cpre y colangioresonancia metaanalisis

A systematic review and economic evaluation of magnetic resonance cholangiopancreatography compared with diagnostic endoscopic retrograde cholangiopancreatography

E Kaltenthaler, Y Bravo Vergel, J Chilcott, S Thomas, T Blakeborough, SJ Walters and H Bouchier

Health Technology Assessment 2004; Vol. 8: No. 10

HTAHealth Technology AssessmentNHS R&D HTA Programme

March 2004

Copyright notice

© Queen's Printer and Controller of HMSO 2004 HTA reports may be freely reproduced for the purposes of private research and study and may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising Violations should be reported to [email protected] Applications for commercial reproduction should be addressed to HMSO, The Copyright Unit, St Clements House, 2–16 Colegate, Norwich NR3 1BQ

How to obtain copies of this and other HTA Programme reports.An electronic version of this publication, in Adobe Acrobat format, is available for downloading free ofcharge for personal use from the HTA website (http://www.hta.ac.uk). A fully searchable CD-ROM isalso available (see below).

Printed copies of HTA monographs cost £20 each (post and packing free in the UK) to both public andprivate sector purchasers from our Despatch Agents.

Non-UK purchasers will have to pay a small fee for post and packing. For European countries the cost is£2 per monograph and for the rest of the world £3 per monograph.

You can order HTA monographs from our Despatch Agents:

– fax (with credit card or official purchase order) – post (with credit card or official purchase order or cheque)– phone during office hours (credit card only).

Additionally the HTA website allows you either to pay securely by credit card or to print out yourorder and then post or fax it.

Contact details are as follows:HTA Despatch Email: [email protected]/o Direct Mail Works Ltd Tel: 02392 492 0004 Oakwood Business Centre Fax: 02392 478 555Downley, HAVANT PO9 2NP, UK Fax from outside the UK: +44 2392 478 555

NHS libraries can subscribe free of charge. Public libraries can subscribe at a very reduced cost of £100 for each volume (normally comprising 30–40 titles). The commercial subscription rate is £300 per volume. Please see our website for details. Subscriptions can only be purchased for the current orforthcoming volume.

Payment methods

Paying by chequeIf you pay by cheque, the cheque must be in pounds sterling, made payable to Direct Mail Works Ltdand drawn on a bank with a UK address.

Paying by credit cardThe following cards are accepted by phone, fax, post or via the website ordering pages: Delta, Eurocard,Mastercard, Solo, Switch and Visa. We advise against sending credit card details in a plain email.

Paying by official purchase orderYou can post or fax these, but they must be from public bodies (i.e. NHS or universities) within the UK.We cannot at present accept purchase orders from commercial companies or from outside the UK.

How do I get a copy of HTA on CD?

Please use the form on the HTA website (www.hta.ac.uk/htacd.htm). Or contact Direct Mail Works (seecontact details above) by email, post, fax or phone. HTA on CD is currently free of charge worldwide.

The website also provides information about the HTA Programme and lists the membership of the variouscommittees.

HTA

A systematic review and economicevaluation of magnetic resonancecholangiopancreatography comparedwith diagnostic endoscopic retrogradecholangiopancreatography

E Kaltenthaler,1* Y Bravo Vergel,2 J Chilcott,1

S Thomas,3 T Blakeborough,4 SJ Walters1

and H Bouchier1

1 ScHARR Rapid Reviews Group, School of Health and Related Research,University of Sheffield, UK

2 Nuffield Institute for Health, University of Leeds, UK3 Northern General Hospital, Sheffield, UK4 Royal Hallamshire Hospital, Sheffield, UK

* Corresponding author

Declared competing interests of authors: none of the authors has any financialinterest in the companies producing the equipment used to undertake magnetic resonancecholangiopancreatography.

Published March 2004

This report should be referenced as follows:

Kaltenthaler E, Bravo Vergel Y, Chilcott J, Thomas S, Blakeborough T, Walters SJ, et al. A systematic review and economic evaluation of magnetic resonancecholangiopancreatography compared with diagnostic endoscopic retrogradecholangiopancreatography. Health Technol Assess 2004;8(10).

Health Technology Assessment is indexed in Index Medicus/MEDLINE and Excerpta Medica/EMBASE.

NHS R&D HTA Programme

The NHS R&D Health Technology Assessment (HTA) Programme was set up in 1993 to ensure that high-quality research information on the costs, effectiveness and broader impact of health

technologies is produced in the most efficient way for those who use, manage and provide care in the NHS.

The research reported in this monograph was commissioned by the HTA Programme and was fundedas project number 02/43/01. Technology assessment reports are completed in a limited time toinform decisions in key areas by bringing together evidence on the use of the technology concerned.

The views expressed in this publication are those of the authors and not necessarily those of the HTA Programme or the Department of Health. The editors wish to emphasise that funding andpublication of this research by the NHS should not be taken as implicit support for anyrecommendations made by the authors.

HTA Programme Director: Professor Tom WalleySeries Editors: Dr Ken Stein, Professor John Gabbay, Dr Ruairidh Milne,

Dr Chris Hyde and Dr Rob RiemsmaManaging Editors: Sally Bailey and Caroline Ciupek

The editors and publisher have tried to ensure the accuracy of this report but do not accept liability for damages or losses arising from material published in this report.

ISSN 1366-5278

© Queen’s Printer and Controller of HMSO 2004

This monograph may be freely reproduced for the purposes of private research and study and may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising.

Applications for commercial reproduction should be addressed to HMSO,The Copyright Unit, St Clements House, 2–16 Colegate, Norwich, NR3 1BQ.

Published by Gray Publishing, Tunbridge Wells, Kent, on behalf of NCCHTA.Printed on acid-free paper in the UK by St Edmundsbury Press Ltd, Bury St Edmunds, Suffolk.

Criteria for inclusion in the HTA monograph seriesReports are published in the HTA monograph series if (1) they have resulted from work commissioned for the HTA Programme, and (2) they are of a sufficiently high scientific quality as assessed by the referees and editors.

Reviews in Health Technology Assessment are termed ‘systematic’ when the account of the search,appraisal and synthesis methods (to minimise biases and random errors) would, in theory, permitthe replication of the review by others.

O2

Objectives: To compare the clinical and cost-effectiveness of magnetic resonancecholangiopancreatography (MRCP) with diagnosticendoscopic retrograde cholangiopancreatography(ERCP) for the investigation of biliary obstruction. Data sources: Electronic bibliographic databases, thereference lists of relevant articles and various healthservices research-related resources.Review methods: The data sources were searchedand selected studies were assessed using qualitycriteria. In total, 28 prospective diagnostic studies wereidentified reporting several suspected conditions plusone of patient satisfaction. Analyses were thenperformed to establish sensitivities, specificities,likelihood ratios and confidence intervals. The relativecost-effectiveness of adopting MRCP scanning in theinvestigation of the biliary tree was undertaken using aprobabilistic economic model. Results: The median sensitivity for choledocholithiasis(13 studies) was 93% and the median specificity 94%.The median likelihood ratio for a positive value was15.75 and for a negative value 0.08. Reportedsensitivities for malignancy were somewhat lower,ranging from 81 to 86%, and specificities ranged from92 to 100%. There was some evidence that MRCP isan accurate diagnostic test in comparison to ERCP,although the quality of studies was moderate.Claustrophobia prevented at least some patients fromhaving MRCP in ten of the 28 studies. The other 18

studies did not mention claustrophobia. The probabilityof avoiding unnecessary diagnostic ERCP is estimated at30%. These patients could avoid the unnecessary riskof complications and death associated with diagnosticERCP, and substantial cost saving would be gained. Theoverall expected cost saving associated with MRCP is£149; the overall expected gain in quality-adjusted life-year is estimated at 0.011.Conclusions: There is some evidence that MRCP is anaccurate investigation compared with diagnostic ERCP,although the values for malignancy compared withcholedocholithiasis were somewhat lower. The qualityof studies was moderate. The limited evidence onpatient satisfaction showed that patients preferredMRCP to diagnostic ERCP. The estimated clinical andeconomic impacts of diagnostic MRCP versusdiagnostic ERCP are very favourable. The baselineestimate is that MRCP may both reduce cost and resultin improved quality of life outcomes compared withdiagnostic ERCP. Further research is suggested tocompare MRCP and diagnostic ERCP with finaldiagnosis and also with the full range of targetconditions; to examine patient satisfaction and ways ofreducing problems with claustrophobia; to look atprotocols to help identify who could most benefit fromMRCP or ERCP; to assess the relative need andurgency of patient access to magnetic resonanceimaging services, and also to determine how demandwould affect availability and potential cost savings.


iii

© Queen’s Printer and Controller of HMSO 2004. All rights reserved.

Abstract

A systematic review and economic evaluation of magneticresonance cholangiopancreatography compared with diagnosticendoscopic retrograde cholangiopancreatography

E Kaltenthaler,1* Y Bravo Vergel,2 J Chilcott,1 S Thomas,3 T Blakeborough,4

SJ Walters1 and H Bouchier1

1 ScHARR Rapid Reviews Group, School of Health and Related Research, University of Sheffield, UK2 Nuffield Institute for Health, University of Leeds, UK3 Northern General Hospital, Sheffield, UK4 Royal Hallamshire Hospital, Sheffield, UK* Corresponding author


v


Glossary and list of abbreviations ............. vii

Executive summary .................................... ix

1 Aim of the review ...................................... 1

2 Background ................................................ 3Description of underlying health problem ...................................................... 3Current service provision ........................... 4MRCP ......................................................... 5

3 Effectiveness ............................................... 9Methods for reviewing effectiveness .......... 9Results ........................................................ 11

4 Economic analysis ...................................... 27Role of diagnostic ERCP and MRCP in the diagnosis of biliary tree obstruction .................................................. 27Decision analytical modelling .................... 28Synthesis of results ..................................... 38Potential methodological strengths andlimitations of the economic analysis .......... 39Conclusions on the economics of diagnostic MRCP versus ERCP .................. 41

5 Factors relevant to the NHS ..................... 43

6 Discussion ................................................... 45Main results: clinical effectiveness ............. 45Main results: cost-effectiveness .................. 45Assumptions, limitations and uncertainties ............................................... 45Need for further research .......................... 46

7 Conclusions ................................................ 47

Acknowledgements .................................... 49

References .................................................. 51

Appendix 1 Search strategies .................. 57

Appendix 2 Other sources searched ....... 59

Appendix 3 Search strategies used ......... 61

Appendix 4 Methodological search filters used in Ovid MEDLINE .................. 63

Appendix 5 Excluded studies .................. 65

Appendix 6 Evidence tables .................... 67

Health Technology Assessment reportspublished to date ....................................... 91

Health Technology Assessment Programme ................................................ 99

Contents


vii


Glossary and list of abbreviations

List of abbreviations2D two-dimensional

3D three-dimensional

CBD common bile duct

CCTR Cochrane Controlled TrialsRegister

CDSR Cochrane Database ofSystematic Reviews

CEAC cost-effectivenessacceptability curve

CI confidence interval

CT computed tomography

EORTC QLQ-30 European Organisation forResearch and Treatment ofCancer Quality of LifeQuestionnaire-30

ERCP endoscopic retrogradecholangiopancreatography

EUS endoscopic ultrasonography

FCE finished clinical episode

FSE fast spin-echo

HASTE half-Fourier single-shotturbo spin echo

HEED Health EconomicsEvaluations Database

HES Hospital Episode Statistics

HRGs Healthcare Resource Groups

ICD-10 International Classificationof Diseases-10

IOC intraoperativecholangiography

� kappa value, used todetermine interobserveragreement

continued

GlossaryAccuracy True positives and true negativesas a proportion of all results; true positives +true negatives/total.

Choledocholithiasis Gallstones in theextrahepatic bile ducts.

Cholelithiasis The presence of microscopiccrystals or large stones in the gallbladder.

Likelihood ratio of a positive test Thelikelihood of a positive test being found in aperson with the condition compared with in aperson without it; sensitivity/(1 – specificity).

Likelihood ratio of a negative test Thelikelihood of a negative test being found in aperson without the condition compared with ina person with it; (1 – sensitivity)/specificity.

Sensitivity The proportion of patients withdisease who test positive; true positives/(truepositives + false negatives).

Specificity The proportion of patientswithout disease who test negative; truenegatives/(false positives + true negatives).

Technical terms and abbreviations are used throughout this report. The meaning is usually clear fromthe context, but a glossary is provided for the non-specialist reader. In some cases, usage differs in the

literature, but the term has a constant meaning throughout this review.

Glossary and list of abbreviations

viii

List of abbreviations continued

LFT liver function test

LRN likelihood ratio negative

LRP likelihood ratio positive

MAICER maximum acceptableincremental cost-effectiveness ratio

MIP maximum intensityprojection

MR magnetic resonance

MRCP magnetic resonancecholangiopancreatography

MRI magnetic resonance imaging

NHS DARE Database of Assessment ofReviews of Effectiveness

NHS EED Economic EvaluationsDatabase

NSRC National Schedule ofReference Costs

PBJ pancreatobiliary junction

PSC primary sclerosingcholangitis

PTC percutaneous transhepaticcholangiography

QALY quality-adjusted life-year

RARE rapid acquisition andrelaxation enhancement

RCR Royal College ofRadiologists

ROC receiver operatingcharacteristic

SD standard deviation

SE standard error

TSE turbo spin echo

TTO time trade-off

All abbreviations that have been used in this report are listed here unless the abbreviation is well known (e.g. NHS), or it has been used only once, or it is a non-standard abbreviation used only in figures/tables/appendices in which case the abbreviation is defined in the figure legend or at the end of the table.


ix


BackgroundMagnetic resonance cholangiopancreatography(MRCP) is an alternative to diagnostic endoscopicretrograde cholangiopancreatography (ERCP) forimaging the biliary tree and investigating biliaryobstruction. MRCP is a purely diagnostic test withno therapeutic value. It does not have the smallbut definite morbidity and mortality associatedwith ERCP.

Biliary obstruction may be due tocholedocholithiasis, tumours or trauma includinginjury after gallbladder surgery, among othercauses. Choledocholithiasis is the most commoncause. Between 5 and 22% of the Westernpopulation has gallstones. The overall prevalencerate for symptomatic gallstones for England andWales in 1991–2 was 182 per 10,000 person-yearsat risk. The incidence rate was 8 for cholelithiasis,9 for other disorders of the gallbladder and 2 forother disorders of the biliary tract per 10,000person-years at risk. At the time ofcholecystectomy for symptomatic cholelithiasis,8–25% of patients under 60 years and 15–60% ofpatients over 60 years also havecholedocholithiasis.

MRCP refers to selective or partially selectivemagnetic resonance imaging (MRI) of thepancreatic and biliary ducts. It was developed in1991 and techniques have progressively improvedsince then. Patients should be fasting and theprocedure takes a few minutes, usually withoutsedation. Claustrophobia is a problem with somepatients. A major feature of MRCP is that it is nota therapeutic procedure, whereas ERCP is used fordiagnosis and treatment. The impact of this is thatif ERCP is necessary after MRCP as a therapeuticintervention, MRCP could have been avoided andpatients would be able to proceed immediately totreatment. However, if no therapeutic interventionis found to be necessary, MRCP avoids thepotential morbidity and mortality associated withERCP. MRCP is particularly useful where ERCP isdifficult, hazardous or impossible. It is also animportant option for patients with failed ERCPs.ERCP and MRCP have different contraindications,allowing them to be used as complementarytechniques.

There are opportunity costs associated with MRCP,in that if an MRI scanner is used for MRCP itcannot be used for other types of imaging.

ObjectiveThe aim of this review is to compare the clinicaland cost-effectiveness of MRCP with diagnosticERCP for the investigation of biliary obstruction.

Number and quality of studiesand direction of evidenceInitially 67 potentially relevant papers wereconsidered for inclusion, of which 38 wereexcluded owing to poor quality or comparatorsother than ERCP. In total, 28 prospectivediagnostic studies were identified comparingMRCP with diagnostic ERCP. One study of patientsatisfaction was also identified. The 28 studiesreported several suspected conditions.Choledocholithiasis was included in 18 studies,malignancy in four, obstruction in three, stricturein two, dilatation in five and primary sclerosingcholangitis (PSC) in two studies.

The quality of the studies was moderate. In all butone study, patients selected to have both MRCPand diagnostic ERCP did not have both and oftenthe reasons why were unclear. Only 13 of the 28 studies reported blinding to both clinicalinformation for patients and ERCP results, andonly six of the 28 studies reported information onagreement of MRCP results for more than oneinvestigator. Nine studies gave no information onother diagnostic tests and most studies did notadequately report inclusion and exclusion criteriaand relevant patient characteristics. Of the 28studies, seven reported results comparing MRCPwith final diagnosis, which included ERCP andother test results. The remaining 21 studiesreported results comparing MRCP with ERCP.

Effectiveness was assessed by condition. Forcholedocholithiasis 15 of the 18 studies reportedadequate data for analysis; two of these wereremoved as they differed in some aspects from theother studies. Owing to statistically significant

Executive summary

heterogeneity between the studies, the medianvalues were considered the most appropriate toreport. The median sensitivity for the 13 studies ofcholedocholithiasis was 0.93 (range 0.81–1.00) andthe median specificity 0.94 (0.83–0.99). Alikelihood ratio describes how many times aperson with disease is more likely to receive aparticular test result than a person withoutdisease. The median positive likelihood ratio was15.75 (range 5.44–64.78) and the median negativelikelihood ratio 0.08 (0.00–0.19).

For malignancy, sensitivity ranged from 81 to94.4% and specificity from 92 to 100%. Positivelikelihood ratios ranged from 10.12 to 43 andnegative likelihood ratios from 0.15 to 0.21. Thesensitivity for dilatation ranged from 87 to 100%and the specificity from 91 to 100%. Forobstruction, both sensitivity and specificity rangedfrom 91 to 100%. Sensitivity for stricture was 100%and specificity ranged from 98 to 99%.

Claustrophobia associated with MRCP in at leastsome patients was reported in ten of the 28 studies,with no information on claustrophobia reported inthe remaining 18 studies. There were no adverseeffects associated with MRCP in any of the studies,although six studies reported adverse effectsassociated with ERCP, including pancreatitis,bleeding and pain. Twenty studies reported noinformation regarding adverse effects.

One study was identified that dealt with patientsatisfaction: most patients preferred MRCP, butthere were still some who preferred ERCP. Nearlyhalf of the patients in this small study complainedof claustrophobia associated with MRCP, althoughonly 5.9% refused MRCP for this reason.

Summary of benefitsThe median sensitivity for choledocholithiasis (13 studies) was 93% (range 81–100%) and themedian specificity 94% (83–99%). The medianlikelihood ratio for a positive value was 15.75(range 5.44–64.78) and for a negative value 0.08(0.00–0.19). Reported sensitivities for malignancywere somewhat lower, ranging from 81 to 86%,and specificities ranged from 92 to 100%.

In the 28 studies, which included 38 subgroups,one positive likelihood ratio was less than 5 andfour negative likelihood ratios were greater than0.2. There is therefore some evidence that MRCP isan accurate diagnostic test in comparison to ERCP,although the quality of studies was moderate.

Claustrophobia prevented at least some patientsfrom having MRCP in ten of the 28 studies. Theother 18 studies did not mention claustrophobia.

Cost-effectivenessThe probability of avoiding unnecessary diagnosticERCP, that is, the probability of a true-negativeMRCP, is estimated at 30% [95% confidenceinterval (CI) 20 to 40%]. These patients couldavoid the unnecessary risk of complications anddeath associated with diagnostic ERCP, andsubstantial cost saving would be gained. Theoverall expected cost saving associated with MRCPis £149 (£325 to –£15); the overall expected gainin quality-adjusted life-year is estimated at 0.011(0.000, 0.030).

ConclusionsThere is some evidence that MRCP is an accurateinvestigation compared with diagnostic ERCP,although the values for malignancy compared withcholedocholithiasis were somewhat lower. Thequality of studies was moderate. The limitedevidence on patient satisfaction showed thatpatients preferred MRCP to diagnostic ERCP.

The estimated clinical and economic impacts ofdiagnostic MRCP versus diagnostic ERCP are veryfavourable. The baseline estimate is that MRCPmay both reduce cost and result in improvedquality of life outcomes compared with diagnosticERCP. The uncertainty analysis, investigating theimpact of parametric uncertainty within themodel, indicates that this result is robust.However, there are marked uncertainties in thestructure and assumptions within the decisionanalytical model that are not captured within thisparametric uncertainty analysis. The resultspresented in this assessment will thus overstate therobustness of the economic outcomes for MRCP.

Recommendations for researchThe following were identified as areas wherefurther research is needed.

� Good quality studies are needed comparingMRCP and diagnostic ERCP with finaldiagnosis, stating inclusion/exclusion criteriaand relevant patient characteristics. This wouldhelp to overcome some of the shortcomings ofcomparisons with diagnostic ERCP.x

Executive summary


xi


� Studies are needed comparing MRCP withdiagnostic ERCP for the full range of targetconditions, in particular differentiation ofbenign and malignant strictures and the impacton management and outcome.

� More research is needed in the area of patientsatisfaction and ways to reduce problems withclaustrophobia and make MRCP moreacceptable to patients.

� Protocols, assessing prior risk, are needed tohelp to identify which patients with which

suspected conditions would most benefit fromMRCP and which would benefit from ERCP.

� To understand the real opportunity costsassociated with MRCP, studies are needed toassess the relative need and urgency of patientaccess to MRI services.

� As the development of MRCP (a non-invasivetest) may result in an increase in requests overwhat would be expected for ERCP (an invasivetest), research is needed to determine how thiswill affect availability and potential cost savings.

Diagnosis of biliary obstruction is usually madeor confirmed on the basis of an initial

ultrasound examination showing dilated bileducts. If the ultrasound shows information such asa stone in the bile duct or an inoperable tumourthen the patient proceeds directly to endoscopicretrograde cholangiopancreatography (ERCP).ERCP is a diagnostic and therapeutic modalitythat has a small but definite morbidity andmortality.1 Magnetic resonancecholangiopancreatography (MRCP) is analternative to diagnostic ERCP for imaging thebile ducts. MRCP is a purely diagnostic test withno therapeutic capability, but does not have therisks associated with ERCP. ERCP may be usedtherapeutically for stone extraction,sphincterotomy and stent insertion. MRCP may be

appropriate where ERCP is indicated fordiagnostic purposes only.

The overall aim of the review is to assess theclinical effectiveness and cost-effectiveness of theuse of MRCP compared with diagnostic ERCP.More specifically, the review aims:

� to evaluate the clinical effectiveness of MRCP interms of test reliability, diagnostic accuracy,diagnostic impact, therapeutic impact andpatient outcomes in comparison with diagnosticERCP

� to evaluate cost-effectiveness in comparison withdiagnostic ERCP

� to estimate the possible overall cost in Englandand Wales.


1


Chapter 1

Aim of the review

Description of underlying healthproblemBiliary obstructionBiliary obstruction can be due to a variety ofcauses, including gallstones, tumours of the bileducts or pancreas, other tumours that have spreadto the biliary system, trauma including injury fromgallbladder surgery, choledochal cysts, enlargednodes in the porta hepatis and inflammation ofthe bile ducts.2 Risk factors for biliary obstructioninclude a history of cholelithiasis (gallstones),chronic pancreatitis or pancreatic cancer, recentbiliary surgery, recent biliary cancer and abdominaltrauma. Symptoms of biliary obstruction includepale-coloured stools, dark urine, jaundice, itching,abdominal pain in the upper right quadrant,fever, nausea and vomiting. Jaundice is the mostcommon presentation of patients with liver andbiliary disease.3 Diseases affecting the gallbladderand bile ducts occur commonly in the elderly.4

One of the most common causes of biliaryobstruction is choledocholithiasis, also known ascommon bile duct (CBD) stones.

Primary sclerosing cholangitis (PSC) is anidiopathic chronic fibrosing inflammatory diseaseof the bile ducts that eventually leads to bile ductobliteration, cholestasis and biliary cirrhosis.5 Theprevalence of this disease of uncertain aetiology is1–6 per 100,000 people.6

CholedocholethiasisCBD stones can be divided into primary stones,which form in the biliary ducts, and secondarystones, which form in the gallbladder and latermigrate into the biliary ducts. Most CBD stones(95%) are secondary.7 It is difficult to estimate theexact prevalence of choledocholithiasis as smallstones can migrate from the CBD into theduodenum without symptoms.7

All patients with symptomatic gallbladder stones(cholelithiasis) need to be assessed for CBD stones,and treatment of CBD stones is usually suggestedas the occurrence of symptoms or complications isunpredictable.8 Patients may remain symptom freebut later develop symptoms and require furthertreatment. Between 5 and 22% of the populationin the Western world have gallstones.7 In the

Morbidity Statistics from General Practice forEngland and Wales in 1991–2, gallstones werefound to be most prevalent among women aged45–64 years. The overall prevalence rate was 182per 10,000 person-years at risk.9 The incidencerate was 8 for cholelithiasis, 9 for other disordersof the gallbladder and 2 for other disorders of thebiliary tract per 10,000 person-years at risk.9

Since the advent of laproscopic cholecystectomythere has been renewed interested in thepreoperative diagnosis of choledocholithiasis.10 Atthe time of cholecystectomy for symptomaticcholelithiasis, 8–15% of patients under 60 years ofage and 15–60% of patients over the age of 60have CBD stones.8 Mortality associated withcholethiasis and cholecystitis for England andWales in 1999 was 679 for underlying cause ofdeath and 1431 for mentioned cause.11

MalignancyPatients with malignant biliary strictures usuallypresent with cholestasis and jaundice.4 Malignantbiliary obstruction usually has a poor prognosisand is usually managed with palliation of thejaundice.12 A small group of patients may besuitable for surgical resection.13 Palliation usuallytakes the form of stent insertion.14 In one study of182 patients with malignant biliary obstruction,management was endoscopic and palliative. At theend of one year 20.4% were still alive.12

The most common malignant cause of biliaryobstruction is carcinoma of the head of thepancreas. Another less common malignancy,cholangiocarcinoma, is a form of adenocarcinoma,primary to the biliary tree.15 Of these, 10% areprimary intrahepatic masses, 25% occur at theconfluence of the right and left ducts (Klatskintumours), 30% arise in the proximal commonducts and 35% arise in the distal common duct.15

Cholangiocarcinoma is an uncommon neoplasm,representing approximately 0.5–1% of all cancersand 30% of hepatic primary malignancies.5 Survivalof patients with Klatskin tumours that are advancedand incurable is usually less than 6 months.14

According to the National Cancer Statistics forEngland, in 1997 there was a total of 1836 newlydiagnosed cases of malignant neoplasms of the


3


Chapter 2

Background

liver and intrahepatic bile ducts, 404 cases ofmalignant neoplasm of the gallbladder, and 590 cases of malignant neoplasm of other andunspecified parts of the biliary tract.16

Significance in terms of ill-healthBiliary obstruction can lead to infections and belife-threatening if not corrected. If the obstructioncontinues for a long period, chronic liver diseasecan result. Obstructions caused by cancerfrequently have a worse outcome. Complicationsinclude infections, sepsis and liver disease such asbiliary cirrhosis if obstructions are left untreated.2

Biliary obstruction presents as an acute or chroniccondition. Jaundice without pain suggests amalignant cause, and prompt diagnosis andtreatment are important. Choledocholithiasis canpresent with severe pain and patients may beadmitted to hospital. Pain may be intermittent andfor this reason the condition is often notimmediately diagnosed. PSC is a chronic condition,although there may be acute exacerbations.

Current service provisionPatients presenting with symptoms of biliaryobstruction (jaundice and/or abdominal pain)initially have laboratory investigations, whichusually include liver function tests.Ultrasonography is the first-line imaginginvestigation in patients with jaundice or rightupper quadrant pain.3 Although ultrasonographyis non-invasive, quick and inexpensive it is veryoperator and patient dependent. Bowel gasfrequently obscures the lower end of the CBD.Computed tomography (CT) may also be used andexperience is required to interpret these images,as well as magnetic resonance imaging (MRI)results. Those patients with a high probability ofCBD stones on the basis of the ultrasoundinvestigations usually proceed directly to ERCP.

ERCPERCP was developed approximately 30 years agoand is both an endoscopic and a radiologicalprocedure.17 It is one of several invasive directcholangiography techniques, along withpercutaneous transhepatic cholangiography(PTC). ERCP is both a diagnostic and therapeuticintervention.18 It is usually performed with the useof conscious sedation. An endoscope is passed tothe ampulla (the opening of the bile andpancreatic ducts), located in the second portion ofthe duodenum. In diagnostic procedures,catheters are passed through the channel of the

endoscope into the duct of interest and contrastmedium is injected to outline the ductalstructures. Sphincter pressure measurements canbe achieved, although this is not routine and has asignificant risk of producing acute pancreatitis.Therapeutic manoeuvres are performed byincising the sphincter muscle at the opening ofthe bile duct or pancreatic duct. Other accessoriesmay be passed through the endoscope channelinto the duct to remove stones, insert stents orablate tissue.17 The results are very dependent onthe skills of the team involved.19

ERCP is currently the ‘gold standard’ for thediagnosis of pancreatic and biliary ductalpathology.20 Approximately 10% of ERCPsperformed are unsuccessful, approximately 15%will demonstrate normal results and many willdemonstrate abnormalities that do not requirefurther endoscopic therapy.21 Since theintroduction of endoscopic ultrasonography (EUS)and MRCP for the diagnosis of CBD stones, it hasbecome clear that ERCP is imperfect andsphincterotomy and balloon or basket trawl of theduct may possibly be a more appropriate goldstandard for diagnosis in the future.22

ERCP failureERCP is a technically demanding procedure, withreported failure rates between 3 and 12%.23 Successrate is highly variable from one centre to anotherdepending on the disease entities being treated, theavailability of dedicated accessories and well-trainedstaff, and the skill of the endoscopists.24 Otherfactors include the proportion of patients withadverse anatomical factors (large diverticula andstones, tortuous ducts).25 PTC and intraoperativecholangiography (IOC) are either of inferiordiagnostic quality or invasive and thereforeassociated with high complication rates.26

ERCP complicationsDiagnostic ERCP has a complication rate of 5–6%and a mortality rate ranging from 0.011 to0.089%.27 Therapeutic ERCP has a complicationrate of 4–10%, although some authors put the rateas high as > 20%,28 and a mortality rate rangingfrom 0.071 to 0.3%.29 Risk factors forcomplications after ERCP are patient related,procedure related or operator related. Patient-related risk factors include underlyingcoagulopathy and suspected dysfunction of thesphincter of Oddi. Procedure-relatedcomplications include difficult bile ductcannulation, injection of radiographic contrastmaterial into the pancreatic duct and precutbiliary sphincterotomy. Complication rates are

Background

4

somewhat lower in diagnostic ERCP than intherapeutic ERCP. Higher post-ERCPcomplications are associated with centres wherelower numbers of ERCPs are performed and whenendoscopists have low ERCP caseloads.17,28 ERCPmay be impossible in some patients, such as thosewho have had a Billroth II gastrectomy, Roux-en-Ydiversions, pancreatic pseudocysts, sclerosingcholangitis or prior serious ERCP complications.20

ERCP is an invasive procedure and is associatedwith risks such as pancreatitis and perforation.17

Risk factors for pancreatitis include patient ageunder 60, use of precut papillotomy and failedclearance of bile duct stones.24 Pancreatitis is themost frequent complication and occurs in about5–10% of cases.4 One large single centre studyfound that pancreatitis constituted 70% ofcomplications.27 There is also a 1% risk ofbleeding, perforation and cholangitis.4 The mainadvantage in using ERCP is that it is also atherapeutic intervention and can follow onimmediately after diagnostic ERCP. Majordisadvantages are that diagnostic ERCP in ahealthy patient may result in the morbidity ormortality described above. In addition, if MRCPfollows diagnostic ERCP in patients withmalignancies, it is difficult to identify andaccurately stage the pancreatic mass, especiallyafter insertion of a stent into the lower CBD.

