Randomized Trial of Medical or Endoscopic Therapy to Prevent Recurrent Ulcer Hemorrhage in Patients With Adherent Clots

GASTROENTEROLOGY 2002;123:407 413 CLINICAL ALIMENTARY TRACT Randomized Trial of Medical or Endoscopic Therapy to Prevent Recurrent Ulcer Hemorrhage in Patients With Adherent Clots DENNIS M. JENSEN,* THOMAS O. G. KOVACS,* ROME JUTABHA,* GUSTAVO A. MACHICADO,*, IAN M. GRALNEK,* THOMAS J. SAVIDES, JAMES SMITH, MARY ELLEN JENSEN,* GWEN ALOFAITULI,* and JEFF GORNBEIN* *CURE Digestive Disease Research Center, UCLA School of Medicine, VA Greater Los Angeles Healthcare System, Los Angeles, California; Northridge Hospital Medical Center, Van Nuys, California; UCSD School of Medicine, San Diego, California; and Oschner Foundation Clinic, New Orleans, Louisiana See editorial on page 632. Background & Aims: Treatment of high-riskpatients with nonbleeding adherent clots on ulcers is controversial. In a previous randomized trial, there was no benefit to endoscopic therapies compared with medical therapy for prevention of ulcer rebleeding. Our purpose was to test the hypothesis that patients treated with combination endoscopic therapy would have significantly lower rebleeding rates than those treated with medical therapy. Methods: In this randomized, controlled trial, 32 high-riskpatients with severe ulcer hemorrhage and nonbleeding adherent clots resistant to target irrigation were randomized to medical therapy or to combination endoscopic therapy (with epinephrine injection, shaving down the clot with cold guillotining, and bipolar coagulation on the underlying stigmata). Physicians blinded to the endoscopic therapy managed all patients. Results: Patients were similar at study entry, except for older age in the medical group and lower platelet count in the endoscopic group. By hospital discharge, significantly more medically treated patients (6/17; 35.3%) than endoscopically treated patients (0/15; 0%) rebled (P 0.011). There were no complications of endoscopic treatment. Conclusions: Combination endoscopic therapy of nonbleeding adherent clots significantly reduced early ulcer rebleeding rates in high-riskpatients compared with medical therapy alone. This endoscopic treatment was safe. Stigmata of peptic (duodenal or gastric) ulcer hemorrhage have been used for more than a decade to stratify patients according to the risk of recurrent hemorrhage with medical treatment. 1 3 Also,endoscopic therapies have been recommended for patients with major stigmata of ulcer hemorrhage,such as active bleeding or nonbleeding visible vessels,by many investigators who performed randomized prospective studies and a National Institutes of Health consensus panel which reviewed them. 4 In contrast,other stigmata of ulcer hemorrhage such as adherent clots were not recommended for endoscopic therapy,based on previous results. 4 Clots found by endoscopists in the base of an ulcer have been characterized as adherent or nonadherent after suctioning and target jet irrigation. 1,5,6 Nonbleeding, adherent clots obscure the underlying stigmata,which may be a nonbleeding visible vessel,a lesser stigmata such as a flat spot,or no stigma. If active bleeding from underneath a clot is not seen during the endoscopy,the stigma underlying the clot is one of the risk factors that determines the rate of rebleeding when patients are treated medically. Accordingly,the rebleeding rates on medical therapy may vary from about 50% for a nonbleeding visible vessel underlying an adherent clot,to 10% for a flat spot,to about 3% for a clean ulcer base. 1,7 In a large study of the natural history of stigmata of ulcer hemorrhage,lau et al. reported that among 778 patients with severe ulcer hemorrhage,104 (13%) had adherent clots on emergency endoscopy. 8 With medical therapy,29% of patients with adherent clots rebled during the hospitalization. This contrasted with a rebleeding rate on medical therapy of 39% of patients with nonbleeding visible vessels,13% with flat spots,and 5% with a clean ulcer base. Only 3 groups that reported randomized trials of medical compared with endoscopic therapy for patients with severe upper gastrointestinal (UGI) hemorrhage Abbreviations used in this paper: ICU, intensive care unit; UGI, upper gastrointestinal. 2002 by the American Gastroenterological Association 0016-5085/02/$35.00 doi:10.1053/gast.2002.34782

