Peptic ulcer bleeding remains the most common cause of hospitalization

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2009;7: Predicting Mortality in Patients With Bleeding Peptic Ulcers After Therapeutic Endoscopy PHILIP W. Y. CHIU,* ENDERS K. W. NG,* FRANCES K. Y. CHEUNG,* FRANCIS K. L. CHAN, W. K. LEUNG, JUSTIN C. Y. WU, VINCENT W. S. WONG, M. Y. YUNG,* KELVIN TSOI, JAMES Y. W. LAU,* JOSEPH J. Y. SUNG, and SYDNEY S. C. CHUNG* *Department of Surgery and Department of Medicine and Therapeutics, Institute of Digestive Disease, Prince of Wales Hospital, Chinese University of Hong Kong, Hong Kong This article has an accompanying continuing medical education activity on page 253. Learning Objective Identify factors associated with mortality in patients with bleeding peptic ulcer treated with therapeutic endoscopy. See CME exam on page 253. See Editorial on page 257. Background & Aims: Despite advances in management of patients with bleeding peptic ulcers, mortality is still 10%. This study aimed to identify predictive factors and to develop a prediction model for mortality among patients with bleeding peptic ulcers. Methods: Consecutive patients with endoscopic stigmata of active bleeding, visible vessels, or adherent clots were recruited, and risk factors for mortality were identified in this deprivation cohort by using multiple stepwise logistic regression. A prediction model was then built on the basis of these factors and validated in the evaluation cohort. Results: From 1993 to 2003, 3220 patients with bleeding peptic ulcers were treated. Two hundred eighty-four of the patients developed rebleeding (8.8%); emergency surgery was performed on 47 of these patients, whereas others were managed with endoscopic retreatment. Two hundred twenty-nine of these sustained in-hospital death (7.1%). In patients older than 70 years, presence of comorbidity, more than 1 listed comorbidity, hematemesis on presentation, systolic blood pressure below 100 mm Hg, in-hospital bleeding, rebleeding, and need for surgery were significant predictors for mortality. Helicobacter pylori related ulcers had lower risk of mortality. The receiver operating characteristic curve comparing the prediction of mortality with actual mortality showed an area under the curve of From 2004 to 2006, data were collected prospectively from a second cohort of patients with bleeding peptic ulcers, and mortality was predicted by using the model developed. The receiver operating characteristic curve showed an area under the curve of Conclusions: Among patients with bleeding peptic ulcers after endoscopic hemostasis, advanced age, presence of listed comorbidity, multiple comorbidities, hypovolemic shock, in-hospital bleeding, rebleeding, and need for surgery successfully predicted in-hospital mortality. Peptic ulcer bleeding remains the most common cause of hospitalization for acute nonvariceal gastrointestinal bleeding. 1 Although endoscopic therapy has been demonstrated to be effective in achieving primary hemostasis for bleeding peptic ulcers, with lower rates of transfusion, hospitalization, and mortality compared with surgery, mortality from bleeding peptic ulcers still amounts to 10%. 1 5 A scoring system for the prediction of mortality in peptic ulcer bleeding would facilitate the identification of at-risk patients so that aggressive management strategies could improve the outcomes. Few studies address the prediction of mortality in patients with bleeding peptic ulcers. This study aims to identify potential predictive factors for hospital mortality among patients with bleeding peptic ulcers. Methods A single-center study was conducted at the endoscopy center of the Prince of Wales Hospital, which serves a population of 1.5 million people in the northeastern part of Hong Kong. Consecutive patients with bleeding peptic ulcers were managed by a dedicated team of endoscopists composed of both gastroenterologists and surgeons. Prospective data from 1993 to 2006 on the endoscopic diagnosis, therapeutic options, and surgical outcomes of these patients were collected. The study consisted of 2 parts; the first part (with data from 1993 to 2003) concentrated on identifying predictive factors and constructing a prediction model for bleeding ulcer related mortality from this derivation cohort, and the second part (with data from 2004 to 2006) validated this prediction model in another evaluation cohort of patients with bleeding peptic ulcers. All patients with clinical diagnosis of upper gastrointestinal bleeding received endoscopy within 24 hours after admission. Peptic ulcers with active bleeding, visible vessels, or adherent clots at the ulcer base were treated. Adherent clots were actively washed away to reveal the base of the ulcer by vigorous irrigation or mechanical removal with a mini-snare. Endoscopic treatment consisted of a single therapy, including epinephrine injection with Abbreviations used in this paper: AUC, area under the curve; ROC, receiver operating characteristic by the AGA Institute /09/$36.00 doi: /j.cgh