Costs and variation in servicesCosts for a diagnostic ERCP examination of thebile duct are approximately £846 (2002 costs).30

When complications are present the cost rises toapproximately £1113 (2002 costs), although thisranges from £570 to £1409.30 Costs fortherapeutic ERCP, including extraction of CBDstones, are approximately £1108 (2002 costs).30

The provision of ERCP within a hospital setting willdepend on several factors. These include demandfor the procedure as this will influence waiting time.The expertise of endoscopists varies widely, whichwill have an impact on the number of failures andcomplication rates. One study reported a meanhospital stay of 2.6 days for post-ERCP pancreatitis,although this was longer in severe cases.31

MRCPMRCP refers to selective or partially selective MRIof the pancreatic and biliary ducts.32 MRCP isused to investigate suspected choledocholithiasis,neoplastic obstruction (tumours), benign andmalignant strictures, chronic pancreatitis, primary

sclerosing cholangitis, mucinous ductal ectasia,anatomical variants and postcholecystectomybiliary disorders.32 Indications for the use ofMRCP include unsuccessful or contraindicatedERCP, patient preference for non-invasiveimaging, patients with a low index of suspicion forpancreatic or biliary disease, patients where theneed for therapeutic ERCP is considered unlikelyand those with suspected neoplastic, pancreatic orbiliary obstruction.32

MRCP was first developed in 1991,33 since whenthere have been significant improvements intechnique.34 MRCP broadly involves two methods,the acquisition of a volume of data that can beacquired at different angles, and multisliceacquisition in coronal/axial planes and theproduction of composite images using maximumintensity projection (MIP). MRCP is based onheavily T2-weighted images, resulting in contrastbetween stationary fluids (bile and pancreatic) andbackground (hepatic and pancreatic parenchymas,abdominal fat). The bile presents a very high signalintensity compared with the low signal intensitybackground. Sequences and saturation pulses areselected to produce no signal from flowing blood.Contiguous thin slices obtained in a coronal planeare reconstructed in multiple angles with MIP.This provides an overview of the biliary tree andpancreatic ducts similar to conventional directcholangiopancreatography. The pancreatic ductand bile duct can be observed from several angles.MRCP techniques involve the following variables:pulse sequences, two- or three-dimensional (2D or3D) acquisition, single-slice or multislice acquisition,coil, respiration (breath-holding or non-breath-holding) and background suppression.35

Initially, T2-weighted 3D gradient echo sequenceswere used.35 2D fast spin-echo (FSE) sequenceswere then reported, but these required a relativelylong breath-holding period (44–60 seconds).Third generation MRCP is a non-breath-holdingFSE technique. With this technique, clear imagesare produced but a long acquisition time isrequired (11–15 minutes), depending onrespiratory rate. Finally, a fourth-generationMRCP technique, single-shot turbo spin echo(TSE) sequence has been developed. The adventof FSE sequences has shortened imaging timesignificantly.36 These are rapid acquisition andrelaxation enhancement (RARE) techniques. Theacquisition of data and the filling of K-space hasallowed these sequences to become breath-holdingusing half-Fourier methods to fill K-space, forexample in the half-Fourier single-shot turbo spinecho (HASTE) sequence.


5


No patient preparation is required for MRCP,although fasting for 2–4 hours is useful inreducing fluid in the gastric antrum andduodenum and filling the biliary tree andgallbladder.18 In most patients sedation is notrequired.34 MRI allows imaging in any plane andany thickness of image section.

Many patients requiring investigations with MRCPare elderly and may find it difficult to hold theirbreath for long periods; therefore, the use of non-breath-holding techniques has been explored.However, major limitations of using non-breath-holding techniques include limitations in theevaluation of intrahepatic ducts due to both motionartefacts and limited spatial resolution of the 2Dtechnique.33 Non-breath-holding techniques areoften based on FSE sequences.37 More recently,investigators have explored the possibility ofMRCP with quiet breathing. Patients lie supineand a flexible torso coil is strapped to the upperabdomen. Initially some breath-holding images areobtained to identify adjacent anatomy and to allowpositioning of subsequent MRCP images. Currentimaging techniques usually require breath-holds of1–15 seconds, but diagnostic quality images canstill be obtained without breath-holding.

Indications and contraindications forthe use of MRCPBiliary obstruction represents the main indicationfor MRCP, owing to the ability of this technique toaccess the presence, site and cause ofobstruction.37 MRCP is particularly useful whereERCP is difficult, hazardous or impossible, such asin patients who have had Billroth II gastrectomy,Roux-en-Y diversions, pancreatic pseudocysts,sclerosing cholangitis and prior serious ERCPcomplications.20 MRCP can be used to determineduct calibre, anomalies, strictures, dilatation,filling defects (calculi) and extraductal collectionsof fluid (cysts, diverticula and fistulae).38 It isimportant to assess which subgroups of patientswill benefit from pre- or postoperative MRCP;39

also, if patients do not have an obstruction, theyhave not suffered unnecessary invasive procedures.MRCP is an important option for patients withfailed ERCPs. Another important advantage ofMRCP is that it can be coupled with MRI ofadjacent viscera for identification, characterisationand staging of malignant strictures.40

The contraindications for ERCP and MRCP aredifferent, allowing them to be employed ascomplementary techniques capable of imaging thepancreaticobiliary ducts in virtually all types ofpatient.41

Contraindications to MRCP, as in all MRI, includecardiac pacemakers, retinal metal fragments and, insome cases, subarachnoid aneurysm ferromagneticsurgical clips.41 Other patients unsuitable forMRCP include those with severe claustrophobia,massive ascites or haemodynamic instability.18

Patient obesity may limit the quality of MRCPimages and prevent patients from being able toenter the MRI scanner.34 There are no known risksassociated with MRCP provided patients havebeen carefully screened.34 Claustrophobia andemotional distress prevent completion of the MRIprocedure in up to 5% of patients.42

In one study exploring the value of MRCP resultsto alter differential diagnosis and to preventdiagnostic and/or therapeutic ERCP, the authorsfound that the value of MRCP information may belimited if patient selection is inappropriate andmay differ depending on the speciality of thephysicians involved.20 Physicians appeared mostconcerned that MRCP might miss small bile ductstones or subtle ductal strictures. When ERCP wasplanned, the addition of MRCP to ERCP did notreduce the differential diagnosis significantly andprevented few diagnostic or therapeutic ERCPs.Further research was recommended to evaluate theusefulness of MRCP to a variety of physicians forpatients with normal examinations or for whomthe pretest need for ERCP was uncertain.

Limitations of MRCPThere are several limitations associated withMRCP. Smaller CBD stones can be missed byMRCP.43 However, usually stones up to 2–3 mm insize are visible. Papilla can only be seen in about40% of patients who have MRCP.43 There may alsobe difficulty in depicting minor narrowing of thecystic and pancreatic ducts.44 Another problemassociated with MRCP is that MIP reconstructedimages may completely obscure small fillingdefects and may demonstrate respiratory motionartefacts. Source images should always beinterrogated, so in practice this is not an issue.The major problem with multislice MRCP (forMIP) is respiratory misregistration. Another issueis T2 weighting, which may vary with differentMRI sequences and influence findings. MRCPmay be associated with diagnostic errors, andseveral technical and interpretive pitfalls havebeen associated with its use.45 Incomplete imagingmay create confusion regarding ductal anatomy ordisease. MRCP yields only static images and mayfail to depict various anomalies.46 It is thereforeimportant that both source images and projectionimages are analysed in order to visualise andevaluate the anatomy of the entire

Background

6

pancreatobiliary tract. MRCPs of diagnostic qualitycan be obtained in 92–97% of patients.23

It should be noted that MRCP is only a diagnosticprocedure. The impact of this is that if ERCP isnecessary afterwards as a therapeutic intervention,MRCP could have been avoided and patientswould be able to proceed immediately totreatment. For this reason, routine use of MRCPfor confirmation of the presence of biliaryobstruction before ERCP is difficult to justify.However, if no therapeutic intervention is found tobe necessary, MRCP avoids the potential morbidityand mortality associated with ERCP.

Personnel, setting and equipmentnecessary for MRCPMRCP is usually performed by a radiologist withtraining in MRCP techniques in a hospital setting.An MRI scanner is needed, with the softwarenecessary to perform MRCP. Such MRI scanners

are usually less than 8 years old, although it ispossible to upgrade scanners so that MRCP can beperformed. Periodic software upgrades areundertaken on MRI scanners. Some mobilescanning units are able to perform MRCP. MRCP,like ERCP, is an elective procedure. An MRCPinvestigation takes approximately 15 minutes ofroom time, and the sequences take seconds tominutes. The number of MRCP scans undertakenmay be strongly influenced by the amount of timethe MRI scanner is allocated to undertake suchinvestigations. Depending on the findings atMRCP, patients may proceed to ERCP, surgery orpalliative treatment. Costs have been estimated tobe £454 per MRCP.30 MRCP is estimated to costabout 30–50% of the cost of ERCP.40 There areopportunity costs associated with MRCP in that ifan MRI scanner is used for MRCP it cannot be usedfor other types of imaging. This will have animpact on the provision of other MRIinvestigations.


7


Methods for reviewingeffectivenessSearch strategyThe search strategy aimed to identify all literaturerelating to the clinical and cost effectiveness of theuse of MRCP. The main searches were conductedin January 2003.

Thirteen electronic bibliographic databases weresearched, covering biomedical, science, healtheconomic and grey literature. A list of databases isprovided in Appendix 1.

In addition, the reference lists of relevant articleswere handsearched and various health servicesresearch-related resources were consulted via theInternet. These included health economics andhealth technology assessment organisations,guideline-producing agencies, generic researchand trials registers, and specialist sites. A list ofthese additional sources is given in Appendix 2.

A combination of free-text and thesaurus terms wasused. ‘Population’ search terms (e.g. biliary, biliarytract, bile, gallbladder, choledocholithiasis) werecombined with ‘intervention’ terms (e.g. magneticresonance imaging, MRI, non-invasive diagnosticimaging). To inform the cost-effectiveness reviewand background to the review, additional searcheswere conducted on the epidemiology of biliaryobstruction within the UK (England and Wales),endoscopic retrograde cholangiography, cost–utility analyses for gastrointestinal cancer and quality of life associated with extrahepatic bileduct cancer. Copies of the main search strategiesused in the major databases are included inAppendix 3.

No language or study/publication-type restrictionswere applied to the main searches. An economicevaluations methodological search filter was usedto identify articles for the cost-effectiveness part ofthe review (refer to Appendix 4)

Inclusion and exclusion criteriaThe titles and abstracts of the papers identifiedthrough the search process outlined above wereassessed for relevance to the study question usingthe following criteria.

Inclusion criteria� Subjects: adult patients with suspected biliary

obstruction or dilatation� intervention: MRCP� comparators: diagnostic ERCP� outcome measures to include:

– sensitivity in different patient groups– specificity in different patient groups– likelihood ratios in different patient groups– acceptability to patients– adverse effects

� methodology to include, where available:– systematic reviews– randomised controlled trials– non-randomised studies: prospective

diagnostic trials– economic evaluations.

Full copies were obtained of all those papers thatappeared to be relevant, or that could not beassessed on the basis of the abstract alone.

Exclusion criteriaStudies involving pancreatic ductal systemabnormalities were excluded as the main focus ofthis review was biliary ductal system abnormalities.Studies were also excluded that compared two ormore forms of MRCP but did not include acomparison with diagnostic ERCP. Review papersthat were not systematic were also excluded.

Initially, 67 papers were considered for inclusionas they included a comparison of MRCP withdiagnostic ERCP. Of these, 38 papers were excluded.

Reasons for exclusion were:

� published before 1995� comparison with failed or unsuccessful ERCP� multiple comparators (apart from PTC, IOC

and surgery)� repetition of trial data from another included

study� MRCP results informed the decision to proceed

to ERCP� case–control design� retrospective study design.

The reason for the exclusion of studies before 1995is that the technology for MRCP has changed so


9


Chapter 3

Effectiveness

rapidly since then that it was felt to beinappropriate to include studies before that date asthey would not be comparable. Studies involving acomparison with failed or unsuccessful ERCP wereexcluded as the aim of the study was to compareoutcomes of MRCP with diagnostic ERCP as thegold standard. If ERCP failed or was unsuccessfulthen this comparison was not possible. Studiesinvolving multiple comparators were excluded onlyif the outcomes for comparison with diagnosticERCP were not reported separately. The last threereasons for exclusion are based on Lijmer andcolleagues47 as these factors have been found toalter the observed diagnostic performance. The 38 excluded studies are listed in Appendix 5.

Figure 1 shows a summary of study selection andexclusion.

Data extraction strategyData were extracted from papers by one researcherusing a standardised data extraction form. Non-English language papers were excluded from thereview.

Where available, the following data were reviewedin relation to MRCP and ERCP:

� study characteristics� suspected condition

� patient characteristics� diagnosis� sensitivity� specificity� likelihood ratios� positive and negative predictive values� accuracy� prevalence.

Quality assessment strategyThe studies were assessed using quality criteria for diagnostic or screening tests.48 The tenquestions outlined in Greenhalgh48 for assessingreports of diagnostic or screening tests are asfollow.

� Is the test potentially relevant?� Has the test been compared with a true gold

standard?� Did the validation study include an appropriate

spectrum of patients?� Has work-up bias been avoided?� Has expectation bias been avoided?� Was the test shown to be reproducible?� What are the features of the test as derived from

this validation study?� Were confidence intervals given?� Has a sensible ‘normal range’ been derived?� Has this test been placed in the context of other

potential tests in the diagnostic sequence?

Effectiveness

10

Potentially relevant studies identifiedand screened for retrieval

N = 1437

Total abstracts screenedN = 311

Studies rejected at titleN = 1126

Studies rejected at abstractN = 139

Rejected full papersN = 105

Studies excluded on thebasis of quality

N = 38

Total full papers screenedN = 172 (plus 4 from

other sources)

Studies potentially relevantN = 67

Included studiesN = 28 (plus 1 study of

patient satisfaction)

FIGURE 1 Summary of flow of study selection and exclusion: clinical effectiveness

Synthesis and presentation of resultsLikelihood ratios were not reported in any of the28 studies. These were therefore calculated usingstandard formulae.49 A likelihood ratio describeshow many times a person with disease is more likelyto have a particular test result than a person withoutdisease. Positive likelihood ratios are usually anumber greater than 1, while negative likelihoodratios are usually a number between 0 and 1.49 Asa guide, positive likelihood ratios greater than 10or negative likelihood ratios less than 0.1 provideconvincing diagnostic evidence. Those above 5and below 0.2 give strong diagnostic evidence, butthis will depend on pretest probability andcontext.49 Likelihood ratios combine sensitivityand specificity [sensitivity/(1 – specificity)].

Positive predictive value refers to the proportion ofthose with positive test results who have thedisease (number of true positives/total positive),while negative predictive value is the proportion ofthose with negative test results who do not havethe disease (number of true negatives/totalnegative). They describe the probabilities thatpositive or negative test results are correct. Theydepend on the prevalence of the disease in thesample, unlike sensitivity, specificity and likelihoodratios. Because disease prevalence is rarelyconstant across studies, positive and negativepredictive values are often associated with anunacceptably high level of heterogeneity, makingthem unsuitable choices for effect measures.49

Ninety-five per cent confidence intervals (95% CI)for the sensitivity, specificity, likelihood ratio for apositive test result and likelihood ratio for anegative test result were calculated for eachindividual study using the Wilson50 method, asrecommended by Altman and colleagues.51

The point estimates and 95% confidence intervalsfor the above summary statistics for selectedstudies were presented graphically as forest plots.The combining of the sensitivities, specificities andlikelihood ratios from studies of diagnosticaccuracy should only be applied in the absence ofvariability in the diagnostic threshold. This can beexamined graphically by plotting receiveroperating characteristic (ROC) curves andscatterplots of sensitivity versus specificity for thestudies. Calculation of the correlation coefficientbetween sensitivities and specificities will testwhether they are related, as would be the case ifthere were variation in the diagnostic threshold.Sensitivities and specificities for selected studieswere pooled using an approximation to theinverse variance approach as described in Deeks.49

Ninety-five per cent confidence intervals for thepooled sensitivity and specificity estimates werealso calculated. The homogeneity of thesensitivities and specificities for the selectedstudies was tested using a standard chi-squaredtest, as both measures are simple proportions.

To avoid computational problems withsensitivities, specificities and likelihood ratios, 0.5was added to all four cells in the 2 × 2 table if anyone of them was zero.

Likelihood ratios are ratios of probabilities, and ina meta-analysis can be treated as risk ratios. Aweighted average of the likelihood ratios can becomputed using the standard Mantel–Haenszelmethod of meta-analysis for risk ratios. Theheterogeneity of the likelihood ratios was alsotested by standard heterogeneity tests aftercombining the statistics in a meta-analysis.

If there is any evidence of heterogeneity or thatthe diagnostic threshold varies between thestudies, then the best summary of the results willbe an ROC curve rather than a single point.Littenberg and Moses52,53 proposed a method offitting a whole family of summary ROC curves thatallow for variation in diagnostic odds ratio withdiagnostic threshold. The Littenberg and Mosesmethod involves regression of the log diagnosticratio (D) on the measure of diagnostic threshold(S). Ordinary least squares linear regression wasused to produce estimates of the parameters a andb from the regression equation, D = a + bS. Theseestimates of a and b were used to calculate asummary ROC curve, which was plotted on agraph alongside the original data.

ResultsQuantity and quality of researchavailableIn total, 28 studies were identified that directlycompared MRCP with diagnostic ERCP (Table 1).In addition to these 28 studies, one study wasidentified that covered patient satisfaction;54 this iscovered later in this chapter (‘Patient satisfaction’,p. 14).

Study characteristicsAppendix 6, Table 15, outlines the studycharacteristics of the 28 included studies.

Sample selectionThe ideal sample in a study to compare diagnostictechniques is a consecutive or randomly selectedseries of patients recruited from a relevant clinical


11


population. Selection bias may be introduced byselecting patients for inclusion in a non-randommethod.49 Of the 28 studies, one reported a non-consecutive sample77 and in 13 studies themethod of sample selection was notreported.21,55–57,59,61,62,64,68,69,71,74,75 Barish andcolleagues58 and Soto and colleagues73 reportrandom selection of referrals, and Macaulay andcolleagues70 a sequential sample selection,although it is not clear whether or not thesequence was consecutive. Therefore, only 13studies used appropriate methods of sampleselection. The data for seven patients wereduplicated in two of the studies.58,73

Sample sizes of the studies also varied. Thesmallest reported sample size was 20 patients whohad MRCP62 and the largest reported sample sizewas 159 who had completed ERCP.77

ProceduresSome studies reported considerably more detailregarding the type of MRCP. Four studies reportedthe use of a non-breath-holding procedure.60,61,70,80

Only one study76 reported that all patients hadboth MRCP and ERCP. Six other studies reportedsome failed MRCP or MRCP not of diagnosticquality.6,58,72,74,78,79 Claustrophobia in patientsassociated with MRCP was reported in tenstudies.55,57,61,63,64,69,71,72,75,78 One study reportedthat one patient was unable to undergo MRCPowing to obesity.69

Failed ERCP procedures were reported in 12studies, with no indication as to whether or notsubsequent investigations wereattempted.6,55,59,60,63,64,65,69,71,74,75,6,77 In theremaining studies, other forms of directcholangiography (PTC, IOC, preoperativecholangiography) or surgery werereported.21,56–58,61,66,68,70,72,73,78–80 In two studies itwas not clear why some patients did not haveERCP.62,67

Time between proceduresThe time between MRCP and ERCP is important,particularly in the diagnosis of choledocholithiasis,as spontaneous stone fragmentation or movement

Effectiveness

12

TABLE 1 Included studies

Study, year; country Suspected condition Total no. of patients

Adamek et al., 199855; Germany CBD obstruction 86Alcaraz et al., 200056; Spain Obstruction of the biliary tree 81Angulo et al., 200057; USA Symptoms consistent with biliary disease, 73

emphasis on PSCBarish et al., 199558; USA Biliary or pancreatic disease 29Calvo et al., 200259; Spain Choledocholithiasis 61Chan et al., 199660; Hong Kong Choledocholithiasis 47Demartines et al., 200021; Switzerland Cholelithiasis 40Dwerryhouse et al., 199861; UK Choledocholithiasis 40Feldman et al., 199762; USA Pancreaticobiliary neoplasm 20Guibaud et al., 199563; Canada Bile duct obstruction 79Hintze et al., 199764; Germany Disorder affecting biliary or pancreatic duct system 78Holzknecht et al., 199865; Germany Not reported (planned ERCP) 61Laokpessi et al., 200166; France Choledocholithiasis 101Lee et al., 199767; South Korea Biliary disease 46Lomanto et al., 199768; Italy Choledocholithiasis, stenosis of bile ducts and 136

biliary–enteric anastomosisLomas et al., 199969; UK Biliary strictures or choledocholithiasis 76Macaulay et al., 199570; USA Biliary obstruction 29Regan et al., 199671; USA Choledocholithiasis 23Reinhold et al., 199872; Canada Bile duct obstruction 110Soto et al., 199673; USA Not reported (planned ERCP) 46Soto et al., 200074; Columbia Choledocholithiasis 57Soto et al., 200075; Columbia Choledocholithiasis 51Stiris et al., 200076; Norway Choledocholithiasis 50Sugiyama et al., 199877; Japan Pancreatobiliary disease 159Taylor et al., 200278; Australia Biliary tract disease in 96% (the other 4% had 146

pancreatic duct disease)Textor et al., 20026; Germany PSC 150Varghese et al., 200079; Ireland Not reported (planned ERCP) 191Zidi et al., 199980; France Choledocholithiasis 70

may result in some stones being diagnosed in theone procedure but not in the other. Ideally, thetime between procedures should be no more thana few hours. In five studies the time betweenprocedures was not reported.21,55,56,62,68 In only tenstudies did patients receive both procedures within24 hours of each other.57,58,60,64,69,71,73,76,78,80 Inone study70 ERCP was performed 109 days beforeMRCP in one patient.

Patient characteristicsPatient characteristics are shown in Appendix 6,Table 16. No information regarding the age ofpatients was reported in two studies.61,62 Threestudies did not report age ranges6,21,59 and sixstudies reported ranges that included children(age less than 17 years),55,63–65,72,77 with one,64

including at least one child aged 5 years. Thelowest reported mean was 48.6 years6 and thehighest was 71 years.80

Most studies reported more females than males,apart from five studies with more males thanfemales21,55,59,60,70 and one study with an equalnumber of males and females.69 Three studies didnot report the gender of patients.61,62,73

All studies apart from three65,73,79 reported thecondition under investigation. Suspectedconditions included obstruction in fivestudies,55,56,63,70,72 biliary disease or disorder in six studies,57,58,64,67,77,78 neoplasm in one,62 PSC inone6 and CBD stones or choledocholithiasis in 12 studies.21,59–61,66,68,69,71,74,75,76,80

Inclusion and exclusion criteria of the studies wereassessed and there was considerable variation inthe amount of detail provided in the studies.Three studies reported no inclusion criteria.77,79,80

Four studies stated only that patients referred forERCP were included58,65,73,78 and one studymentioned only that patients included were 18years of age or older.75 Twelve of the includedstudies did not report exclusioncriteria.6,56,60,62,64,68,70,71,73,76,77,80

Quality assessmentAll 28 studies included in the review had mostpatients receiving the gold standard diagnostictest, ERCP. Appendix 6, Table 17, shows thequality assessment criteria applied to each study. Itwas difficult to determine whether or not thestudies included an appropriate spectrum ofpatients. All studies included patients with biliarytree abnormalities of some description. Studiesreporting both pancreatic and biliary treeabnormalities were included if the biliary

abnormality results were reported separately. Asstated above, some studies did not providedetailed inclusion and exclusion criteria or statethe suspected condition, making it difficult toascertain exactly what type of patient was includedin the study.

Work-up bias refers to whether or not all patientshad both ERCP and MRCP. Ideally, all patientswho were selected should have had both tests. Nostudies where the MRCP results informed thedecision to proceed to ERCP were included in thisreview. ERCP is associated with a high failure rate,so it would be expected that some patients whowere included in a study would not have asuccessful ERCP. Failure rates of between 3 and 10%have been reported in the literature. With regardto MRCP, claustrophobia is a problem in somepatients; therefore, it would be expected that notall patients in every study would be able to haveMRCP. In only one study76 did all patients haveboth MRCP and ERCP. There is no mention ofERCP failures or claustrophobia in patients havingMRCP in this study. Therefore, work-up bias wasnot apparent in any of the included studies andinformation regarding number of patients fromthe original sample not receiving both tests ispresent if it was reported in the studies.

Expectation bias refers to whether or not blindedassessment was undertaken. Of the 28 studies, 25reported at least partial blinded assessment. Onestudy68 did not report any information onblinding, one study77 reported that blinding didnot take place and another study reported thatresults were assessed independently.69 Thirteenstudies reported that clinicians were blinded tofindings of other investigations as well as clinicalinformation,6,21,56–58,60,62,63,70,73,74,75,80 while theother studies either reported partial blinding (onlyof the other diagnostic test underinvestigation)55,59,72,79 or made no mention ofother tests.61,64,65–67,71,76,78

Most studies made no mention of the issue ofreproducibility, meaning that if the same personperforms the same test on two occasions on apatient whose characteristics remain the same theresults would not vary. It is also important toconfirm that the findings of two different peopleperforming the same test on the same patientwould be similar. This second question wasaddressed by six studies that reported kappavalues relating to agreement betweenreviewers.6,56,63,67,72,74 One study reported theimaging interpretation results for both an off-siteradiologist and an on-site radiologist.65


13


Features of the tests refer to sensitivity, specificity,likelihood ratios, and positive and negativepredictive values. These were calculated for allstudies where sufficient raw data were reported inthe studies. The results are reported in theOutcomes section (see below). Nine studiesreported confidence intervals.60,63,66,69,71,72,74,75,78

To place MRCP and ERCP results in context it isimportant to know what other tests were conductedand what the results of the tests were. Potentiallyimportant investigations include ultrasound, CT,clinical presentation and laboratory tests. Somestudies do not report the use of any othertests,56,58,62,65,67,70,73,74,77 whereas some studiesreport that other tests took place but do not givetheir results6,55,57,59–61,63,64,66,68,69,71,72,76,80 andothers report some of these test results.21,75,78,79

Inclusion and exclusion criteria were reportedearlier in this chapter. When little or noinformation regarding inclusion and exclusioncriteria is reported in studies, it is difficult toascertain whether or not an appropriate spectrumof patients has been included in the study. It isalso difficult to determine whether or not studiesare comparable, as different patient groups mayhave been included.

Diagnoses and reported sensitivities andspecificitiesIn Appendix 6, Table 18, the final diagnosesreported in the studies, as well as sensitivity,specificity and adverse events reported in thestudies, are shown. Although most studies providedetails of the diagnosis of most patients takingpart in the study, five report only the number ofpatients with the diagnosis under investigation,giving no indication of the other diagnosesobtained.21,61,68,75,76

If sensitivities and specificities were not reportedin the studies, they were calculated provided theraw data were published in the study. Calculatedvalues are reported in parentheses in Appendix 6,Table 18. No studies reported likelihood ratios.These have been calculated and are reportedbelow (Outcomes).

ComparatorsSome studies calculated sensitivities andspecificities in comparison to ERCP or equivalent(PTC, IOC and surgery) and some to finaldiagnosis. Information on final diagnosisdefinition for the studies is given in Appendix 6,Table 18. Seven studies compared MRCP withfinal diagnosis6,55,62,66,67,70,71 and of these three

also reported sensitivities and specificities forERCP compared with final diagnosis.66,67,71 In all cases ERCP formed the major component offinal diagnosis, apart from one.6 The remaining21 studies compared MRCP with ERCP orequivalent.21,56–61,63–65,68,69,72–80 Zidi andcolleagues80 included sonography as a comparatoras well as direct cholangiography and surgery.

For the three studies comparing ERCP with finaldiagnosis as well as MRCP, the results are asfollows. Lee and colleagues67 reported ERCP tohave a sensitivity of 71% and a specificity of 92%for malignancy. Laokpessi and colleagues66

reported ERCP to have a sensitivity of 95% and aspecificity of 100% for CBD stones. Finally, Reganand colleagues71 reported ERCP to have both asensitivity and a specificity of 100% for CBD stones.