408 JENSEN ET AL. GASTROENTEROLOGY Vol. 123, No. 2 have included nonbleeding adherent clots on peptic ulcers. 3,9 11 Swain et al. stratified patients by stigmata of ulcer hemorrhage in several trials and compared the rates of further bleeding of patients treated with endoscopic lasers (argon or neodymium-yttruim aluminum garnet) or medical therapy. 3,9 Rates of recurrent bleeding of ulcer patients with adherent clots were 10% 25% for laser or for medical therapy,but depended on whether bleeding was seen under the clot. Swain cautioned about the use of lasers for treatment of adherent clots that could not be irrigated from the ulcer base because induction of active bleeding was common with laser treatment and the rebleeding rate on medical therapy was relatively low. In a previous randomized,prospective blinded trial of different treatments for nonbleeding adherent clots,the Center for Ulcer Research and Education (CURE) Hemostasis Research Group reported rates of further bleeding with medical therapy as 30%,with heater probe coagulation as 33.3%,and with injection of alcohol as 37.5%. 10 In that study,induction of severe bleeding sometimes resulted when clots were forcibly removed to expose the underlying stigma. 10 A cost analysis of the study revealed that endoscopic therapy did not result in any cost savings. More recently,bleau et al. reported in abstract form that combination therapy (with epinephrine injection and heater probe) significantly reduced rebleeding rates of patients with nonbleeding adherent clots compared with medical therapy alone: 4.8% vs. 34.3%,respectively. 11 The purposes of this study were: (1) to test the hypothesis that patients treated with endoscopic combination therapy for nonbleeding adherent clots on ulcers would have significantly lower rebleeding rates than those treated with medical therapy alone and (2) to report a safe and effective method to remove most of the adherent clot so that endoscopic hemostasis of the underlying stigmata was efficacious. Materials and Methods General Methods and Criteria for Entry and Rebleeding This was a multicenter,prospective randomized trial. The study was approved by Human Subjects Protection Committees at each hospital before initiating the study. Patients with severe UGI bleeding and ulcers were assessed for a multicenter trial of endoscopic hemostasis of active arterial bleeding,nonbleeding visible vessels,or nonbleeding adherent clots. Patients in the last subgroup are the subjects of this report. Entry criteria included: (1) severe UGI bleeding requiring intensive care unit (ICU) or monitored unit admission for resuscitation and a documented hematocrit drop of at least 8% or transfusion of 2 or more units of packed red blood cells,(2) written informed consent before emergency endoscopy to be randomized if an adherent clot was found,and (3) a nonbleeding adherent clot on a peptic ulcer resistent to water jet irrigation and suctioning and obscuring the underlying stigma. Exclusion criteria were: (1) previous endoscopic therapy for this bleeding ulcer,(2) age less than 18 years,(3) pregnancy,(4) life expectancy of less than 30 days because of very severe comorbid medical or surgical problems exclusive of ulcer hemorrhage,(5) anticoagulation therapy required or uncorrectable severe coagulopathy (defined as platelet count less than 50,000 or prothrombin time more than 4 seconds prolonged despite transfusions of platelets or fresh frozen plasma, respectively),and (6) uncooperative or combative patients. At the time of emergency endoscopy for severe UGI hemorrhage,32 consecutive patients who met entry criteria and had nonbleeding adherent clots resistant to target irrigation on endoscopy were entered into this study. Patients were randomized at each site by opening a sealed opaque envelope that designated endoscopic treatment or medical-sham endoscopic treatment. Physicians who were blinded to the endoscopic therapy managed all patients before and after the random assignment. Criteria for rebleeding were established before the study began. All 3 of the following criteria were used to define rebleeding,which occurred 6 or more hours after randomization: (1) recurrence of hematochezia,melena,or hematemesis; (2) decrease in hemoglobin by at least 2 grams; and (3) concurrent tachycardia and/or hypotension. For all patients whose physicians documented rebleeding by these criteria,the previous therapy was declared a failure and the code was broken. All patients with rebleeding were then offered,in consultation with their managing physicians,the choice of further medical therapy alone,repeat endoscopy with endoscopic treatment of major stigmata,or ulcer surgery. Treatment Techniques: Endoscopic and Medical Groups Before initiation of this study,there was a meeting of investigators to standardize techniques for diagnosis and therapy (Table 1). Figures,endoscopic pictures,and videos of nonbleeding adherent clots on ulcers were used to teach investigators to distinguish them from nonbleeding visible ves- Table 1. Study Criteria for Distinguishing Nonbleeding Adherent Clots and NonbleedingVisible Vessels Visible vessel Adherent clot Size a 4mm 6 mm Color Variable or translucent Red Form Discrete Amorphous Relation to stigma Is the stigma Covers up stigma Rebleedingrate b About 50% About 35% a Size refers to diameter and height above ulcer base. b Rebleedingrate is on medical therapy (histamine two receptor antagonist drugs or proton pump inhibitors, transfusions, and correction of coagulopathies).