2 312 CHIU ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 3 1:10,000 dilution, injection of alcohol, thermal therapy with a 3.2-mm heater probe, or a combination of injection plus thermal therapy. The heater probe (CD-10Z; Olympus, Tokyo, Japan) was introduced through a dual channel endoscope (2T-10 or 2T-200; Olympus, Hong Kong) and targeted the ulcer base with 30 J per goal until the achievement of a coaptive effect. The presence of Helicobacter pylori was confirmed with both a rapid urease test and biopsy taken at the antrum. After therapeutic endoscopy, the patients were transferred to a designated gastroenterology ward for close observation. Permanent hemostasis was defined as successful initial hemostasis without the development of rebleeding. Rebleeding after therapeutic endoscopy is defined as follows: fresh hematemesis, hypotension (systolic blood pressure 90 mm Hg) with tachycardia (pulse 110 beat/min) or melena, or a total transfusion of more than 4 units to maintain a hemoglobin level greater than 10 g/dl within 72 hours. During the first derivation period, several comparative clinical trials were conducted. These included a randomized trial comparing injection alone against injection plus heater probe for bleeding peptic ulcers (total number of patients, 276), 6 a randomized trial comparing repeat endoscopy or surgery for peptic ulcer rebleeding (total number of patients, 92), 7 a randomized trial comparing the use of adjunctive high-dose omeprazole infusion against a placebo for peptic ulcer rebleeding (total number of patients, 240), 8 and a randomized trial investigating the need for endoscopic treatment for bleeding ulcers with visible vessels or adherent clots after high-dose omeprazole infusion (total number of patients, 162). 9 The total number of patients involved in these randomized studies in the database was 770. During the second validation period, one randomized trial was conducted on the use of omeprazole infusion before endoscopy in patients with acute nonvariceal upper gastrointestinal bleeding. In this study, a total of 150 patients presented with bleeding peptic ulcers that required endoscopic treatment, and 60 patients received intravenous omeprazole infusion before endoscopic therapy. 10 Patients baseline characteristics and details of the endoscopic therapy were recorded on a pro forma basis by endoscopists at the end of the procedure. Baseline demographics, clinical presentation of the bleeding peptic ulcers, severity of the bleeding, endoscopic findings, treatments performed, and the clinical outcomes were noted. Comorbidity is defined as the presence of any one of the following diseases, including (1) cardiac diseases including myocardial infarction and congestive heart failure; (2) chronic pulmonary diseases; (3) acute and chronic liver disease including liver failure and cirrhosis; (4) gastrointestinal or biliary diseases including previous history of peptic ulcer and gallstone diseases; (5) acute and chronic renal diseases; (6) vascular disorders including peripheral vascular diseases and aortic aneurysm; (7) cerebrovascular disease; (8) diabetes mellitus and endocrine diseases; (9) trauma; (10) hematologic disorders including leukemia and lymphoma; (11) autoimmune diseases including systemic lupus erythematous, scleroderma, and rheumatic arthritis; (12) chronic bone pain and diseases; (13) presence of any malignancy; and (14) burns. Data entries were checked by full-time research nurses to confirm the accuracy, and these data were entered daily into the computerized gastrointestinal bleeding registry. The clinical outcomes of all patients were tracked by research nurses to document rebleeding, hospital stay, transfusion requirement, and in-hospital mortality. Statistical analysis was performed with the SPSS program v.14 (SPSS