AGA Abstracts clinical intervention before endoscopy. Rockall score (RS) is used to identify patients with UGIB who have adverse outcome. These risk score are useful for predict outcome and recommended to assesse as soon as possible. However, these risk score are not widely used in clinical practice in Japan because of complicated calculation. Recently reported AIMS65 is the simple risk scoring system that identified five factors associated with increased risk for in-hospital mortality, length of stay and cost of admission: serum albumin ,3.0 g/dL, an international normalized ratio .1.5, altered mental status, systolic blood pressure ≤90 mmHg and age .65. We retrospectively evaluated the usefulness of the AIMS65 for assessing the need for clinical intervention, re-bleeding and death in cases of UGIB. Patients and method: In 387 patients who underwent urgent endoscopic examination due to UGIB symptoms such as hematemesis or melena between January 2003 and December 2009 at Tottori University Hospital, 215 patients were satisfied by the evaluation of AIMS65. These patients were assessed by GBS, RS and AIMS65, and the score defined as high-risk: GBS of greater than 0, RS of greater than 2 and AIMS65 of greater than 1. The patients who needed the clinical intervention or died in-hospital within 30 days were defined as high-risk. Result: Of the 387 evaluated patients, 166(43%) were need for hemostasis, 29(7%) were re-bleeding or unsuccessful endoscopic hemostasis, and 31(8%) died, but only 3 patients died of bleeding. In our total 387 patients, 181(47%) were categorized as those with high-risk for clinical intervention or poor outcome. GBS identified 369(95%) of the 387 patients as those with high-risk, and 2 of 18 low-risk patients were performed hemostasis, and neither re-bleeding nor death in low-risk patient. RS identified 332(86%) of the 387 patients as high-risk, and only one of 55 low-risk patients was performed hemostasis, and neither re-bleeding nor death in low-risk patient. AIMS65 identified 111(52%) of the 215 patients as those with high-risk, and 43 were performed clinical intervention and 9 were re-bleeding in 104 low- risk patients. Low-risk score (0-1) could not estimate unnecessary for hemostasis, however, mortality rate is 3%(3/104) and the negative predict value is 97%. Furthermore, in 12 high- score (4-5) patients, 5(42%) patients died of comorbidities. Conclusion: AIMS65 is useful scoring system for stratifying patients with UGIB into high and low risk categories for mortality. High-score patients via AIMS65 especially should be treated carefully with or without clinical intervention. Su1921 Predictors of Fatal Duodenal Diverticular Bleeding Jasbir S. Makker, Sridhar Chilimuri Background: Duodenal diverticular bleeding is a rare cause of upper gastrointestinal bleeding. The true incidence of duodenal diverticular bleeding is unclear in literature but incidence as high as 22% has been reported in autopsy studies. Diagnosis of bleeding duodenal diverticula continues to be a challenge and as a result, significant delays often occur in its diagnosis leading to increased length of stay, morbidity and health care costs. Objective: To identify etiology, clinical features, diagnostic methods and therapeutic modalities for fatal duodenal diverticular bleeding. Methods: A comprehensive search of National Library of Medicine Pubmed database for reported cases of duodenal diverticular bleeding in patients 18 and older was done for the period 1981 to 2012. Cases were divided into two groups based on blood pressure at presentation. Hypotensive patients were categorized under the fatal duodenal diverticular bleeding group and normotensive patients in the non-fatal duodenal diverticular bleeding group. Chi-square test and t-test were used to analyze categorical and continuous variables respectively in the two groups. Results: Review of literature reported in English language yielded 65 duodenal diverticular bleeding cases (Table 1). A total of 29 patients with fatal duodenal diverticular bleeding predominantly male (p= 0.0005) and mean age 60 (SD±23) were identified. A higher proportion of diverticula identified in fatal duodenal bleeding group were solitary (93% vs 83%, p=0.04) and located in third and fourth duodenal segments (50% vs 26%, p=0.0008). Esophagogastroduodenoscopy (EGD) was often the first diagnostic modality used in both the groups. However, in