August 2002 RANDOMIZED TRIAL OF ADHERENT CLOTS ON ULCERS 409 sels and flat spots. Furthermore,the endoscopic treatment was standardized,based on our previous trial 10 and a subsequent feasibility study of patients with nonbleeding adherent clots. 12 Our treatment technique was recently described for nonbleeding adherent clots,and this was taught by diagrams,slides, and videos to investigators before the study began. 13 Briefly, the pedicle of the adherent clot was injected with 1-mL aliquots of dilute epinephrine (1:10,000 in normal saline) in 4 quadrants. Then a disposable polyp snare without electrocoagulation was used to shave down the clot by cold guillotining off segments of the clot and to deliver them away from the ulcer. The technique was to shave the clot down to 3 4 mm above the ulcer base without forcibly removing the clot from its attachment because disruption of the pedicle could be associated with induction of active bleeding. 10 Target washing with a probe was used to help visualize the underlying stigma. More injections of epinephrine were used if any bleeding was induced. The small adherent clot remnant and/or the underlying stigmata were then treated with endoscopic coaptive coagulation,as previously described. 1,7,13,14 The treatment parameters for coaptive coagulation of ulcers were: a large diameter (3.2 mm) bipolar probe (Gold probe; Microvasive Boston Scientific,Natick,MA),generator setting of 12 15 watts,firm tamponade pressure on the stigmata or residual clot,and 10 seconds of coagulation per application before moving the probe. The endpoint was flattening the residual small clot or visible vessel,effective coagulation with white coagulum,and no bleeding of the ulcer or stigma. The endoscopist used as many probe applications as necessary to achieve these endpoints. For patients randomly assigned to medical therapy,sham endoscopic therapy was conducted. This consisted of further gentle target irrigation of the ulcer base and clot,without contact of the bipolar probe or endoscope to the clot and without any endoscopic therapy. Forceful suctioning on top of the clot was not used. All patients received similar general medical therapy consisting of: (1) intensive care unit or monitored bed observation for at least 24 hours after randomization,(2) correction of coagulopathies (if present),(3) full liquid diet and proton pump inhibitors twice per day,starting immediately after recovery from the conscious sedation of the emergency endoscopy,(4) transfusion of packed red cells for severe anemia (with hematocrit less than 25) or clinically severe rebleeding,and (5) treatment of Helicobacter pylori infection if present. H. pylori treatment was started within 72 hours of randomization and when the patient was hemodynamically stable and ingesting their usual diet. All patients were discharged from the hospital when there was no evidence of rebleeding for at least 72 hours, when they were medically stable,and when they had resumed their usual level of activity and eating. Sample Size Calculation and Data Analysis Before starting this study,the sample size was estimated. This was based on conservative estimates of rebleeding rates before discharge of 35% for the medical group and 5% for the endoscopic treatment group. These estimates were from our previous randomized controlled trial of medical vs. monotherapies for the medical treatment rebleeding rate 10 and a subsequent feasibility study of combination therapy for nonbleeding adherent clots in patients with severe ulcer hemorrhage and comorbid medical-surgical conditions for the endoscopic treatment rebleeding rate. 12 With an 80% power (ß of 0.20) using 2-tailed testing and an alpha of 0.05,27 patients in each group were calculated to be the sample size. For the final analysis, P values of 0.05 for the primary outcome (rebleeding rate) and other outcomes were chosen at the study onset as statistically significant. An interim analysis was planned when about 60% of the enrollment was completed. The interim analysis results were previously published as an abstract. 