Inc, Chicago, IL). Clinical data collected including demographics and endoscopic findings were first analyzed by using the 2 test or Fisher exact test, as appropriate. The potential predictive factors of mortality risk were identified if the P value was less than.15 on initial univariate analysis. These potential factors were then subjected to multivariate analysis by using multiple stepwise logistic regression. The predictive power of the combined factors identified from the logistic regression was validated in a separate prospectively collected database. A predictive model was developed from the previously identified risk factors, and the risk factors were weighted according to the representation from the effect of their odds ratio. The calculated predictive score from the formula was analyzed by using the receiver operating characteristic (ROC) curve and represented by using the area under the curve (AUC). An AUC of 0.5 indicates no predictive power, whereas a value of 1.0 indicates excellent predictive power. Results From 1993 to 2003, 9032 patients presented to the Endoscopy Center, Prince of Wales Hospital, Chinese University of Hong Kong with bleeding peptic ulcers. Of these patients, 3220 formed the derivation cohort who had bleeding peptic ulcers with endoscopic stigmata, including active bleeding, visible vessels, or adherent clots. There were 1300 gastric ulcers, 1808 duodenal ulcers, and 112 anastomotic ulcers. Of the 3220 patients, 2950 were treated with endoscopic therapy (91.6%). One thousand one hundred ninety-two patients had a history of peptic ulcers (37.0%), 336 patients (10.4%) were aspirin users, and 1107 were NSAID users (34.4%). Primary endoscopic therapy failed to achieve successful hemostasis in 19 patients (0.59%), and immediate surgical hemostasis was successful in 18 of them. Rebleeding occurred in 284 patients (8.8%). Emergency surgery was performed for 47 of these patients with rebleeding, whereas the others were managed by a second endoscopic therapy. Of those patients with endoscopic stigmata including acute bleeding, visible vessels, or adherent clots, 229 died (in-hospital mortality of 7.1%). Univariate analysis showed that a significantly higher proportion of nonsurvivors were elderly patients older than 70 years of age (P.01; odds ratio, 2.07; 95% confidence interval, ). The nonsurvivor group had a larger proportion of having the listed comorbidities, a higher number of comorbidities, and a significantly higher incidence of ischemic heart disease (Table 1). A significantly higher number of patients in the nonsurvivor group presented with hematemesis, shock, and a hemoglobin level less than 8 g/dl. There was a significantly higher risk of mortality for patients with in-hospital bleeding, those who developed rebleeding, and those treated with surgery and endoscopic therapy by using a single modality. Patients with H pylori infection had a lower risk of mortality (P.01; odds ratio, 0.15; 95% confidence interval, ). Multiple stepwise logistic regression analysis showed that age 70, the presence of listed comorbidities, more than 1 comorbidity, hematemesis on presentation, initial systolic blood pressure 100 mm Hg, in-hospital ulcer bleeding, rebleeding, and need for surgery were significant predictors of mortality (Table 1). H pylori related ulcers were associated with a lower risk of mortality. The ROC curve comparing the prediction of mortality with the actual mortality shows an AUC of (Figure 1). Pre-endoscopic prediction score for mortality with exclusion of predictive factors including H pylori status, rebleeding, and need of surgery showed an AUC of for the derivation cohort (Figure 2).

3 March 2009 PREDICTION MODEL FOR MORTALITY AMONG PATIENTS WITH BLEEDING PEPTIC ULCERS 313 Table 1. Comparison of Clinical and Endoscopic Parameters Between Survivors and Nonsurvivors for the Derivation Cohort Univariate analysis Multivariate analysis Variables Survivors (n 2991) Nonsurvivors (n 229) P value Odds ratio (95% confidence interval) P value Odds ratio (95% confidence interval) Age 70 y 1213 (40.5%) 134 (58.5%).01 a 2.07 ( ).01 a 1.47 ( ) Male 2046 (68.4%) 157 (68.6%).962 Comorbidities 1628 (54.4%) 202 (88.2%).001 a 6.26 ( ).001 a 4.12 ( ) Ischemic heart disease 358 (11.9%) 49 (21.4%).001 a 2.00 ( ) More than 1 comorbidity 814 (27.2%) 130 (56.8%).001 a 3.51 ( ).014 a 1.70 ( ) NSAID user 1029 (34.4%) 78 (34.1%) ( ) Aspirin user 309 (10.3%) 27 (11.8%) ( ) H pylori