components factor analysis with varimax rotation was utilized to identifydiscrete domains embedded within each survey. Cronbach’s alpha assessedinternal consistency of items within each domain.Results: Eighty subjects participated in item generation. Pre– and post–procedure surveys derived from their responses were piloted in another 60subjects. Each survey was reduced by rank order to 25 items with responsesbased on a 5–point Likert scale; these were administered to an additional81 subjects. Factor analysis extracted 4 domains with eigenvalues greaterthan 1 (pain, anxiety, logistical issues and concern over results) accountingfor 57.3% of the variance in the pre–procedure survey, while 3 domainswere identified in the post–procedure survey (pre–procedural discomfort,intra–procedural pain, and procedure environment) accounting for 59.9%of the total variance. Internal consistency was fair to good (Cronbach’salpha 0.52–0.65 and 0.12–0.72, pre– and post–procedure, respectively).Conclusions: Patient–based experience surveys may be used to determinepreference for gastrointestinal procedures. These instruments may assesstolerance for new techniques such as unsedated endoscopy, and comparepatient preference among competing endoscopic or radiological modalitiesof gastrointestinal cancer screening.

735

AN UPDATE ON THE UPPER–GI SAFETY PROFILE OFROFECOXIB (VIOXX) COMPARED WITH NSAIDsDouglas J. Watson, Ph.D., Qinfen Yu, M.S., James A. Bolognese,M.Stat., Alise S. Reicin, M.D., Richard H. Holmes, B.S. and Thomas J.Simon, M.D.*. Epidemiology, Merck Research Laboratories, WestPoint, PA; Clinical Biostatistics, Merck Research Laboratories, Rahway,NJ; Clinical Research, Merck Research Laboratories, Rahway, NJ andClinical Research, Merck Research Laboratories, West Point, PA.

Purpose: To update a previously published pre–specified analysis of theincidence of upper–GI perforations, ulcers, and bleeds (PUBs) in clinicaltrials of rofecoxib (12.5, 25, or 50 mg daily) compared with NSAIDs(naproxen, diclofenac, ibuprofen, or nabumetone).Methods: Investigators in 20 randomized, double–blind clinical trials ofpatients with osteoarthritis (OA, 15 trials) and rheumatoid arthritis (RA, 5trials) reported potential PUBs for adjudication by an external, blindedcommittee that used pre–specified criteria to confirm events. A post–hoccombined analysis of all reported PUBs during active–treatment periods orwithin 14 days of stopping treatment was performed. It included eventsdiagnosed during unscheduled procedures, but excluded those diagnosedduring planned surveillance procedures in endoscopy trials (4 trials). Theanalysis was done according to the intention–to–treat principle. Time tofirst PUB was evaluated using survival analysis. Incidence rates wereexpressed as the number of events per 100 person–years. Relative risks(RR) were estimated with the Cox proportional hazards model with pro-tocol as a stratification variable.Results: Ten thousand twenty–six patients were treated with once dailyrofecoxib (12.5, 25, or 50 mg) and 7046 were treated with NSAID(naproxen 1000 mg, diclofenac 150 mg, ibuprofen 2400 mg, or nabum-etone 1500 mg daily). One hundred three patients with investigator–reported PUBs (87 confirmed) were eligible for the analysis. The rate/100patient–years for investigator–reported PUBs was 0.77 with rofecoxib vs.2.01 for NSAIDs (RR�0.36; 95% CI 0.24, 0.53; p�0.001), while that forconfirmed PUBs was 0.67 with rofecoxib vs. 1.66 for NSAIDs (RR�0.36;95% CI 0.24, 0.56; p�0.001). The rate/100 patient–years for confirmed,complicated (by perforation, obstruction, or major bleeding) PUBs was0.17 with rofecoxib vs. 0.45 for NSAIDs (RR�0.33; 95% CI 0.14, 0.76;p�0.01).Conclusions: In this combined analysis of 20 randomized, double–blind,clinical trials of patients with OA and RA, patients treated with the COX–2selective inhibitor rofecoxib had an approximately 65% lower incidence ofinvestigator–reported PUBs, confirmed PUBs, and confirmed, complicatedPUBs compared with patients treated with NSAIDs.