15 Meanwhile,enrollment into the trial continued. Subsequently,an external advisory board was scheduled to review the results of the whole study (e.g.,active bleeding,visible vessel,and adherent clots) and to decide whether the whole study or any parts of it (e.g., adherent clots) should be stopped early,if statistically significant differences were found,according to the criteria set forth by O Brien and Fleming. 16 A P value of 0.02 was selected at the study onset for stopping early. Between the time of the interim analysis results 15 and the meeting of the external advisory board,2 additional clot patients had been randomized into the medical,rebled,and completed the study. They were also considered in the deliberations of the advisory board, which recommended stopping the adherent clot study early. Data on demographics,transfusions,rebleeding,further endoscopic therapy,hospital and ICU days,and complications until hospital discharge were prospectively collected by a research coordinator. In addition,data on ulcer surgery and mortality up to 30 days after randomization were compiled. SAS was used for data management. For statistical analysis,background variables and results were compared with 2 or analysis of variance. 17 A P value 0.05 was considered significant for the final analysis. The 95% confidence intervals were also calculated for the risk differences in rebleeding rates and further endoscopic treatment. Results During the entire study (October 1995 to March 2001),147 patients hospitalized with severe ulcer hemorrhage and major stigmata of ulcer hemorrhage on emergency endoscopy were randomized. These included 38 with active arterial bleeding (25.9% of the total),77 with nonbleeding visible vessels (52.4%),and 32 with nonbleeding adherent clots (21.8%). The former 2 subgroups for which 2 different endoscopic treatments were compared are reported upon elsewhere. 18,19 Only the adherent clot subgroup,for which medical treatment was compared with endoscopic therapy,is included in this report. One patient who had been included in the interim analysis of nonbleeding adherent clots in the treat-

410 JENSEN ET AL. GASTROENTEROLOGY Vol. 123, No. 2 Table 2. Background Characteristics of Patients by Treatment Group Medical group Endoscopic treatment Number of patients 17 15 Age a 70.1 4.3 55.8 3.3 b M/F 14/3 13/2 Hypotension on admission 7 (41.2%) 5 (33.3%) ASA and/or NSAID ingestion 15 (88.2%) 10 (66.7%) H. pylori positive 7 (41.2%) 10 (66.7%) Dyspepsia before bleed 7 (41.2%) 4 (26.7%) Recent alcohol consumption 3 (17.7%) 5 (33.3%) Smoker 2 (11.8%) 3 (20%) Musculoskeletal disease 11 (64.7%) 4 (26.7%) b Chronic liver disease 2 (11.8%) 4 (26.7%) Bleedingas reason for admission 8 (47.1%) 12 (80%) Other reason for admission 9 (52.9%) 3 (20%) Duodenal/gastric ulcers 8/9 6/9 Ulcer diameter (mm) a 11.8 1.1 12.6 2.5 Baseline hematocrit a 23.5 1.0 24.9 1.0 Comorbidity score 2.3 0.3 1.9 0.3 Platelet count a 258, 118 30950 178,267 17603 b Prothrombin time (seconds) a 14.6 1.5 13.0 0.9 Partial thromboplastin time (seconds) a 30.9 3.0 25.2 1.7 Transfusions before randomization Red cells a 3.2 0.4 2.9 0.7 Fresh frozen plasma a 0.5 0.3 1.1 0.6 Platelets 0 0 a Mean SEM. b P 0.05. ment arm 15 on review of study forms before the final analysis was found to have a nonadherent clot that was washed off to reveal a nonbleeding visible vessel. This patient did not rebleed and had an uneventful recovery. He was excluded from the final analysis of the adherent clot group but was included with the nonbleeding visible vessel group. 