positive 1187 (39.7%) 20 (8.7%).001 a 0.15 ( ).001 a 0.20 ( ) History of peptic ulcer 1145 (38.3%) 47 (20.5%).001 a 0.42 ( ) Hematemesis 435 (14.5%) 55 (24.0%).001 a 1.86 ( ).039 a 1.45 ( ) Systolic blood pressure 100 mm Hg 554 (18.5%) 90 (39.3%).001 a 2.85 ( ).001 a 2.24 ( ) Hemoglobin 8 g/dl 949 (31.7%) 100 (43.7%).001 a 1.67 ( ) Size of ulcer 2 cm 553 (18.5%) 56 (24.5%).026 a 1.43 ( ) In-hospital bleeder 327 (10.9%) 90 (39.3%).001 a 5.28 ( ).001 a 2.46 ( ) Gastric ulcer 1192 (39.9%) 108 (47.2%).030 a 1.35 ( ) Posteroinferior duodenal ulcer 90 (3.0%) 5 (2.2%) ( ) Lesser curve gastric ulcer 46 (1.5%) 5 (2.2%) ( ) Active spurter on primary 224 (7.5%) 19 (8.3%) ( ) esophagogastroduodenoscopy Endoscopic therapy (single) 646 (90.1%) 71 (9.9%).001 a 1.63 ( ).072 Endoscopic therapy (combination) 2345 (93.7%) 158 (6.3%) Rebleeding 237 (7.9%) 47 (20.5%).001 a 3.00 ( ).016 a 1.63 ( ) Surgery 118 (3.9%) 42 (18.3%).001 a 5.47 ( ).001 a 4.60 ( ) a Factors with P value.05. A predictive scoring system was then derived on the basis of the predictors identified through multiple stepwise logistic regression (Table 2). A weight was assigned to each predictive factor according to the odds ratio to balance the influence of each factor. Validation of the prediction model for mortality was conducted in another prospectively collected cohort. From January 2004 to October 2006, 4406 patients presented to the Endoscopy Center with acute upper gastrointestinal bleeding. Figure 1. ROC curve on the prediction of in-hospital mortality by the model combining various predictive factors for the derivation cohort (age 70, the presence of comorbidities, number of comorbidities, hematemesis on presentation, initial systolic blood pressure 100 mm Hg, in-hospital ulcer bleeding, non H pylori related ulcers, rebleeding, and need for surgery) identified in the derivation cohort by using stepwise logistic regression. The AUC is Figure 2. ROC curve on the prediction of in-hospital mortality with the combination of pre-endoscopic predictive factors only for the derivation cohort (including age 70, presence of comorbidities, number of comorbidities, hematemesis on presentation, initial systolic blood pressure 100 mm Hg, in-hospital ulcer bleeding). The AUC is

4 314 CHIU ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 3 Table 2. Predictors of Mortality for Patients With High-Risk Bleeding Peptic Ulcers and the Associated Score Component Values From the Derivation Cohort Predictors of mortality Odds ratio (95% confidence interval) Score risk factor value Clinical factors (pre-endoscopy factors) Age 70 y 1.47 ( ) 1 Presence of listed comorbidities 4.12 ( ) 2 More than 1 listed comorbidity 1.70 ( ) 1 Hematemesis 1.45 ( ) 1 Initial systolic blood pressure 2.24 ( ) mm Hg In-hospital bleeders 2.46 ( ) 1 Outcomes Presence of H pylori 0.20 ( ) 1 Development of rebleeding 1.63 ( ) 1 Need for operation 4.60 ( ) 2 NOTE. The CU prediction score is calculated with the addition of all the score component values 1. Of these patients, 634 had bleeding peptic ulcers and required treatment with endoscopic therapy. Of these patients, 66 (10.4%) died within 30 days after hospitalization. The previously identified predictive factors were used to construct a mortality prediction score, and the prediction score was calculated for each patient. The correlation of the prediction score to the mortality for the evaluation cohort is shown in Figure 3. All the patients in the evaluation cohort survived when the score was less than 3. The risk of mortality increased with each increment of the prediction composite score. The ROC curve Figure 4. ROC curve on the prediction of mortality comparing the Chinese University ulcer bleed score with the Rockall score for the evaluation cohort of 634 patients with bleeding peptic ulcers. The AUC is for the CU prediction score (black line) and for the Rockall score (gray line). correlating this predictive score to actual mortality shows an AUC of (Figure 4). The Rockall score was also calculated for each of the patients, and the ROC curve correlating the Rockall score to mortality shows an AUC of (Figure 4). Figure 3. Distribution of cases of in-hospital mortality in relation to the scores attained from the evaluation cohort of 634 patients with bleeding peptic ulcers. Figure 5. ROC curve on the prediction of mortality in the evaluation cohort with pre-endoscopic predictive factors only. The AUC is

