem with MRCP was claustrophobia ornoise (n� 15), and the differences were more striking in thesubgroup without this problem. The subgroup undergoingpurely diagnostic ERCPs showed clear preferences forMRCP.

CONCLUSIONS: In many respects, MRCP is well tolerated,and certain subgroups, especially those undergoing diagnos-tic ERCPs, prefer MRCP over ERCP. As an endoscopist,one needs to be aware of the limitations of MRCP and relaythese to the patient, as it seems that patients find MRCP

more difficult than anticipated, and a significant number stillprefer ERCP over MRCP. Patient satisfaction may be fur-ther improved by reducing noise and claustrophobia withselective premedication, earplugs, and the use of the newquieter fenestrated magnetic resonance imaging scanners.(Am J Gastroenterol 2001;96:2646–2650. © 2001 by Am.Coll. of Gastroenterology)

INTRODUCTION

Magnetic resonance cholangiopancreatography (MRCP) is anewly developed application of magnetic resonance imag-ing (MRI), which offers the opportunity of providing bothhigh-quality cross-sectional images of extraductal structuresand projectional (coronal) images of the biliary tree andpancreatic duct (1). The early experience with MRCP dem-onstrates a high diagnostic accuracy in the evaluation ofboth the biliary and pancreatic systems, comparable withthat of diagnostic ERCP (1–3). Unlike MRCP, ERCP carriesa definite morbidity and may cause patient discomfort. Cur-rently, little data are available on patient acceptability ortolerance of both procedures. This study compared MRCPwith ERCP with respect to patient satisfaction and prefer-ence. The clinical role and test performance of MRCP,although an interesting question being addressed by severalother studies, was not the aim of this particular study.

PATIENTS AND METHODS

Over a period of 2 months, 34 patients referred for ERCPagreed to participate and were enrolled. Each patient under-went both MRCP and ERCP, to compare each patient’ssatisfaction with the two procedures. MRCP was alwaysperformed before ERCP. Patients for whom it was impos-sible to organize the two tests because of urgency of theERCP procedure were excluded. After completion of bothtests, patients filled out a validated questionnaire using aseries of seven-point Likhert scales designed to measure thedegree of 1) anxiety, 2) pain, 3) discomfort, 4) expectations,and 5) willingness to repeat each test (4). As well, a secondset of Likhert scales was used to allow patients to directlycompare ERCP with MRCP in terms of the first three of the

THE AMERICAN JOURNAL OF GASTROENTEROLOGY Vol. 96, No. 9, 2001© 2001 by Am. Coll. of Gastroenterology ISSN 0002-9270/01/$20.00Published by Elsevier Science Inc. PII S0002-9270(01)02679-X

above five dimensions and one extra dimension: preferenceof procedure. Patients were then asked to identify the mostdifficult aspect of each procedure, as well as their preferencebetween MRCP and ERCP coded as a binary response.Patients were asked to choose the one aspect (if any) of eachprocedure they found most difficult.

All MRCPs were performed with a General Electric (Mil-waukee, WI) Signa scanner using two-dimensional T2-weighted fast spin-echo sequences. Coronal, axial, and obliqueviews were obtained. No contrast media or sedation was ad-ministered at MRCP although Buscopan was given as needed.Total examination time was approximately 30 min duringwhich patients were lying supine with their arms raised.

All ERCPs were performed with the patient in proneposition. Pharyngeal anesthesia with a local xylocaine spraywas administered to all patients. Premedication using i.v.meperedine and midazolam was given, and additional doseswere given as needed to maintain patient comfort. Buscopan(hyoscine butylbromide) was administered i.v. if necessary.An Olympus (Olympus America, Melville, NY) duodeno-scope (video or fiberoptic, diagnostic or therapeutic) wasused for each case. Therapeutic interventions were carriedout when required at the same setting.