19 Thirty-two patients with nonbleeding adherent clots on ulcers were randomly assigned. Table 2 shows the background features of these patients according to their treatment group. The mean age of the patients and prevalence of musculoskeletal disorders (such as arthritis) were significantly higher for the medical than the endoscopic treatment group. In contrast,the mean platelet count was significantly lower in the endoscopic treatment group. No other differences in background variables were significant. For the nonbleeding adherent clot group,the trial was stopped early at the recommendation of the external advisory board after they reviewed the interim analysis results and subsequent enrollment and rebleeding rates. Table 3 presents the outcomes after randomization of all the study patients from the final analysis. Before hospital discharge,the medically treated group had an ulcer rebleeding rate after randomization of 35.3% (6/ 17),which was significantly higher than 0% rebleeding rate of the endoscopic group (P 0.011). A 95% confidence interval for the true difference in rebleeding rate is 12.6% 58.0% (Figure 1). The rate of further endoscopic treatment of 23.5% (4/17) in the medical Table 3. Outcomes After Randomization by Treatment Group Medical treatment Endoscopic treatment P value Patients 17 15 Rebleedingbefore discharge 6 (35.3%) 0 0.011 Further endoscopic treatment 4 (23.5%) 0 0.045 Units RBCs transfused a 2.3 0.9 1.0 1.0 0.35 Hospital days, median 4 4 0.33 ICU days, median 1 1 0.44 Ulcer surgery, 30 days 2 (11.8%) 1 (6.7%) 0.62 Mortality, 30 days 1 (5.9%) 1 (6.7%) 0.93 a Expressed as mean SEM.

August 2002 RANDOMIZED TRIAL OF ADHERENT CLOTS ON ULCERS 411 Figure 1. The recurrence of ulcer hemorrhage following randomization to medical vs. endoscopic therapy, up to the time of hospital discharge for patients with nonbleeding adherent clots. *Indicates a significant difference (P 0.011). group was significantly higher than the 0% rate in the endoscopic group (P 0.045). A 95% confidence interval of the true difference in endoscopic treatment rates is 3.3% 43.7%. There were no significant differences between treatments in units of red blood counts transfused after randomization or median post-randomization hospital or ICU days. For 30 days after randomization,there were no significant differences in ulcer surgery or deaths. There were no complications of endoscopic therapy. Specifically,in no case was severe hemorrhage induced nor did perforation result. Discussion Hemorrhage from peptic ulcers is the most common cause of severe UGI bleeding. 1 However,the overall prevalence of nonbleeding adherent clots in benign-appearing peptic ulcers during emergency endoscopy of patients with severe UGI hemorrhage is low and reported as 10% 13%. 1,7,8 Among patients assessed for the present study with major stigmata of hemorrhage (e.g., active arterial bleeding,nonbleeding visible vessels,or clots),the adherent clot subgroup was the smallest, representing about 22% of those randomly assigned. Nevertheless,nonbleeding adherent clots are an important subgroup of patients,which is a challenge to gastroenterologists and endoscopists because the choice of medical or endoscopic treatment of patients with this stigma of ulcer hemorrhage is controversial and the previous results of endoscopic monotherapies (such as heater probe or sclerotherapy alone) have not been promising. 10 Swain and others have reported that the nonbleeding visible vessel is a small clot in a rent of the underlying artery. 2,3 Nevertheless,most investigators including Swain distinguish nonbleeding visible vessels endoscopically as a stigmata of recent hemorrhage from larger adherent clots that cover all or part of the ulcer base and obscure the underlying stigmata. Although some investigators may disagree with the arbitrary differentiation of clots and visible vessels based on size,obscuring what is beneath,and other features (Table 1),this standardization was deemed necessary and useful before this study started. The reasons were that in most prospective trials reported,patients with nonbleeding visible vessels (as defined herein) have higher rebleeding rates (about 50%) on medical therapy than patients with adherent clots (rebleed rate about 30% 35%). There are limited controlled trial data about the adherent clot group (as defined herein) and endoscopic treatment is controversial. Therefore,the focus of this study was on adherent clots. The independent risk factors for recurrence of rebleeding from peptic ulcers before hospital discharge are listed in Table 4. 