AGA Abstracts Su1928 OUTCOMES OF INFLAMMATORY BOWEL DISEASE IN PATIENTS WITH EOSINOPHIL-PREDOMINANT COLONIC INFLAMMATION Tarik Alhmoud, Anas Gremida, Diego Colom Steele, Meng-Jun Xiong, Audra Kerwin, Imaneh Fallahi, Wa'el Tuqan, Nina Nandy, Mahmoud Ismail, Barakat Aburajab Altamimi, David Martin Background: Inflammatory bowel disease (IBD) is characterized by acute intestinal mucosal inflammation with chronic inflammatory features. Along with the typical acute inflammation (neutrophil-predominant), IBD patients might have various degrees of intestinal eosinophilic inflammation. The effect of intestinal eosinophils on the outcomes of IBD remains unclear. Aim: To assess the association between intestinal eosinophilic inflammation (at the time of IBD diagnosis), and the risk of developing a future IBD flare. Methods: This is a retrospective study performed in a single university hospital. Newly diagnosed IBD patients with established IBD clinic follow-up were included. Archived intestinal mucosal biopsy specimens (obtained at the time of initial diagnosis) were re-examined by two expert pathologists. The number of eosinophils per high power field (HPF) was counted, and the mucosal inflammation was classified according to the eosinophilic inflammatory patterns. The primary outcome was the presence disease flare(s) during the follow-up period. Logistic regression analysis with correction for length of follow-up and demographic variables was performed. Results: We included 167 patients with newly diagnosed IBD and confirmed clinic follow-up. Patients were 54% males and mainly Whites (63%), with a mean age of 40 at the time of diagnosis. Most patients had ulcerative colitis (77%) and were followed for 4 years on average. 40% of patients had at least one disease flare (table 1). Most patients (85%) had lower endoscopy with biopsies from the left side of the colon at the time of diagnosis. The characteristics of eosinophilic-inflammation are described in table 2. Patients without eosinophil-predominant inflammation were more likely to get a flare [odds ratio (OR) 4, 95% confidence interval (CI): 1.1-14]. Patients with mixed inflammation were not at higher risk for a flare compared to patients with eosinophil-predominant inflammation; odds ratio 2.2, 95%CI: 0.7-7). Higher tissue eosinophil counts and the presence of eosinophilic cryptitis were not associated with flare(s) risk; OR 1.9, 95%CI: 0.5-6.4 and OR 0.78, 95%CI: 0.27-2.2, respectively. Conclusion: Eosinophil-predominant inflammation might be associated with lower risk of IBD flares in this pilot study. Future studies with a larger number of patients and longer follow-up are warranted; to assess IBD outcomes in patients with the eosinophil-predominant inflammation phenotype, and to evaluate clinical response to different treatments in this group of patients. S-636 AGA Abstracts Su1929 THE INCIDENCE OF DEEP REMISSION INCLUDING HISTOLOGIC NORMALIZATION IN LONGSTANDING UC Seth Shaffer, Charles N. Bernstein Background: The primary treatment goal in UC is to maintain clinical remission, and endoscopic mucosal healing. Up to 40% of those with mucosal healing have persistent histologic inflammation. Reduced histologic inflammation is associated with decreased risks of relapse, hospitalization, corticosteroid use, colectomy, and neoplasia. Some persons with UC may completely normalize their colon histology. It is unknown what patient characteris- tics, disease extent, or treatment characteristics are associated with histologic normalization and whether normalization can impact on clinical outcomes. Aims: To assess the rate of dysplasia in patients with UC who achieve endoscopic and histologic normalization on at least two consecutive colonoscopies compared to the rest of the cohort. Methods: A retrospec- tive chart review was undertaken, of a referral clinic's patients from 1994 to 2017. Patients included were those diagnosed with UC, with at least 1 colonoscopy undertaken in the referral clinic. Data extracted from the chart included: age, sex, date of diagnosis, date of last follow-up, endoscopic extent of disease, severity of disease, colon histology, medication history, all dysplasia surveillance colonoscopies, and presence/absence of dysplasia. Dedicated GI pathologists then reviewed biopsies to confirm normalization. Results: A total of 350 individuals' charts were reviewed, with 294 meeting the inclusion criteria. 108 patients had at least 1 normal colonoscopy with normal histology. 28 of these 108 had two consecutive normal colonoscopies and histology. Comparing those with 2 normal colonoscopies to all others (including with 1 normal colonoscopy) there was no statistical difference in the medications used, duration of medications or age at diagnosis, but disease duration was longer for those with 2 consecutive normal colonoscopies (20.5 years vs 15.7, p = 0.02). The mean number of colonoscopies was 5.4 for those with two consecutive normal colonosco- pies, vs 3.1 (p<0.001). No dysplasia occurred after 2 consecutive normal endoscopies compared to 18 (6.8%) cases of dysplasia in the comparison cohort. For those with at least one normal colonoscopy, 5 (6.3%) developed dysplasia at some point. Conclusions: Persons who develop deep remission manifested by 2 consecutive colonoscopies with normal endos- copy and normal histology have a lower risk of dysplasia (0 in this cohort). Having 1 colonoscopy with normal histology was not protective against developing dysplasia. In persons with 2 consecutive normal colonoscopies by endoscopy and histology dysplasia surveillance intervals should be lengthened. Su1930 MEDICARE INSURED ULCERATIVE COLITIS PATIENTS HAVE SIGNIFICANTLY HIGHER IN-HOSPITAL MORTALITY AS COMPARED TO COMMERCIALLY INSURED PATIENTS Artin Galoosian, Mona Rezapour, Benny Liu, Taft Bhuket, Robert J. Wong Background and Aims: Frequent hospitalizations among patients with ulcerative colitis (UC) are associated with an increase in morbidity and mortality. However, it is not clear if recurrent hospitalizations and its effect on mortality affects all UC patients equally. We aim to evaluate the disparities in UC-related hospitalization rates and in-hospital mortality over a seven- year period in the U.S. Methods: Using data from the 2007-2013 Nationwide Inpatient Sample, adults that were hospitalized with a primary diagnosis of UC were evaluated to determine hospitalization rates and in-hospital mortality stratified by age, sex, race/ethnicity, insurance status and other co-morbidities. Hospitalization rates were calculated by dividing the number of UC hospitalizations by the total number of hospitalizations and were compared across groups using incidence rate ratios (RR) and the z-statistic. Multivariate linear regression models evaluated for predictors of in-hospital mortality. Results: Among 49,871 UC hospitali- zations (53% female, mean age of 45.0 ± 21, 72.5% non-Hispanic White, 11.2% African American, 10.6% Hispanic, with 51.9% using commercial insurance, 23.2% Medicare, 11.9% Medicaid), overall hospitalization rates among UC patients was 0.92 per 1,000-persons. Compared to females, males with UC had a significantly higher hospitalization rate (RR: 1.24, 95% CI 1.23-1.24, p<0.001). Compared to non-Hispanic whites with UC, African