Descriptive variables are expressed as a mean � 1 SD fordescriptive continuous variables, mean with 95% CIs wheninferences were being drawn from a continuous variable,and as a proportion with 95% CI for categorical variables.Statistics were performed using the �2 test and the two-tailed Student’s t test where appropriate. Paired t-tests wereused to analyze repeated measures on the same patient. Asignificance level of � � 0.05 was used to determine sta-tistical significance. Posthoc subgroup analyses were per-formed to further investigate and explain interesting andclinically significant trends that arose in the above mainanalysis.

Informed consent was obtained from all patients accord-ing to institutional review board approval and the humaneand ethical principles set forth in the Helsinki guidelines.

RESULTS

Two patients (5.9%) refused MRCP because of claustro-phobia. The remaining 32 patients underwent MRCP fol-lowed by ERCP, with 94% of them undergoing both pro-

cedures on the same day. Average age of patients was 56 �18 yr (range 18–83), and 66% (22 of 33) were women.ERCP was performed because of varying degrees of suspi-cion for biliary obstruction in 23 patients (on the basis ofabnormal clinical, laboratory, sonographic, and/or CT ab-normalities), and because of previous pancreatitis in theother nine subjects. All i.v. lines were inserted by a seniorgastroenterology trainee, and ERCP procedures were firstattempted by this same trainee. After what was believed tobe a reasonable attempt by the trainee, the attending ER-CPist took over the procedure.

ERCP was unsuccessful, in terms of cannulation of theduct of interest, in five patients (15.6%; 3.0–28.2%), four ofwhom had a second attempt with successful outcome. Thesecases include one patient who had had a Billroth II gastrec-tomy and another with partial opacification of the commonduct. Satisfaction comparisons included only the first ERCPattempt. Adequate images were obtained in all patients whounderwent MRCP. Difficulty at MRCP was attributed to theenclosed nature of the examining table in 15 patients (47%;29–65%). Nine patients (28%; 14–47%) complained ofhaving to hold their arms up in the air, and two patients (6%;1–21%) of noise during the examination. The most difficultaspect of ERCP was passing the endoscope in 15 patients(47%; 29–65%). In addition, discomfort from lying prone,gas pains, and insertion of the i.v. catheter were each re-ported in three patients (9%; 2–25%). Six patients (21%;7–36%) reported no difficulty with MRCP versus eightpatients (25%; 12–43%) with ERCP (ns).

Individual Assessment of ProceduresThe first set of postprocedure Likhert scales (questionsassessing each test separately, without reference to the othertest) revealed no significant difference in degree of anxietybetween MRCP and ERCP (Table 1). Patients reported asignificantly lower degree of discomfort with MRCP ascompared with ERCP (2.47 � 1.6 vs 3.09 � 1.7, respec-tively; 95% CI for difference � 0.01–1.5; p � 0. 047), anda lower degree of pain (1.3 � 0.8 vs 2.7 � 1.8; 95% CI fordifference � 0.6–2.1; p � 0.001). MRCP was found to be“more difficult than expected” to a greater degree than wasERCP (�0.7 � 1.5 vs �1.3 � 1.5; 95% CI for difference ��1.6 to �0.2; p � 0.012). Patients were equally willing torepeat either procedure, although there was a trend favoringMRCP in this regard (p � 0.09).

Table 1. First Set of Postprocedure Patient Satisfaction Scores Rating Each Test Independently

ScaleMRCP(�SD)

ERCP(�SD) p

Anxiety not anxious 1 . . . 7 extremely anxious 2.44 � 1.9 2.78 � 2.0 0.35Pain no pain 1 . . . 7 extreme pain 1.34 � 0.8 2.69 � 1.8 �0.001*Discomfort no discomfort 1 . . . 7 extreme discomfort 2.47 � 1.6 3.09 � 1.7 0.047*Difficulty relative to expectations easier �3 . . . 3 more difficult �0.72 � 1.5 �1.30 � 1.5 0.01*Willingness to repeat same test very willing 1 . . . 7 not willing 2.19 � 1.8 2.75 � 1.7 0.09

* Statistical significance for the difference between the two tests (paired t-test) is denoted by an asterisk.