Adverse effectsAdverse effects are reported in Appendix 6, Table 18. None of the 28 studies reported anyadverse effects associated with MRCP. Six studiesreported adverse effects associated withERCP.6,55,57,59,72,76 Two studies reported that noadverse events occurred21,56 and the remaining 20 studies did not report any informationregarding adverse effects. None of the 28 studiesreported mortality associated with ERCP.

OutcomesTables 2–5 show the sensitivity, specificity withconfidence intervals, likelihood ratios withconfidence intervals, and positive and negativepredictive values, prevalence and accuracy. Threestudies were not included in these calculations asthey did not provide sufficient data for thenecessary calculations56,62,64 The remaining 25 studies are listed in the tables. If sufficient datawere provided to calculate sensitivities andspecificities, these are presented for everycondition for which the data were provided.

Holzknecht,65 Macaulay,70 Reinhold72 and Sotoand colleagues74 provided results for differentreviewers (but the same patients). Soto75 alsoprovided data on the same patients using differentMRCP techniques. These data were combined anda mean calculated for each study to generate anoverall value for sensitivity and specificity bycondition for each individual study where the rawdata were provided.

Patient satisfactionThe 28 studies included in this review did notreport information on patient satisfaction.However, one study was identified that dealt with

Effectiveness

14

patient satisfaction only.54 The study did notattempt to define clinical outcome of performanceof MRCP versus ERCP. This study recruited 34patients who were to undergo ERCP. MRCP was tobe performed before ERCP. However, two patientswere unable to undergo MRCP owing toclaustrophobia, leaving 32 patients who underwentboth procedures.

Patients were asked to complete validatedquestionnaires using a series of seven-point Likertscales to measure the degree of anxiety, pain,discomfort, expectations and willingness to repeateach test. The first set of postprocedure Likertscales assessed each test separately withoutreference to the other test. No significant

differences in degree of anxiety were reportedbetween MRCP and ERCP. Patients reported asignificantly lower degree of discomfort withMRCP than with ERCP (2.47 ± 1.6 vs 3.09 ± 1.7,respectively; p = 0.047) and a lower degree ofpain (1.3 ± 0.8 vs 2.7 ± 1.8; p < 0.001). MRCPwas found to be ‘more difficult than expected’ to agreater degree than was ERCP (–0.7 ± 1.5 vs –1.3 ± 1.5; p = 0.012). Patients were equallywilling to repeat either procedure, although thetrend favoured MRCP (p = 0.09).

In the second set of Likert scales the questionsmade a direct comparison between MRCP andERCP. With regard to anxiety, pain anddiscomfort, patients had a greater satisfaction after


15


TABLE 2 Sensitivity

Study Condition Sensitivity Lower CI Upper CI

Adamek55 Abnormality 0.89 0.77 0.95Adamek55 Malignancy 0.81 0.63 0.92Angulo57 Normal 0.86 0.67 0.95Angulo57 Dilatation CBD 0.93 0.81 0.97Angulo57 Obstruction 1.00 0.91 1.00Angulo57 CBD stones 0.50 0.24 0.76Angulo57 PSC 0.83 0.63 0.93Barish58 Dilatation 0.87 0.62 0.96Calvo59 Choledocholithiasis 0.91 0.76 0.97Chan60 Choledocholithiasis 0.95 0.75 0.99Demartines21 CBD stones 1.00 0.83 1.00Dwerryhouse61 CBD stones 0.88 0.53 0.98Guibaud63 Obstruction 0.91 0.83 0.96Guibaud63 Choledocholithiasis 0.81 0.65 0.91Guibaud63 Malignancy 0.86 0.60 0.96Holzknecht65 Choledocholithiasis 0.86 0.60 0.96Holzknecht65 Dilatation 0.94 0.81 0.98Holzknecht65 Stenosis 0.86 0.71 0.94Holzknecht65 Overall 0.91 0.80 0.97Laokpessi66 CBD stones 0.93 0.87 0.96Lee67 Malignancy 0.81 0.60 0.92Lomanto68 Choledocholithiasis 0.92 0.74 0.98Lomas69 Choledocholithiasis 1.00 0.70 1.00Lomas69 Stricture 1.00 0.83 1.00Macaulay70 Dilatation 1.00 0.82 1.00Regan71 Choledocholithiasis 0.93 0.70 0.99Reinhold72 Obstruction 0.91 0.82 0.95Reinhold72 Choledocholithiasis 0.90 0.74 0.97Soto73 Dilatation 0.96 0.82 0.99Soto74 Choledocholithiasis 0.96 0.80 0.99Soto75 Choledocholithiasis 0.96 0.81 0.99Stiris76 CBD stones 0.88 0.72 0.95Sugiyama77 Anomalous PBJ 0.82 0.52 0.95Taylor78 Choledocholithiasis 0.98 0.89 1.00Taylor78 Stricture 1.00 0.76 1.00Textor6 PSC 0.88 0.73 0.95Varghese79 Choledocholithiasis 0.91 0.77 0.97Zidi80 Choledocholithiasis 0.57 0.43 0.70

PBJ, pancreatobiliary junction.

MRCP, although discomfort associated with ERCPcompared with MRCP did not reach statisticalsignificance. Scores were 0.6 (95% CI 0.02 to 1.2),0.9 (0.5 to 1.3) and 0.4 (–1.2 to 1.0), respectively.Two patients expressed no preference betweenMRCP and ERCP, whereas 19 (59%, 95% CI41–76%) preferred MRCP to ERCP. Using theLikert scales patients showed a significantly higherpreference for MRCP than for ERCP (mean score= 0.8, 95% CI 0.2 to 1.4; p = 0.01).

Several subgroup analyses were performed. Thefirst was in patients who did not complain ofclaustrophobia or noise (n = 15; 46.8%). Theresults were more striking in this subgroup withoutclaustrophobia. In the subgroup undergoing purelydiagnostic ERCP there were clear preferences for

MRCP. In this study, MRCP was always performedfirst and this may have influenced the results. Onthe whole patients preferred MRCP over ERCP,although there was still some patients whopreferred ERCP to MRCP. They suggest ways toovercome problems associated with MRCP,including better patient selection, fenestratedscanners, earplugs and selective sedation.

In addition to the above study, one of the 28studies described above66 reported that 4% ofpatients in that study found the noise of theMRCP investigation to be disturbing.

Assessment of effectivenessA total of 28 studies was included in thissystematic review. Of these, 25 provided enough

Effectiveness

16

TABLE 3 Specificity

Study Condition Specificity Lower CI Upper CI

Adamek55 Abnormality 0.92 0.67 0.99Adamek55 Malignancy 1.00 0.90 1.00Angulo57 Normal 0.96 0.86 0.99Angulo57 Dilatation CBD 0.93 0.78 0.98Angulo57 Obstruction 0.91 0.76 0.97Angulo57 CBD stones 0.98 0.91 1.00Angulo57 PSC 0.98 0.89 1.00Barish58 Dilatation 1.00 0.61 1.00Calvo59 Choledocholithiasis 0.83 0.55 0.95Chan60 Choledocholithiasis 0.85 0.66 0.94Demartines21 CBD stones 0.90 0.71 0.97Dwerryhouse61 CBD stones 0.93 0.79 0.98Guibaud63 Obstruction 1.00 0.92 1.00Guibaud63 Choledocholithiasis 0.98 0.93 0.99Guibaud63 Malignancy 0.98 0.94 1.00Holzknecht65 Choledocholithiasis 0.94 0.83 0.98Holzknecht65 Dilatation 0.93 0.77 0.98Holzknecht65 Stenosis 0.88 0.70 0.96Holzknecht65 Overall 0.80 0.55 0.93Laokpessi66 CBD stones 1.00 0.90 1.00Lee67 Malignancy 0.92 0.75 0.98Lomanto68 Choledocholithiasis 1.00 0.91 1.00Lomas69 Choledocholithiasis 0.97 0.89 0.99Lomas69 Stricture 0.98 0.90 1.00Macaulay70 Dilatation 0.91 0.62 0.98Regan71 Choledocholithiasis 0.88 0.53 0.98Reinhold72 Obstruction 1.00 0.90 1.00Reinhold72 Choledocholithiasis 0.96 0.90 0.99Soto, 199673 Dilatation 0.94 0.73 0.99Soto74 Choledocholithiasis 0.96 0.80 0.99Soto75 Choledocholithiasis 1.00 0.87 1.00Stiris76 CBD stones 0.94 0.74 0.99Sugiyama77 Anomalous PBJ 1.00 0.97 1.00Taylor78 Choledocholithiasis 0.89 0.81 0.94Taylor78 Stricture 0.99 0.95 1.00Textor6 PSC 0.99 0.95 1.00Varghese79 Choledocholithiasis 0.98 0.95 0.99Zidi80 Choledocholithiasis 1.00 0.85 1.00

data to calculate sensitivities, specificities,likelihood ratios and confidence intervals. Thesensitivities and specificities are presented here bycondition.

Assessment of effectiveness by conditionCholedocholithiasisThere were 18 studies investigatingcholedocholithiasis or CBD stones. The data from15 of these studies are presented graphically inFigures 2–12. Three of the original 18 studies withresults for choledocholithiasis were not included inthis analysis, two owing to lack of blindedassessment68,69 and one because no information

was provided regarding the age and gender ofpatients.61

Figure 2 shows the ROC curve for the 15 studies.Two studies (Zidi80 and Angulo57) stand out ashaving sensitivities somewhat lower than the other13 studies.

Figure 3 shows a scatterplot of sensitivity versusspecificity for the 15 studies. There does notappear to be any correlation between theseestimates. One can crudely test for a diagnosticthreshold effect by computing the correlationbetween sensitivities and specificities across the


17


TABLE 4 Likelihood ratios

Study Condition LRP LRP lower LRP upper LRN LRN lower LRN upper CI CI CIa CIa

Adamek55 Abnormality 11.62 1.76 76.57 0.12 0.02 0.76Adamek55 Malignancy 54.64 3.47 861.18 0.20 0.01 3.14Angulo57 Normal 20.73 5.28 81.31 0.14 0.04 0.56Angulo57 Dilatation CBD 13.44 3.52 51.33 0.08 0.02 0.30Angulo57 Obstruction 11.00 3.74 32.36 0.00 – –Angulo57 CBD stones 30.00 3.90 230.73 0.51 0.07 3.91Angulo57 PSC 38.83 5.53 272.37 0.18 0.03 1.25Barish58 Dilatation 11.81 0.81 172.17 0.17 0.01 2.45Calvo59 Choledocholithiasis 5.44 1.53 19.36 0.11 0.03 0.40Chan60 Choledocholithiasis 6.16 2.48 15.26 0.06 0.03 0.15Demartines21 CBD stones 10.50 2.81 39.24 0.00 – –Dwerryhouse61 CBD stones 13.13 3.35 51.36 0.1 0.03 0.52Guibaud63 Obstruction 87.00 5.52 1372.23 0.09 0.01 1.49Guibaud63 Choledocholithiasis 38.19 9.60 151.97 0.19 0.05 0.76Guibaud63 Malignancy 48.00 11.96 192.72 0.15 0.04 0.58Holzknecht65 Choledocholithiasis 14.00 4.58 42.78 0.15 0.05 0.47Holzknecht65 Dilatation 13.18 3.46 50.23 0.06 0.02 0.24Holzknecht65 Stenosis 7.18 2.46 20.91 0.16 0.05 0.46Holzknecht65 Overall 4.57 1.66 12.63 0.11 0.04 0.29Laokpessi66 CBD stones 64.78 4.13 1015.86 0.08 0.00 1.19Lee67 Malignancy 10.12 2.64 38.86 0.21 0.05 0.79Lomanto68 Choledocholithiasis 70.20 4.46 1105.97 0.10 0.01 1.60Lomas69 Choledocholithiasis 30.00 7.68 117.19 0.00 – –Lomas69 Stricture 50.00 7.18 348.04 0.00 – –Macaulay70 Dilatation 11.00 1.70 71.28 0.00 – –Regan71 Choledocholithiasis 7.47 1.19 46.94 0.08 0.01 0.48Reinhold72 Obstruction 63.18 4.03 991.27 0.10 0.01 1.55Reinhold72 Choledocholithiasis 24.00 7.86 73.31 0.10 0.03 0.32Soto73 Dilatation 16.37 2.44 109.77 0.04 0.01 0.26Soto74 Choledocholithiasis 23.96 3.50 163.78 0.04 0.01 0.30Soto75 Choledocholithiasis 49.11 3.15 765.62 0.06 0.00 0.88Stiris76 CBD stones 15.75 2.33 106.28 0.13 0.02 0.89Sugiyama77 Anomalous PBJ 235.92 14.60 3811.82 0.21 0.01 3.38Taylor78 Choledocholithiasis 9.02 4.86 16.74 0.02 0.01 0.05Taylor78 Stricture 118.00 16.76 830.78 0.00 – –Textor6 PSC 95.79 13.56 676.70 0.12 0.02 0.86Varghese79 Choledocholithiasis 47.72 15.48 147.06 0.09 0.03 0.28Zidi80 Choledocholithiasis 25.08 1.60 392.61 0.44 0.03 6.89

a With likelihood ratios of zero, the data are unsuitable for the calculation of 95% CI.LRP, likelihood ratio positive; LRN, likelihood ratio negative.

15 studies. The estimated correlation was –0.43,which was not statistically significant (p = 0.11).

Sensitivities and specificities along with 95% CI forthese estimates for the 15 studies are presented inFigures 4 and 5. From the graphs all of theconfidence intervals overlap, although theconfidence intervals for some studies are wide,reflecting the small sample sizes and variability inthe sensitivity estimates. Two studies, Angulo57 andZidi,80 have point estimates (Figure 4) of sensitivitythat are clearly different from the other 13 studies,suggesting, as with the ROC plot (Figure 2), thatthese studies may be outliers.

SensitivityThe overall estimate of mean sensitivity is 0.87

(95% CI: 0.85 to 0.91) from the 15 studies.However, a chi-squared test confirms the statisticalsignificance of the heterogeneity observed inFigure 4 (�2 = 70.8.8, df = 14, p < 0.0001). Thelarge between-study heterogeneity is clearlyevident in Figure 6, with the results of two studies(Angulo57 and Zidi80) lying some distance from thesummary sensitivity estimate of 0.87. In such asituation it is probably inappropriate to considerpooling sensitivities at all, and it may be better tonote the heterogeneity by describing the mediansensitivity (0.91) and the range (0.50 to 1.00)between which the sensitivities are seen to vary.

If the Angelo and Zidi studies are excluded thenthe overall pooled estimate across the 13 studiesnow becomes 0.92 (95% CI 0.89 to 0.95). The

Effectiveness

18

TABLE 5 Predictive values, accuracy and prevalence

Study Condition Positive predictive Negative predictive Accuracy Prevalencevalue value

Adamek55 Abnormality 0.98 0.71 0.90 0.78Adamek55 Malignancy 1.00 0.87 0.92 0.45Angulo57 Normal 0.90 0.94 0.93 0.31Angulo57 Dilatation CBD 0.95 0.90 0.93 0.59Angulo57 Obstruction 0.93 1.00 0.96 0.53Angulo57 CBD stones 0.83 0.92 0.91 0.14Angulo57 PSC 0.95 0.92 0.93 0.33Barish58 Dilatation 1.00 0.75 0.90 0.71Calvo59 Choledocholithiasis 0.94 0.77 0.89 0.73Chan60 Choledocholithiasis 0.82 0.96 0.89 0.42Demartines21 CBD stones 0.90 1.00 0.95 0.48Dwerryhouse61 CBD stones 0.78 0.97 0.92 0.21Guibaud63 Obstruction 1.00 0.87 0.94 0.63Guibaud63 Choledocholithiasis 0.93 0.94 0.94 0.25Guibaud63 Malignancy 0.86 0.98 0.97 0.11Holzknecht65 Choledocholithiasis 0.80 0.96 0.92 0.22Holzknecht65 Dilatation 0.94 0.93 0.94 0.55Holzknecht65 Stenosis 0.91 0.81 0.87 0.59Holzknecht65 Overall 0.93 0.75 0.89 0.76Laokpessi66 CBD stones 1.00 0.81 0.95 0.77Lee67 Malignancy 0.89 0.85 0.87 0.46Lomanto68 Choledocholithiasis 1.00 0.95 0.97 0.39Lomas69 Choledocholithiasis 0.82 1.00 0.97 0.13Lomas69 Stricture 0.95 1.00 0.99 0.28Macaulay70 Dilitation 0.95 1.00 0.97 0.62Regan71 Choledocholithiasis 0.93 0.88 0.91 0.65Reinhold72 Obstruction 1.00 0.83 0.94 0.69Reinhold72 Choledocholithiasis 0.90 0.96 0.95 0.27Soto73 Dilatation 0.96 0.94 0.95 0.61Soto74 Choledocholithiasis 0.96 0.96 0.96 0.49Soto75 Choledocholithiasis 1.00 0.96 0.98 0.51Stiris76 CBD stones 0.97 0.81 0.90 0.64Sugiyama77 Anomalous PBJ 1.00 0.99 0.99 0.07Taylor78 Choledocholithiasis 0.83 0.99 0.92 0.36Taylor78 Stricture 0.92 1.00 1.00 0.09Textor6 PSC 0.97 0.96 0.96 0.23Varghese79 Choledocholithiasis 0.91 0.98 0.97 0.18Zidi80 Choledocholithiasis 1.00 0.50 0.70 0.70


19


1 Angulo57 9 Reinhold72

2 Calvo59 10 Soto74

3 Chan60 11 Soto75

4 Demartines21 12 Stiris76

5 Guibaud63 13 Taylor78

6 Holzknecht65 14 Varghese79

7 Laokpessi66 15 Zidi80

8 Regan71

15

1

15

14

119

12 6

13 48 3

2

107

0.000.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

0.02 0.04 0.06 0.08

1 – specificity (false-positive rate)

Sens

itivi

ty (t

rue-

posit

ive

rate

)

0.10 0.12 0.14 0.16 0.18

FIGURE 2 ROC curve for 15 studies

Specificity1.000.950.900.850.80

Sens

itivi

ty

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

15

14

13

12

1110

98

76

5

32

1

1 Angulo57 9 Reinhold72

2 Calvo59 10 Soto74

3 Chan60 11 Soto75

4 Demartines21 12 Stiris76

5 Guibaud63 13 Taylor78

6 Holzknecht65 14 Varghese79

7 Laokpessi66 15 Zidi80

8 Regan71

4

FIGURE 3 Scatterplot of sensitivities versus specificities (n = 15, r = –0.43, p = 0.11)

Effectiveness

20

0.0 0.2 0.4 0.6Sensitivity

Stud

y

0.8 1.0 1.2

Zidi80

Varghese79

Taylor78

Stiris76

Soto75

Soto74

Reinhold72

Regan71

Laokpessi66

Holzknecht65

Guibaud63

Demartines21

Chan60

Calvo59

Angulo57

Sensitivity

95% CI

Lower limit

Upper limit

FIGURE 4 Forest plot of estimated sensitivities (n = 15)

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

Sensitivity

Stud

y

Zidi80

Varghese79

Taylor78

Stiris76

Soto75

Soto74

Reinhold72

Regan71

Laokpessi66

Holzknecht65

Guibaud63

Demartines21

Chan60

Calvo59

Angulo57

Sensitivity

95% CI

Lower limit

Upper limit

FIGURE 5 Forest plot of estimated specificities (n = 15)

sensitivities for these 13 studies now show noevidence of significant heterogeneity (�2 = 10.7,df = 12, p =0.55). The median sensitivity for the13 studies is 0.93, with a range 0.81 to 1.00.

SpecificityFollowing the same calculations as for thesensitivities, the overall estimate of meanspecificity is 0.95 (95% CI 0.94 to 0.97) from the15 studies. However, a chi-squared test confirmsthe statistical significance of the heterogeneityobserved in Figure 7 (�2 = 27.8, df = 14, p < 0.015). The large between-study heterogeneityis clearly evident in Figure 7, with the results ofmany studies lying some distance from thesummary specificity estimate of 0.95. In such asituation it is probably inappropriate to considerpooling specificities at all, and it may be better tonote the heterogeneity by describing the medianspecificity (0.96) and the range (0.83 to 0.99)between which the specificities are seen to vary.

As before, if the Angulo and Zidi studies areexcluded then the overall estimate of meanspecificity is 0.95 (95% CI 0.93 to 0.97) from the13 studies. However, a chi-squared test confirmsthe statistical significance of the heterogeneity (�2 = 24.3, df = 12, p < 0.018). The medianspecificity for the 13 studies is 0.94 and the range,as before, is 0.83 to 0.99.

Likelihood ratiosThe positive likelihood ratios show no evidence ofsignificant heterogeneity (�2 = 19.6, df = 14, p = 0.14), the Mantel–Haenszel pooled estimatefor 15 studies being 16.3 (95% CI 11.5 to 23.2)(Figure 8). However, owing to the hetereogeneityreported for both sensitivities and specificitiesabove it is probably more appropriate to reportthe median values for positive likelihood ratios.The median positive likelihood ratio is 23.96 withthe range between 5.44 and 64.78.

Combining negative likelihood ratios for the 15studies using the Mantel–Haenszel method yieldsan overall estimate of 0.13 (95% CI 0.10 to 0.16)(Figure 9). However, as with the sensitivities andspecificities, there is significant heterogeneity innegative likelihood ratios between studies (�2 = 86.4, df = 14, p < 0.001). The estimate of0.13 is outside the 95% confidence intervals fortwo of the 15 studies (Angulo and Zidi). Themedian negative likelihood ratio for the 15 studiesis 0.09, with a range of 0.00 to 0.51.

As before, if the Angelo and Zidi studies areexcluded from the analysis, then the positivelikelihood ratios show no evidence of significantheterogeneity (�2 = 18.8, df = 12, p =0.093), theMantel–Haenszel pooled estimate for 13 studiesbeing 15.9 (95% CI 11.2 to 22.7) (Figure 10). The


21


Sensitivity

95% CI

Lower limit

Upper limit

0.2 0.4 0.6 0.8 1.0 1 .2Sensitivity

Stud

y

Zidi80

Varghese79

Taylor78

Stiris76

Soto75

Soto74

Reinhold72

Regan71

Laokpessi66

Holzknecht65

Guibaud63

Demartines21

Chan60

Calvo59

Angulo57

FIGURE 6 Forest plot of estimated sensitivities with summary estimate (n = 15)

Effectiveness

22

Study

Zidi80

Overall (95% CI)

Varghese79

Taylor78

Stiris76

Soto75Soto74

Reinhold72Regan71

Laokpessi66

Holzknecht65Guibaud63

Demartines21

Chan60

Calvo59

Angulo57

Risk ratio (95% CI)

25.08 (1.60 to 392.59)

16.33 (11.50 to 23.19)

47.72 (15.48 to 147.06) 9.02 (4.86 to 16.74)15.75 (2.33 to 106.28)49.11 (3.15 to 765.58)23.96 (3.50 to 163.77)24.00 (7.86 to 73.30) 7.47 (1.19 to 46.94)64.78 (4.13 to 1015.81)14.00 (4.58 to 42.78)38.19 (9.60 to 151.97)10.50 (2.81 to 39.24) 6.16 (2.48 to 15.26) 5.44 (1.53 to 19.36)30.00 (3.90 to 230.72)

% Weight

2.74.2

25.25.02.03.86.4

5.13.05.24.07.5

13.311.4

1.1

0.000984 1Risk ratio

1015.81

FIGURE 8 Forest plot of estimated likelihood ratios for a positive test result (n = 15)

Stud

y

Zidi80

Varghese79

Taylor78

Stiris76

Soto75

Soto74

Reinhold72

Regan71

Laokpessi66

Holzknecht65

Guibaud63

Demartines21

Chan60

Calvo59

Angulo57

Sensitivity

95% CI

Lower limit

Upper limit

0.5 0.6 0.7 0.8Specificity

0.9 1.0 1.1

FIGURE 7 Forest plot of estimated specificities with summary estimate (n = 15)


23


Study

Zidi80

Overall (95% CI)

Varghese79

Taylor78

Stiris76

Soto75

Soto74

Reinhold72

Regan71

Laokpessi66

Holzknecht65Guibaud63

Demartines21

Chan60Calvo59

Angulo57

Risk ratio (95% CI)

0.43 (0.31 to 0.59)

0.13 (0.10 to 0.16)

0.09 (0.03 to 0.27)0.02 (0.00 to 0.17)0.13 (0.05 to 0.33)0.04 (0.01 to 0.26)0.04 (0.01 to 0.30)0.10 (0.04 to 0.30)0.08 (0.01 to 0.52)0.77 (0.04 to 0.14)0.15 (0.04 to 0.55)0.19 (0.00 to 0.39)0.03 (0.00 to 0.44)0.06 (0.01 to 0.42)0.11 (0.04 to 0.34)0.51 (0.27 to 0.95)

% Weight

6.612.311.8

4.95.75.39.42.0

11.74.6

10.54.24.23.33.8


549.648

FIGURE 9 Forest plot of estimated likelihood ratios for a negative test result (n = 15)

Study

Overall (95% CI)

Varghese79

Taylor78

Stiris76

Soto75

Soto74

Reinhold72

Regan71

Laokpessi66

Holzknecht65

Guibaud63

Demartines21

Chan60

Calvo59

Risk ratio (95% CI)

15.92 (11.17 to 22.70)

47.72 (15.48 to 147.06)

9.02 (4.86 to 16.74)15.75 (2.33 to 106.28)49.11 (3.15 to 765.58)23.96 (3.50 to 163.77)

24.00 (7.86 to 73.30) 7.47 (1.19 to 46.94)

64.78 (4.13 to 1015.81)14.00 (4.58 to 42.78)38.19 (9.60 to 151.97)10.50 (2.81 to 39.24)

6.16 (2.48 to 15.26)

5.44 (1.53 to 19.36)

% Weight

4.4

26.25.22.14.0

6.75.3

3.15.44.17.8

13.8

11.9


1015.81

FIGURE 10 Forest plot of likelihood ratios for a positive test result (n =13)

median positive likelihood ratio for the 13 studiesis 15.75, with a range of 5.44 to 64.78. Similarly,the negative likelihood ratios show no evidence ofsignificant heterogeneity (�2 = 10.3, df = 12, p = 0.587), the Mantel–Haenszel pooled estimatefor 13 studies being 0.09 (95% CI 0.06 to 0.12)(Figure 11). The median negative likelihood ratiofor the 13 studies was 0.08, with a range between0.00 and 0.19.

Differences between studies in patient groups, testexecution and study design can introducevariability in diagnostic thresholds. Finally,because of the suspected heterogeneity in thestudies (even after excluding the Angulo and Zidi

studies) and the suspected variation in diagnosticthreshold, it was felt appropriate to calculate asummary ROC curve for the studies.

Summary ROC curvesRegression of the log diagnostic odds ratio (D) onthe measure of diagnostic threshold (S) for the 13studies produces estimates of the parameters aand b from the regression equation, D = a + bS.The parameter estimates are reported in Table 6.

The r2 for the regression model is 0.095 and theoverall F-statistic (df 1,11) is 1.15 (p = 0.31). Thenon-significant result for the S coefficient (p = 0.306) suggests that there is no reliable

Effectiveness

24

Study

Overall (95% CI)

Varghese79

Taylor78

Stiris76

Soto75

Soto74

Reinhold72

Regan71

Laokpessi66

Holzknecht65

Guibaud63

Demartines21

Chan60

Calvo59

Risk ratio (95% CI)

0.09 (0.06 to 0.12)

0.09 (0.03 to 0.27)

0.02 (0.00 to 0.17)0.13 (0.05 to 0.33)0.04 (0.01 to 0.26)0.04 (0.01 to 0.30)

0.10 (0.04 to 0.30)0.08 (0.01 to 0.52)

0.07 (0.04 to 0.14)0.15 (0.04 to 0.55)0.19 (0.09 to 0.39)0.03 (0.00 to 0.44)

0.06 (0.01 to 0.42)

0.11 (0.04 to 0.34)

% Weight

13.7

13.25.46.45.9

10.52.3

13.05.1

11.74.6

4.6

3.6


549.648

FIGURE 11 Forest plot of likelihood ratios for a negative test result (n =13)

TABLE 6 Parameter estimates for Littenberg–Moses summary ROC curve (n = 13 studies)

Coefficienta

Unstandardised coefficients 95% CI for B

Model B SE Sig. Lower bound Upper bound

1 (Constant) 5.386 0.264 20.437 0.000 4.806 5.966S –0.213 0.199 –1.073 0.306 –0.650 0.224

a Dependent variable: log (diagnostic odds ratio).

statistical evidence that the diagnostic odds ratiochanges with threshold (p = 0.306). A similaranalysis including the two previously excludedstudies with high specificities and low sensitivitiesgave very similar results.

Figure 12 shows the results of the Littenberg–Mosesmethod for the estimation of a summary ROCcurve. However, there is no unique joint summaryestimate of sensitivity and specificity from this plot.It is only possible to obtain a summary estimate ofone value conditional on the value of the other. Tointerpret both diagnostic odds ratios and summaryROC curves it is necessary to have some knowledgeof either the sensitivity or the specificity of the testin the population to which it is to be applied. TheLittenberg–Moses curve shows the relationshipbetween sensitivities and specificities across the 13studies. The ROC curve presentation is unlikely tobe useful for individual clinicians in practice as,without knowledge of the diagnostic thresholdbeing applied, it is impossible to know where onthe ROC curve an individual is operating. For thepurposes of an economic analysis of the technologyacross a broad population of clinicians, however,the ROC curve, if it can be generated, would bethe most appropriate method of analysis.

However, the results from the linear regressionsuggest that there is no statistically significantevidence that the diagnostic odds ratio changeswith threshold. Furthermore the fact that theoverall fit of the model is very poor, with only 10%of the variability being explained by the measure of

threshold, implies that there is insufficient evidencewithin this data set to quantify the expectedrelationship between sensitivity and specificity.