1,6,7 These include clinical features such as age greater than 65 years,in-hospital onset of hemorrhage, presentation with hypotension or red blood emesis,and severe comorbidity. 1,6,7 These are the clinical parameters for all patients with peptic ulcer hemorrhage from multivariate analyses,regardless of stigmata of ulcer hemorrhage. It is not known whether age alone is an independent predictor of rebleeding in subgroups with different stigmata,such as adherent clots. Endoscopic major stigmata of hemorrhage (active bleeding or nonbleeding visible vessels) and large ulcers ( 2 cm) also increased the risk of ulcer rebleeding in some studies. 1,5,13 In the present study,the lower platelet count in the endoscopic treatment group could increase the risk of rebleeding and actually bias the results against this treatment. In contrast,musculoskeletal disorders,which had a higher prevalence in the medical group,are not reported to be an independent risk factor for ulcer rebleeding,although these disorders may be associated with increased nonsteroidal anti-inflammatory drug (NSAID) or aspirin use. However,the prevalence of NSAID and/or aspirin use Table 4. Independent Risk Factors for Early Recurrence of Ulcer Hemorrhage Clinical and laboratory Age 65 yr Inpatient hemorrhage Hypotension or shock Red blood emesis or hematochezia Severe comorbidity Severe coagulopathy Endoscopic Major stigmata of hemorrhage (active arterial bleeding or nonbleedingvisible vessel) Large ulcer ( 2 cm diameter) Data from references 1, 6, 7, and 13.

412 JENSEN ET AL. GASTROENTEROLOGY Vol. 123, No. 2 before the ulcer hemorrhage was not significantly different in the study groups. Among the patients treated in this study,the only other significant difference in independent risk factors was in the age,which was higher in the medical group and might influence rebleeding rates. However,in our previous studies of patients with ulcers, and UGI hemorrhage,age was not an independent risk factor for rebleeding for the subgroup of patients with nonbleeding adherent clots. 10,12 Rebleeding rates for patients on medical therapy with nonbleeding adherent clots vary in the literature and seem to depend on other risk factors than the clot alone. 1,3,6,8 12 Elderly patients with large ulcers (e.g.,at least 12 mm in diameter) and severe comorbid conditions,hospitalized for severe UGI hemorrhage,having nonbleeding adherent clots on endoscopy,and treated medically were reported to have rebleeding rates of about 35% in our 2 prior randomized trials. 1,10 The rebleeding rate of 35.3% in this study for the medical therapy group (with high-dose oral proton pump inhibitors) is very similar to the rebleeding rates for medical therapy in these earlier reports. In contrast,the rebleeding rate on medical therapy of low-risk,younger patients with smaller ulcers and no comorbid conditions was 8% in another recent prospective study. 5 Based on the potential of inducing bleeding by endoscopic treatment or removal of the clot, 2,3,10 the poor results of earlier randomized prospective endoscopic trials, 2,3,10 and the low rebleeding rate reported for young, low-risk patients on medical therapy, 5 the standard of care in most major medical centers previously has been medical therapy alone for ulcer patients with severe UGI hemorrhage and nonbleeding adherent clots. However,both the endoscopic therapy and medical therapies have recently improved. In a feasibility study, combination therapy as described herein offered significant promise in reducing rebleeding rates of high-risk patients with nonbleeding adherent clots on ulcers. 12 Both the ulcer rebleeding rate and the need for further endoscopic treatment after randomization were significantly reduced in the present study. Another group of investigators in a multicenter trial has also reported that combination epinephrine and heater probe significantly reduced the early rebleeding rate compared with medical therapy. 