2647AJG – September, 2001 Satisfaction After MRCP and ERCP

Comparative Assessment of ProceduresThe results from the second set of Likhert scales (questionsmaking direct comparisons between MRCP and ERCP) aresummarized in Figure 1. The scales range from �3 to � 3,and scores above 0 favor MRCP. They revealed that, withregards to anxiety, pain, and discomfort, patients had agreater satisfaction after MRCP, although the third dimen-sion (discomfort) did not reach statistical significance: meanscores (95% CI) were 0.6 (0.02–1.2), 0.9 (0.5–1.3), and 0.4(�0.2–1.0), respectively. Nineteen patients (59%; 41–76%)preferred undergoing MRCP over ERCP, with two patientsexpressing no preference. Using the Likhert scales, patientsdisplayed a significantly higher preference for MRCP com-pared with ERCP (mean score � 0.8 [0.2–1.4]; p � 0.01).

Subgroup AnalysesThe advent of newer, quieter, fenestrated machines and theuse of selective sedation and earplugs may render claustro-phobia and noise modifiable factors in the overall satisfac-tion with MRCP; therefore, it was decided to look at thesubgroup without claustrophobia separately. In those whodid not complain of claustrophobia or noise (n � 15), thedifference in patients’ willingness to repeat either procedurewas now statistically significant (mean difference � 1.2units [95% CI � 0.5–1.9], p � 0.002). As well, the meandifference in scores rating the difficulty of the procedure(compared with expectations) dropped from �0.9 (favoringERCP) to �0.1 (95% CI � �1.1–0.9), which was no longer

significant (p � 0.80). The results from the Likhert scalesconstructed as direct comparisons did not change. Thisanalysis supported the generation of a hypothesis that if thedisadvantages of noise and claustrophobia were able to beeliminated from MRCP using the techniques describedabove, then patient satisfaction after MRCP (compared withtheir expectations) and their willingness to repeat the testwould both improve.

A second subgroup that was believed to be of interest wasthe group of patients undergoing therapeutic rather thandiagnostic ERCPs. In this subgroup (n � 16), the onlysignificant difference in individual assessments was in painscores, with ERCP scoring higher than MRCP (95% CI fordifference � 0.09–2.2, p � 0.04); all other dimensions werenot significant, as were the difficulty of the procedures(compared with expectations) and the willingness to repeatthe procedures. The comparative scales showed similar re-sults, with greater pain for ERCP (mean � 0.8; 0.3–1.4, p �0.007) and other dimensions being not significant. Therewas no significant preference for either procedure in thissubgroup (p � 0.58), and although the proportion statingthat ERCP would be chosen over MRCP was lower (43.8%)than the corresponding proportion for the entire study co-hort, this difference was not significant (p � 0.1). On theother hand, the subgroup with purely diagnostic ERCPs(n � 16) showed a clear preference for MRCP (p � 0.002).

The third subgroup comprised those patients who re-

Figure 1. Patient-rated procedure differences and preferences according to direct comparison Likhert scales‡. Error bars represent SEs.Statistical significance is denoted by an asterisk.

2648 Menon et al. AJG – Vol. 96, No. 9, 2001

ported that their most difficult aspect of the ERCP wasrelated to a bad experience with i.v. insertion. This analysiswas performed because it was believed that this problemcould be modified with the use of dedicated personnel forthe purpose of starting i.v., and that it was in a sense unfairto attribute pain from i.v. insertion to the ERCP procedureitself. However, elimination of these three patients did notalter any of our results, including the comparisons relatingto procedural pain or discomfort. The only change notedwas that the mean difference in scores for willingness torepeat either test narrowed to 0.59 (95% CI � �0.1–1.3),still marginally favoring MRCP.