MalignancyThree studies presented sensitivities andsensitivities for malignancy.55,63,67 Adamek andcolleagues55 reported malignant strictures in 27patients due to pancreatic carcinoma, ampullarycarcinoma or cholangiocarcinoma. Guibaud andcolleagues63 reported biliary malignant biliaryobstruction in 14 patients, with six due toampullary tumours, four to pancreatic cancers,three to cholangiocarcinomas and one to externalcompression from metastasis. Lee and colleagues67

did not report the type of malignancy, butpresented sensitivity and specificity fordistinguishing malignant from benign lesions.Sensitivities ranged from 81% for Adamek55 andLee67 to 86% for Guibaud.63 Specificities rangedfrom 92% for Lee67 to 100% for Adamek.55

Although raw data were not provided in the studyby Feldman and co-workers,62 the reportedsensitivity for malignancy in this study was 94.4%.

DilatationFive studies present figures for dilatation,57,58,65,70,73

with sensitivities ranging from 87%58 to 100%70

and specificities ranging from 91%70 to 100%.58

ObstructionThree studies present data on obstruction,57,63,72

with sensitivities ranging from 91%63,72 to 100%57

and specificities ranging from 91%57 to 100%.63,72


25


0 0.05 0.1False-positive rate (1 – specificity)

True

-pos

itive

rat

e (s

ensit

ivity

)

0.15 0.20.8

0.85

0.9

0.95

1

Littenberg–Moses methodD = 5.386 – 0.213S

FIGURE 12 Littenberg–Moses summary ROC curve and actual data for 13 studies

StrictureTwo studies present data on strictures, with bothreporting a sensitivity of 100%69,78 and specificitiesof 98%69 and 99%.78

PSCTwo studies present data on PSC.6,57 Sensitivityranges from 83%57 to 88%6 and specificity rangesfrom 98%57 to 99%.6

The remaining studies present data for anyabnormality55 (sensitivity 89%, specificity 92%),stenosis65 (sensitivity 86%, specificity 88%),anomalous PBJ77 (sensitivity 82%, specificity100%) and overall figures65 (sensitivity 91%,specificity 80%).

Likelihood ratiosLikelihood ratios [sensitivity/(1 – specificity) for apositive test and (1 – sensitivity)/specificity for anegative test] describe how many times a personwith disease is more likely to have a particular testresult compared with a person without the disease.Positive likelihood ratios greater than 5 andnegative likelihood ratios less than 0.2 give strongdiagnostic evidence that a test is accurate. In the28 studies in this review, which included 38subgroups, only one positive likelihood ratio wasless than 5.65 This value was based on the meandata from two reviewers. In total, four negativelikelihood ratios in the 38 subgroups were greaterthan 0.2.57,67,77,80 These included one studylooking at a subgroup of CBD stones, althoughthe main objective of the study was to investigatePSC,57 one study investigating patients withsuspected malignancy,67 one study that involvedsonography as a comparator as well as diagnosticERCP,80 and one study investigating diagnosis ofan anomalous PBJ.77

Factors affecting resultsSeveral factors may have influenced the results inthese studies. First, comparisons were made withERCP in 21 studies and with final diagnosis inseven studies, making comparisons between allstudies difficult. ERCP is not a perfect goldstandard, so differences in diagnosis betweenMRCP and ERCP may be due not to MRCP giving an incorrect result but rather to ERCPgiving an incorrect result. Reporting in the

studies was poor as varying definitions for finaldiagnosis were given and it was not entirely clearthat those studies stating ERCP as the comparator did not incorporate other test results as well.

Another factor affecting the results is the date ofthe investigations. MRCP techniques arecontinually improving. Studies taking place in1995–8 may have used less accurate techniquesthan those undertaken more recently.

The quality of the studies was moderate and thiswould have a potential impact on the results. In all but one study, selected patients did not haveboth MRCP and diagnostic ERCP, and often thereasons why not were unclear. Only 13 of the 28studies reported adequate blinding and only six of the 28 studies reported information onagreement of MRCP results by more than oneinvestigator. Nine studies gave no information onother diagnostic tests and most studies did notadequately report inclusion and exclusion criteria.

Summary and conclusionsThe median sensitivity for choledocholithiasis (n = 13 studies) was 93% (range 81–100%) andspecificity 94% (range 83–99%). The medianlikelihood ratio for a positive value was 15.75 andfor a negative value 0.08. For malignancy reported sensitivities were somewhat lower,ranging from 81 to 86%, and specificities rangedfrom 92 to 100%. All positive likelihood ratios,apart from one, were greater than 5, and allnegative likelihood ratios, apart from four, wereless than 0.2.

No studies reported any adverse effects associatedwith MRCP, although six studies reported adverseeffects associated with ERCP. Twenty studiesreported no information regarding adverse effects.Claustrophobia prevented at least some patientsfrom having MRCP in ten of the 28 studies. Theother 18 studies make no mention of claustrophobia.In the one study identified dealing with patientsatisfaction, on the whole patients preferred MRCPto ERCP. There is, therefore, some evidence thatMRCP is an accurate diagnostic test in comparisonto ERCP, although study quality was moderate.

Effectiveness

26

Role of diagnostic ERCP andMRCP in the diagnosis of biliarytree obstructionERCP has established itself as the conventionalmodality for imaging the pancreaticobiliary treeand affording the opportunity for therapeuticintervention.

However, the procedure remains operatordependent and carries significant morbidity andmortality rates. Whereas the complication rate fortherapeutic ERCP is acceptable, considering thatthe complications of surgical management aresignificantly higher, diagnostic ERCP can beassociated with very serious and even fatalcomplications.23 Furthermore, with reported initialfailure rate for cannulation of the duct between 3and 12%,23,41 repeated ERCP significantlycontributes to overall workload and morbidity.

MRCP is becoming increasingly available toclinicians for non-invasive evaluation of the biliaryand pancreatic ducts. The technique can providediagnostic information comparable to diagnosticERCP, both with regard to the presence ofobstruction and in a range of common clinicalsituations, such as suspected choledocholithiasis orstricture.21,56 The evidence provided by thesystematic review suggests that MRCP is anaccurate diagnostic test in comparison to ERCP(see Chapter 3).

Although the spatial resolution of MRCP does notyet match that of ERCP and potential pitfalls inMRCP interpretation should be taken intoaccount,46 its projectional display and ability todemonstrate obstructed as well as normal ductsegments make it an appropriate test forevaluating patients with suspected biliary diseaseand may limit ERCP to those who requiretherapy.21 MRCP does not require contrast mediainjection, unlike ERCP, providing particularadvantages where this is not technically possible ormay introduce infection.

However, the main criticism of MRCP, whencompared with ERCP or PTC, has been the factthat no therapeutic option can be offered at thesame time and that it may only add further to the

cost of the diagnostic work-up in these patients.This is the main reason why, in an era of costawareness, routine use of MRCP for theconfirmation of presence of biliary obstructionbefore ERCP is difficult to justify.

Potential clinical applications of MRCP have beenidentified in the literature.

� MRCP can be especially useful in cases whereERCP fails, such as biliary–enteric anastomosis,or is inconclusive, for example in cases ofcomplete ductal obstruction.21

� In the case of patients with low risk ofcholedocholithiasis, on the basis of clinicalhistory, liver blood tests and abdominalultrasound examination.21

� In the case of strictures in general, conventionalMR images obtained with MRCP may addspecificity by allowing visualisation of theextraductal anatomy, particularly forcholangiocarcinoma. However, continuedinvestigation is needed to determine the abilityof MRCP with conventional MRI to differentiatebetween benign and malignant disease. Many ofthese patients will also require endoscopicbiliary stenting.81

� In the evaluation of malignant strictures, MRCPcan provide a detailed map of the biliary treeabove the stricture, also showing isolatedsegmental ducts not evaluated directly at ERCP,playing an important role in planning surgery.81

However, MRCP in malignant strictures shouldbe considered as part of a complete upperabdominal evaluation, together withconventional MR images.33

It can be concluded that MRCP may substantiallydecrease the need for purely diagnostic ERCP. IfMRCP can replace a fraction of the ERCPscurrently being obtained for diagnostic purposes,it is reasonable to assume that the number ofcomplications and the cost of diagnosticevaluation per patient would decrease. In otherwords, the potential exists not only forimprovement in patient management but also fora reduction in the cost of diagnostic work-up. Thedecision analytical model seeks to providequantified estimates of these clinical and economictrade-offs.


27


Chapter 4

Economic analysis

Ideally, if the aetiology of the obstruction could bepredicted with accuracy, based on the clinicalarguments reported above, the decision regardingwhich diagnostic test to choose would be less of adilemma. However, decisions on whether topursue diagnostic ERCP, MRCP or PTC (especiallyfor proximal biliary obstruction) as defined at theinitial ultrasound examination in the patient arenot recommendable. Because of its ability todelineate the level of biliary obstruction and theadded benefit of being non-invasive, free ofcomplication and relatively inexpensive,ultrasound is widely advocated as the initial non-invasive imaging study in evaluating suspectedbiliary obstruction, to guide further radiographicevaluation. However, ultrasound fails to define theaetiology of biliary obstruction, especiallycholedocholithiasis and cholangiocarcinoma, inapproximately two out of three cases.82

In short, clinical judgement based on liverfunction tests (LFT) or previous ultrasoundexamination cannot provide certainty as noperfect test exists to evaluate the presence andunderlying causes of biliary obstruction. That isthe main reason why the pool of patientsconsidered for the modelling comprises adultpatients with suspected biliary obstruction, with orwithout preliminary diagnosis after initialultrasound examination. (In younger patients theduct systems are frequently of smaller calibre andcontain less bile, making them more difficult tovisualise.) The economic impact of MRCP anddiagnostic alternative strategies with differentprobabilities of CBD stones or strictures associated with ultrasound and LFT results, isestimated later in this chapter (‘Synthesis ofresults’, p. 38), based on work by SM Everett andcolleagues (unpublished conference abstract,UGW, 2002) (see Table 14).

Decision analytical modellingThe primary objective of modelling was toevaluate the relative cost-effectiveness of MRCPcompared with the conventional practice ofdiagnostic ERCP for the investigation of biliarytree obstruction, for those patients for whom achoice is available; for example, excluding patientssuffering from severe claustrophobia and thosewith previous gastric surgery that preventendoscopic access to the ampulla.

Evaluation of the relative cost-effectiveness ofadopting MRCP scanning in the investigation ofthe biliary tree will be undertaken using a

probabilistic economic model. Uncertainty analysiswill include presentation of a cost-effectivenessacceptability curve (CEAC) and analysis of theimpact of different risks of CBD stones.

PerspectiveIdeally, cost-effectiveness analyses should take abroad societal perspective, that is, include thehealth outcomes and costs for everyone affected bythe intervention. This analysis uses the perspectiveof the healthcare provider, excluding informalcarers because of their relatively minor importancegiven the characteristics of both diagnostictechnologies.

Decision treeThe pool of patients considered for the modellingcomprises adult patients with suspected biliaryobstruction, with or without preliminary diagnosisafter initial ultrasound examination. The reasonfor this decision was justified above.

Two possible conditions are considered:

� Choledocholithiasis (i.e. gallstones in theextrahepatic bile duct), also known as CBDstones.

� Biliary strictures: these can be due to a numberof underlying conditions, the most commonbeing benign and malignant strictures (cancer).

These are the most frequent conditions affectingthe biliary tree, where MRCP can providediagnostic information comparable to diagnosticERCP.21,23,41,56

From the results presented later in the chapter, itcan be concluded that there is moderate evidencethat MRCP is an accurate investigation incomparison to diagnostic ERCP.

The structure of the decision analytical modelincludes both CBD stones and benign andmalignant strictures, to approximate the level ofuncertainty to clinical practice. For further detailson causes of biliary obstruction see Chapter 2.

Treatment options for patients with the mostcommon cause of obstruction, symptomatic CBDstones, include therapeutic ERCP (for clearance ofthe bile duct), endoscopic decompression by internalstent, and dissolution. The first two options havebeen simplified as therapeutic ERCP withsphincterotomy, for the sake of simplicity of themodel. Dissolution is not included in this analysis asits efficacy is still debatable. Very small stones thatonly require regular checks have not been modelled.

Economic analysis

28

Pancreatic head lesions, and in particular cancerof the head of the pancreas, are the most commonmalignant neoplasm in the hepatobiliary area,causing in many cases an indirect dilatation in thebiliary ducts. Periampullary carcinoma, in theregion of the ampulla of Vater, is less commonthan head of the pancreas cancer and is associatedwith lower mortality. Although more uncommonthan malignant neoplasm of the head of pancreas,cholangiocarcinoma occurs most frequently in theextrahepatic biliary tree but can also arise fromintrahepatic bile ducts. Extrahepatic lesions maypresent as biliary stricture, involving the CBD(30–36%), the common hepatic duct (15–30%) andthe biliary bifurcation with the typical features ofKlatskin’s tumour (10–26%).33,81

For the purposes of modelling,cholangiocarcinoma in the intrahepatic bile ductand PSC are excluded, because they areuncommon conditions and usually more associatedwith liver problems and liver treatments.

Treatment options for patients with malignantstrictures are either curative or palliative. Ingeneral, the probability that curative resection isappropriate is quite low, and aggressive surgicaltherapy has to be carefully considered taking intoaccount age and mortality risk. With the exceptionof cholangiocarcinoma in the CBD, whereresection is possible in around 30% of cases,palliative treatment through endoscopic stentplacement is considered the standard treatmentfor extrahepatic malignant biliary stricture.83

Benign extrahepatic strictures are normally theresult of a major complication of laparoscopiccholecystectomy or chronic pancreatitis. Surgicalreconstruction is required in most cases (about85%) and can be safely accomplished with minimalmorbidity and excellent long-term outcomes.84

Regarding the potential economic benefits ofMRCP versus diagnostic ERCP, MRCP may avertthe need for invasive diagnostic testing in patientswho ultimately do not require therapeutic ERCPintervention, so minimising the costs of failedexaminations and morbidity and mortalityassociated with diagnostic ERCP. Thus, MRCP ispotentially most beneficial in patients with a highprobability of having a non-obstructive dilatation(true negatives) and in patients with suspectedmalignant strictures (i.e. cancer), who can benefitfrom its important role in planning surgery.

In contrast, diagnostic information provided byERCP allows immediate therapeutic intervention

required for the removal of CBD stones andintroduction of stents to relieve biliary obstruction,and also further diagnostic tests such as biopsies.So, especially in the case of patients with a highprobability of choledocholithiasis, the risks ofmorbidity and mortality associated with ERCPcould be compensated in terms of healthoutcomes, by an immediate and accuratetherapeutic intervention, and also in terms of costsavings. The waiting time between MRCP andtherapeutic ERCP is crucial, particularly in thecase of stone extraction, as spontaneous stonefragmentation and natural expulsion may result inunnecessary therapeutic ERCP, with an impact onboth costs and avoidable risks for the patient.These potential benefits from both diagnostictechnologies are summarised in Table 7.

The clinician is faced with the problem ofdeciding between these possible diagnosticstrategies to maximise health outcomes. Thedecision problem is illustrated in the structure ofthe decision tree (Figure 13) that follows the actualchronology of the acts, states and outcomes.

The MRCP or ERCP test option is illustrated atthe main decision node. The product of the healthstate utilities and probabilities are combined atchance nodes. The treatment decisions (i.e.therapeutic ERCP, open surgery or palliative stenttreatment) are illustrated at chance nodes becausethey are not part of the main decision, and aredependent on the clinician’s perception of theclinical state of the individual patient.

Patients who are incorrectly diagnosed will betreated at some point in the future, but during thistime they will experience reduced quality of life.This is especially true for patients suffering frommalignant lesions of the biliary tree, for whom aconsiderable improvement in quality of life relatedto pain relief is not provided. The reduced qualityof life for patients suffering from chest and backpain relating to CBD stones, mainly intermittent,is modelled as the negation of pain relief. Thesepatients will be diagnosed and treated at somepoint in the future. However, given that theoptions of treatment after diagnosis do not havean important impact on survival, future costs forthe system as well as follow-up treatments are notmodelled (Figure 14).

The prognosis for most patients suffering frompancreatic head lesions, the most common of allmalignant lesions of the biliary tree, is very poorand more than three-quarters die within a year ofdiagnosis.85 This is the main reason why the


29


survival horizon of the modelling has been limitedto 12 months. In the case of patients with CBDstones, the relief of pain is experienced in theshort term after the removal of the stones.

AssumptionsTo facilitate the modelling, the followingassumptions have been introduced.

� Repeated MRCP, therapeutic ERCP and surgeryare not modelled. Although a realistic option ofclinical practice, repeated diagnostic ERCP afterfailed cannulation is not modelled either.

� In most cases therapeutic ERCP is performedwith sphincterotomy. For the sake ofsimplification the option ERCP withoutsphincterotomy is not modelled.

� The patient either is sick (i.e. biliary obstruction)or healthy (i.e. normal ducts). There is some riskassociated with both the treatments and theERCP diagnostic test, so a healthy patient who istested or treated with ERCP has a loweroutcome than a healthy patient who is not.

� MRCP is generally considered risk free, so theutility of a healthy patient who is tested withMRCP equals 1. However, one should take intoaccount in the discussion of results that in termsof patient satisfaction, MRCP scores lower thanexpected by clinicians because of reasonsrelating to noise and claustrophobia.54

� Patients who present contraindications for anMRCP test (i.e. usual general exclusions forMRI, such as claustrophobia and cardiacpacemakers) or ERCP (i.e. previous gastricsurgery which may prevent endoscopic access tothe ampulla) are excluded from this analysisbecause they do not have a real choice betweentechnologies.

Base case values and parametersThe parameters used in the model and the sourcesfrom which these are derived are given in Table 8.

Annual incidenceHospital Episode Statistics (HES) were used toestimate the annual incidence of patients withsuspected biliary obstruction and incidence ofdifferent biliary tree disorders. Finished clinicalepisode (FCE) records for primary diagnosis (i.e.the main condition investigated during therelevant episode of healthcare) were used toestimate the pool of patients with suspected biliaryobstruction/disorder. Appropriate FCE records for2001/2 for England and Wales, grouped by fourcharacter International Classification of Diseases(ICD-10) codes were identified and are presentedin Table 9.

It is recognised that there are importantlimitations in using this database to estimate

Economic analysis

30

TABLE 7 Potential economic benefits of MRCP versus diagnostic ERCP

Advantages of MRCP Disadvantages of ERCP

Cancer patients treated with MRCP can benefit from its In cancer patients treated a priori with diagnostic ERCP, important role in planning surgery the invasive nature of the diagnostic therapy can affect

imaging MRCP

ERCP may also affect the ability to stage the tumour accurately and thus affect management decisions, especially in the context of pancreatic cancers

MRCP demonstrates ducts proximal to stricture or ERCP often does not demonstrate these ducts; if it does, obstruction, so the cause of the obstruction can be they need to be drained because of the high risk of determined with higher level of accuracy ensuing infection and cholangitis

In the case of true negatives (i.e. non-obstructive A healthy patient tested using the ERCP diagnostic dilatation) MRCP represents the advantage that it is option has a lower health outcome than a healthy patient risk free for the patient, so there is no health tested with MRCP, because of the 4–5% associated outcome loss morbidity (haemorrhage, sepsis, pancreatitis, bile

leakage) as well as recognised mortality

Disadvantages of MRCP Advantages of ERCP

No therapeutic option offered at the same time. If MRCP results ERCP diagnostic information allows immediate and in disease (true-positive), when ERCP is a recommendable accurate therapeutic intervention when required (e.g. therapeutic intervention MRCP costs could have been avoidable. removal of CBD stones, introduction of stents)In the case that the waiting time between MRCP diagnostic and intervention is considerable, spontaneous stone fragmentation can change the accuracy of the diagnostic into false-positive, with associated impact in costs and disutilities

incidence; for example, only a proportion ofpotential patients with biliary disorders areadmitted to hospital, and there may be more thanone record for the same person, depending on thenumber of transfers or multiple visits during theyear. However, following consultation with theCollege of the British Society of Gastroenterologyand the Royal College of Radiologists (RCR), andin the absence of a better source of information,this is the best estimate available. These estimateswere validated with data from GP patientconsulting9 for new and first ever episode rates forbiliary tree and pancreatic lesions (Table 10). Giventhat the approximate population in England andWales is 50 million people, based on the overallannual incidence rate presented in Table 10 (12per 10,000 person-years at risk) this gives a totalannual incidence of 60,000 for biliary tree andpancreatic lesions. This figure is very similar to thetotal of 52,617 annual incidence based on FCEestimates.

Probabilities of eventsProportions for the different disorders wereestimated using FCE records (see Table 9) andvalidated by estimates from the literature reviewedand by clinical judgement.

The probabilities of death and overallcomplications after diagnostic and therapeuticERCP were estimated from the literature review and checked with estimates from local data.

The clinical-effectiveness review did not provideany useful data regarding the proportion ofpatients with malignant strictures who undergoopen surgery or palliative treatment. Estimateswere based on local data and clinical judgementaccording to the type of cancer, ranging between10 and 30%. However, given that the mostcommon cancer affecting bile duct obstruction isthe head of the pancreas, and that its chances of


31


~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

Patient with suspectedbiliary obstruction

DiagnosticERCP

MRCP

CBD stone – ERCP

malignant stricture

benign stricture

True-negative

True-positive

False-positive

True-positive

False-positive

True-positive

False-positive

True-positive

False-positive

True-positive

Endoscopic stent

Resection

False-positive

True-positive

False-negative

True-positive

False-positive

True-positive

False-positive

False-negative

CBD stone – ERCP

CBD stone

malignant stricture

malignant stricture

benign stricture

benign stricture

CBD stone

malignant stricture

benign stricture

True-negative

False-negative

T–

T+

T–

T+

Alive

Dead

Endoscopic stent

Resection

FIGURE 13 Basic decision tree

Economic analysis

32

~

~

~

~

~

~

~

~

~

~

~

~

MRCP

T+

T–

True-negative

benign stent

malignant stent

CBDS-ERCP

CBD stone

True-positive

False-positive

= Cot MRCP= T

True-positive

Endoscopic stent

Resection

False-positive

True-positive

False-positive

True-positive

False-positive

= Cost MRCP= T × U pain CBD stone

= Cost MRCP + Cost CT + Cost surg + catheter= T – U surgery

= Cost MRCP + Cost CT + Cost surg + catheter= T – U surgery

= Cost MRCP + Cost CT + Cost therapeutic ERCP + Cost stent= T × U post stenting – U therapeutic ERCP

= Cost MRCP + Cost CT + Cost therapeutic ERCP= T – U therapeutic ERCP

= Cost MRCP + Cost CT + Cost surgery + Cost stent= T × U post surgery – U surgery

= Cost MRCP + Cost CT + Cost surgery= T – U surgery

= Cost MRCP + Cost therapeutic ERCP= T – U therapeutic ERCP

= Cost MRCP + Cost therapeutic ERCP= T – U therapeutic ERCP

= Cost MRCP= T × U extrahep. malignant stent

= Cost MRCP= T × U pain

malignant stent

benign stent

False-negative

FIGURE 14 Model: MRCP and ERCP sections with costs and utilities calculation

Health Technology Assessm

ent2004; Vol. 8: No. 10

33

© Q

ueen’s Printer and Controller of H

MSO

2004. All rights reserved.

FIGURE 14 Model: MRCP and ERCP sections with costs and utilities calculation (cont’d)

~

~

~

~

~

~

~

~

~

~

~

~

~

DiagnosticERCP

Alive

Dead = Cost diag.ERCP + r × diag ERCP w cc

T–

T+

CBDS

malignant

benign st.

True –

False –

True +

False +

stent

Resection

True +

False +

CBDS

malignant

benign st.

True-positive

= Cost therapeutic ERCP= T – therap. ERCP

= Cost therapeutic ERCP= T – U therap. ERCP

= Cost diag.+r × Cost diag ERCP w cc+Cost CT+Cost therap.ERCP+Cost stent= T × U post stenting – U diag.ERCP – r × U diag.ERCP w cc. – U therap. ERCP

= Cost diag.ERCP+r × Cost diag ERCP w cc+Cost CT+Cost surgery+Cost stent= T × U post surgery – U diag.ERCP – r × U diag.ERCP w cc. – U surgery

= Cost diag.ERCP+r × Cost diag ERCP w cc+Cost CT+Cost surgery= T – U diag.ERCP – r × U diag.ERCP w cc. – U surgery

= Cost diag.+r × Cost diag ERCP w cc+Cost CT+Cost therap.ERCP= T – U therap.ERCP – U diag.ERCP – r × U diag.ERCP w cc

= Cost diag.ERCP+r × Cost diag.ERCP w cc+Cost surg+Cost catheter= T – U diag. ERCP – r × U diag ERCP w cc – U surg= Cost diag.ERCP+r × Cost diag.ERCP w cc+Cost surg+Cost catheter= T – U diag. ERCP – r × U diag ERCP w cc – U surg

= C diag.+r × C diag wc= T – U diag. – r × U diagwc

= Cost diag.ERCP+r × Cost diag ERCP w cc= T × U pain CBD stone – U diag.ERCP – r × U diag ERCP w cc

= Cost diag.ERCP+r × Cost diag ERCP w cc= T × U extrahep.malignant – U diag.ERCP – r × U diag ERCP w cc

= Cost diag.ERCP+r × Cost diag ERCP w cc= T × U pain – U diag.ERCP – r × U diag ERCP w cc

False-positive

True-positive

False-positive

Economic analysis

34

TABLE 8 Prior information and sources for key parameters

Mean value SD Range Distribution Source

Probabilitiesa

Probability of CBD stones 0.37 Beta HES

Probability of malignant stricture 0.17 Beta HES

Probability of benign stricture 0.15 Beta HES

Death after diagnostic ERCP (%) 0.085 0.005 0.07–0.1 Beta Cotton et al., 199427

Overall complications after diagnostic ERCP (%) 5.5 0.16 5–6 Beta Cotton et al., 199427

Resection open surgery for malignant stricture 0.10 Fixed parameter Clinical judgement

Endoscopic stent placement for malignant stricture 0.90 Fixed parameter Clinical judgement

UtilitiesHealth-related quality of life general population 1

MRCP examination 1

Chest and back pain relating to CBD stones and strictures in the extrahepatic bile ductb 0.89 0.003 0.88–0.90 Normal. Range: severe (periodical) Cook et al., 199486; Community, TTO

pain, moderate pain

Diagnostic ERCP procedure only 0.9904 Beta (a 1, b 99) Gregor et al., 199687; Clinicians, TTO

Therapeutic ERCP with sphincterotomy 0.95 Beta (a 4, b 96) Bass et al., 199388; patients, standardgamble

Biliary stricture surgery 0.884 0.038 0.77–0.998 Normal. Range: complication of Bass et al., 199388; patients, standard surgery, surgical scar gamble

ERCP papillotomy/other complications 0.759 0.016 0.7115–0.808 Normal. Range: papillotomy Gregor et al., 199687; Clinicians, TTOcomplication, other complication

Patient with extrahepatic malignant stricture 0.37 0.041 0.25–0.50 Normal Luman et al., 199789; EORTC QLQ-30

Postintervention for extrahepatic malignant biliary stricture 0.61 0.041 0.50–0.75 Normal Luman et al., 199789; EORTC QLQ-30

continued


ent2004; Vol. 8: No. 10

35

© Q


MSO


TABLE 8 Prior information and sources for key parameters (cont’d)

Mean value SD Range Distribution Source

Costs (£ 2002)c

MRI: medical gastroenterologyd £454 £26 £376–532 Normal HRGs 2002 (F03op)

CT: medical gastroenterologyd £4340 £20.16 £279–400 Normal HRGs 2002 (F04op)

Diagnostic ERCP examination of bile duct £846 £100.83 £514–1119 Normal HRGs 2002 (G17)Diagnostic ERCPwith complications £1113 £139.83 £570–1409 Normal HRGs 2002 (G16)

Therapeutic ERCP, extraction of CBD stones £1108 £130.83 £750–1535 Normal HRGs 2002 (G15)

Surgery malignant neoplasm of extrahepatic bile ducts > 69 years or with complications £2004 £134.33 £1649–2455 Normal HRGs 2002 (G13)

Memotherm biliary stent: palliative, malignant stricture £800 £10 Normal Sheffield Northern General Hospital and Bradford Royal Infirmary

Bilioplasty balloon catheter £80 £10 Normal Data from suppliers

a Sensitivity and specificity for diagnosing the cause of the biliary obstruction. Obtained against final diagnosis, made on the basis of surgical findings, percutaneous biopsy, clinicalfollow-up and others.

b Approximated same discomfort relating to gallbladder, according to clinical judgement.c Elective procedures, relating to Hepato-biliary and Pancreatic System HRGs section. HRG codes in parentheses. Reporting range for 50% of NHS Trusts.d Primary Care Trusts Outpatient HRG Data, 2001/2. Reporting range for 50% of NHS Trusts.EORTC QLQ-30, European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire-30.

resectability are around 10%, this figure was usedas a general simplification. FCE records of mainoperations affecting the hepatobiliary organs (i.e.normally the most resource-intensive procedureperformed during a theatre session) provide anoverview of the proportions of this type ofintervention (Table 11).

Probabilities of events are proportions, such as thepercentage of patients suitable for curativetreatment, so the beta distribution has been used.

Estimates of sensitivity and specificity taken fromthe review of clinical effectiveness are summarisedin Table 12.

Economic analysis

36

TABLE 9 FCEs for Primary Diagnosis biliary tree and pancreatic head lesions

Primary diagnosis ICD-10a All NHS Trusts All NHS Trusts in Englandb in Walesc

Malignant neoplasm of extrahepatic bile duct C240 562 32Malignant neoplasm of ampulla of Vater C241 1,059 30Malignant neoplasm of biliary tract, unspecified C248–C249 235 3Malignant neoplasm of head of pancreas C250 6,652 325Benign neoplasm of extrahepatic bile duct D135 152 17Obstruction of bile duct K831 4,431 136Other specified diseases of biliary tract K838 2,432 146Disease of biliary tract, unspecified K839 348 19Calculus of bile duct with cholangitis K803 1,502 52Calculus of bile duct with cholecystitis K804 1,292 81Calculus of bile duct without cholangitis or cholecystitis K805 15,860 768Other cholelithiasis K808 667 93Acute cholecystitis K810 4,902 52Chronic cholecystitis K811 5,008 1,376Other cholecystitis K818 80 7Cholecystitis, unspecified K819 3,991 185Perforation of bile duct K832 24 1Fistula of bile duct K833 49 2Spasm of sphincter of Oddi K834 27 1Biliary cyst K835 17 1

Total disorders of the biliary tree and pancreatic head lesions 49,290 3,327

% Patients with malignant stricture 17.26 11.72% Patients with benign stricture 14.93 9.55% Patients with CBD stones 37.84 27.08% Cholelithiasis, cholecystitis or other 35.59 56.59

a FCEs are grouped by four character ICD-10 codes. b 2001/2 figures.c 2000/1 figures.