11 Our technique of combination therapy also has been successfully and safely applied to endoscopic treatment of patients with severe diverticular hemorrhage when adherent clots obscuring the underlying stigma were found on urgent colonoscopy after colonic purge. 20,21 The endoscopic combination therapy that was used in this study seems to be safer and significantly more effective than previous endoscopic monotherapies. 10 Severe bleeding was not induced by the combination therapy, and rebleeding before hospital discharge did not occur in the present study. In contrast,previous monotherapies (heater probe or alcohol injection alone) along with clot removal from ulcers of patients similar to those in the present study,were sometimes associated with induction of severe active bleeding,and the overall clinically significant rebleeding rates with these treatments exceeded 35%. 10 In our opinion,both the preinjection of epinephrine around the clot pedicle and subsequently the technique of shaving down the clot without pulling it off its attachment to the ulcer base in this present study contributed to the lack of severe induced bleeding. Furthermore,these techniques plus shaving down the clot to a small remnant and/or exposing the underlying stigmata contributed to the success of coaptive coagulation of the underlying artery and to effective hemostasis. 1,14 From this study,we conclude that: (1) high-risk patients with nonbleeding adherent clots on ulcers treated with medical therapy (e.g.,proton pump inhibitors twice a day) had severe ulcer rebleeding 35.3% of the time; (2) similar patients treated endoscopically had a significantly lower rebleeding rate (0% in this study); (3) the technique of combination endoscopic therapy (with epinephrine injection,shaving down the clot with cold guillotining,and coaptive coagulation with bipolar probe) was efficacious,did not induce severe bleeding,and was safe; and (4) endoscopic combination therapy seemed to be significantly more effective than our previous endoscopic monotherapies (heater probe or alcohol injection) for treatment of similar patients with adherent clots on ulcers. 10 Because of potential concerns about the small sample size,our inability to enlarge the current study,and the lack of differences in other outcomes such as hospital stay,an element of caution is recommended. Although rates of rebleeding and repeat endoscopy with therapy for rebleeding were significantly lower in the endoscopic than medical group,further studies may be warranted to confirm these results and evaluate the influence of therapies on other outcomes such as hospital stay,surgery, and cost of care. References 1. Jensen DM. Endoscopic control of non-variceal upper gastrointestinal hemorrhage. In: Yamada T, Alpers D, Laine L, Owyang C, Powell D. Textbook of gastroenterology. 3rd ed. Philadelphia: Lippincott, 1999:2857 2879. 2. Swain CP, Storey DW, Bown SG, Heath J, Mills TN, Salmon PR, Northfield TC, Kirkham JS, O Sullivan JP. Nature of the bleeding vessel in recurrently bleeding gastric ulcers. Gastroenterology 1986;90:595 608. 3. Swain CP, Salmon PR, Kirkham JS, Bown SG, Northfield TC.

August 2002 RANDOMIZED TRIAL OF ADHERENT CLOTS ON ULCERS 413 Controlled trial of Nd: YAG laser photocoagulation in bleeding peptic ulcer. Lancet 1986;1:1113 1116. 4. National Institutes of Health Consensus Development Conference. Therapeutic endoscopy and bleedingulcers. JAMA 1989; 262;1369 1372. 5. Laine L, Stein C, Sharma V. A prospective outcome study of patients with clot in an ulcer and the effect of irrigation. Gastrointest Endosc 1996;43:107 110. 6. Freeman ML. The current endoscopic diagnosis and intensive care unit management of severe ulcer and other nonvariceal upper gastrointestinal hemorrhage. Gastrointest Endosc Clin N Am 1991;1:209 239. 7. Jensen DM. Current diagnosis and treatment of severe ulcer hemorrhage. Clinical update. In: Lieberman D, ed. ASGE 1999; 6:1 4. 8. Lau JY, ChungSC, LeungJW, Lo KK, YungMY, Li AK. The evolution of hemorrhage in bleeding peptic ulcers: a sequential endoscopic study. Endoscopy 1998:30:513 518. 