DISCUSSION

This study did not attempt to define the clinical outcome ofthe performance of MRCP in comparison with ERCP, butrather to assess patient satisfaction. There are no otherpublished studies to date formally measuring tolerability,patient satisfaction, or quality of life after MRCP. In thisstudy, MRCP had higher clinical scores for patient satisfac-tion in terms of anxiety, discomfort, and pain. The two typesof tests (individual ratings and comparative ratings) were allconcordant with respect to the direction of their point esti-mates, speaking to the validity of the analysis. Two dimen-sions, anxiety and discomfort, were significant using oneapproach but not with the other; however, the directions ofthe trends were again concordant. The nonsignificant resultson one set of tests may be caused by a type II error becauseof our sample size, confirmed by the fact that the CIs for thenonsignificant estimates for these two indicate that a clini-cally important difference (one point on the seven-pointscale) may have been missed. The comparative scales mayhave been more sensitive to differences as was seen in theassessment of overall procedural preference. Finally, al-though the posthoc subgroup analyses were helpful in ex-plaining some of the results and elucidating potential meth-ods for improving procedural satisfaction, they involvedsmaller numbers of patients and were, therefore, associatedwith reduced power, and were prone to some bias, as thesubgroups were not defined a priori.

There was a nonsignificant trend toward patients beingmore willing to repeat MRCP rather than ERCP, but thedifference was small. In addition, only 59% clearly pre-ferred MRCP over ERCP. These results are somewhat sur-prising, as MRCP is a noninvasive test with one of itsproposed advantages lying in augmented patient satisfac-tion. Therefore, one would have expected a more dramaticpreference of MRCP over ERCP. Also, patients seemed tohave been better prepared for the adversities of ERCP thanthose of MRCP, as the patients in this study believed thatMRCP was in general more often more difficult than theyhad expected. A more adequate warning of the negativeaspects of MRCP than is currently routine may be war-ranted.

Melendez and McCrank reported that up to 30% of pa-

tients undergoing MRI experience considerable apprehen-sion, and that 5–10% of patients experience severe anxietyreactions (5). Such reactions may furthermore result in poorquality images because of motion artifact or incompletestudies (6). The most common difficulty associated withMRCP was claustrophobia, reported in 15 (47%) of patients,and the selective use of sedation in this subgroup mayincrease patient satisfaction with this procedure (7). The useof fenestrated MRI scanners may also reduce the incidenceof claustrophobia. The noise level during MRCP, reportedas the most difficult aspect of the procedure by 6% ofpatients, may be improved by quieter scanners and/or theprovision of disposable earplugs (6). In this study, the pro-cedural preference was clearer in those patients who did notsuffer from excessive noise or claustrophobia, so that ad-dressing these potentially modifiable problems should resultin a higher patient satisfaction with MRCP compared withERCP.

Using an approximate costing estimation method, whichhas been described previously (8, 9), the average prelimi-nary estimates for the total direct costs for MRCP and ERCPare approximated as $442 ($557 if enhanced MRI exami-nation also performed) and $220 ($340 if sphincterotomyperformed), respectively (year 2000 Canadian dollars).

Overall, of the 32 patients who underwent ERCP, 50%were diagnostic, whereas the remainder required therapeuticintervention such as stone extraction or stent insertion. Thepatients requiring a diagnostic study alone appeared to havethe greatest preference for MRCP over ERCP, in subgroupanalysis. It is important to remember that this study wasperformed early on in our MRCP experience, and was notdesigned to determine the optimal method of cholangiogra-phy, but rather to compare the satisfaction with each pro-cedure in a paired cohort of patients. As a result also, thestudy did not directly address the possible cost savings orimpact on patient care of a strategy involving the use ofMRCP as an initial screening tool. At least one randomizedtrial is currently underway to assess the formal cost-effec-tiveness and overall patient outcome of this strategy (10).

One unavoidable limitation of this study design is that the“cross-over,” for ethical and practical reasons, is in onedirection only, i.e., MRCP was always performed beforeERCP. One is, therefore, unfortunately, not able to analyzefor the possibility of so-called “order interactions” regardingthe relationship between the sequence in which the two testswere performed and the responses on the questionnaire (11).One can find reasons to predict why the second test per-formed might be rated more harshly than the first, and yetequally find reasons why the opposite could be true. There-fore, it is not possible to accurately predict the direction ofany bias that may be present because of this interaction.