TABLE 10 Summary table: annual incidence estimates for biliary tree and pancreatic head lesions

New and first ever episode ratesa

Cholelithiasis 8 per 10,000 person-years at riskOther disorders of the biliary tract 2 per 10,000 person-years at riskDiseases of the pancreas 2 per 10,000 person-years at risk

Hospital Episode Statisticsb

Total disorders of the biliary tree and pancreatic head lesions (England and Wales): 52,617primary diagnosis

Patients with malignant stricture 17%Patients with benign stricture 15%Patients with CBD stones 37%Cholelithiasis, cholecystitis or other 37%

a Source OPCS (1992).9b HES England figures for 2001/2; Wales figures for 2000/1.

Sensitivity and specificity of ERCP were assumedto be 100%. The sensitivity and specificity ofMRCP were assumed to have beta distributionsbased upon means and standard errors drawnfrom the clinical review. No correlation wasassumed between MRCP and ERCP; therefore, theestimates of the difference in sensitivity andspecificity will have a greater uncertainty.

Estimation of costsThe costs of healthcare resources used as inputs tothe diagnostic tests and subsequent treatmentoptions were estimated using the National Scheduleof Reference Costs (NSRC), in particular for NHSTrust Elective Inpatient Healthcare Resource Group(HRG) data (2002, v.3). HRGs are groups ofinpatient FCEs which are purported to have similarhealthcare resource requirements. Each individualrecord in the HES data set is assigned to a singleHRG based on the data contained in the record.90

The costs included in the NSRC reflect the actualcosts incurred in the 2001/2 financial year andrepresent national figures for England.

Reported costs are averages, and they includeoverheads and capital costs. They correspond toelective procedures, relating to hepatobiliary and

pancreatic system HRGs. Instead of reporting costvariation in all NHS Trusts, the 50% interquartilerange has been included to improve the skewnessnormally present in this data set.

Depending on the type of procedure, complicationcosts are also included in the HRG estimation (seeTable 8); however, the follow-up and treatment ofcomplications are not included.

Where more than one estimate for each cost itemwas obtained (e.g. ERCP with or without


37


TABLE 11 FCEs for main operations on the biliary tree and pancreatic head lesions

Main operations ICD-10a All NHS Trusts All NHS Trusts in Englandb in Walesc

Diagnostic ERCP examination of CBD and/or pancreatic duct J431–J459 14,018 751

Excision of bile duct J271–J279 27 –Extirpation of lesion of bile duct J281–J289 37 1Connection of CBD J301–J309 180 11Open introduction of prosthesis into CBD J312–J319 59 –Reconstruction of CBD J321–J329 55 –Open removal of calculus from CBD J331–J339 200 3Operations on ampulla of Vater J361–J368 19 2Other open operations on CBD J371–J379 101 4Endoscopic incision of sphincter of Oddi J381–J389 11,663 421Other therapeutic endoscopic operations on ampulla of Vater J391–J399 268 18ERCP placement of prosthesis in CBD J401–J409 5,122 165Other therapeutic ERCP operations in CBD J411–J419 1,308 110Therapeutic PTC insertion operations into CBD J461–J489 820 43Other operations in bile duct J491–J529 717 10Excision of head of pancreas J561–J569 618 33

Total main operations on the biliary tree and pancreatic head lesions 35,212 1,572

% Diagnostic ERCP 39.81 47.77% Therapeutic ERCP 52.14 45.41% Therapeutic PTC 2.32 2.73

a FCEs are grouped by four character ICD-10 codes. b 2001/2 figure.c 2000/1 figures.

TABLE 12 Diagnostic test performance characteristics used inthe model

CI Beta distribution

Mean Lower Lower a b

95% 95%

MRCPSensitivity 93% 81% 100% 37 3Specificity 94% 83% 99% 32 2

ERCPSensitivity 100% – – – –Specificity 100% – – – –

complications), the range of values was used tocalculate the distribution based on the minimum,median and maximum values. The standarddeviation was estimated based on an approximate99% confidence interval between the minimumand maximum values of the range.

Local activity data from the Bradford RoyalInfirmary and the Sheffield Teaching HospitalNHS Trust were used to check that there were nostriking differences from the national costdatabase. Missing data for cost variables wereobtained from suppliers (e.g. the current price ofMemotherm metal stents).

Utility estimatesThe tree structure combines very different types ofpossible biliary disorders in terms of severity andassociated mortality, and for these reasons cost percase avoided or cost per life-year gained could notbe used as outcome measures. Utility values forthe health states were assigned as the end-pointsof the decision tree. The states of perfect healthand death were used as anchor states, with valuesof 1 and 0, respectively. Utility scores for thedifferent health states were obtained from theHarvard CUA database, with the exception ofutilities related to biliary tree malignantneoplasms, which required an additional literaturesearch. Those utilities reported by clinicians andcalculated using the methodologies of standardgamble and time trade-off (TTO) were preferredto those reported as utility scores using a ratingscale. The peculiarities of patient satisfaction withdiagnostic ERCP and MRCP were discussed inChapter 3 (‘Patient satisfaction’, p. 14).

The probability of ending in one of the healthstates included in the tree structure is determinedby previous events. The predictive value of diseasegiven a positive or negative test, the predictive valueof no disease given a positive or negative test, andthe different probabilities of biliary tree disordersplay key roles in determining final outcomes.

Model uncertaintyAn overall sensitivity analysis for a population ofnormal risk is presented using Monte-Carlosimulation, assigning a probabilistic structure(prior distribution) to each of the model inputsand generating a CEAC. The probabilisticsensitivity analysis was undertaken for all modelparameters simultaneously, to determine theoverall impact of uncertainty within the model.

The economic value of diagnostic MRCP, however,is dependent on the risks of disease in the

population in question. At an individual patientlevel, the economic value of MRCP is directlyrelated to the prior assessment of risk of CBDstones and/or malignancies. A sensitivity analysishas been undertaken to investigate the impact ofprior risk assessment based on ultrasound andLFT results on the decision to opt for MRCP.

Synthesis of resultsThe key clinical and economic results, togetherwith uncertainty estimates (based upon 1000Monte-Carlo simulations) are shown in Table 13.

Cost-effectiveness results have been displayed onthe cost-effectiveness plane (Figure 15), which plotsincremental costs and incremental quality-adjustedlife-years (QALYs) in a 2D plane. The cost-effectiveness plane from 1000 samples shows thatthe uncertainty in the cost difference betweenMRCP and ERCP is much larger than theuncertainty in the QALY difference. This result isnot surprising: with a similar sensitivity andspecificity to ERCP for the most common biliarytree disorders, and despite all of its recognisedadvantages over ERCP in terms of morbidity andmortality, the main criticism of this test has beenthat it may only add further to the cost of thediagnostic work-up in patients who needtherapeutic intervention. This is especially true inthe case of patients with CBD stones (see theopening section to this chapter).

As shown in Table 13, the probability of avoidingunnecessary diagnostic ERCP, that is, the probabilityof a true-negative MRCP, is estimated at 31% (95%CI 20 to 40%). These patients could avoid theunnecessary risk of complications and deathassociated with diagnostic ERCP. Furthermore,substantial cost saving would be gained: theoverall expected cost saving associated with MRCPis £149 (£325 to –£15) and the overall expectedQALY gain is estimated at 0.011 (0.000 to 0.030).

The probability that MRCP is cost saving isestimated at approximately 96.5%, with anexpected cost saving of £149 (£325 to –£15). Fromthis result, one can infer that the cost savings interms of avoided unnecessary diagnostic ERCP,associated complications and deaths cancompensate for the added diagnostic costs inthose patients who need therapeutic intervention.

Table 14 presents the key clinical and economicresults for populations at lower risk of CBD stonesor strictures based on the results of ultrasound and

Economic analysis

38

LFTs according to the classifications in the workby Everett and colleagues (unpublished). Figure 17presents the estimated cost savings from MRCPover a range of risks of CBD stones. It should benoted that these results assume that the risks ofmalignant and benign strictures are held constantat the values used in the baseline model.

Potential methodological strengthsand limitations of the economicanalysisStrengthsThis model used data from all published sourcesidentified in the systematic review undertaken to

compare sensitivity and specificity of bothtechnologies, with diagnostic ERCP as acomparator. Additional searches were conductedto identify utilities related to biliary tree malignantneoplasms and costs of MRI technology. Data werecross-checked with clinical judgement and localhospital data where available. Literaturerecommended by clinicians in the field was readand analysed.

The basic tree structure and its main assumptionswere discussed and checked by experts in the maindisciplines related to biliary tree disorders andupper digestive organs in general: a consultantgastroenterologist, two consultant radiologists anda consultant biliary pancreatic surgeon.


39


TABLE 13 Key economic and clinical results

Mean 95%

Lower Upper

Key clinical resultsProbability of true-negative MRCP (i.e. avoiding unnecessary diagnostic ERCP) 31% 20% 41%Probability of true-positive MRCP CBD stones (i.e. necessary therapeutic ERCP) 34% 27% 42%Probability of death diagnostic ERCP 0.1% 0.1% 0.1%

Key economic results MRCP vs ERCPIncremental QALYs 0.012 0.002 0.028Incremental costs –£163 –£340 £7Incremental net benefit (threshold of £20,000 per QALY) £400 £159 £755Cost-effectiveness MRCP vs ERCP Dominant Dominant £259.23Probability of cost saving 0.970 – –Probability of incremental QALYs positive 0.993 0.002 0.028Probability of cost-effectiveness better than £20,000 per QALY 1.000 – –

–0.02 0.02 0.04

Incremental QALY (MRCP – ERCP)

Incr

emen

tal c

ost (

MRC

P –

ERC

P)

0.06 0.08 0.1

200

100

0

–100

–200

–300

–400

–500

FIGURE 15 Cost-effectiveness plane for MRCP versus diagnostic ERCP

Potential weaknessesMissing information and unknown parametersMany of the weaknesses in the technologyassessment relate to the poor design of studiesavailable, in particular on diagnostic testcharacteristics for final clinical outcomes and impacton clinical practice. Although these are primarilyof clinical interest these impacts were only assessedwithin the decision analytical economic model. Theclinical systematic review focused separately on the

test characteristics for different disease types;although no significant differences were identified(sensitivities and specificity overlap for thedifferent diseases), minor differences were found that were consistent with clinical judgement.It would, therefore, be preferable to modelseparately the test sensitivity and specificity forCBD stones and strictures. However, owing to thepoor design of the studies, there is insufficientinformation to incorporate this separately within

Economic analysis

40

TABLE 14 Economic results with different probabilities of CBD stones or strictures associated with ultrasound and LFT resultsa

Ultrasound normal, Ultrasound and UltrasoundLFTs abnormal LFTs abnormal abnormal,

LFTs abnormal

Probability of CBD stone or stricture 21% 32% 37%

Key clinical resultsProbability of true-negative MRCP 45% 34% 30%

(i.e. avoiding unnecessary diagnostic ERCP)Probability of true-positive MRCP CBD stones 19% 30% 34%

(i.e. necessary therapeutic ERCP)

Key economic results MRCP vs ERCPIncremental QALYs 0.014 0.011 0.011Incremental costs –£250 –£185 –£149Incremental net benefit (threshold of £20,000 per QALY) £4524 £218 £364Cost-effectiveness MRCP vs ERCP Dominant Dominant £259.23Probability of cost saving 0.991 0.980 0.965

a Estimations based on Everett et al. (Everett S, Hamlin J, Beckett C, Bzeizi K. MRCP – an important investigation in patientsat low risk of pancretico-biliary disease. Conference abstract, UGW, 2002, unpublished).

£0 £5000 £10,000 £15,000

Threshold (MAICER)

Prob

. E[IN

B] >

0

£20,000 £25,0000.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

FIGURE 16 Cost-effectiveness acceptability curve for MRCP. (MAICER, maximum acceptable incremental cost-effectiveness ratio.)

the model. The model, therefore, uses thesystematic review results for the most commondisorder, CBD stones. The results for malignanciesare prone to bias.

Ideally, the decision analytical model would use anROC curve analysis of sensitivity and specificity tocapture the impact of variation in diagnosticthresholds. Sensitivity and specificities should behandled similarly. Owing to the shortcomings inthe design and/or reporting of the studies this isnot captured fully in the model.

Where standard deviations or confidence intervalswere not given, the uncertainty surrounding themedians we estimated based on the minimum andmaximum range, approximated as a 99%confidence interval.

The impact of the diagnostic technology on a widerange of types of malignancy made difficult theidentification of appropriate utilities for thepurpose of the modelling.

Some parameters (e.g. proportion of patients withmalignant strictures who undergo palliativetreatment) proved difficult to estimate based on areview of the available literature, so clinicaljudgement and local data were the only feasiblealternatives.

Limitations of the modelIn some cases, incomplete imaging may createconfusion regarding ductal anatomy or disease,and repeat MRCP can help to avoid pitfalls ininterpretation. However, for the sake ofsimplification, neither the possibility of pitfalls inMRCP interpretation nor the option of repeatedMRCP was modelled. For the same reason,repeated diagnostic ERCP after cannulationfailure was not modelled.

Conclusions on the economics ofdiagnostic MRCP versus ERCPThe estimated clinical and economic impacts ofdiagnostic MRCP versus diagnostic ERCP are veryfavourable. The baseline estimate is that MRCPwould both be cost saving and result in improvedquality of life outcomes compared with diagnosticERCP. The uncertainty analysis, investigating theimpact of parametric uncertainty within themodel, indicates that this result is robust. It shouldbe noted, however, that there are markeduncertainties in the structure and assumptionswithin the decision analytical model that are notcaptured within this parametric uncertaintyanalysis. The results presented in this assessmentwill thus overstate the robustness of the economicoutcomes for MRCP.


41


0% 10% 20% 30% 40%

Probability of CBD stones

Cos

t sav

ings

of M

CRP

ove

r ER

CP

50% 60% 70%£0

£100

–£100

£200

£300

FIGURE 17 Cost savings from MRCP for different risks of CBD stones


43


ERCP is often associated with significantmorbidity and mortality.1 Little information

on adverse effects was provided in the studies,with only eight studies providing any data onadverse effects associated with ERCP. However,reports in the literature show diagnostic ERCP tohave a complication rate of 5–6% and a mortalityrate ranging from 0.01%1 to 0.089%.27

Therapeutic ERCP is reported to have acomplication rate of 4–10%, although someauthors put the rate as high as > 20%,28 and amortality rate ranging from 0.07%1 to 0.3%.91 Ifpatients receive diagnostic ERCP when MRCP mayhave been more appropriate, legal issues maybecome important if complications arise from thisunnecessary invasive procedure. This fact may beaffecting the current provision of MRCP. In reality,following the British Society of GastroenterologyWorking Party Report 2001,92 it seems that thepattern of provision of ERCP is changing, with theprovision moving away from surgeons andradiologists, and towards medicalgastroenterologists.

The potential sources of economic benefit forMRCP compared with diagnostic ERCP areimportant but highly dependent on access to andwaiting lists for adequate MRI technology athospital level. Furthermore, the potentialeconomic benefits of MRCP are affected bydiagnostic ERCP operator ability and skills.However, MRCP is also a diagnostic technologyhighly dependent on skilful interpretation.

There has been a substantial increase in theworkload of departments of clinical radiology overthe past few years. Department of Health figuresdemonstrate an increase of 7.4% between 1996and 1998. This overall rise in workload has beencompounded by the increased complexity of theinvestigations and the increase in interventional

work that has been required of departments ofclinical radiology. During the same period, a largenumber of radiology posts have been advertisedwhere an appointment has not been possiblebecause of a lack of applicants (in 1999 this meantthat over 40% of advertised consultant posts werenot filled at the first attempt). The benchmarkingfigures produced by the RCR demonstrate that theindividual consultant’s workload is substantiallyhigher than the RCR 1993 recommendations of12,500 examinations per radiologist.93

In short, even in cases where MRCP is deemed tobe the most appropriate diagnostic technique toundertake, it may not be possible to receive it.Some hospitals in the UK may not have thenecessary MRI scanner and there may not besufficient time set aside for MRCP investigationseven in hospitals in possession of the appropriatescanners.

Taking into account the above infrastructure andprofessional limitations, protocols are needed todetermine which patients are most suitable forMRCP. To present some illustrative figures: theutilisation rate of MRCP over all MRI scanscarried out in the unit of radiology of the RoyalHallamshire Hospital, Sheffield, during the periodApril 2001 to April 2002 was 5.42%, in comparisonwith 41% for spine lesions, 9% for knee, 8% forbrain and 4% for pelvis, among others.

This is where the real opportunity cost ofperforming MRCP on a routine basis could beidentified, estimating MRI waiting lists by severityand type of condition and reinterpretingpriorities. The elderly may be one group that mayparticularly benefit from MRCP, as invasiveprocedures are often avoided owing to frailty. Thiskind of estimation is well beyond the objective ofthis report.

Chapter 5

Factors relevant to the NHS


45


Main results: clinical effectivenessThe median sensitivity (n = 13 studies) forcholedocolithiasis was 93% (range 81–100%) andspecificity 94% (range 83–99%). The medianlikelihood ratio for a positive value was 15.75 andfor a negative value was 0.08. For malignancy,reported sensitivities were somewhat lower,ranging from 81 to 86%, and specificities rangedfrom 92 to 100%, with positive likelihood ratiosranging from 10.12 to 48 and negative likelihoodratios ranging from 0.15 to 0.21.

In the 28 studies in this review one positivelikelihood ratio was less than 565 and four negativelikelihood ratios were greater than 0.2.57,67,77,80 Nostudies reported any adverse effects associatedwith MRCP, although six studies reported adverseeffects associated with ERCP and two studiesreported no adverse effects associated with ERCP.Twenty studies reported no information regardingadverse effects.

Main results: cost-effectivenessFor the general population with a probability ofCBD stones of 37%, the probability of avoidingunnecessary diagnostic ERCP, that is, theprobability of a true-negative MRCP, is estimatedat 30% (95% CI 20% to 40%). These patientscould avoid the unnecessary risk of complicationsand death associated with diagnostic ERCP.Furthermore, substantial cost saving would begained: the overall expected cost saving associatedwith MRCP is £149 (£325 to –£15) and the overallexpected QALY gain is estimated at 0.011 (0.000to 0.030). It should be noted that uncertainties inthe structural assumptions within the economicmodel are not captured within this parametricuncertainty analysis.

For populations with a lower risk of CBD stonesthe economics are further improved; for example,in a population with a 21% risk of CBD stones theestimated cost savings would be £250. Forpopulations with a risk of CBD stones in the orderof 60%, MRCP is approximately cost neutral whencompared with ERCP.

Assumptions, limitations anduncertaintiesThe reporting in the studies was of moderatequality, with little information regarding patientcharacteristics and why some patients did notreceive both tests. ERCP has been assumed to bethe gold standard, although this test isimperfect.22 It was often unclear whether thosestudies comparing MRCP with ERCP incorporatedother tests as well.

Information on adverse effects was not reported in20 of the 28 studies, making it difficult todetermine the extent to which they occur.

Seven studies reported MRCP results comparedwith final diagnosis, one of which did not provideadequate data for calculations. Three of thesestudies provided results for both ERCP and MRCP.The remaining 21 studies reported resultscomparing MRCP versus ERCP. Thus, differenttypes of study are being compared in the analysis.Sensitivities and specificities for those studiescomparing MRCP with final diagnosis did notdiffer to any extent from those comparing MRCPwith ERCP. Final diagnosis was not always clearlydefined, but usually included ERCP and othercomponents such as histological results.

There are several problems associated with usingsummary ROC curves. Although the production ofa summary ROC curve allows the computation of asummary estimate of diagnostic performance, theresults cannot be directly applied to clinicalpractice.49 Many important aspects of study designwere not provided in the studies and the standardsof reporting were poor. There is little informationas to what impact publication bias has on theresults as there is no equivalent to the funnel plotfor studies of diagnostic accuracy. The need tosummarise information with a summary ROCtechnique, owing to variability and interdependencebetween the observed sensitivities and specificities,can be considered to indicate a problem with theapplication of a diagnostic technology. This isespecially the case when an ROC-like relationshipis observed for a test that purports not to haveexplicit variation in cut-points. Although the

Chapter 6

Discussion

Discussion

46

summary ROC method allows for this variation, itdoes not attempt to characterise or explain it, sothat the meta-analysis fails to provide informationthat will assist an operator in using the technologyin the most accurate manner.49

Recent methodological research94 hasrecommended that value of informationapproaches to analysing uncertainty should beused wherever possible in identifying researchrequirements in health technology assessment.This technique, however, only captures parametricuncertainty within a model, and caution should beexercised where there is marked uncertainty in theunderlying structural assumptions. A value ofinformation analysis has therefore not beenpresented and the further research requirementsare based primarily on addressing the structuraluncertainty within the assessment.

Need for further researchThe following were identified as areas wherefurther research is needed.

� Good quality studies are needed comparingMRCP and diagnostic ERCP to final diagnosis,

stating inclusion/exclusion criteria and relevantpatient characteristics. This would help toovercome some of the shortcomings ofcomparisons with diagnostic ERCP.

� Studies are needed comparing MRCP withdiagnostic ERCP for the full range of targetconditions, in particular differentiation ofbenign and malignant strictures and the impacton management and outcome.

� More research is needed in the area of patientsatisfaction and ways to reduce problems withclaustrophobia and make MRCP moreacceptable to patients.

� Protocols, assessing prior risk, are needed tohelp to identify which patients with whichsuspected conditions would most benefit fromMRCP and which would benefit from ERCP.

� To understand the real opportunity costsassociated with MRCP, studies are needed toassess the relative need and urgency of patientaccess to MRI services.

� As the development of MRCP (a non-invasivetest) may result in an increase in requests overwhat would be expected for ERCP (an invasivetest), research is needed to determine how this will affect availability and potential costsavings.


47


There is some evidence that MRCP is anaccurate investigation in comparison to

diagnostic ERCP. The quality of the studies wasmoderate. As MRCP provides diagnosticinformation only, there will always be cases wherediagnostic ERCP is the preferred method astreatment can follow immediately.

The limited evidence on patient satisfactionshowed that patients preferred MRCP todiagnostic ERCP; however, some patients hadproblems with MRCP, owing to claustrophobia andnoise.

Many of the weaknesses in the economic analysisrelate to the interpretation of the availableinformation on diagnostic test characteristics forfinal clinical outcomes and impact on therapeuticinterventions. Although these final outcomes areprimarily of clinical interest, their impact can onlybeen assessed within the decision analytical model

presented within the economic section of thisassessment.

The estimated clinical and economic impacts ofdiagnostic MRCP versus diagnostic ERCP are veryfavourable. The baseline estimate is that MRCPwould both be cost saving and result in improvedquality of life outcomes compared with diagnosticERCP. The uncertainty analysis, investigating theimpact of parametric uncertainty within themodel, indicates that this result is robust. It shouldbe noted, however, that there are markeduncertainties in the structure and assumptionswithin the decision analytical model that are notcaptured within this parametric uncertaintyanalysis. The results presented in this assessmentwill thus overstate the robustness of the economicoutcomes for MRCP. The economic modelindicated that MRCP remains potentially costsaving in populations with relatively high risks ofCBD stones.

Chapter 7

Conclusions


49


Andrea Shippam provided secretarial support.Catherine Beverley provided advice on the

literature searching. Mr WEG Thomas (ConsultantBiliary Pancreatic Surgeon and Clinical Director ofSurgery for Sheffield Teaching Hospitals) and DrConrad Beckett (Consultant Gastroenterologist,Bradford Royal Infirmary) provided clinicaladvice. Dr MB Sheridan (Consultant Radiologistand Lead Clinician for Radiology Leeds East,Department of Clinical Radiology, St James’sUniversity Hospital, Leeds), Dr Jerry Healy(Consultant Radiologist, Department of Radiology,Chelsea and Westminster Hospital, London), DrKarl P Claxton (Senior Lecturer, Centre for HealthEconomics, University of York) and Dr MarieWestwood (Reviewer, CRD, University of York)peer reviewed the report.

All responsibility for the contents of the reportremains the authors.

Contributions of authorsHelen Bouchier undertook the electronicliterature searches. Eva Kaltenthaler carried outthe review of clinical effectiveness. Yolanda Bravoand Jim Chilcott carried out the economicanalysis. Steven Thomas and Tony Blakeboroughprovided clinical advice.

About ScHARRThe School of Health and Related Research(ScHARR) is one of the four Schools that comprise

the Faculty of Medicine at the University ofSheffield. ScHARR brings together a wide rangeof medical- and health-related disciplinesincluding public health, general practice, mentalhealth, epidemiology, health economics,management sciences, medical statistics,operational research and information science. Itincludes the Sheffield unit of the Trent Institutefor Health Services Research, which is funded byNHS R&D to facilitate high-quality health servicesresearch and capacity development.

The ScHARR Technology Assessment Group(ScHARR-TAG) synthesises research on theeffectiveness and cost-effectiveness of healthcareinterventions for the NHS R&D HealthTechnology Assessment Programme on behalf of arange of policy makers, including the NationalInstitute of Clinical Excellence. ScHARR-TAG ispart of a wider collaboration of six units fromother regions. The other units are: SouthamptonHealth Technology Assessment Centre (SHTAC),University of Southampton; Aberdeen HealthTechnology Assessment Group (Aberdeen HTAGroup), University of Aberdeen; LiverpoolReviews & Implementation Group (LRiG),University of Liverpool; Peninsular TechnologyAssessment Group (PenTAG), University of Exeter;NHS Centre for Reviews and Dissemination,University of York; and West Midlands HealthTechnology Assessment Collaboration(WMHTAC), University of Birmingham.

Acknowledgements

1. Misra SP, Dwivedi M. Complications of endoscopicretrograde cholangiopancreatography andendoscopic sphincterotomy: diagnosis,management and prevention. Natl Med J India2002;15:27–31.

2. Medlineplus Medical Encyclopedia. BiliaryObstruction. MEDLINE 2003. www.nlm.nih.gov/medlineplus/ency/article/000263.htm

3. Kim JH, Kim MJ, Park SI, Chung JJ, Song S, Kim KS, et al. MR cholangiography in symptomaticgallstones: diagnostic accuracy according toclinical risk group. Radiology 2002;224:410–16.

4. Siegel JH, Kasmin FE. Biliary tract diseases in theelderly: management and outcomes. Gut 1997;41:433–5.

5. Vitellas KM, Keogan MT, Freed KS, Enns RA,Spritzer CE, Baillie JM, Nelson RC. Radiologicmanifestations of sclerosing cholangitis withemphasis on MR cholangiopancreatography[review]. Radiographics 2000;20:959–75.

6. Textor HJ, Flacke S, Pauleit D, Keller E, Neubrand M, Terjung B, et al. Three-dimensionalmagnetic resonance cholangiopancreatographywith respiratory triggering in the diagnosis ofprimary sclerosing cholangitis: comparison withendoscopic retrograde cholangiography. Endoscopy2002;34:984–90.

7. Kalloo AN, Kantsevoy SV. Gallstones and biliarydiseases. Primary Care Clinics in Office Practice2001;28:591–606.

8. Diagnosis and treatment of common bile ductstones (CBDS). Results of a consensusdevelopment conference. Scientific Committee ofthe European Association for Endoscopic Surgery(E.A.E.S.). Surg Endosc 1998;12:856–64.

9. RCGP, OPCS, and DHSS Morbidity Statistics fromGeneral Practice. Fourth national study1991–1992. Series MB5 No. 3; 1995. London:Office of Population, Censuses and Surveys, HMSO.

10. Mehta SN, Reinhold C, Barkun AN. Magneticresonance cholangiopancreatography. GastrointestEndosc Clin North Am 1997;7:247–70.

11. Anonymous National Statistics (England and Wales):Mortality Statistics Cause. National Statistics SeriesDH2, No. 26. London: Office of Population,Censuses and Surveys, HMSO.

12. Weaver SA, Stace, BSF, Hayward SJ, Taylor GJ,Rooney NI, Robertson DAF. Endoscopic palliationand survival in malignant biliary obstruction. Dig Dis Sci 2001;46:2147–53.

13. Doglietto GB, Alfieri S, Pacelli F, Mutignani M,Costamagna G, Carriero C, et al. Extrahepatic bileduct carcinoma: a western experience with 118consecutive patients. Hepato-Gastroenterology2000;47:349–54.

14. Hintze RE, Abou-Rebyeh H, Adler A, Veltzke-Schlieker W, Felix R, Wiedenmann B. Magneticresonance cholangiopancreatography-guidedunilateral endoscopic stent placement for Klatskintumors. Gastrointest Endosc 2001;53:40–6.

15. Kim PN, Outwater EK, Mitchell DG. Mirizzisyndrome: evaluation by MRI imaging. Am JGastroenterol 1999;94:2546–50.

16. Cancer Statistics Registrations 1995–1997England. National Statistics Series, Series MB1,No. 28. London: The Stationery Office; 2001.

17. Baron TH, Fleischer DE. Past, present and future of endoscopic retrogradecholangiopancreatography: perspectives on theNational Institutes of Health ConsensusConference. Mayo Clin Proc 2002;77:407–12.

18. Ahrendt SA, Cameron JL, Pitt HA. Currentmanagement of patients with perihilarcholangiocarcinoma [review]. Adv Surg1996;30:427–52.

19. Tanner AR, Dwarakanath AD, Tait NP. Thepotential impact of high-quality MRI of the biliary tree on ERCP workload. Eur J GastroenterolHepatology 2000;12:773–6.

20. Sahai AV, Devonshire D, Yeoh KG. The decision-making value of magnetic resonancecholangiopancreatography in patients seen in areferral center with suspected biliary andpancreatic disease. Am J Gastroenterol 2001;96:207–80.

21. Demartines N, Eisne, L, Schnabel K, Fried R,Zuber M, Harder F. Evaluation of magneticresonance cholangiography in the management ofbile duct stones. Arch Surg 2000;135:148–52.