5 March 2009 PREDICTION MODEL FOR MORTALITY AMONG PATIENTS WITH BLEEDING PEPTIC ULCERS 315 The pre-endoscopic prediction score for mortality in the validation cohort showed an AUC of (Figure 5). Discussion Despite advances in the management of bleeding peptic ulcers during the past decade, the mortality rate remains noteworthy. 1,4,9,11,12 The reported rate of in-hospital mortality for bleeding peptic ulcers ranges from 10% to 15%. The prediction of mortality from bleeding peptic ulcers helps to identify at-risk patients, for whom intensive monitoring and aggressive management strategies might prevent this catastrophic event. Baradarian et al 13 reported that early intensive resuscitation reduced length of stay in intensive care unit, surgical intervention, and mortality from upper gastrointestinal bleeding. A number of scoring systems have been developed to predict the outcomes for patients with upper gastrointestinal hemorrhage Rockall et al 14 focused on the prediction of mortality among those with acute upper gastrointestinal hemorrhage, whereas the predictive score of Blatchford et al 17 identifies patients who are in need of intervention. However, few of these predictive scores target the prediction of mortality for bleeding peptic ulcers alone. 18,19 The present study focused on the development of a scoring system to predict ulcer mortality by using a large-scale database. The validity of the system was confirmed in another, separate group of patients. It was found that the elderly often had poor outcomes when they presented with bleeding ulcers. 20 The severity of bleeding was represented by the development of hypotension. In-hospital bleeders are also known to have poorer outcomes, because they usually have significant disease that warrants in-hospital management. The additional bleeding further complicates the management of such disease. 21 Endoscopic stigmata of recent hemorrhage correlated with peptic ulcer rebleeding and mortality in numerous predictive models. 14,15,22 Although high-risk stigmata (Forrest I, IIa, and IIb) have been shown to predict mortality, our study showed that there is no difference in mortality between those with active bleeding and those with visible vessels or adherent clots. It is noteworthy that most predictive models have been developed on the basis of a whole group of patients with acute gastrointestinal hemorrhage, whereas our group was highly selected to include only patients with high-risk Forrest classification Ia, Ib, IIa, or IIb ulcers. Calvet et al 23 showed that addition of a second endoscopic treatment after epinephrine injection improves outcomes in high-risk bleeding ulcers. Combined endoscopic therapy, however, did not incur a survival advantage in their meta-analysis. Although univariate analysis showed that our patients receiving single endoscopic therapy had higher mortality, type of endoscopic therapy was not a significant predictor of mortality in subsequent multivariate analysis. Combined endoscopic therapy reduces the rate of rebleeding and need for surgery. Although the majority of our patients (77.7%) received combination therapy, we did not find a significant effect on mortality. The first evaluation cohort included 3220 patients with bleeding peptic ulcer on endoscopy, and 2950 required endoscopic therapy. The second deprivation cohort included 801 patients with bleeding peptic ulcer, and 634 required endoscopic therapy. Hence, the requirement for therapeutic endoscopy reduced from 91.6% in the first cohort to 79.1% (634 of 801) in the second. This might be partly due to the conduction of a prospective randomized study on the use of intravenous omeprazole before endoscopy, which showed a lower requirement of therapeutic endoscopy (19.1% in omeprazole group versus 28.4% in placebo group). 10 Peptic ulcer rebleeding and the need for surgery remained significant predictors of mortality. Peptic ulcer rebleeding has been reported to be associated with a 6-fold to 12-fold increase in risk of mortality. The adjunctive use of intravenous highdose proton pump inhibitors has been confirmed to significantly reduce peptic ulcer rebleeding. 7 In addition, meta-analysis confirmed that the use of intravenous proton pump inhibitors reduced all-cause mortality in Asian randomized trials. 24 We previously reported that hypotension, hemoglobin less than 10 g/dl, fresh blood in stomach, active bleeding, and large ulcers were predictors of ulcer rebleeding. 