patients with fatal duodenal diverticular bleeding, EGD failed to detect any lesion in 54% patients as compared to 37% patients in the non-fatal duodenal diverticular bleeding group (p=0.02). Using a side viewing endoscope or doing a repeat EGD similarly failed in 53% patients in fatal duodenal diverticular bleeding group as compared to 38% in non-fatal duodenal diverticular bleeding group (p= 0.04). Dieulafoy lesion and duodenal ulcers were less frequently identified (29% vs 57%, p= 0.0001) in fatal duodenal diverticular bleeding group. Treatments with endoscopic modalities were used uniformly in both groups but surgical treatments were used more often in fatal duodenal diverticular bleeding group (38% vs 19%, p=0.004). Conclusions: Duodenal diverticulum bleeding should be strongly considered as a cause in hypotensive patients with upper gastrointestinal bleeding of unknown origin. This data suggests that the endoscope should be passed beyond the second part of duodenum in hypotensive patients presenting with upper gastrointestinal bleeding. Table 1: Comparison between fatal and non-fatal duodenal diverticular bleed S-510 AGA Abstracts Su1922 Patterns of Antithrombotic Therapy Use Following Upper Gastrointestinal Bleeding: A Descriptive Study in UK Primary Care Maria E. Saez, Antonio Gonzalez-Perez, Saga Johansson, Péter Nagy, Luis A. Garcia Rodriguez Introduction. The importance of antithrombotics such as acetylsalicylic acid (ASA) in second- ary prevention of cardiovascular (CV) events is well-established, but side-effects such as upper gastrointestinal bleeding (UGIB) have been associated with this therapy. We assessed antithrombotic therapy after UGIB in routine clinical practice. Methods. The Health Improve- ment Network UK primary care database was used to identify patients aged 40-84 years with a UGIB diagnosis, from 2000-2007 (n=2036). Patients within this study cohort were followed up from start date (initial date of recorded UGIB) for 1 year. Current use of each drug was defined as having a recorded prescription that lasted until, or ended in the 7 days prior to, the start date. Chronic use was defined as current use with .1 year of continuous prescription history prior to start date. Represcribing rates were estimated at 30, 90, 180 and 365 days using the product-limit method for each drug, and reported as proportion of current users of each drug who received a new prescription during the designated time interval. Results. Represcription rates for current users appeared lower at each time point for ASA than for other antithrombotics (cumulative represcription rates at 365 days [95% confidence intervals]: ASA, 0.43 [0.39-0.47]; warfarin, 0.66 [0.57-0.75]; clopidogrel, 0.69 [0.57-0.81]; dipyridamole, 0.49 [0.31-0.71]) (Table 1). Represcription rates among a subset of chronic users of antithrombotics tended to be similar to those among overall current users (cumulative represcription rates at 365 days: ASA, 0.46 [0.41-0.51]; warfarin, 0.66 [0.55-0.78]; clopidogrel, 0.71 [0.55-0.85]; dipyridamole 0.56 [0.33-0.82]). When analyzed by age, cumulative ASA represcription rates 1 year after UGIB tended to be lower in older patients (0.58 [0.40-0.78], 0.44 [0.37-0.52] and 0.42 [0.37-0.47] in patients aged 40-54, 55-69 and 70-85 years, respectively).When analyzed by calendar year, ASA represcribing rates after UGIB appeared to increase during the study period from 0.37 (0.24-0.53) in 2000 to 0.48 (0.38-0.60) in 2007. Similar represcribing rates for ASA were observed between male and female patients. Conclusions. Less than half of current ASA users received a new prescription in the year following an episode of UGIB. Represcription rates appeared higher among users of antithrombotics such as clopidogrel and warfarin (although based on small numbers of patients) than ASA users. The trend for increased represcription in the younger age group may reflect reluctance to prescribe antithrombotics, particularly ASA, to older patients after UGIB despite an increased CV risk in these patients. It appears that antithrombo- tic use falls after UGIB. The determinants of represcribing, patterns of switching antithrombo- tics and strategies that could ensure antithrombotic adherence should be further analyzed. Table 1. Represcribing rates and 95% confidence intervals at the different time points of follow-up, and for chronic users at 365 days of follow-up.