22. Sheridan MB. Endoscopic retrogradecholangiopancreatography should no longer beused as a diagnostic test: the case in favour. DigLiver Dis 2002;34:370–4.

23. Farrell RJ, Noonan N, Mahmud N, Morrin MM,Kelleher D, Keeling PW. Potential impact ofmagnetic resonance cholangiopancreatography onendoscopic retrograde cholangiopancreatographyworkload and complication rate in patientsreferred because of abdominal pain. Endoscopy2001;33:668–75.


51


References

24. Ponchon T, Pilleul F. Diagnostic ERCP. Endoscopy2002;34:29–42.

25. Cotton PB. Endoscopic retrogradecholangiopancreatography and laparoscopiccholecystectomy. Am J Surg 1993;165:474–8.

26. Adamek HE, Weitz M, Breer H, Jakobs R, SchillingD, Riemann JF. Value of magnetic-resonancecholangio-pancreatography (MRCP) afterunsuccessful endoscopic-retrograde cholangio-pancreatography (ERCP). Endoscopy 1997;29:741–4.

27. Cotton PB, Jowell PS, Baillie J. Spectrum ofcomplications after diagnostic ERCP and effects ofcomorbidities [abstract]. Gastrointest Endosc 1994;40abs P18.

28. Baillie J. Predicting and preventing post-ERCPpancreatitis. Curr Gastroenterol Rep 2002;4:112–19.

29. Freeman ML, Nelson DB, Sherman S, Haber GB,Herman ME, Dorsher PJ, et al. Complications ofendoscopic biliary sphincterotomy. N Engl J Med1996;335:909–18.

30. Department of Health. Department of HealthReference Costs 2002. London: Department ofHealth; 2003.

31. Cohen SA, Siegel JH, Kasmin FE. Complicationsof diagnostic and therapeutic ERCP. AbdomImaging 1996;21:385–94.

32. Coakley FV, Qayyum A. Magnetic resonancecholangiopancreatography. Gastrointest Endosc2002;55:S2–12.

33. Pavone P, Laghi A, Catalano C, Panebianco V,Fabiano S, Passariello R. MRI of the biliary andpancreatic ducts [review]. Eur Radiol 1999;9:1513–22.

34. Fulcher AS. MRCP and ERCP in the diagnosis ofcommon bile duct stones. Gastrointest Endosc2002;56:S178–82.

35. Watanabe Y, Dohke M, Ishimori T, Amoh Y, Oda K, Okumura A, et al. High-resolution MRcholangiopancreatography. Crit Reviews inDiagnostic Imaging 1998;39:115–258.

36. Takehara Y. Can MRCP replace ERCP? J MagnReson Imaging 1998;8:517–34.

37. Pavone P, Laghi A, Panebianco V, Catalano C,Lobina L, Passariello R. MR cholangiography:techniques and clinical applications. Eur Radiol1998;8:901–10.

38. Czako L, Endes J, Takacs T, Boda K, Lonovics J.Evaluation of pancreatic exocrine function bysecretin-enhanced magnetic resonancecholangiopancreatography. Pancreas2001;23:323–8.

39. Barkum AN. Common bile duct stones: magneticresonance cholangiopancreatography vs.endoscopic retrograde cholangiopancreatographyfor detection. J Gastrointest Surg 2000;4:332–3.

40. Jalan R, Sargentoni J, Coutts GA, Bell JD, Rolles K, Burroughs AK, Taylor Robinson SD.Hepatic phosphorus-31 magnetic resonancespectroscopy in primary biliary cirrhosis and itsrelation to prognostic models. Gut 1996;39:141–6.

41. Lomas DJ, Gimson A. Magnetic resonancecholangiopancreatography [review]. Hosp Med2000;61:395–9.

42. Catalano C, Pavone P, Laghi A, Panebianco V,Scipioni A, Fanelli F, et al. Pancreaticadenocarcinoma: combination of MR imaging, MRangiography and MR cholangiopancreatographyfor the diagnosis and assessment of resectability.Eur Radiol 1998;8:428–34.

43. Kalra M, Sahani D, Ahmad A, Saini S. The role ofmagnetic resonance cholangiopancreatography inpatients with suspected biliary obstruction [review].Cur Gastroenterol Rep 2002;4:160–6.

44. Urakami A, Majima T, Tsunoda T. A pitfall ofMRCP for preoperative cholangiography: reportof a case. Am J Gastroenterol 2000;95:1601–2.

45. Fulcher AS, Turner MA. Pitfalls of MRcholangiopancreatography (MRCP). J Comput AssistTomogr 1998;22:845–50.

46. Irie H, Honda H, Kuroiwa T, Yoshimitsu K, Aibe H, Shinozaki K, Masuda K. Pitfalls in MRcholangiopancreatographic interpretation.Radiographics 2001;21:23–37.

47. Lijmer JG, Mol BW, Heisterkamp S, Bonsel GJ,Prins MH, vanderMeulen JH, Bossuyt PM.Empirical evidence of design-related bias instudies of diagnostic tests. JAMA 1999;282:1061–6.

48. Greenhalgh T. How to read a paper: papers thatreport diagnostic or screening tests. BMJ 1997;315:540–3.

49. Deeks JJ. Systematic reviews of evaluations ofdiagnostic and screening tests. Eggers M, DaveySmith G, Altman DG editors. In Systematic reviewsin health care: meta-analysis in context. London: BMJPublishing Group; 2001. pp. 248–82.

50. Wilson EB. Probable inference, the law ofsuccession, and statistical inference. J Am Stat Assoc1927;22:209–12.

51. Newcombe RG, Altman DG, Machin D, BryantTN, Gardner MJ, editors. In Statistics withconfidence. Confidence intervals and statisticalguidelines. 2nd ed. London: BMJ Books; 2000.

52. Moses LE, Littenberg B, Shapiro D. Combiningindependent studies of a diagnostic test into asummary ROC curve: data-analytical approachesand some additional considerations. Stat Med1993;12:1293–316.

53. Littenberg B, Moses LE. Estimating diagnosticaccuracy from multiple conflicting reports: a newmeta-analytical method. Med Decis Making1993;13:313–21.

References

52

54. Menon K, Barkun AN, Romagnuolo J, Friedman G,Mehta SN, Reinhold C, Bret PM. Patientsatisfaction after MRCP and ERCP. Am JGastroenterol 2001;96:2646–50.

55. Adamek HE, Albert J, Weitz M, Breer H, Schilling D, Riemann JF. A prospective evaluationof magnetic resonance cholangiopancreatographyin patients with suspected bile duct obstruction.Gut 1998;43:680–3.

56. Alcaraz MJ, De la Morena EJ, Polo A, Ramos A,De la Cal MA, Gonzalez Mandly A. A comparativestudy of magnetic resonance cholangiography anddirect cholangiography. Revista Espanola deEnfermedades Digestivas 2000;92:427–38.

57. Angulo P, Pearce DH, Johnson CD, Henr, JJ,LaRusso NF, Petersen BT, Lindor KD. Magneticresonance cholangiography in patients with biliarydisease: its role in primary sclerosing cholangitis. J Hepatol 2000;33:520–7.

58. Barish MA, Yucel EK, Soto JA, Chuttani R,Ferrucci JT. MR cholangiopancreatography:efficacy of three-dimensional turbo spin-echotechnique. AJR Am J Roentgenol 1995;165:295–300.

59. Calvo MM, Bujanda L, Calderon A, Heras I,Cabriada JL, Bernal A, et al. Role of magneticresonance cholangiopancreatography in patientswith suspected choledocholithiasis. Mayo Clin Proc2002;77:422–8.

60. Chan YL, Chan AC, Lam WW, Lee DW, Chung SS,Sung, JJ, et al. Choledocholithiasis: comparison ofMR cholangiography and endoscopic retrogradecholangiography. Radiology 1996;200:85–9.

61. Dwerryhouse SJ, Brown E, Vipond MN.Prospective evaluation of magnetic resonancecholangiography to detect common bile ductstones before laparoscopic cholecystectomy. Br JSurg 1998;85:1364–6.

62. Feldman DR, Kulling DP, Kay CL, Cole DJ,Cunningham JT, Hawes RH, et al. Magneticresonance cholangiopancreatography: a novelapproach to the evaluation of suspectedpancreaticobiliary neoplasms. Ann Surg Oncol1997;4:634–8.

63. Guibaud L, Bret PM, Reinhold C, Atri M, Barkun AN. Bile duct obstruction andcholedocholithiasis: diagnosis with MRcholangiography. Radiology 1995;197:109–15.

64. Hintze RE, Adler A, Veltzke W, Abou-Rebyeh H,Hammerstingl R, Vogl T, Felix R. Clinicalsignificance of magnetic resonancecholangiopancreatography (MRCP) compared toendoscopic retrograde cholangiopancreatography(ERCP). Endoscopy 1997;29:182–7.

65. Holzknecht N, Gauger J, Sackmann M, Thoeni RF,Schurig J, Holl J, et al. Breath-hold MRcholangiography with snapshot techniques:prospective comparison with endoscopic retrogradecholangiography. Radiology 1998;206:657–64.

66. Laokpessi A, Bouillet P, Sautereau D, Cessot F,Desport JC, Le Sidaner A, Pillegand B. Value ofmagnetic resonance cholangiography in thepreoperative diagnosis of common bile ductstones. Am J Gastroenterol 2001;96:2354–9.

67. Lee MG, Lee HJ, Kim MH, Kang EM, Kim YH,Lee SG, et al. Extrahepatic biliary diseases: 3D MRcholangiopancreatography compared withendoscopic retrograde cholangiopancreatography.Radiology 1997;202:663–9.

68. Lomanto D, Pavone P, Laghi A, Panebianco V,Mazzocchi P, Fiocca F, et al. Magnetic resonance-cholangiopancreatography in the diagnosis ofbiliopancreatic diseases. Am J Surg 1997;174:33–8.

69. Lomas DJ, Bearcroft PW, Gimson AE. MR cholangiopancreatography: prospectivecomparison of a breath-hold 2D projectiontechnique with diagnostic ERCP. Eur Radiol1999;9:1411–17.

70. Macaulay SE, Schulte SJ, Sekijima JH, Obregon RG,Simon HE, Rohrmann Jr CA, et al. Evaluation of anon-breath-hold MR cholangiography technique.Radiology 1995;196:1–232.

71. Regan F, Fradin J, Khazan R, Bohlman M,Magnuson T. Choledocholithiasis: evluation withMR cholangiography. AJR Am J Roentgenol1996;167:1441–5.

72. Reinhold C, Taourel P, Bret PM, Cortas GA, Mehta SN, Barkun AN, et al. Choledocholithiasis:evaluation of MR cholangiography for diagnosis.Radiology 1998;209:435–42.

73. Soto JA, Barish MA, Yucel EK, Siegenberg D,Ferrucci JT, Chuttani R. Magnetic resonancecholangiography: comparison with endoscopicretrograde cholangiopancreatography.Gastroenterology 1996;110:589–97.

74. Soto JA, Barish MA, Alvarez O, Medina S.Detection of choledocholithiasis with MRcholangiography: comparison of three-dimensional fast spin-echo and single- andmultisection half-Fourier rapid acquisition withrelaxation enhancement sequences. Radiology2000;215:737–45.

75. Soto JA, Alvarez O, Munera F, Velez SM, Valencia J, Ramirez N. Diagnosing bile ductstones: comparison of unenhanced helical CT, oralcontrast-enhanced CT cholangiography, and MRcholangiography. AJR Am J Roentgenol 2000;175:1127–34.

76. Stiris MG, Tennoe B, Aadland E, Lunde OC. MR cholangiopancreaticography and endoscopicretrograde cholangiopancreaticography in patientswith suspected common bile duct stones. ActaRadiol 2000;41:269–72.

77. Sugiyama M, Baba M, Atomi Y, Hanaoka H,Mizutani Y, Hachiya J. Diagnosis of anomalous


53


pancreaticobiliary junction: value of magneticresonance cholangiopancreatography. Surgery1998;123:391–7.

78. Taylor AC, Little AF., Hennessy OF, Banting SW,Smith PJ, Desmond PV. Prospective assessment ofmagnetic resonance cholangiopancreatography fornoninvasive imaging of the biliary tree. GastrointestEndosc 2002;55:17–22.

79. Varghese JC, Liddell RP, Farrell MA, Murray FE,Osborne DH, Lee MJ. Diagnostic accuracy ofmagnetic resonance cholangiopancreatographyand ultrasound compared with directcholangiography in the detection ofcholedocholithiasis. Clin Radiol 2000;55:25–35.

80. Zidi SH, Prat F, Le Guen O, Rondeau Y, Rocher L,Fritsch J, et al. Use of magnetic resonancecholangiography in the diagnosis ofcholedocholithiasis: prospective comparison with areference imaging method. Gut 1999;44:118–22.

81. Pavone P, Laghi A, Passariello R. MR cholangiopancreatography in malignantbiliary obstruction. Semin Ultrasound CT MR1999;20:317–23.

82. Blackbourne LH, Earnhardt RC, Sistrom CL,Abbitt P, Jones RS. The sensitivity and role ofultrasound in the evaluation of biliary obstruction[review]. Am Surg 1994;60:683–90.

83. Luman W, Cull A, Palmer KR. Quality of life inpatients stented for malignant biliary obstructions.Eur J Gastroenterol Hepatol 1997;9:481–4.

84. Robins TN, Stiegmann GV, Durham JD, Johnson SI,Wachs ME, Serra AD, Kumpe DA. Management ofmajor bile duct injury associated with laparoscopiccholecystectomy. Surg Endosc 2001;15:1381–5.

85. NHS Executive Guidance on CommissioningCancer Services. Improving outcomes in uppergastro-intestinal cancers. The manual. 2001.London: NHS Executive Health ServicesDirectorate.

86. Cook J, Richardson J, Street A. A cost utilityanalysis of treatment options for gallstone disease:methodological issues and results. Health Econ1994;3:157–68.

87. Gregor J, Ponich T, Detsky A. Should ERCP beroutine after an episode of ‘idiopathic’pancreatitis? A cost–utility analysis. GastrointestEndosc 1996;44:118–23.

88. Bass E, Pitt HA, Lillemoe K. Cost-effectiveness oflaparoscopic cholecystectomy versus opencholecystectomy. Am J Surg 1993;165:466–71.

89. Luman W, Cull A, Palmer KR. Quality of life inpatients stented for malignant biliary obstructions.Eur J Gastroenterol Hepatol 1997;9:481–4.

90. Dawson D, Street A. Reference costs and thepursuit of efficiency in the ‘new’ NHS. Universityof York, CHE Discussion Paper; 1998. p. 161.

91. Freeman ML. Toward improving outcomes ofERCP: endoscopist’s experience: does it make adifference? Gastrointest Endosc 1998;48:96–102.

92. British Society of Gastroenterology EndoscopyCommittee. Provision of endoscopy relatedservices in district general hospitals. London: BSGWorking Party Report; 2001.

93. Royal College of Radiologists. Two national auditsof radiographic reporting services. London: RCR;2000.

94. Chilcott JB, Brennan A, Booth A, Karnon J,Tappenden P. The role of modelling in prioritisingand planning clinical trials. Health Technol Assess2003;7:23.

95. Arslan A, Geitung JT, Viktil E, Abdelnoor M,Osnes M. Pancreaticobiliary diseases. Comparisonof 2D single-shot turbo spin-echo MRcholangiopancreatography with endoscopicretrograde cholangiopancreatography. Acta Radiol2000;41:621–6.

96. Bearcroft PW, Gimson A, Lomas DJ. Non-invasivecholangio-pancreatography by breath-holdmagnetic resonance imaging: preliminary results.Clin Radiol 1997;52:345–50.

97. Becker CD, Grossholz M, Becke, M, Mentha G, de Peyer R, Terrier F. Choledocholithiasis and bileduct stenosis: diagnostic accuracy of MRcholangiopancreatography. Radiology1997;205:523–30.

98. Boraschi P, Neri E, Braccini G, Gigoni R,Caramella D, Perri G, Bartolozzi C.Choledocolithiasis: diagnostic accuracy of MRcholangiopancreatography. Three-year experience.Magn Reson Imaging 1999;17:1245–53.

99. Fulcher AS, Turne, MA, Capps GW, Zfass AM,Baker KM. Half-Fourier RARE MRcholangiopancreatography: experience in 300subjects. Radiology 1998;207:21–32.

100. Fulcher, AS, Turner MA, Franklin KJ, ShiffmanML, Sterling RK, Luketic VA, Sanyal AJ. Primarysclerosing cholangitis: evaluation with MRcholangiography – a case–control study. Radiology2000;215:71–80.

101. Georgopoulos SK, Schwartz LH, Jarnagin WR,Gerdes H, Breite I, Fong Y, et al. Comparison ofmagnetic resonance and endoscopic retrogradecholangiopancreatography in malignantpancreaticobiliary obstruction. Arch Surg1999;134:1002–7.

102. Hall-Craggs MA, Allen CM, Owens CM, Theis BA,Donald JJ, Paley M, et al. MR cholangiography:clinical evaluation in 40 cases. Radiology1993;189:423–7.

103. Hochwald SN, Dobryansky MBA, Rofsky NM, Naik KS, Shamamian P, Coppa G, Marcus SG.Magnetic resonance cholangiopancreatography

References

54

accurately predicts the presence or absence ofcholedocholithiasis. J Gastrointest Surg 1998;2:573–9.

104. Irie H, Honda H, Jimi M, Yokohat, K, Chijiiwa K,Kuroiwa T, et al. Value of MRcholangiopancreatography in evaluatingcholedochal cysts. AJR Am J Roentgenol 1998;171:1381–5.

105. Irie H, Honda H, Tajima T, Kuroiwa T, Yoshimitsu K, Makisumi K, Masuda K. OptimalMR cholangiopancreatographic sequence and itsclinical application. Radiology 1998;206:379–87.

106. Ishizaki Y, Wakayama T, Okada Y, Kobayashi T.Magnetic resonance cholangiography forevaluation of obstructive jaundice. Am JGastroenterol 1993;88:2072–7.

107. Laghi A, Pavone P, Catalano C, Messina A,Panebianco V, Passariell, R. Choledocholithiasis: adefinite role for MR cholangiography. Radiology1996;201:353, abs 1222.

108. Liu TH, Consorti ET, Kawashima A, Ernst RD,Black CT, Greger PH Jr, et al. The efficacy ofmagnetic resonance cholangiography for theevaluation of patients with suspectedcholedocholithiasis before laparoscopiccholecystectomy. Am J Surg 1999;178:480–4.

109. Magnuson TH, Bender JS, Duncan MD, AhrendtSA, Harmon JW, Regan F. Utility of magneticresonance cholangiography in the evaluation ofbiliary obstruction. J Am Coll Surg 1999;189:63–71.

110. Matos C, Nicaise N, Deviere J, Cassart M, Metens T, Struyven J, Cremer M. Choledochalcysts: comparison of findings at MRcholangiopancreatography and endoscopicretrograde cholangiopancreatography in eightpatients. Radiology 1998;209:443–8.

111. Mendler MH, Bouillet P, Sautereau D,Chaumerliac P, Cessot F, Le Sidaner A, PillegandB. Value of MR cholangiography in the diagnosisof obstructive diseases of the biliary tree: a studyof 58 cases. Am J Gastroenterol 1998;93:2482–90.

112. Miyazaki T, Yamashita Y, Tsuchigame T, Yamamoto H, Urata J, Takahashi M. MRcholangiopancreatography using HASTE (half-Fourier acquisition single-shot turbo spin-echo)sequences. AJR Am J Roentgenol 1996;166:1297–303.

113. Musella M, Barbalace G, Capparelli G, Carrano A,Castaldo P, Tamburrini O, Musella S. Magneticresonance imaging in evaluation of the commonbile duct. Br J Surg 1998;85:16–19.

114. Ng KK, Wan YL, Lui KW, Wong HF, Hung CF,Kong MS, Chiu CT. Three-dimensional magneticresonance cholangiopancreatography forevaluation of obstructive jaundice. J Formos MedAssoc 1997;96:586–92.

115. Pavone P, Laghi A, Catalano C, Broglia L, Messina A,Scipioni A, et al. MR cholangiopancreatography(MRCP) at 0.5 T: technique optimisation andpreliminary results. Eur Radiol 1996;6:147–52.

116. Pavone P, Laghi A, Catalano C, Broglia L, Fiocca F,Passariello R. Non-invasive evaluation of thebiliary tree with magnetic resonancecholangiopancreatography: initial clinicalexperience. Ital J Gastroenterol 1996;28:63–9.

117. Pavone P, Laghi A, Catalano C, Broglia L,Panebianco V, Messina A, et al. MRcholangiography in the examination of patientswith biliary-enteric anastomoses. AJR Am JRoentgenol 1997;169:807–11.

118. Regan F, Smith D, Khazan R, Bohlman M,Schultze-Haakh H, Campion J, Magnuson TH.MR cholangiography in biliary obstruction usinghalf-Fourier acquisition. J Comput Assist Tomogr1996;20:627–32.

119. Rösch T, Meining A, Fruhmorgen S, Zillinger C,Schusdziarra V, Hellerhoff K, et al. A prospectivecomparison of the diagnostic accuracy of ERCP,MRCP, CT, and EUS in biliary strictures.Gastrointest Endosc 2002;55:870–6.

120. Sarli L, Costi R, Gobbi S, Sansebastiano G,Roncoroni L. Asymptomatic bile duct stones:selection criteria for intravenous cholangiographyand/or endoscopic retrograde cholangiographyprior to laparoscopic cholecystectomy. Eur JGastroenterol Hepatol 2000;12:1175–80.

121. Schwartz LH, Coakley FV, Sun Y, Blumgart LH,Fong Y, Panicek DM. Neoplastic pancreaticobiliaryduct obstruction: evaluation with breath-hold MRcholangiopancreatography. AJR Am J Roentgenol1998;170:1491–5.

122. Sugiyama M, Atomi Y, Hachiya J. Magneticresonance cholangiography using half-Fourieracquisition for diagnosing choledocholithiasis. AmJ Gastroenterol 1998;93:1886–90.

123. Taourel P, Bret PM, Reinhold C, Barkun AN, Atri M. Anatomic variants of the biliary tree:diagnosis with MR cholangiopancreatography.Radiology 1996;199:521–7.

124. Topal BV. The value of magnetic resonancecholangiopancreatography in predicting commonbile duct stones in patients with gallstone disease.Br J Surg 2003;90:42–7.

125. Tripathi RP, Batra A, Kaushik S. Magnetic resonancecholangiopancreatography: evaluation in 150patients. Indian J Gastroenterol 2002;21:105–9.

126. Varghese JC, Liddell RP, Farrell MA, Murray FE,Osborne H, Lee MJ. The diagnostic accuracy ofmagnetic resonance cholangiopancreatographyand ultrasound compared with directcholangiography in the detection ofcholedocholithiasis [comment]. [Published


55


References

56

erratum appears in Clin Radiol 2000;55:657.] Clin Radiol 1999;54:604–14.

127. Varghese JC, Farrell MA, Courtney G, Osborne H,Murray FE, Lee MJ. A prospective comparison ofmagnetic resonance cholangiopancreatographywith endoscopic retrogradecholangiopancreatography in the evaluation ofpatients with suspected biliary tract disease. ClinRadiol 1999;54:513–20.

128. Vitellas KM, El Dieb A, Vaswani KK, Bennett WF,Tzalonikou M, Mabee C, et al. MRcholangiopancreatography in patients withprimary sclerosing cholangitis: interobservervariability and comparison with endoscopicretrograde cholangiopancreatography. AJR Am JRoentgenol 2002;179:399–407.

129. Yamashita Y, Abe Y, Tang Y, Urata J, Sumi S,Takahashi M. In vitro and clinical studies of imageacquisition in breath-hold MRcholangiopancreatography: single-shot projectiontechnique versus multislice technique. AJR Am JRoentgenol 1997;168:1449–54.

130. Yeh TS, Jan YY, Tseng JH, Chiu CT, Chen TC,Hwang TL, Chen MF. Malignant perihilar biliaryobstruction: magnetic resonancecholangiopancreatographic findings. Am JGastroenterol 2000;95:432–40.

131. Zidi SH, Prat F, Le Guen O, Rondeau Y, Pelletier G. Performance characteristics ofmagnetic resonance cholangiography in thestaging of malignant hilar strictures. Gut2000;46:103–6.


57


Electronic bibliographicdatabases searched1. BIOSIS previews (the new online version of

Biological Abstracts)2. CCTR (Cochrane Controlled Trials Register)3. CDSR (Cochrane Database of Systematic

Reviews)4. CINAHL5. EMBASE

6. HEED (Health Economic EvaluationsDatabase)

7. MEDLINE8. NHS DARE (Database of Assessments of

Reviews of Effectiveness)9. NHS EED (Economic Evaluations Database)10. NHS HTA 11. Pre-MEDLINE12. Science Citation Index13. Social Sciences Citation Index

Appendix 1

Search strategies


59


1. AHRQ (Agency for Healthcare Research andQuality), USA

2. Bandolier3. British Dietetic Association

a. CCOHTA (Canadian Coordinating Officefor Health Technology Assessment)

4. CenterWatch5. CHE (Centre for Health Economics), York6. CliniWeb7. CMA (Canadian Medical Association)

InfoBase8. COIN (DoH)9. Current Controlled Trials10. Development and Evaluation Committee

(Department of Health)11. DES Reports (West Midlands Health

Technology Assessment Collaboration)12. DoH13. eGuidelines14. EMEA (The European Agency for the

Evaluation of Medicinal Products)15. Google16. HSRU (Health Services Research Unit),

Aberdeen18. INAHTA (International Network of Agencies

for Health Technology Assessment)Clearinghouse

19. Index to Theses (Sheffield University)20. MDChoice21. MeRec

22. MRC Trials Register23. National Assembly for Wales24. National Guidelines Clearinghouse25. National Research Register (2002 Issue 2)26. NCCHTA (National Co-ordinating Centre for

Health Technology Assessment)27. NHS CRD (Centre for Reviews and

Dissemination), University of York28. NeLH (National Electronic Library for

Health)29. New Zealand Health Technology Clearing

House for Health Outcomes and HealthTechnology Assessment (NZHTA)

30. NICE (National Institute for ClinicalExcellence)

31. OMNI32. POINT (DoH)33. RAND34. ReFeR (Research Findings Register)35. ScHARR Library catalogue36. SIGN (Scottish Intercollegiate Guidelines

Network)37. TRIP (Turning Research into Practice)

Database38. TWGAP (Trent Institute Working Group on

Acute Purchasing)39. US FDA (Federal Food and Drug

Administration)40. WHO

Appendix 2

Other sources searched

BIOSIS Previews1985–2003SilverPlatter WebSPIRSSearch undertaken January 2003

#1 mri or mri scan* or mri imag*#2 ‘nuclear-magnetic-resonance-imaging’#3 #1 or #2#4 biliary tract* or biliary tree* or biliary#5 ‘hepatobiliary-system’#6 #4 or #5#7 #3 and #6

CDSR, CCTR and DAREOvid OnlineSearch undertaken January 2003

1 exp *magnetic resonance imaging/2 mri.tw3 magnetic resonance.tw4 mrcp.tw5 ((noninvasive$ or non-invasive$) adj3

(diagnos$ or imag$)).tw6 or/1-57 bilary.tw8 biliary.tw9 exp *biliary tract/10 exp *biliary tract diseases/11 bile.ti12 *bile/13 gall bladder$.tw14 cholestat$.tw15 choledocholethias$.tw16 cholangiopancreatography, endoscopic

retrograde/17 or/7-1618 6 and 17

CINAHL1982–2003Ovid OnlineSearch undertaken January 2003

1 exp *magnetic resonance imaging//2 mri.tw

3 magnetic resonance.tw4 mrcp.tw5 ((noninvasive$ or non-invasive$) adj3

(diagnos$ or imag$)).tw6 or/1-57 bilary.tw8 biliary.tw9 exp *biliary tract/10 exp *biliary tract diseases/11 bile.ti12 *bile/13 gall bladder$.tw14 cholestat$.tw15 choledocholethias$.tw16 cholangiopancreatography, endoscopic


Citation Indexes (Science andSocial Sciences)1981–2003Web of ScienceSearch undertaken January 2003

(Magnetic resonance imag* OR mri) AND biliary*AND (guideline* OR systematic review* OR trial*OR economic* OR pricing* OR cost*)

CRD databases (NHS DARE,EED, HTA)CRD website: complete databasesSearch undertaken January 2003

[(magnetic resonance imag OR mri) AND (bilaryOR biliary)]/all fields

EMBASE1980–2003SilverPlatter WebSPIRSSearch undertaken January 2003

#1 mri#2 ‘nuclear-magnetic-resonance-imaging’


61


Appendix 3

Search strategies used

Appendix 3

62

#3 #1 or #2#4 biliary tract* or biliary tree* or biliary#5 ‘hepatobiliary-system’#6 #4 or #5#7 #3 and #6

HEEDCD ROM versionSearch undertaken January 2003

Search termsBiliaryBilaryMRIMagnetic resonance imag*

Fields searchedQuick search – All data

MEDLINE1966–2003Ovid OnlineSearch undertaken January 2003

1 exp *magnetic resonance imaging/2 mri.tw3 magnetic resonance.tw4 mrcp.tw5 ((noninvasive$ or non-invasive$) adj3

(diagnos$ or imag$)).tw6 or/1-5

7 bilary.tw8 biliary.tw9 exp *biliary tract/10 exp *biliary tract diseases/11 bile.ti12 *bile/13 gall bladder$.tw14 cholestat$.tw15 choledocholethias$.tw16 cholangiopancreatography, endoscopic


Pre-MEDLINEJanuary 2003Ovid OnlineSearch undertaken January 2003

1 mri.tw2 magnetic resonance.tw3 mrcp.tw4 ((noninvasive$ or non-invasive$) adj3

(diagnos$ or imag$)).tw5 or/1-46 bilary.tw7 biliary.tw8 bile.ti9 gall bladder$.tw10 cholestat$.tw11 choledocholethias$.tw12 or/6-1113 5 and 12


63


Guidelines1 guideline.pt2 practice guideline.pt3 exp guidelines/4 health planning guidelines/5 or/1-4