19 The only overlapping factors between prediction of rebleeding and mortality were the presentation of hypotension and clinical evidence of ongoing bleeding. Mortality from bleeding peptic ulcers might not be accountable solely by risk factors of ulcer rebleeding. Because old age, comorbidities, and in-hospital bleeding are predictors of mortality, a catastrophic outcome might actually be the result of complications or worsening comorbid conditions rather than continuous bleeding. Surgery is now reserved as salvage treatment for ulcer rebleeding. 6 Perioperative mortality, however, amounts to 20%. 25,26 In the era of advanced endoscopic treatment and adjunctive proton pump inhibitor therapy, peptic ulcer rebleeding occurs in patients who are poor surgical candidates because of advanced age and multiple comorbidities. These factors significantly increased the risk of mortality in patients who receive surgical hemostasis. H pylori related ulcers are associated with better survival. There is a significantly lower risk of rebleeding in patients with H pylori related ulcers. Labenz et al 27 showed that H pylori infection is associated with a significantly higher intragastric ph after omeprazole treatment compared with that after H pylori eradication. Lin et al 28 demonstrated that intragastric ph was also higher in H pylori related bleeding ulcers after adjunctive omeprazole infusion. Moreover, patients with H pylori related ulcers were generally younger. These factors could improve the clinical outcomes of those with H pylori related bleeding ulcers treated with adjunctive proton pump inhibitors. Because the initial model can only be applied to prediction of mortality after endoscopy, we attempted to calculate a preendoscopic scoring excluding the H pylori status, ulcer rebleeding, and need for surgery. The prediction of peptic ulcer mortality by using the pre-endoscopic factors resulted in an AUC of 0.686, and the pre-endoscopic scoring would also be useful to identify at-risk patients. Our predictive model achieved an acceptable performance in predicting mortality from bleeding peptic ulcers as compared with the Rockall score for local Chinese population. This model, however, might not be able to achieve similar performance when applied to other ethnic populations. Our study is limited by the fact that several randomized trials were conducted during the period for the development of the prediction model and the validation period There might be a bias in treatment because half of the patients who were recruited in one of the randomized studies received an inferior treatment. Three thousand two hundred twenty patients were recruited in our study, and 23.9% of these patients participated in one of these trials. Hence, only 11.9% of all the

6 316 CHIU ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 3 patients received a potentially inferior treatment regime. This might incur a limited effect to the prediction model. Validation studies of the applicability of the current prediction model will be needed in different parts of the world because this model is based on a local Chinese group of patients with bleeding peptic ulcers. There is also a possibility of overfitting the regression model because we have 22 variables in univariate and multivariate analysis. In conclusion, advanced age, presence of listed comorbidities, multiple comorbidities, hypovolemic shock, in-hospital bleeding, rebleeding, and need for surgical intervention successfully predicted in-hospital mortality of patients with bleeding peptic ulcers after endoscopic hemostasis. References 1. Rockall TA, Logan RF, Devlin HB, et al. Incidence of and mortality from acute upper gastrointestinal hemorrhage in the United Kingdom. Br Med J 1995;311: Sacks HS, Chalmers TC, Blum AL, et al. Endoscopic hemostasis: an effective therapy for bleeding peptic ulcers. JAMA 1990;264: Cook DJ, Guyatt GH, Salena BJ, et al. Endoscopic therapy for acute non-variceal upper gastrointestinal hemorrhage: a metaanalysis. Gastroenterology 1992;102: Ohmann C, Imhof M, Ruppert C, et al. Time-trends in the epidemiology of peptic ulcer bleeding. Scand J Gastroenterol 2005;40: van Leerdam ME, Vreeburg EM, Rauws EA, et al. Acute upper GI bleeding: did anything change? time trend analysis of incidence and outcome of acute upper GI bleeding between 1993/1994 and Am J Gastroenterol 2003;98: Chung SC, Lau JY, Sung JJ, et al. Randomized comparison