736

EARLY ONSET OF IMPROVED MODERATE OR SEVEREUPPER GI TOLERABILITY OF CELECOXIB VERSUSTRADITIONAL NSAIDS: RESULTS FROM SUCCESS I, ADOUBLE BLIND RANDOMIZED TRIAL OF 13,274 PATIENTSGurkipal Singh, M.D.*, George Triadafilopoulos, M.D. and John Fort,M.D. Rheumatology, Stanford University Medical Center, Palo Alto,CA and Global Medical Affairs, Pharmacia Corporation, Peapack, NJ.

Purpose: NSAID–induced dyspepsia is a common cause for discomfortand poor quality of life. Previous studies have shown that celecoxib has abetter upper gastrointestinal (UGI) tolerability profile compared toNSAIDs. This analysis compared the incidence of moderate to severe UGIsymptoms during the first 6 weeks of therapy in patients with osteoarthritis(OA) treated with Celecoxib or NSAIDs in a large clinical trial (SUCCESSI).Methods: SUCCESS–1 is a 12–week double–blinded randomized trial thatenrolled 13,274 patients in 39 countries in 6 continents. A total of 13,194patients took at least one dose of the study medication. Of these, 8800 wererandomized to celecoxib (200mg or 400 mg per day), and 4,394 wererandomized to the most commonly used NSAID for each country(naproxen in US and Canada, diclofenac in the rest of the world). UGIsymptoms (abdominal pain, nausea, or dyspepsia) were rated as mild (nolimitation of usual activities), moderate (some limitation of usually activ-ities) or severe (inability to carry out usual activities). For this study, weevaluated the incidence of moderate and severe UGI symptoms, since thesewere more likely to cause at least some limitation in the patient’s usualactivities.Results: A total of 3.70% treated with celecoxib and 5.46% treated withNSAIDs reported moderate or severe UGI event at 6 weeks of therapy(p�0.0001; 32.2% risk reduction). In patients receiving concomitant lowdose aspirin (ASA), 3.56% Celecoxib patients (n�621) and 8.38% NSAIDpatients (n�315) experienced moderate or severe UGI event at 43 days oftherapy (p�0.028; 57.5% risk reduction). In patients not receiving ASA,3.71% Celecoxib (n�8179) and 5.23% NSAID patients (n�4079) expe-rienced moderate or severe UGI event at 43 days of therapy (p�0.0001;29.1% risk reduction). All results were consistent with end–of–studyresults at 12 weeks in each of the subgroups. Results were similar if milddyspepsia was also included in the analysis.Conclusions: The onset of moderate or severe UGI intolerance appears tooccur early in the treatment of traditional NSAIDs or celecoxib. Patients oncelecoxib therapy have a significantly reduced rate of early UGI symptomscompared to those receiving traditional NSAIDs. This difference persistedeven in the presence of concomitant low dose ASA therapy.

Sponsored by Pharmacia Corporation and Pfizer Incorporated.

737

IRRITABLE BOWEL SYNDROME IN AN EMPLOYED U.S.POPULATIONBonnie Dean, Ph.D., Daniel Aguilar, Victoria Barghout, M.S.P.H.,Feride Frech, M.P.H., Shoshanna Nakelsky, M.P.H., David Groves,Ph.D., Michael Crowell, Ph.D. and Joshua Ofman, M.D.*. Zynx HealthInc, Los Angeles, CA; Novartis Pharmaceuticals Corporation, EastHanover, NJ; Comerica, Detroit, MI and Departments of Medicine andHealth Services Research, Cedars–Sinai Medical Center, Los Angeles,CA.