Systematic reviews1 meta-analysis/2 exp review literature/3 (meta-analy$ or meta analy$ or

metaanaly$).tw4 meta analysis.pt5 review academic.pt6 review literature.pt7 letter.pt8 review of reported cases.pt9 historical article.pt10 review multicase.pt11 or/1-612 or/7-1013 11 not 12

Clinical trials1 Clinical.pt

Economic evaluations1 economics/2 exp “costs and cost analysis”/3 economic value of life/

4 exp economics, hospital/5 exp economics, medical/6 economics, nursing/7 economics, pharmaceutical/8 exp models, economic/9 exp “fees and charges”/10 exp budgets/11 ec.fs12 cost$.ti13 (economic$ or pharmacoeconomic$ or price$

or pricing).ti14 or/1-13

Quality of life1 exp quality of life/2 quality of life.tw3 life quality.tw4 hql.tw5 (sf 36 or sf36 or sf thirtysix or sf thirty six or

short form 36 or short formthirty six or shortform thirtysix or shortform 36).tw

6 qol.tw7 (euroqol or eq5d or eq 5d).tw8 qaly$.tw9 quality adjusted life year$.tw10 hye$.tw11 health$ year$ equivalent$.tw12 health utilit$.tw13 hui.tw14 quality of wellbeing$.tw15 quality of well being.tw16 qwb.tw17 (qald$ or qale$ or qtime$).tw18 or/1-17

Appendix 4

Methodological search filters used in Ovid MEDLINE


65


Study Reason for exclusion

Adamek et al., 199726 Comparison with failed ERCPArslan et al., 200095 Retrospective reviewBearcroft et al., 199796 Includes pancreatic duct abnormalitiesBecker et al., 199797 Retrospective reviewBoraschi et al., 199998 Multiple comparators, only subgroup had ERCP and almost half had no comparator at allFulcher et al., 199899 Multiple comparators, less than half had ERCPFulcher et al., 2000100 Case–control designGeorgopoulos et al., 1999101 Retrospective reviewHall-Craggs et al., 1993102 Before 1995Hochwald et al., 1998103 Retrospective reviewIrie et al., 1998104 Retrospective reviewIrie et al., 1998105 Paper deals with diagnosing pancreatic abnormalitiesIshizaki et al., 1993106 Before 1995Laghi et al., 1996107 Same data as Lomanto et al.68

Liu et al., 1999108 MRCP results affected decision to proceed with ERCPMagnuson et al., 1999109 Unclear as to number of patients who had ERCP, multiple comparatorsMatos et al., 1998110 Paper dealt with pancreatitisMendler et al., 1998111 Multiple comparators, less than half had ERCPMiyazaki et al., 1996112 Less than half had ERCP, main comparator was PTCMusella et al., 1998113 Patients separated into groups after MRCP, only one of which had ERCP. Small subset of

patients (n = 9) had ERCPNg et al., 1997114 Retrospective reviewPavone et al., 1996115 Same data as Lomanto et al.68

Pavone et al., 1996116 Same data as Lomanto et al.68

Pavone et al., 1997117 Same data as Lomanto et al.68

Regan et al., 1996118 Trial data repeated71

Rösch et al., 2002119 Number of patients having ERCP is not stated separately from the ERCP/PTC groupSarli et al., 2000120 Case–control designSchwartz et al., 1998121 No data on ERCP reportedSugiyama et al., 1998122 No data comparing ERCP with MRCPTaourel et al., 1996123 Overlap with Reinhold et al.72

Topal et al., 2003124 Retrospective reviewTripathi et al., 2002125 Retrospective reviewVarghese et al., 1999126 Same data as Varghese et al.79

Varghese et al., 1999127 Patient overlap with Varghese et al.79

Vitellas et al., 2002128 Retrospective reviewYamashita et al., 1997129 No results comparing MRCP and ERCP diagnosis are reportedYeh et al., 2000130 Retrospective reviewZidi et al., 2000131 Patients referred for stenting rather than diagnosis

Appendix 5

Excluded studies


67


Appendix 6

Evidence tables

Appendix 6

68

TABLE 15 Study characteristics

Study, country Sample selection Comparison (type of MRCP and Description of patients and procedures; study period Time between ERCP reference tests used) and MRCP

Adamek et al.,199855;Germany

Not reported RARE and HASTE MRCP compared withERCP

86 patients entered the study; 8 were excluded due tobiliary–enteric anastomoses, of the remaining 78, 16 hadunsatisfactory ERCP and 2 had unsatisfactory MRCP(claustrophobia), leaving 60 patients who had both

January–December 1996

Not reported

Alcaraz et al.,200056; Spain

Not reported T2-weighted HASTE and RARE MRCPcompared with ERCP, PTC and surgery

81 patients had MRCP, 70 had ERCP, 7 had PTC and 4 hadsurgery

October 1997–February 1998

Not reported

Angulo et al.,200057; USA

Not reported FSE pulse sequence compared with ERCPand PTC

Initially 74; 1 did not receive MRCP owing to claustrophobia,73 had MRCP, 68 had ERCP, 2 had PTC and 3 had neither

Study period not stated

MRCP performed within24 hours before thescheduled ERCP

Barish et al.,199558; USA

Random selection fromreferrals

3D TSE MRCP compared with ERCP andPTC

30 patients initially selected; 1 patient did not receive MRCPowing to the presence of ascitic fluid in the upper abdomen; 3 had PTC owing to failed ERCP, 8 of the 29 patients did nothave ERCP or PTC

Study period not reported

ERCP performed 8 hoursafter MRCP

Calvo et al.,200259; Spain

Not reported Two HASTE sequences MRCP comparedwith ERCP

116 patients with suspected biliopancreatic pathology initially,of these 61 patients were selected with suspectedcholedocholithiasis, failure in 1 patient for ERCP

November 1996–February 1998

MRCP within 72 hoursbefore ERCP

Chan et al.,199660; Hong Kong

Consecutive sample T2-weighted TSE sequence (non-breath-holding, fat-suppressed) MRCP comparedwith ERCP

47 had MRCP, 45 had ERCP (two failures)

May–August 1995

ERCP within 5 hours afterMRCP

Demartines et al., 200021;Switzerland

Not reported 3 acquisition techniques of MRCP used,including T2/T1 weighted, single-shot TSEand HASTE heavy sequence comparedwith ERCP (high-risk patients) or IOC(moderate-risk patients)

40 patients received ERCP and MRCP, and 30 received IOCand MRCP

April 1997–September 1998

Not reported

continued


ent2004; Vol. 8: No. 10

© Q


MSO


TABLE 15 Study characteristics (cont’d)


Dwerryhouse et al., 199861;UK

Not reported T2-weighted TSE with non-breath-holdingMRCP compared with ERCP and PTC

Initially 405 patients who underwent laparoscopiccholecystectomy; 278 had no known risk factors for CBDstones, 87 underwent early ERCP and were excluded. 40patients with risk factors for CBD stones underwent MRCP. 2patients had failed MRCP owing to claustrophobia, ERCP wasunsuccessful in 4 patients, who then had perioperativecholangiography

February 1996–January 1998

All patients underwentERCP within 1 week afterERCP

Feldman et al.,199762; USA

Not reported Fat-saturated heavily T2-weighted FSEMRCP compared with ERCP

20 patients had MRCP and 17 had ERCP


MRCP either before orafter ERCP, but timebetween the two notreported

Guibaud et al.,199563; Canada

Consecutive 2D FSE MRCP compared with ERCP, PTC,T-tube cholangiography, surgery andautopsy

Initially 198 patients; 72 were excluded owing to no proof ofbile duct obstruction (n = 42), unsuccessful ERCP (n = 12),unsuccessful MRCP owing to claustrophobia (n = 6),inadequate ERCP (n = 10) or MRCP (n = 2), leaving 126patients

September 1992–March 1993

Time between MRCP andfinal diagnosis < 6 hoursin 105 cases, < 1 week in15 cases and > 1 week insix cases

Hintze et al.,199764;Germany

Not reported T2-weighted and fat-suppressed MRCPcompared with ERCP

78 patients examined with both MRCP and ERCP; 1 patientexcluded because of claustrophobia making MRCP impossible,and did not have ERCP owing to a malignant duodenalstenosis; of the 78 patients, 55 had examination of the biliaryduct systems (the other 36 had examination of the pancreaticduct systems)

September 1995–September 1996

ERCP within 24 hoursafter MRCP

Holzknecht et al., 199865;Germany

Consecutive RARE and half-Fourier RARE MRCPcompared with ERCP

66 patients were eligible, 2 were excluded because ofpacemakers, 3 had failed ERCP after MRCP, leaving 61patients who had both MRCP and ERCP

June 1995–April 1996

MRCP performed beforeERCP (patients were dueto have ERCP within thenext 2 days)

continued

69

Appendix 6

70



Laokpessi et al.,200166; France

Consecutive FSE and heavily T2-weighted single-shotFSE sequences with fat-suppression MRCPcompared with ERCP or IOC

Initially 166 inpatients, but only 147 patients had MRCP; ofthese 101 had ERCP and 45 had IOC and cholecystectomy.Those in the group receiving ERCP had a past history ofcholecystectomy or a high surgical or anaesthetic risk. 21removed from study for refusal to sign protocol (n = 3),refusal to undergo MRCP (n = 4) or ERCP (n = 7), orexcessive time between MRCP and final diagnosis (n = 7)

November 1997–December 1999

Average time betweenMRCP and final diagnosis10 hours (range 3–48 hours); if > 48hours between MRCPand final diagnosis,patients were removedfrom the study

Lee et al.,199767; SouthKorea

Consecutive 3D steady-state free-precession MRCPcompared with ERCP

71 patients; 25 were excluded (8 because ERCP was notperformed, 15 who were evaluated for intrahepatic stones, 1 for peripheral type of intrahepatic cholangiocarcinoma, 1 suspected mucinous ductal ectasia of the pancreas), leaving46 patients who had both MRCP and ERCP

January–March 1995

33 patients had MRCPbefore ERCP, rangingfrom 6 hours to 5 days.The remaining 31 patientshad ERCP first, rangingfrom 3 to 16 days

Lomanto et al.,199768; Italy

Not reported T2-weighted TSE sequence MRCPcompared with ERCP and PTC

136 patients referred for MRCP, 62 had MRCP forcholedocholithiasis (the other 74 were: 48 for stenosis of thebiliary tract, 15 with previous hepaticojejunostomy andcholedochojejunostomy, and 11 with chronic pancreatitis), 60 of these patients had ERCP and 2 had PTC

September 1994–October 1995

Not reported

Lomas et al.,199969; UK

Not reported Hybrid four-shot RARE (FSE) sequenceand single-shot half-Fourier RAREsequence compared with ERCP

76 referrals, 2 did not have MRCP (1 was obese and 1 wasclaustrophobic), 5 did not have ERCP (1 died, 1 refused and in3 patients the operator was unable to cannulate the CBD),leaving 69 referrals in 66 patients

18-month period, dates not stated

MRCP took place firstwithin 4 hours of ERCP

Macaulay, et al.,199570; USA

Sequential T2-weighted TSE MRCP (non-breath-holding) compared with ERCP, PTC andIOC

28 patients initially had MRCP, 24 patients had 28 directcholangiographic studies (21 had ERCP, 6 had PTC and 1 hadIOC)


continued

ERCP took place within1–4 hours in 15 patients,4 were within 5–7 daysafter MRCP, 1 was 11days before and 1 was109 days before MRCP, allPTC studies were within2 days after MRCP andthe 1 IOC precededMRCP by 5 days


ent2004; Vol. 8: No. 10

71

© Q


MSO




Regan et al.,199671; USA

Not reported HASTE MRCP compared with ERCP andsonography

26 patients; 2 had unsuccessful ERCP and 1 did not haveMRCP owing to claustrophobia, leaving 23 patients

MRCP performed justbefore ERCP in 18 patients and within 24 hours in 5 patients

Reinhold et al.,199872; Canada

Consecutive FSE MRCP compared with ERCP, IOC andsurgery

Initially 159 patients; 49 were excluded for the followingreasons: 34 lack of diagnosis, 10 unsuccessful ERCP, 3unsuccessful MRCP claustrophobia (n = 1), inadequate ERCP(n = 1) or MRCP (n = 1), leaving a sample of 110 patients.101 patients had ERCP, 2 had IOC and 7 had surgery

5-month study period, dates not reported

MRCP performed firstand ERP or equivalentwas ≤ 6 hours later in 97patients, < 1 week in 7patients and > 1 week in6 patients

Soto et al.,199673; USA

Randomly recruited 3D FSE MRCP compared with ERCP andPTC

46 patients, 7 of whom were included in Barish et al., 199558;45 had ERCP and 1 had PTC

May 1994–April 1995

ERCP/PTC within 24 hours after MRCP

Soto et al.,200074;Columbia

Not reported 3D FSE, single-section half-Fourier RAREand multisection half-Fourier RARE MRCPcompared with ERCP

Initially 59 patients, 10 were excluded for the followingreasons: 2 owing to MRCP contraindications, 4 because 1 ormore of the 3 MRCP sequences could not be completed, and4 because ERCP could not be completed

August 1997–May 1998

MRCP before ERCP,within 72 hours

Soto et al.,200075;Columbia

Not reported Breath-hold, single-shot half-Fourier rapidacquisition and non-breath-holding 3D FSEMRCP compared with ERCP, CT and oralcontrast-enhanced CT cholangiography

Initially 68 patients; 12 did not meet inclusion or exclusioncriteria, 2 did not have MRCP because of claustrophobia, in 3ERCP was not attempted or completed, leaving 51 patientswho had all 4 studies

April 1998–March 1999

MRCP within 48 hoursbefore ERCP

Stiris et al.,200076; Norway

Consecutive HASTE fat-suppressed breath-holdingMRCP compared with ERCP

50; all patients had both techniques

Study period not stated

MRCP performed first,followed by ERCP within12 hours

Sugiyama et al.,199877; Japan

Non-consecutive HASTE MRCP compared with ERCP 187 patients were recruited; 19 underwent onlycholangiography or pancreatography on ERCP, in 8 thecommon channel could not be identified clearly and there wasfailure of cannulation in 2 patients, leaving 159 patients withCBD, main pancreatic duct and common channel depicted

June 1994–August 1996

MRCP 0–14 days beforeERCP

continued

Appendix 6

72



Taylor et al.,200278;Australia

Consecutive HASTE MRCP compared with ERCP, PTCor surgery

Initially 149 procedures (146 patients); MRCP unsuccessful in8 owing to claustrophobia and in 1 patient owing to poorimage quality, 5 were excluded because MRCP was more than24 hours before ERCP, in 20 ERCP was unsuccessful (3 hadsubsequent ERCP, 2 had surgery and 2 had PTC and wereincluded). In 2 patients ERCP and MRCP were bothunsuccessful, leaving 129 patients who had both MRCP andERCP (or equivalent)

November 1998–December 1999

MRCP performed within24 hours before ERCP

Textor et al.,20026; Germany

Consecutive 3D T2-weighted FSE MRCP comparedwith ERCP

150 patients initially; 146 had successful MRCP, 3 patients withPSC had unsuccessful ERCP and 1 failed owing to abilidigestive anastomosis

January 1996–December 2000

ERCP performed 1–14 days before MRCP(mean 3.2 days)

Varghese et al.,200079; Ireland

Consecutive T2-weighted 2D multislice FSE MRCPcompared with ERCP, PTC or IOC

256 patients initially; 64 excluded because ultrasound reportor ERCP hard-copy images were not available (n = 30), directcholangiography was not performed after failed ERCP (n = 22), MRCP not performed owing to contraindications (n = 5) or MRCP images were of non-diagnostic quality (n = 7), resulting in 191 patients [of these 34 hadcholedocholithiasis diagnosed by ERCP (n = 29), IOC (n = 3)and PTC (n = 2)]

18-month period, dates not stated

MRCP performed beforeERCP, within 4 hours to 2 weeks (mean 18 hours)

Zidi et al.,199980; France

Consecutive Non-breath-holding fat-suppressed TSEMRCP compared with ERCP (with orwithout sphincterotomy), endosonographyor IOC

70 inpatients were included, 63 had ERCP, 5 had sonographyand 2 had IOC

12-month period, dates not reported

MRCP performed within12 hours before ERCP


ent2004; Vol. 8: No. 10

73

© Q


MSO


TABLE 16 Patient characteristics

Study Age (years) Gender (male/female) Suspected condition Patient inclusion criteria Exclusions

Adamek et al.,199855

Mean 64.4(range 11–78)

31/29 CBD obstruction Raised alkaline phosphatase or γ-glutamyltranspeptidase more than twicenormal value and serum bilirubin > 2 �g/dl or morphological features onabdominal ultrasonography

8 of the original 86 patients wereexcluded owing to former operationswith bilary–enteric anastomosis (Roux-en-Y gastrojejunostomy orWhipple’s procedure)

Alcaraz et al.,200056

Mean notreported (range40–90)

25/56 Obstruction of the biliary tree Obstruction of the biliary tree based onclinical, laboratory and/or ultrasoundfindings

Not reported

Angulo et al.,200057

Mean 56 (range19–94)

33/40 Symptoms consistent with biliarydisease (cholestasis)

Males and females, age ≥ 18 years;clinical and/or biochemical evidence ofcholestasis

Usual contraindications to MR scanning

Barish et al.,199558


8/22 Suspected biliary or pancreaticdisease (reported separately)

ERCP referrals MRI contraindicated

Calvo et al.,200259

Mean 67 years(range notreported) (inoriginal 116patients)

Ratio 1.08:1 (inoriginal 116 patients)

Choledocholithiasis Patients with suspected bilipancreaticpathology requiring ERCP betweenNovember 1996 and February 1998; age> 18 years; all patients in whom ERCPwas started were included; all patients inwhom MRCP was started wereincluded; informed written consent

Patients with at least 1 absolutecontraindication to either technique;patients with degenerative or ankyloticconditions, senile dementia orimpossibility of patient cooperation inMRCP; patients with severe clinicalconditions with urgent therapeuticrequirements

Chan et al.,199660


27/20 Choledocholithiasis Hospital inpatients referred forendoscopy because of right upperquadrant or epigastric pain, jaundice ordark-coloured urine, fever orbiochemical jaundice

Not reported

Demartines et al., 200021

Mean 59.6 ±15.4 (for all 70patients in thestudy), 62.6 ±18.2 (high-riskgroup who hadERCP) (rangenot reported)

Ratio 3:1 (for all 70patients in the study),ratio 6:1 (for high-riskgroup who had ERCP)

Symptomatic cholelithiasis andsuspected CBD stones

Elevation of bilirubin level ≥ 26 �mol/l(<1.5 mg/dl); alkaline phosphatase level> 216 U/l; CBDS demonstrated by anyimaging modality; CBD diameter ≥ 8 mm (sonography); biliary pancreatitis

3 patients excluded because ERCP wastechnically not possible (1 each withduodenal stenosis, papillary oedema andlarge duodenal diverticulum)

continued

Appendix 6

74

TABLE 16 Patient characteristics (cont’d)


Dwerryhouse et al., 199861

Not reported Not reported CBD stones Abnormal LFT results, previous mildgallstone pancreatitis, dilated CBD of 7 mm or more, previous jaundice

Patients with jaundice, cholangitis orsevere acute gallstone pancreatitis

Feldman et al.,199762

Not reported Not reported Pancreaticobiliary neoplasm Clinical diagnosis of suspectedpancreaticobiliary neoplasm

Not reported

Guibaud et al.,199563


50/76 Bile duct obstruction Patients presented with clinicalsymptoms and/or results of biochemicalstudies consistent with bile ductobstruction

As stated in Description of patients(Table 15)

Hintze et al.,199764


38/40 Disorder affecting biliary orpancreatic duct system

Not previously undergone MRCP orERCP and no definite diagnosis

Not reported

Holzknecht et al., 199865

Mean 55.8 ±17.9 (range14–84)

30/31 Not reported Patients due to have ERCP within thenext 2 days

Patients with contraindications for MRI

Laokpessi et al.,200166

Mean 59.8(range 18–94)for whole group

67/80 for whole group Choledocholithiasis Past history of cholecystectomy or thosewith a high surgical or anaesthetic risk ingroup receiving ERCP and MRCP

Absolute contraindications to MRI, timespan of greater than 48 hours betweenMRCP and final diagnosis; refusal to signprotocol, refusal to undergo MRCP orERCP

Lee et al.,199767


22/24 Biliary disease Suspected biliary disease As stated in Description of patients(Table 15)

Lomanto et al.,199768

Mean 56.3 ± 6(range 20–83)

24/38 Choledocholithiasis Recurrent episodes of jaundice, pain,elevation in bilirubin, alaninetransaminase, aspartate transaminase,alkaline phosphatase, γ-glutamytranspeptidase, amylase,ultrasonographic finding of dilated bileducts (> 6 mm) or suspicion of CBDstones

Not reported

continued


ent2004; Vol. 8: No. 10

75

© Q


MSO




Lomas et al.,199969

Mean notreported(range: 21–92)

33/33 Biliary strictures orcholedocholithiasis

Sonographic evidence of calculi withinthe gallbladder combined with a dilatedcommon bile duct, history of priorjaundice with known gallbladder calculi,known gallbladder calculi and abnormalbiochemical liver function (other causesexcluded), recurrent pain or dilatedCBD or abnormal liver function aftercholecystectomy; strictures suspected inpatients with dilatation with or withoutevidence of a mass lesion or abnormalLFTs

Patients were excluded if the diagnosiswas known and therapeuticinterventions were planned. Alsoexcluded were those patients with theusual contraindications for MRI

Macaulay, et al.,199570


27/1 Biliary obstruction Patients with suspected biliaryobstruction scheduled to undergo ERCP

Not reported

Regan et al.,199671


10/13 CBD stones Clinical suspicion or sonographicevidence of CBD stones

Not reported

Reinhold et al.,199872

Mean 55 ±10.5 (range11–89)

47/63 CBD obstruction Clinical symptoms and/or biochemicalstudy results consistent with CBDobstruction


Soto et al.,199673


Not reported Not reported Patients referred for elective ERCP Not reported

Soto et al.,200074


14/35 Choledocholithiasis Patients with suspectedcholedocholithiasis referred for ERCP


Soto et al.,200075


19/32 Choledocholithiasis 18 years or older Bilirubin > 5 mg/dl, knownhyperuricaemia, creatinine level > 1.3 mg/dl, contraindications for MRCP

Stiris et al.,200076


13/37 CBD stones Patients with clinically and laboratorysuspected CBD stone disease

Not reported

continued

Appendix 6

76



Sugiyama et al.,199877


90/97 Pancreatobiliary disease Not reported Not reported

Taylor et al.,200278

Mean 60 (range17–94

60/69 Biliary tract disease in 96% (theother 4% had pancreatic ductdisease)

Patients scheduled for ERCP Less than 16 years old, inability to giveinformed consent, contraindication toMRCP, patients unable to followinstructions or hold breath for 20 seconds in supine position

Textor et al.,20026

Mean 48.6(range notreported)

67/83 PSC Progressive fatigue, pruritus followed byicterus and/or elevated values foralkaline phosphatase and serumaspartate transaminase and occasionallyan elevated serum concentration ofbilirubin

Not reported

Varghese et al.,200079


76/115 Patients referred for diagnosticERCP

Not reported As stated in Description of patients(Table 15)

Zidi et al.,199980

Mean 71 ±15.5 (range30–93)

34/36 Inpatients with suspected CBDstones

Not reported Not reported


ent2004; Vol. 8: No. 10

77

© Q


MSO


TABLE 17 Quality assessment

Study Appropriate Work-up bias (all patients Expectation bias Reproducibility CIs reported Placed in context of spectrum had both tests) (blinded assessment) other testsof patients


Yes No, of the 78 original patients16 had unsuccessful ERCP and 2 had unsuccessful MRCP

Results of ultrasoundexamination were known, butoperators were unaware of eachother’s findings or clinicaldiagnosis when MRCP imageswere interpreted

Not reported No Yes, other testsmentioned


Yes No, 70 of 81 had ERCP, 7 hadPTC and 4 had surgery

Yes, examinations were readindependently by 2 radiologistsunaware of the patients’ clinicalinformation

Yes, kappa valuesreported for dilatation(� = 0.79), location (� = 0.80) and cause (� = 0.74)

No No other testsmentioned


Yes No, 68 of 73 patients had ERCP,2 had PTC but not clear whathappened to the other 3

Yes, unaware of clinical andbiochemical data and otherresults



Yes No, only 21 of 29 had ERCP Yes, 2 clinicians evaluated imageswithout clinical or radiologicalinformation

Not reported No No other tests reported

Calvo et al.,200259

Yes, although nomention of other 55patients withoutsuspectedcholedocholithiasis

No, 60 out of 61 had ERCP Radiologists were blinded toERCP results, but not torelevant clinical data and resultsof other imaging studies such asultrasonography


Chan et al.,199660

Yes No, 2 out of 47 did not haveERCP

Radiologists who performed andinterpreted MRCP results wereblinded to all clinical,biochemical and imaging findings

Not reported Yes Yes, other testsmentioned


Yes No, 3 were excluded becauseERCP was technically notpossible

Yes, radiologists were unawareof laboratory and ERCP results

Not reported No Yes, other test resultsreported

continued

Appendix 6

78

TABLE 17 Quality assessment (cont’d)



Yes, although a largenumber of patientswas excluded

No, of the 40 chosen for ERCP,2 did not have MRCP and 4 didnot have ERCP

Yes, the ERCP operator wasunaware of the MRCP result,but no mention of clinical orother imaging findings



Yes No, 3 did not have ERCP,reasons not stated

Yes, radiologists were blinded tothe results of the other tests

Not reported No No other testsmentioned


Yes No, of the original sample, 22did not have ERCP results and 8did not have MRCP results

Yes, reviewers were blinded toclinical parameters, results ofother imaging tests and the finaldiagnosis

Yes, kappa valuesreported for diagnosisof bile duct obstruction(� = 0.90),choledocholithiasis (� = 0.77), malignantobstruction (� = 0.82)and all causes (� = 0.82)

Yes Yes, other testsmentioned


Not clear but all hadsuspected disorder ofpancreatic/biliarysystem

No of the original 78 patients, 1 did not have MRCP and 1 didnot have ERCP

Yes, radiologists interpretedresults on a blinded basis, butnot clear whether blinded toclinical information



Not clear as suspectedcondition not reported

No, 2 of the original sample didnot have both MRCP and 3 didnot have ERCP

Yes, radiologists andendoscopists were blinded toeach others reports

Yes, an off-siteradiologist read all 61MRCP images

No No other tests reported


Only patients withpast history ofcholecystectomy orhigh surgical oranaesthetic riskincluded

No, 147 had MRCP and only101 had ERCP out of original166 patients

Yes, MRCP results were readwithout knowledge of ERCPresults, but no informationregarding blinding of clinical data

Not reported Yes Yes, other testsmentioned

continued


ent2004; Vol. 8: No. 10

79

© Q


MSO




Lee et al.,199767

Yes No, 25 of the original samplewere excluded

Yes, radiologists were blinded toERCP results, andgastroenterologist andradiologist were blinded toMRCP results, but no mentionof clinical data

Yes, 2 readersinterpreted MRCP andERCP results andagreement was good (� = 0.487 for MRCP, � = 0.702 for ERCP)



Yes No, 2 of 62 had PTC Not reported Not reported No Yes, other testsmentioned

Lomas et al.,199969

Yes No, of the original 76 patientsonly 66 had both tests

Not clear, results were assessed‘independently’ but investigatorswere aware of clinicalpresentation, prior imaging andlaboratory studies

Not reported, althougha �-value was reportedfor agreement betweenMRCP and ERCP; � =0.88 (CI 0.82 to 0.94)


Macaulay, et al.,199570

Yes No, of the original 28 patients,only 291 underwent ERCP

Yes, ERCP, PTC and IOC wereread by 2 radiologists blinded topatient history and diagnosis,MRCP images were read by 2radiologists blinded to diagnosis;clinical history and findings ofother imaging studies

No, 2 readers, butinter-rater agreementwas not reported


Regan et al.,199671

Yes No, 26 initially, of whom 2 didnot have ERCP and 1 did nothave MRCP

Radiologists were unaware ofthe results of the ERCP andeach other’s results, but noinformation about other tests

2 readers disagreed on3 cases

Yes Yes, tests used for finaldiagnoses mentioned, aswell as sonography


Yes No, of 159 initial patients, 101had ERCP

One reviewer was blinded toclinical findings and otherresults. The other reviewer wasblinded to the results of thedirect cholangiography, but hadaccess to clinical findings

Kappa values werereported forcholedocholithiasis (� = 0.82)


continued

Appendix 6

80



Soto et al.,199673

No, not clear whatsuspected conditionwas

No, 1 had PTC, not clearwhether patients were excludedfrom initial sample

Yes, radiologists did not haveaccess to clinical information orother imaging studies

2 radiologists, but noinformation on inter-rater agreement


Soto et al.,200074

Yes No, only 49 of initial 59 patientshad both tests

Yes, radiologists were blinded toclinical and laboratory data andresults of other imaging tests

Yes, kappa valuesreported for 3D FSEsequence � = 0.92,multisection half-FourierRARE sequence (� =0.84) and single-sectionhalf-Fourier RAREsequence (� = 0.80).ROC curves for bothradiologists alsoreported and diagnosticperformance of the 3sequences for bothradiologists wasreported to be excellent

Yes No other testsmentioned

Soto et al.,200075

Yes No, of initial 68 patients only 51had both tests

Yes, radiologists were blinded toresults of other diagnosticstudies and clinical information.The sequence of interpretationof studies was randomised

Not reported Yes Yes, other test resultsreported


Yes Yes Yes, those performing ERCPwere unaware of MRCP results,but no mention of blinding toclinical data



Yes, but included anypancreatobiliarydisease

No, only 159 of original 187patients had both tests

No, although the endoscopistinterpreting the ERCP imageswas not aware of the MRCPresults, the radiologists wereaware of clinical information andother imaging findings

Not reported No No other test resultsmentioned

continued


ent2004; Vol. 8: No. 10

81

© Q


MSO





Yes No, only 129 of original 146patients had both tests

Yes, the radiologist interpretingMRCP images was blinded toERCP findings, but no mentionof other images and clinicalinformation