between adrenaline injection alone and adrenaline injection plus heat probe treatment for actively bleeding ulcers. Br Med J 1997; 314: Lau JYW, Sung JJY, Lam YH, et al. Endoscopic retreatment compared with surgery in patients with recurrent bleeding after initial endoscopic control of bleeding ulcers. N Engl J Med 1999;340: Lau JYW, Sung JJY, Lee KKC, et al. Effect of intravenous omeprazole on recurrent bleeding after endoscopic treatment of bleeding peptic ulcers. N Engl J Med 2000;343: Sung JJ, Chan FK, Lau JY, et al. The effect of endoscopic therapy in patients receiving omeprazole for bleeding ulcers with nonbleeding visible vessels or adherent clots: a randomized comparison. Ann Intern Med 2003;139: Lau JY, Leung WK, Wu JC, et al. Omeprazole before endoscopy in patients with gastrointestinal bleeding. N Engl J Med 2007;356: Blatchford O, Davidson LA, Murray WR, et al. Acute upper gastrointestinal hemorrhage in west of Scotland: case ascertainment study. Br Med J 1997;315: Longstreth GF. Epidemiology of hospitalization for acute upper gastrointestinal hemorrhage: a population-based study. Am J Gastroenterol 1995;90: Baradarian R, Ramdhaney S, Chapalamadugu R. Early intensive resuscitation of patients with upper gastrointestinal bleeding decreases mortality. Am J Gastroenterol 2004;99: Rockall TA, Logan RFA, Devlin HB, et al. Risk assessment after acute upper gastrointestinal hemorrhage. Gut 1996;39: Saeed ZA, Winchester CB, Michaletz PA, et al. A scoring system to predict rebleeding after therapeutic endoscopic therapy of non-variceal upper gastrointestinal hemorrhage with a comparison of heat probe and ethanol injection. Am J Gastroenterol 1993;88: Ananya D, Wong RCK. Prediction of outcome of acute GI hemorrhage: a review of risk scores and predictive models. Gastrointest Endosc 2004;60: Blatchford O, Murray WR, Blatchford M. A risk score to predict need for treatment for upper gastrointestinal hemorrhage. Lancet 2000;356: Chiu PW, Joeng HKM, Choi CLY, et al. Prediction of peptic ulcer rebleeding after scheduled second endoscopy: clinical or endoscopic factors? Endoscopy 2006;38: Wong SK, Yu LM, Lau JY, et al. Prediction of therapeutic failure after adrenaline injection plus heater probe treatment in patients with bleeding peptic ulcer. Gut 2002;50: Hasselgren G, Blomqvist A, Eriksson S, et al. Short and long term course of elderly patients with peptic ulcer bleeding: analysis of factors influencing fatal outcome. Eur J Surg 1998;164: Klebl FH, Bregenzer N, Schofer L, et al. Comparison of inpatient and outpatient upper gastrointestinal hemorrhage. Int J Colorectal Dis 2005;20: Wara P. Endoscopic prediction of major rebleeding: a prospective study of stigmata of hemorrhage in bleeding ulcer. Gastroenterology 1985;88: Calvet X, Vergara M, Brullet E, et al. Addition of a second endoscopic treatment following epinephrine injection improves outcome in high risk bleeding ulcers. Gastroenterology 2004;126: Leontiadis GI, Sharma VK, Howden CW. Systematic review and meta-analysis: enhanced efficacy of proton-pump inhibitor therapy for peptic ulcer bleeding in Asia: a post hoc analysis from the Cochrane Collaboration. Aliment Pharmacol Ther 2005;21: Towfigh S, Chandler C, Hines OJ, et al. Outcomes from peptic ulcer surgery have not benefited from advances in medical therapy. Am Surg 2002;68: Ripoll C, Banares R, Beceiro I, et al. Comparison of transcatheter arterial embolization and surgery for treatment of bleeding peptic ulcer after endoscopic treatment failure. J Vasc Interv Radiol 2004;15: Labenz J, Tillenburg B, Peitz U, et al. Helicobacter pylori augments the ph-increasing effect of omeprazole in patients with duodenal ulcer. Gastroenterology 1996;110: Lin HJ, Tseng GY, Hsieh YH, et al. Will Helicobacter pylori affect short-term rebleeding rate in peptic ulcer bleeding patients after successful endoscopic therapy? Am J Gastroenterol 1999;94: Reprint requests Address requests for reprints to: Philip W. Y. Chiu, Department of Surgery, Institute of Digestive Disease, Chinese University of Hong Kong, Hong Kong. philipchiu@surgery.cuhk.edu.hk; fax: (852) Conflicts of interest The authors disclose the following: James Lau received consulting fees and lecture fees from AstraZeneca. Francis Chan received consultation fees from Pfizer, lecture fees from Takeda, Pfizer, and AstraZeneca, and grant support from Pfizer. Joseph Sung received lecture fees from Astra- Zeneca and GlaxoSmithKline.