Purpose: Irritable Bowel Syndrome (IBS) affects up to 20% of adults andoften afflicts those in their most productive years of life. While IBSsymptoms impact quality of life (QoL) and worker productivity, little hasbeen done to characterize the nature and impact of IBS on an employedpopulation. Our objective was to characterize the frequency, severity, andimpact of IBS among an employed population in the U.S.Methods: A survey was mailed to the workforce of a large U.S. bank withemployees in 4 states (MI, CA, TX, and FL). We report the results of thefirst phase of the survey, which was conducted to measure IBS according

S241AJG – September, Suppl., 2002 Abstracts

to the ROME II criteria and to describe workers with and without IBS(demographics, predominant symptoms, and employer information). Self–reported frequency, severity, and distress due to symptoms were assessedalong with the occurrence of abdominal surgeries. Comparisons betweenthe IBS and non–IBS groups were made using chi–square and t–tests.Results: A total of 2,536 (21.5%) of the 11,806 employees responded to thesurvey. The majority of respondents were full–time employees (91%) andfemale (77%) with a mean age of 41�/–10 years. The mean duration ofsymptoms was 11�/–10 years. Thirty–nine percent of respondents met theRome II criteria for IBS during the past 12 months. Thirty–four percent ofIBS sufferers met Rome II criteria for diarrhea, 28% met criteria forconstipation, and the remaining 38% were undifferentiated. The symptomsof abdominal pain, diarrhea, constipation, gas and bloating were eachsignificantly more frequent among respondents with IBS compared to thosewithout (p�0.05). Among the 999 respondents with IBS, over 50% re-ported moderate to severe symptoms with 67% reporting moderate tosevere abdominal pain. Abdominal symptoms, with the exception of bloat-ing, were reported as more severe among respondents with IBS (p�0.05).The IBS group reported greater levels of distress (moderate to extreme)compared to those without IBS (p�0.05). The IBS group also had asignificantly higher frequency of abdominal surgeries (p�0.05).Conclusions: Symptoms of IBS are frequent, severe and cause moderate tosevere distress among an employed population. For most abdominal com-plaints, those with IBS reported greater frequency, severity and distress, aswell as more abdominal surgeries. Further work will describe the impact onHRQoL and work–related productivity.

738

MRCP VS. ERCP: EVOLUTION INTO A CLINICALCOMMUNITY PRACTICEIdriz Kovacevic, M.D., Kyle P. Etzkorn, M.D.*, Sanjivani Joglekar,M.D., Antony Maniatis, M.D. and Mark Fleisher, M.D. ClinicalResearch Group, The Borland Groover Clinic, Jacksonville, FL.

Purpose: MRCP is quickly evolving as an alternative to ERCP in thediagnosis of hepato–biliary disease. Much has been described at academiccenters with respect to its indications, use and impact on the use of ERCP.Little is known about MRCP and its role and impact in clinical practiceoutside of academic centers.Methods: From the intoduction into our community of MRCP in January2000 till December of 2001 all ERCP & MRCP where reviewed retro-spectively. Tabulations of monthly volume, ordering physician & specialtytype, indication and corelation between MRCP & ERCP results wherestudied.Results: During a 23 month period, a total of 48 physicians ordered MRCP.21(43%) where gastroenterologist(GI),10(21%) general surgeons(GS),10(21%)General Internal Medicine & Family Practice(GIM&FP),7(14%)Oncologists(ONC). Of Gastroenterologist, the majority of MRCP(76%) where ordered by three GI, Two of whom had third tier training, andthe third individual had trained in an era prior to ERCP. Initial use ofMRCP, months 1–8 was by GI, but thereafter GS ordered followed byGIM&FP and ONC. Indication for MRCP by GI:58% High risk patients forERCP, 19% Failed Canulation, 14% Pancreatic Duct evaluation,5%Other.Of Patients who had both successful ERCP & MRCP within a 3 weekwindow, clinical correlation of findings was excellent(98%). Indications forMRCP by GS: Post Cholysystecomy Pain 61%, Jaundice 19%, Abdominalpain(?)20%. GIM&FP indications: 76% Abnormal LFT, 13% Abdominalpain,11% other. Over a 23 month period, the number of diagnostic ERCPvs. therapeutic ERCP saw a slight decline with the introduction of MRCP.Conclusions: The Introduction of MRCP has seen a decline in diagnosticERCP, while utilization of MRCP has increased in time by GS andGIM&FP. Further study of clinical impact on outcome, cost savings andreduction of the incidence of complications from ERCP with the introduc-tion of MRCP are needed in the community setting as varied subsepecial-ties order MRCP.