Not reported Yes Yes, other test resultsreported

Textor et al.,20026

Yes No, of 150 patients only 146had both tests

Yes, radiologists were blinded topatients’ gender, age, clinical andmedical history

Yes, kappa valuesreported for bile ductabnormalities (� = 0.98and 0.978)

No Yes, other testsmentioned


Not clear whatsuspected diagnosiswas

No, 64 of the original 256patients were excluded from thestudy, resulting in a final sampleof 191 patients; not clear whohad ERCP or IOC or PTC, apartfrom those diagnosed withcholedocholithiasis

Patient clinical information andultrasound findings wereavailable to the endoscopists,but they were unaware of theMRCP findings at the time ofERCP

Not reported No Yes, other tests, such asultrasound, reported

Zidi et al.,199980

Yes No, only 63 of original 70 hadERCP

Yes, radiologists were blinded toERCP results and previousinvestigations


Appendix 6

82

TABLE 18 Results

Study Diagnosis Sensitivity Specificity Adverse effects


Final diagnosis No. of patients

Normal 13Benign stricture 15Choledocholithiasis 3Hepatolithiasis 1Choledochal cyst 1Malignant stricture 23Cholangiocarcinoma 4

Compared with final diagnosis (ERCPplus histological findings or follow-up)

Any abnormality 42/47 (89%)Detection of malignancy 22/27 (81%)

ERCP

Any abnormality 91%Malignancy 93%

Compared with final diagnosis

Any abnormality 12/13 (92%)Detection of malignancy 33/33 (100%)

ERCP

Any abnormality 92%Malignancy 94%

3 patients had mildacute pancreatitisafter ERCP


Diagnosis

Normal 11Choledocholithiasis 20Neoplasms 15Benign biliary stricture 23Chronic pancreatitis 8Other 4

Compared with ERCP (with histologicalexamination) Location of obstruction

Intrahepatic/hilar 100%Suprapancreatic 92%Ampullary 86%Intrapancreatic 69%

Cause of obstruction

Choledocolithiasis 89%Malignant obstruction 92%Benign stricture 63%Pancreatitis 50%

Compared to ERCP (with histologicalexamination)Location of obstruction

Intrahepatic/hilar 100%Suprapancreatic 94%Ampullary 91%Intrapancreatic 92%

Cause of obstruction

Choledocholithiasis 90%Malignant obstruction 88%Benign stricture 90%Pancreatitis 99%

No complicationsoccurred duringthe procedures


DiagnosisBenign biliary disease 45 (58%)

including PSC 23 (32%)Malignant biliary disease 9 (12%)Normal biliary tree 22 (30%)

Compared with ERCP

Normal ducts 19/22 (86%)Dilatation:Hepatic ducts 40/42 (95%)CBD 38/41 (93%)Obstruction 37/37 (100%)Biliary stones* 5/10 (50%)PSC 19/23 (83%)*Sensitivity of MRCP in detection of bile ductstones was greater in patients without PSC(80%) than in patients with PSC (20%)

Compared with ERCP

Normal ducts 46/48 (96%)Dilatation:Hepatic ducts 24/28 (86%)CBD 27/29 (93%)Obstruction 30/33 (91%)Biliary stones 59/60 (98%)PSC 46/47 (98%)

6 of 68 patientswho underwentERCP (8.8%)developedcomplications suchas abdominal pain(n = 3) requiringhospitalisation forat least 24 hours,pancreatitis (n = 2) andperforation of theCBD (n = 1)

continued


ent2004; Vol. 8: No. 10

83

© Q


MSO


TABLE 18 Results (cont’d)



Diagnosis according to MRCP

Normal 6Periampullary stricture 5Calculous obstruction 3Structural duct anomaly 1

(Caroli’s disease)

Solitary stricture:(chronic pancreatitis) 2hepaticojejunostomy

stricture 1cystic lesion (biliary

cystadenocarcinoma) 1

Compared with ERCP

Diagnosis of CBD 19/21 (90%)Dilatation of the CBD 13/15 (87%)

Compared with ERCP

Dilatation of the CBD 6/6 (100%) Not reported

Calvo et al.,200259

High-probability patients (n = 49)

ERCP MRCP

Gallstones 32 29Normal 1

Diagnostic doubt between gallstone and aerobilia 1Diagnostic doubt between 1gallstone lodged in papilla and ampulloma

Papillitis 4 2Diagnostic doubt 2

Ampulloma 2 1Diagnostic doubt between 1ampulloma and obstructive choledocholithiasis

Cholangiocarcinoma 1 1Normal 10 8Diagnostic doubt with aerobilia 2

Intermediate-probability patients (n = 9)Gallstones 3 3Papillitis 2 1

Diagnostic doubt 1Normal 4 4

Compared with ERCP91%

Compared with ERCP 84% No morbidityassociated withdiagnostic ERCP,but thoseundergoingtherapeutic ERCPhad a morbidity of4% (2 of 49patients), whichconsisted of mildacute pancreatitisand digestive tractbleeding secondaryto sphincterotomythat requiredendoscopicsclerotherapy

continued

Appendix 6

84



Chan et al.,199660

ERCP MRCP

Absence of ductal dilatation 16 16CBD dilatation 29 28Choledocholithiasis 19 18Without choledocholithiasis 26 22

Compared with ERCP

95% (CI 0.8449 to 1.000)

Compared with ERCP

85% (CI 0.7046 to, 0.9877)

Not reported


ERCP MRCP

CBD stones 19 21

Compared with ERCP

19/19 (100%)

Compared with ERCP

19/21 (calculated 90.5%)

No complications


8 patients with CBD stones Compared with ERCP

7/8 (88%)

Compared with ERCP

28/30 (93%)

Not reported


ERCP MRCP

Bile duct dilatation 14 14Normal ducts 3 3Malignancy18 malignant pancreaticobiliary neoplasms

Compared with final diagnosis (ERCPplus pathological diagnosis)

Dilatation 14/14 (100%)Malignancy 17/18

(calculated 94.4%)

Not reported


Diagnosis

Bile duct obstruction 79Choledocholithiasis 32Malignant obstruction 14

Compared with ERCP

Bile duct obstruction 72/79 (91%) (CI 85 to 100%)

Choledocholithiasis 26/32 (81%) (CI 68 to 95%)

Malignant obstruction 12/14 (86%) (CI 67 to 100%)

Compared with ERCP

Bile duct obstruction 100%Choledocholithiasis 98%

(CI 95 to 100%)Malignant obstruction 98%

(CI 96 to 100%)

Not reported


Final diagnosis

Cholangiocarcinoma (Klatskin) 14Papillary stenosis 7Normal 7Choledocholithiasis 6Liver metastasis 6Juxtapapillary duodenal diverticulum 6Liver cirrhosis 4Primary sclerosising cholangitis 3Surgical ligation of bile duct 2Caroli syndrome 1Ischaemic type biliary lesion 1

Compared with ERCP

Normal duct 5/7 (71%)Recognition of dilatation 20/24 (83%)Recognition of stricture 22/26 (85%)Correct stricture location 20/26 (77%)Diagnosis of benign stricture 6/12 (50%)Diagnosis of malignant stricture 8/10 (80%)Diagnosis of stones 4/5 (80%)Overall 89%

Compared with ERCP

Overall 78%

Not reported

continued


ent2004; Vol. 8: No. 10

85

© Q


MSO





Diagnosis

ERCP MRCPCholangiolithiasis 13 12Dilatation 34 32Stenosis 36 32

Compared with ERCP

Cholangiolithiasison-site 12/13 (92.3%)off-site 11/13 (84.6%)

Dilatationon-site 32/34 (94.1%)off-site 32/34 (94.1%)

Stenosison-site 32/36 (88.9%)off-site 30/36 (83.3%)

Overallon-site 42/46 (91.3%)off-site 43/46 (93.5%)

Compared with ERCP

Cholangiolithiasison-site 46/48 (95.8%)off-site 45/48 (93.7%)

Dilatationon-site 25/27 (92.6%)off-site 26/27 (96.3%)

Stenosison-site 22/25 (84%)off-site 22/25 (84%)

Overallon-site 12/15 (80%)off-site 12/15 (80%)

Not reported


Diagnosis

CBD free of obstruction 15CBD stones (including stones 113 (15)

< 3 mm)Malignant strictures of the papilla 5Cystic lesions of CBD 1Adenocarcinoma of head of pancreas 2Papillary stenosis 7Hilum cholangiocarcinoma 2Benign strictures of CBD 2

Compared with final diagnosis (stoneextraction with ERCP or IOC)

MRCP 93% (CI 86.1 to 96.7) ERCP 95% (CI 87.3 to 98.4)


MRCP 100% (CI 87.4 to 100) ERCP 100% (CI 79.1 to 100)

Not reported

Lee et al.,199767

Diagnosis (n = 46)

Normal 1Choledocholithiasis 11Choledochal cyst 1Non-specific biliary dilitation 12Klatskin tumour 4Proximal CBD carcinoma 1Ampullary carcinoma 8Pancreatic head carcinoma 5Hepatoma with biliary tumour emboli 3

Compared with final diagnosis (ERCPplus surgical findings)

MRCP 17/21 (81%)ERCP 15/21 (71%)


MRCP 23/25 (92%)ERCP 23/25 (92%)

Not reported

continued

Appendix 6

86




24 of 62 patients were positive for stones Compared with ERCP

22/24 (91.6%)

Compared with ERCP

100%

Not reported

Lomas et al.,199969

Diagnosis

Strictures 20Choledocholithiasis 11Other 12Normal 25Chronic pancreatitis 1

Compared with ERCPStrictures 19/19 100%Choledocholithiasis 9/9 100%

Compared with ERCP

Strictures 98% (CI 94 to 100%)Choledocholithiasis 97% (CI 93 to 100%)

Not reported

Macaulay et al.,199570

Diagnosis

Obstruction 14Duct stones or sludge 7Intrahepatic ducts

Dilated 18Non-dilated 11

All 4 hepatic segmentsDilated 18Non-dilated 11

Compared with final diagnosis (directcholangiography plus endoscopic orfluoroscopic observation)

Obstruction 14/14 100%Duct stones or sludge 5/7 (calculated

71.4%)


(91%)

Not reported

Regan et al.,199671

Diagnosis

CBD stones 15CBD dilatation 12

Compared with final diagnosis (ERCP,endoscopic balloon or basket extractionor surgical removal of stones)

Choledocholithiasis

ERCP MRCP

No. of patients 15 14100% 93% (CI 70

to 100%)DilatationMRCP 100%

Compared with ERCPCholedocholithiasis

ERCP MRCP100% 89% (CI 52

to 100%)DilatationMRCP 100%

Not reported

continued


ent2004; Vol. 8: No. 10

87

© Q


MSO





Diagnosis

Choledocholithiasis 30Normal biliary tract 32Postcholecystectomy dilatation

or oddities 27Pancreatic carcinoma 14Ampullary carcinoma 2Cholangiocarcinoma 1Metastases 2Chronic pancreatitis 2

Compared with ERCP

Choledocholithiasis

Reviewer 1 90% (CI 79 to 100%)Reviewer 2 90% (CI 70 to 100%)Bile duct obstruction

91% (CI 84 to 97%)

Compared with ERCP

CholedocholithiasisReviewer 1 100%Reviewer 2 93% (CI 87 to 98%)Bile duct obstruction

100%

3 of 28 patientswithcholedocholithiasisdevelopedcomplications(11%) (2 withpancreatitis and 1with postsphincterbleeding), amongthose negative forcholedocholithiasis4 of 80 (5%)developedcomplications(pancreatitis in all 4)

Soto et al.,199673

Diagnosis

Normal 17Ampullary stenosis 12Choledocholithiasis 6Chronic pancreatitis with stricture 4Pancreatic head adenocarcinoma 2

with strictureMid-CBD cholangiocarcinoma 1Periportal adenopathy 1Sclerosing cholangitis 1Common bile duct polyp 1Intrahepatic cholangiocarcinoma 1

Compared with ERCP

Bile duct dilatation 26/27 (96.3%)Bilary strictures 9/10 (90%)Intraductal abnormalities 7/7 (100%)

Compared with ERCP

Normal bile ducts 16/17 (94.1%)

Not reported

Soto et al.,200074

DiagnosisBile duct dilatation with onlycholedocholithiasis 18Bile duct dilatation with 4

choledocholithiasis and hepatolithiasis

Choledocholithiasis with 2normal-calibre ducts

Bile duct dilatation without stones 19Normal biliary ductal system 6

Compared with ERCP

3D FSERadiologist 1 96% (CI 90 to 100%)Radiologist 2 92% (CI 84 to 99%)

Single-section half-Fourier RARERadiologist 1 100% (CI 99 to 100%)Radiologist 2 92% (CI 84 to 99%)

Multi-section half-Fourier RARERadiologist 1 92% (CI 84 to 99%)Radiologist 2 96% (CI 90 to 100%)

Compared with ERCP

3D fast SERadiologist 1 96% (CI 90 to 100%)Radiologist 2 100% (CI 99 to 100%)

Single-section half-Fourier RARERadiologist 1 96% (CI 90 to 100%)Radiologist 2 96% (CI 90 to 100%)

Multi-section half-Fourier RARERadiologist 1 92% (CI 84 to 99%)Radiologist 2 92% (CI 84 to 99%)

Not reported

continued

Appendix 6

88



Soto et al.,200075

Diagnosis

Bile duct stones 26

Compared with ERCP

96% (CI 78 to 99%)

Compared with ERCP

100% (CI 84 to100%)

Not reported forERCP or MRCP


CBD stones 32 Compared with ERCP

28/32 (87.5%)

Compared with ERCP

17/18 (94.4%)

1 of 50 patientsdevelopedmoderatepancreatitis (2%)


Diagnosis

No. of No. detectedpatients with MRCP

Anomalous 11 9PBJ

Congenital 7 7choledochal cyst

Mucosal hyperplasia 5 0of gallbladder

Carcinoma of 1 0gallbladder

Gallbladder stone 1 1CBD stone 2 1

Compared with ERCP

Anomalous PBJ 82%

Without PBJ:

Pancreatobiliary lesions in 22/24 (92%)pancreatic carcinoma

Pancreatic cystic disease 13/13 (100%)Chronic pancreatitis 13/15 (87%)Bile duct carcinoma 11/12 (92%)Polypoid gallbladder lesions 5/7 (71%)Gallbladder stones 23/27 (85%)Choledocholithiasis 20/21 (95%)

Compared with ERCP

Anomalous PBJ 100%

Not reported


Diagnosis

Choledocholithiasis 46Stricture 12Normal 47Dilated biliary tree 22Other 4

Compared with ERCP

45/46 (97.8%) (CI 88.5 to 99.9%)

Compared with ERCP

74/83 (89.1%) (CI 80.4 to 94.9%)

Not reported

continued


ent2004; Vol. 8: No. 10

89

© Q


MSO




Textor et al.,20026

Diagnosis

MRCP Final

Normal 71 (47.3%) 72 (48.0%)Choledocholithiasis 3 (2.0%) 5 (3.3%)Cholangiocellular 36 (24.0%) 39 (26.0%)

carcinomaPrimary sclerosing 29 (19.3%) 34 (22.7%)

cholangitis

Compared with final diagnosis (clinicalpresentation and biochemical tests)

For diagnosing PSC 29/33 (88%)


For diagnosing PSC 108/109 (99%)

1 patient had mildpancreatitisfollowing ERCP


Diagnosis

Final ERCP MRCP

Choledocholithiasis 34 29 31Strictures 47Normal ducts 100

Compared with ERCP

Choledocholithiasis (3 false positives and 3 false negatives for choledocholithiasis)

31/34 (91%)


Choledocholithiasis 98%

Not reported

Zidi et al.,199980

DiagnosisUndilated CBD 24Dilated CBD 46CBD stones 49

Compared with ERCP

Choledocholithiasis 28/49 (57.1%)

Compared with ERCP

Choledocholithiasis 21/21 (100%)

Not reported


99

Health Technology AssessmentProgramme

Prioritisation Strategy GroupMembers

Chair,Professor Tom Walley, Director,NHS HTA Programme &Professor of ClinicalPharmacology, University of Liverpool

Professor Bruce Campbell,Consultant Vascular & GeneralSurgeon, Royal Devon & ExeterHospital

Professor Shah Ebrahim,Professor in Epidemiology of Ageing, University of Bristol

Dr John Reynolds, ClinicalDirector, Acute GeneralMedicine SDU, RadcliffeHospital, Oxford

Dr Ron Zimmern, Director,Public Health Genetics Unit,Strangeways ResearchLaboratories, Cambridge

HTA Commissioning BoardMembers

Programme Director, Professor Tom Walley, Director,NHS HTA Programme &Professor of ClinicalPharmacology, University of Liverpool

Chair,Professor Shah Ebrahim,Professor in Epidemiology ofAgeing, Department of SocialMedicine, University of Bristol,Canynge Hall, WhiteladiesRoad, Bristol

Deputy Chair, Professor Jenny Hewison,Professor of Health CarePsychology, Academic Unit ofPsychiatry and BehaviouralSciences, University of LeedsSchool of Medicine, Leeds

Professor Douglas Altman,Professor of Statistics inMedicine, Centre for Statisticsin Medicine, Oxford University,Institute of Health Sciences,Cancer Research UK MedicalStatistics Group, Headington,Oxford

Professor John Bond, Professorof Health Services Research,Centre for Health ServicesResearch, University ofNewcastle, School of HealthSciences, Newcastle upon Tyne

Professor John Brazier, Directorof Health Economics, SheffieldHealth Economics Group,School of Health & RelatedResearch, University ofSheffield, ScHARR RegentCourt, Sheffield

Dr Andrew Briggs, PublicHealth Career Scientist, HealthEconomics Research Centre,University of Oxford, Instituteof Health Sciences, Oxford

Dr Christine Clark, MedicalWriter & Consultant Pharmacist,Cloudside, Rossendale, LancsandPrincipal Research Fellow,Clinical Therapeutics in theSchool of Pharmacy, BradfordUniversity, Bradford

Professor Nicky Cullum,Director of Centre for EvidenceBased Nursing, Department ofHealth Sciences, University ofYork, Research Section,Seebohm Rowntree Building,Heslington, York

Dr Andrew Farmer, SeniorLecturer in General Practice,Department of Primary HealthCare, University of Oxford,Institute of Health Sciences,Headington, Oxford

Professor Fiona J Gilbert,Professor of Radiology,Department of Radiology,University of Aberdeen, LilianSutton Building, Foresterhill,Aberdeen

Professor Adrian Grant,Director, Health ServicesResearch Unit, University ofAberdeen, Drew Kay Wing,Polwarth Building, Foresterhill,Aberdeen

Professor Alastair Gray, Director,Health Economics ResearchCentre, University of Oxford,Institute of Health Sciences,Headington, Oxford

Professor Mark Haggard,Director, MRC ESS Team, CBUElsworth House, Addenbrooke’sHospital, Cambridge

Professor F D Richard Hobbs,Professor of Primary Care &General Practice, Department ofPrimary Care & GeneralPractice, University ofBirmingham, Primary Care andClinical Sciences Building,Edgbaston, Birmingham

Professor Peter Jones, Head ofDepartment, UniversityDepartment of Psychiatry,University of Cambridge,Addenbrooke's Hospital,Cambridge

Professor Sallie Lamb, ResearchProfessor in Physiotherapy/Co-Director, InterdisciplinaryResearch Centre in Health,Coventry University, Coventry

Dr Donna Lamping, SeniorLecturer, Health ServicesResearch Unit, Public Healthand Policy, London School ofHygiene and Tropical Medicine,London

Professor David Neal, Professorof Surgical Oncology, OncologyCentre, Addenbrooke's Hospital,Cambridge

Professor Tim Peters, Professorof Primary Care Health ServicesResearch, Division of PrimaryHealth Care, University ofBristol, Cotham House, CothamHill, Bristol

Professor Ian Roberts, Professorof Epidemiology & PublicHealth, Intervention ResearchUnit, London School ofHygiene and Tropical Medicine,London

Professor Peter Sandercock,Professor of Medical Neurology,Department of ClinicalNeurosciences, University ofEdinburgh, Western GeneralHospital NHS Trust, BramwellDott Building, Edinburgh

Professor Martin Severs,Professor in Elderly HealthCare, Portsmouth Institute ofMedicine, Health & Social Care,St George’s Building,Portsmouth

Dr Jonathan Shapiro, SeniorFellow, Health ServicesManagement Centre, ParkHouse, Birmingham

Current and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)


Health Technology Assessment Programme

100

Diagnostic Technologies & Screening PanelMembers

Chair,Dr Ron Zimmern, Director ofthe Public Health Genetics Unit,Strangeways ResearchLaboratories, Cambridge

Dr Paul Cockcroft, ConsultantMedical Microbiologist/Laboratory Director, PublicHealth Laboratory, St Mary’s Hospital, Portsmouth

Professor Adrian K Dixon,Professor of Radiology,Addenbrooke’s Hospital,Cambridge

Dr David Elliman, Consultant inCommunity Child Health,London

Dr Andrew Farmer, SeniorLecturer in General Practice,Institute of Health Sciences,University of Oxford

Dr Karen N Foster, ClinicalLecturer, Dept of GeneralPractice & Primary Care,University of Aberdeen

Professor Jane Franklyn,Professor of Medicine,University of Birmingham

Professor Antony J Franks,Deputy Medical Director, TheLeeds Teaching Hospitals NHSTrust

Mr Tam Fry, HonoraryChairman, Child GrowthFoundation, London

Dr Susanne M Ludgate, MedicalDirector, Medical DevicesAgency, London

Dr William Rosenberg, SeniorLecturer and Consultant inMedicine, University ofSouthampton

Dr Susan Schonfield, CPHMSpecialised ServicesCommissioning, CroydonPrimary Care Trust

Dr Margaret Somerville,Director of Public Health,Teignbridge Primary Care Trust,Devon

Mr Tony Tester, Chief Officer,South Bedfordshire CommunityHealth Council, Luton

Dr Andrew Walker, SeniorLecturer in Health Economics,University of Glasgow

Professor Martin J Whittle,Head of Division ofReproductive & Child Health,University of Birmingham

Dr Dennis Wright, ConsultantBiochemist & Clinical Director,Pathology & The KennedyGalton Centre, Northwick Park& St Mark’s Hospitals, Harrow

Pharmaceuticals PanelMembers

Chair,Dr John Reynolds, ClinicalDirector, Acute GeneralMedicine SDU, OxfordRadcliffe Hospital

Professor Tony Avery, Professorof Primary Health Care,University of Nottingham

Professor Iain T Cameron,Professor of Obstetrics &Gynaecology, University ofSouthampton

Mr Peter Cardy, ChiefExecutive, Macmillan CancerRelief, London

Dr Christopher Cates, GP andCochrane Editor, Bushey HealthCentre, Bushey, Herts.

Mr Charles Dobson, SpecialProjects Adviser, Department ofHealth

Dr Robin Ferner, ConsultantPhysician and Director, WestMidlands Centre for AdverseDrug Reactions, City HospitalNHS Trust, Birmingham

Dr Karen A Fitzgerald,Pharmaceutical Adviser, Bro TafHealth Authority, Cardiff

Professor Alastair Gray,Professor of Health Economics,Institute of Health Sciences,University of Oxford

Mrs Sharon Hart, ManagingEditor, Drug & TherapeuticsBulletin, London

Dr Christine Hine, Consultant inPublic Health Medicine, Bristol South & West PrimaryCare Trust

Professor Robert Peveler,Professor of Liaison Psychiatry,Royal South Hants Hospital,Southampton

Dr Frances Rotblat, CPMPDelegate, Medicines ControlAgency, London

Mrs Katrina Simister, NewProducts Manager, NationalPrescribing Centre, Liverpool

Dr Ken Stein, Senior Lecturer inPublic Health, University ofExeter

Professor Terence Stephenson,Professor of Child Health,University of Nottingham

Dr Richard Tiner, MedicalDirector, Association of theBritish Pharmaceutical Industry,London

Professor Dame Jenifer Wilson-Barnett, Head of FlorenceNightingale School of Nursing& Midwifery, King’s College,London



101

Therapeutic Procedures PanelMembers

Chair, Professor Bruce Campbell,Consultant Vascular andGeneral Surgeon, Royal Devon& Exeter Hospital

Dr Mahmood Adil, Head ofClinical Support & HealthProtection, Directorate ofHealth and Social Care (North),Department of Health,Manchester

Professor John Bond, Head ofCentre for Health ServicesResearch, University ofNewcastle upon Tyne

Mr Michael Clancy, Consultantin A & E Medicine,Southampton General Hospital

Dr Carl E Counsell, SeniorLecturer in Neurology,University of Aberdeen

Dr Keith Dodd, ConsultantPaediatrician, DerbyshireChildren’s Hospital, Derby

Professor Gene Feder, Professorof Primary Care R&D, Barts &the London, Queen Mary’sSchool of Medicine andDentistry, University of London

Ms Bec Hanley, FreelanceConsumer Advocate,Hurstpierpoint, West Sussex

Professor Alan Horwich,Director of Clinical R&D, TheInstitute of Cancer Research,London

Dr Phillip Leech, PrincipalMedical Officer for PrimaryCare, Department of Health,London

Mr George Levvy, ChiefExecutive, Motor NeuroneDisease Association,Northampton

Professor James Lindesay,Professor of Psychiatry for theElderly, University of Leicester

Dr Mike McGovern, SeniorMedical Officer, Heart Team,Department of Health, London

Dr John C Pounsford,Consultant Physician, NorthBristol NHS Trust

Professor Mark Sculpher,Professor of Health Economics,Institute for Research in theSocial Services, University ofYork

Dr L David Smith, ConsultantCardiologist, Royal Devon &Exeter Hospital

Professor Norman Waugh,Professor of Public Health,University of Aberdeen


Health Technology Assessment Programme

102Current and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)

Expert Advisory NetworkMembers

Mr Gordon Aylward, Chief Executive, Association of British Health-Care Industries, London

Ms Judith Brodie, Head of Cancer SupportService, Cancer BACUP, London

Mr Shaun Brogan, Chief Executive, RidgewayPrimary Care Group, Aylesbury,Bucks

Ms Tracy Bury, Project Manager, WorldConfederation for PhysicalTherapy, London

Mr John A Cairns, Professor of Health Economics,Health Economics ResearchUnit, University of Aberdeen

Professor Howard Stephen Cuckle, Professor of ReproductiveEpidemiology, Department ofPaediatrics, Obstetrics &Gynaecology, University ofLeeds

Professor Nicky Cullum, Director of Centre for EvidenceBased Nursing, University of York

Dr Katherine Darton, Information Unit, MIND – TheMental Health Charity, London

Professor Carol Dezateux, Professor of PaediatricEpidemiology, London

Professor Martin Eccles, Professor of ClinicalEffectiveness, Centre for HealthServices Research, University ofNewcastle upon Tyne

Professor Pam Enderby,Professor of CommunityRehabilitation, Institute ofGeneral Practice and PrimaryCare, University of Sheffield

Mr Leonard R Fenwick, Chief Executive, Newcastleupon Tyne Hospitals NHS Trust

Professor David Field, Professor of Neonatal Medicine,Child Health, The LeicesterRoyal Infirmary NHS Trust

Mrs Gillian Fletcher, Antenatal Teacher & Tutor andPresident, National ChildbirthTrust, Henfield, West Sussex

Ms Grace Gibbs, Deputy Chief Executive,Director for Nursing, Midwifery& Clinical Support Servs., WestMiddlesex University Hospital,Isleworth, Middlesex

Dr Neville Goodman, Consultant Anaesthetist,Southmead Hospital, Bristol

Professor Robert E Hawkins, CRC Professor and Director ofMedical Oncology, Christie CRCResearch Centre, ChristieHospital NHS Trust, Manchester

Professor F D Richard Hobbs, Professor of Primary Care &General Practice, Department ofPrimary Care & GeneralPractice, University ofBirmingham

Professor Allen Hutchinson, Director of Public Health &Deputy Dean of ScHARR,Department of Public Health,University of Sheffield

Professor Rajan Madhok, Medical Director & Director ofPublic Health, Directorate ofClinical Strategy & PublicHealth, North & East Yorkshire& Northern Lincolnshire HealthAuthority, York

Professor David Mant, Professor of General Practice,Department of Primary Care,University of Oxford

Professor Alexander Markham, Director, Molecular MedicineUnit, St James’s UniversityHospital, Leeds

Dr Chris McCall, General Practitioner, TheHadleigh Practice, CastleMullen, Dorset

Professor Alistair McGuire, Professor of Health Economics,London School of Economics

Dr Peter Moore, Freelance Science Writer,Ashtead, Surrey

Dr Andrew Mortimore, Consultant in Public HealthMedicine, Southampton CityPrimary Care Trust

Dr Sue Moss, Associate Director, CancerScreening Evaluation Unit,Institute of Cancer Research,Sutton, Surrey

Professor Jon Nicholl, Director of Medical CareResearch Unit, School of Healthand Related Research,University of Sheffield

Mrs Julietta Patnick, National Co-ordinator, NHSCancer Screening Programmes,Sheffield

Professor Chris Price, Visiting Chair – Oxford, ClinicalResearch, Bayer DiagnosticsEurope, Cirencester

Ms Marianne Rigge, Director, College of Health,London

Professor Sarah Stewart-Brown, Director HSRU/HonoraryConsultant in PH Medicine,Department of Public Health,University of Oxford

Professor Ala Szczepura, Professor of Health ServiceResearch, Centre for HealthServices Studies, University ofWarwick

Dr Ross Taylor, Senior Lecturer, Department ofGeneral Practice and PrimaryCare, University of Aberdeen

Mrs Joan Webster, Consumer member, HTA –Expert Advisory Network

The National Coordinating Centre for Health Technology Assessment,Mailpoint 728, Boldrewood,University of Southampton,Southampton, SO16 7PX, UK.Fax: +44 (0) 23 8059 5639 Email: [email protected]://www.ncchta.org ISSN 1366-5278

FeedbackThe HTA Programme and the authors would like to know

your views about this report.

The Correspondence Page on the HTA website(http://www.ncchta.org) is a convenient way to publish

your comments. If you prefer, you can send your comments to the address below, telling us whether you would like

us to transfer them to the website.

We look forward to hearing from you.

cpre y colangioresonancia metaanalisis

Documents