AGA Abstracts Table 2: Predictors of GIB Mo1257 RISK OF GASTROINTESTINAL BLEEDING HOSPITALIZATION IN ESRD PATIENTS WITH NEW DIAGNOSIS OF ATRIAL FIBRILLATION Aref Qureini, Laith Numan, Khalil M. Abuamr Introduction: Atrial fibrillation (AF) increases the risk for thromboembolic disease especially strokes, for that reason high-risk AF patients should be on anticoagulation. Warfarin is one of the most commonly used medication with atrial fibrillation. There is current clinical equipoise regarding the benefits and harms of using warfarin for anticoagulation for patients with atrial fibrillation and End-Stage Renal Disease (ESRD). Methods: This study utilized the HCUP database to do a retrospective study overlooking hospital encounters in the state of New York. We used ICD-9 codes to find our patient population and apply exclusions. The cohort time frame was between 2007-2014. All patients have a diagnosis of CKD, are above 18 years of age. Our index event was the diagnosis of atrial fibrillation and a CHADS- VASC score above 2. We analyzed the incidence of major gastrointestinal (GI) bleeding 90 days post index event. We compared outcomes of GI bleed incidence between CKD (stages 1-4) vs ESRD (stage 5). The analysis was done using Kaplan Meier Survival & Hazard Function. We also calculated the overall incidence of GI bleeding in both patient populations as well. Results: 83,354 patients were included. 69,175 had CKD 1-4 & 14,179 had ESRD. The incidence of GI Bleed within the first 90 days of hospitalization after a primary diagnosis of atrial fibrillation among CKD stage 1 – 4 patients was approximately 158 per 10,000 and among ESRD patients approximately ~240 per 10,000 (risk difference of 82 per 10,000). Kaplan Meier (KM) curve comparison of two groups revealed different KM curve that was statistically significant (P<0.0001). Hazard rate of GI bleed hospitalization peaks at 10 days and lessens after 20 days. Finally, we found that the ESRD group has a consistently higher rate of readmission for GI bleeding. Discussion: Our study results show that ESRD patients on warfarin for a new AF, have an increased risk of GI bleeding compared with CKD patients from a hospital admission database. Results: revealed that for every 122 patients with a new diagnosis of AF and ESRD, CHADSVASC score ≥2, one patient will have a GI bleed hospital admission within the first 90 days post discharge. In addition to that, ESRD patients had a consistently higher rate of readmissions due to GI bleeding. For those reasons, it is important to evaluate the risks and benefits for ESRD patients with new AF before starting anticoagulation. S-736 AGA Abstracts Kaplan Meier (KM) curve comparison of two groups revealed different KM curve that was statistically significant (P<0.0001) Hazard rate of GI bleed hospitalization peaks at 10 days and lessens after 20 days, and levels down till 80 days. ESRD group has a consistently higher rate of readmission for GI bleeding. Mo1258 ONLINE SYMPTOM CHECKERS FOR GASTROINTESTINAL BLEEDING: A REAL-LIFE EMERGENCY ROOM EXPERIENCE PITTING DOCTOR VERSUS TECHNOLOGY Andrew C. Berry, Nicholas A. Berry, Madhuri S. Mulekar, Bin Wang, Kyle Yuquimpo, Ann S. Monardo, Gilad Shapira, Ashlynn Swaney, Vicki Sharma, Ramez Kouzy, Hayden Hamby, Vikash Pernenkil, Fareed Rifai, Heidi E. Henson, Jamie L. Skrove, Javier Sobrado, Bruce B. Berry In today's tech-savvy world, online symptom checkers provide a means for patients to conveniently self-diagnose and triage their healthcare needs. We aimed to analyze the diagnostic accuracy for online symptom checkers in regards to GI bleeding and, most importantly, compare its diagnostic accuracy for GI bleeding to other GI diagnoses and non- GI diagnoses. IRB approval was obtained for an ER chart analysis from an academic 300- bed hospital over a 2-year period for a random 10-day block each month on 1617 patients. Excluded patients were those who came as a direct transfer, sent from clinic, unable to consciously make a decision, carrying a non-specific ER physician-determined diagnosis, or those who left without being seen. Five online symptom checkers were utilized for diagnosis, three with triage capabilities. Patients presented to the ER and subsequent physician-deter- mined diagnosis and patient symptoms were recorded, along with demographics. ER physi- cian-determined diagnosis was classified into 1 of 4 groups: GI-Bleeding Oral, GI-Bleeding Rectal, Other GI Dx, and Non-GI Dx. Retrospectively, study personal entered all unique patient symptoms into each online symptom checker. Whether the ER physician-determined diagnosis was listed by each symptom checker Top1, Top3, or Top10 choices was recorded. All patients presented to the ER, so the proportion of symptom checker diagnoses with triage advice not suggesting an ER emergent visit was also recorded (to not justify an ER visit) A total of 591 patients were enrolled (Figure 1A). Overall triage accuracy is listed in Figure 1B. No significant differences were noted in regards to diagnostic accuracy superiority of symptom checkers for GI bleeding orally versus GI bleeding rectally, for either of the five symptom checkers or whether the diagnosis was listed Top1 or Top3 (Figure 2A). Overall, ability of symptom checkers to accurately diagnose GI bleeding (orally or rectally) remained quite poor and inconsistent. Regardless of symptom checker, the correct diagnosis