MRCP represents an attractive alternative to diagnosticERCP as it is noninvasive and well tolerated by most pa-tients. Overall, procedural satisfaction appeared to be higherwith MRCP compared with ERCP; however, the difference

2649AJG – September, 2001 Satisfaction After MRCP and ERCP

in the willingness to repeat either test was not significantand, in direct comparison, there was a significant proportionof patients who still preferred ERCP over MRCP, despiteMRCP’s noninvasive nature. Patients undergoing MRCPneed to be better counseled regarding the inconveniencesassociated with the test (raising arms, claustrophobia,noise). The MRCP procedure was often more difficult thanpatients anticipated. It appears that the patients requiringpurely diagnostic cholangiography, and those who did notsuffer from claustrophobia and/or noise, had the greatestdifference in satisfaction between MRCP and ERCP. Betterpatient selection, fenestrated scanners, earplugs, and selectivesedation may, therefore, overcome some of the problems withsatisfaction in these subgroups. As MRCP becomes more re-fined and widespread, its role with respect to ERCP and otherimaging modalities, such as endoscopic ultrasound or helicalCT, will require further characterization of comparative cost-effectiveness and impact on patient outcome.

ACKNOWLEDGMENTS

The authors appreciate the administrative assistance in pre-paring this manuscript provided by Ms. A. Khalili. Dr.Barkun and Dr. Reinhold are research scholars funded bythe Fonds de la Recherche en Sante du Quebec. Dr. Ro-magnuolo is a research fellow funded by the Alberta Heri-tage Foundation for Medical Research.

Reprint requests and correspondence: Alan N. Barkun, M.D.,F.R.C.P.(C.), M.Sc. (Epid.), F.A.C.P., Division of Gastroenterol-ogy (Room D7-148), Montreal General Hospital, 1650 Cedar Av-enue, Montreal, Quebec, Canada H3G 1A4.

Received Nov. 2, 2000; accepted Apr. 30, 2001.

REFERENCES

1. Mehta SN, Reinhold C, Barkun AN. Magnetic resonancecholangiopancreatography (MRCP). Gastrointest Endosc ClinNorth Am 1997;7:247–70.

2. Adamek HE, Albert J, Weitz M, et al. A prospective evalua-tion of magnetic resonance cholangiopancreatography in pa-tients with suspected bile duct obstruction. Gut 1998;43:680–3.

3. Takehara Y. Can MRCP replace ERCP? J MRI 1998;8:517–34.

4. Mangelsdorff AD. Patient satisfaction questionnaires. MedCare 1979;17:86–90.

5. Melendez JC, McCrank E. Anxiety-related reactions associ-ated with magnetic resonance imaging examinations. JAMA1993;270:745–7.

6. Mackenzie R, Sims C, Ownes RG, et al. Patients’ perceptionsof magnetic resonance imaging. Clin Radiol 1995;50:137–41.

7. Murphy KJ, Brunberg JA. Adult claustrophobia, anxiety andsedation in MRI. MRI 1997;15:51–4.

8. Barkun JS, Caro J, Barkun AN. Cost-effectiveness of laparo-scopic versus mini-cholecystectomy. Surg Endosc 1995;9:1221–4.

9. Barkun AN, Barkun JS, Sampalis JS, et al. The costs andeffectiveness of extracorporeal gallbladder stone shock wavelithotripsy versus laparoscopic cholecystectomy — a random-ized clinical trial. Int J Technol Assess Health Care 1997;13:589–601.

10. Romagnuolo J, Barkun AN, Joseph L, et al. A randomizedclinical trial comparing ERCP and MRCP in an effective-ness setting. Gastrointest Endosc 2000;51:AB199 (ab-stract).

11. Louis TA, Lavori PW, Bailar JC. Crossover and self-con-trolled designs in clinical research. N Engl J Med 1984;310:24–31.

2650 Menon et al. AJG – Vol. 96, No. 9, 2001