9. Matthewson K, Swain CP, Bland M, Kirkham JS, Bown SG, Northfield TC. Randomized comparison of Nd:YAG laser, heater probe, and no endoscopic therapy for bleedingpeptic ulcers. Gastroenterology 1990;98:1239 1244. 10. Jensen DM, Kovacs TOG, Jutabha R, Randall GM, ChengS, Jensen ME, Freeman M, Gralnek I, Sue M, Machicado GA, Gornbein J. Final results and cost assessment of endoscopic vs. medical therapies for prevention of recurrent ulcer hemorrhage from adherent clots in a randomized controlled trial (abstr). Gastrointest Endosc 1995;41:279. 11. Bleau BF, Gostout CJ, Shaw MJ, Keate RF, Harford WV, Bracy W, Magee D, Fleiseher DE, and the Adherent Clot Study Group. Final results: rebleedingfrom peptic ulcers associated with adherent clots: a prospective randomized controlled study comparingendoscopic with medical therapy (abstr). Gastrointest Endosc 1997;45:AB87. 12. Jensen DM, Kovacs TOG, Jutabha R, Machicado GA, Savides T, Smith J. A safe and effective technique for endoscopic removal of adherent clots from GI lesions: cold guillotining after epinephrine injection (abstr). Gastrointest Endosc 1996;43:A25. 13. Jensen DM. Thermal probe or combination therapy for nonvariceal UGI hemorrhage. Tech Gastrointest Endosc 1999;1: 107 114. 14. Johnston JH, Jensen DM, Auth D. Experimental comparison of endoscopic Yttrium-aluminum-garnet laser, electrosurgery, and heater probe for canine gut artery coagulation: the importance of vessel compression and avoidance of tissue erosion. Gastroenterology 1987;92:1101 1108. 15. Jensen DM, Kovacs TOG, Jutabha R, Machicado GA, Gralnek IM, Savides JT, Smith J, Lam F, Fontana L, ChengS, Jensen ME, Alofaituli G. Randomized controlled trial of medical therapy compared to endoscopic therapy for prevention of recurrent ulcer hemorrhage in patients with non-bleeding adherent clots (abstr). Gastrointest Endosc 2000;51:AB131. 16. Fleming TR, Harrington DP, O Brien PC. Designs of group sequential tests. Control Clin Trials 1984:5:348 361. 17. Bailar JC III, Mosteller F. Medical uses of statistics. 2nd ed. Waltham, MA: N Engl J Med Books, 1992:183 184. 18. Jensen DM, Smith J, Savides TJ, Kovacs TOG, Jutabha R, Gralnek IM, Machicado GA, Lam F, Fontana L, ChengS, Jensen ME, Gornbein J. Randomized controlled study of combination epinephrine injection and Gold probe compared to Gold probe alone for hemostasis of actively bleedingpeptic ulcers (abstr). Gastrointest Endosc 2000:51:AB201. 19. Jensen DM, Kovacs TOG, Jutabha R, Gralnek IM, Machicado GA, Savides TJ, Smith J, ChengS, Fontana L, Lam F, Jensen ME, Gornbein J. Randomized prospective study of bipolar coagulation alone compared to combination epinephrine injection and bipolar coagulation for prevention of rebleeding from ulcers with nonbleedingvisible vessels (abstr). Gastrointest Endosc 2000;51: AB201. 20. Jensen DM, Machicado GA. Colonoscopy for diagnosis and treatment of severe lower gastrointestinal bleeding: routine outcomes and cost analysis. Gastrointest Endosc Clin N Am 1997;7:477 498. 21. Jensen DM, Machicado GA, Jutabha R, Kovacs TOG. Urgent colonoscopy for the diagnosis and treatment of severe diverticular hemorrhage. N Engl J Med 2000:342:78 82. Received February 18, 2002. Accepted February 28, 2002. Address requests for reprints to: Dennis M. Jensen, M.D., CURE Digestive Disease Research Center, Building 115, Room 318, VA Greater Los Angeles Healthcare System, 11301 Wilshire Boulevard, Los Angeles, California 90073. e-mail: djensen@mednet.ucla.edu; fax: (310) 794-2908. Supported in part by NIH RO1 DK 33273, NIH 1K24DK 02650, and NIH DK 41301 (CURE Human Studies CORE), by Microvasive-Boston Scientific Corporation, and by a Veterans Administration HSR & D Career Development Award (to I.M.G.). The authors thankthe three study coordinators who carefully followed all patients in this study: Florence Lam, RN, Lana Fontana, RN, and Susie Cheng, RN. We also thankjulie Pham for word processing this manuscript and Ken Hirabayashi for making the figure.