739

CHRONIC ABDOMINAL WALL PAIN IN REFERRALPRACTICE: LONG–TERM OUTCOMEChristopher D. Costanza, M.D., George F. Longstreth, M.D.,FACG*and Janis F. Yao, M.S. Gastroenterology, University of California atSan Diego, San Diego, CA; Gastroenterology, Kaiser Permanente, SanDiego, CA and Research and Education, Kaiser Permanente, Pasadena,CA.

Purpose: Chronic abdominal wall pain (CAWP) is a common, oftenunrecognized syndrome that can result in unnecessary diagnostic tests. Itsfrequency in practice and long–term outcome are unknown. We retrospec-tively compared the frequencies of CAWP and irritable bowel syndrome(IBS) in a gastroenterology practice and assessed diagnostic accuracy andlong–term outcome.Methods: We identified patients with CAWP from the consultation recordsfor 1996 to 2000 of a single gastroenterologist in a large group–modelHMO. In a subset of patients, we obtained demographic, clinical, diagnos-tic, treatment and follow–up data from medical records and telephoneinterviews.Results: Of 2,709 patients referred during five years, 1,004 (37.1%) weresent for endoscopic indications instead of symptoms (e.g., heme–positivestool, personal or family history of colonic neoplasm, iron deficiencyand/or colorectal cancer screening), and 1,705 (62.9%) were referred forgastrointestinal symptoms. CAWP was diagnosed in 144 (5.3% of allreferrals and 8.4% of symptomatic patients) based on the history andphysical examination, especially the presence of Carnett’s sign. IBS wasdiagnosed by Rome criteria in 278 patients (10.3% of all referrals and16.3% of symptomatic patients). Of 62 patients studied, 4 were excludedand 58 patients were in the analysis, the referring physicians had diagnosedCAWP in only 2 (3.4%). The patients’ age (mean � SD) was 51.1 � 12.9years; 45 (77.6%) were women. They had experienced pain for 12.4 � 13.7months. During 12 months before referral, there were 3.64 � 2.0 visits forabdominal pain and 1.38 � 1.60 imaging tests. The gastroenterologist’streatment comprised explanation, reassurance, analgesics and local heatapplication; none received local anesthesia. During the 12 months after thediagnosis, there were 1.07 � 1.59 visits and 0.62 � 1.20 imaging tests. Inthe 58 patients contacted 54.4 � 15.6 months after diagnosis, pain haddisappeared in 31 (53.4%); in 27 (46.6%) pain persisted at 68.7% � 20.2%of its worst severity.Conclusions: CAWP is common in gastroenterology practice and occursabout one–half as often as IBS. Referred patients are seldom diagnosed byprimary care physicians and have had much diagnostic imaging. A historyand physical examination usually permits accurate diagnosis without ad-ditional radiologic or endoscopic procedures. CAWP tends to disappear ordiminish in intensity over time.

740

DOES CONCURRENT ESOPHAGOGASTRODUODENOSCOPY(EGD) AND COLONOSOCOPY (COL) SUBJECT PATIENTS TOA HIGHER RISK OF PROCEDURE RELATED ADVERSEEVENTS?Aparna Kulkarni, M.D.*, Harish Gagneja, M.D., Madhukar Kaw, M.D.,Joyce Roquemore and Jeffrey Morris. Dept of Gastroenterology,University of Texas M D Anderson Cancer Center, Houston, TX.

Purpose: Back to back EGD and COL are performed in clinical practice.Does doing these procedures concurrently (thereby increasing sedationtime and possible therapeutic interventions) subject patients to increasedrisk of adverse events beyond that seen when doing either procedure alone?Methods: We evaluated the number of adverse events seen in 1479 patientswho had EGD alone, COL alone or both procedures done back to back atour institution from 1–1–2001 to 9–30–2001. The following adverseevents were recorded: 1.Respiratory compromise as noted by drop inPaO2 � 90%, need for reversal of medications or for endotracheal intu-bation 2. Cardiac complications such as hypotension, dysarrhythmias or

S242 Abstracts AJG – Vol. 97, No. 9, Suppl., 2002