AGA Abstracts was carried out in two hospitals in the south of Spain. The databases of the two hospitals were reviewed over two periods of time (from 1995 to 2000 and from 2001 to 2014) and were compared. The following variables were tested: age, sex, smoking, localization of disease at diagnosis and family history of IBD and were compared between the two periods of time. Inclusion criteria were age more than 14 years old and a clinical diagnosis of IBD. Results: The population of both hospitals was 1.011.555 habitants and a total of 2519 patients were collected. In the first period, 430 patients were registered. The total incidence rate was 7,07 per 100,000 person-years. Ulcerative Colitis (UC) incidence was 3,57 per 100,000 person- years and Crohn Disease (CD) incidence was 3,50 per 100,000 person-years. UC was diagnosed in 50,4%(n=217). In the second period, 2089 patients were registered. The incidence of IBD was 14.7 per 100,000 person-years. The incidence of UC was 7.6 per 100,000 person-years and 7.1 cases per 100,000 person-years for CD patients. UC was diagnosed in 51.7% (n=1081) Epidemiological data are shown in Image1. Comparing both periods of time, the incidence has doubled with statistically significant differences: Image2. Regarding the epidemiological data, statistically significant differences between both periods was found in the following variables: age, family history of IBD and localization of UC at diagnosis (p <0.05). Conclusions The incidence of IBD in the south of Spain has dramatically increased in the last two decades and it is similar to the data reported in other European countries. Some differences have been found between the two periods (age at diagnosis, family history and localization). Due to the fact that it is a retrospective study, selection bias cannot be ruled out. Mo1832 Validation of Administrative Data for Capturing Crohn's Disease Patients Requiring Surgical Bowel Resection Christopher Ma, Remo Panaccione, Gordon W. Moran, Eric Benchimol, Cynthia H. Seow, Yvette Leung, Kerri L. Novak, Marietta Iacucci, Subrata Ghosh, Gilaad Kaplan Background: Administrative databases have been widely used to evaluate surgical outcomes in Crohn's disease (CD) patients but the validity of administrative data for defining the diagnosis of CD and CD-related bowel resections has not been adequately validated. Aims: To evaluate the accuracy of International Classification of Disease (ICD) coding in identifying patients who are admitted for CD and undergo bowel resection. Methods: Population-based surveillance was conducted in the Calgary Health Zone between January 1 and December 31, 2011 using the Discharge Abstract Database to identify adults ( ≥18 years) admitted for CD who underwent surgical resection using Canadian Classification of Health Intervention (CCI) coding. Surgical resection codes were stratified by site of resection, surgical approach, surgical urgency, and post-surgical anatomy (anastomosis, stoma, or pouch). The administra- tive data was validated against chart review and reported as a positive predictive value (PPV) with 95% confidence interval (CI). Results: The administrative database identified 104 admissions of CD requiring bowel resection and correctly identified the diagnosis of CD in 101/104 patients (97.1%). Administrative data was highly predictive for small bowel (PPV 0.86 [95% CI: 0.70-0.95]) and large bowel CD (PPV 1.00 [0.80-1.00]), but was less accurate for ileocolonic CD (PPV 0.67 [0.46-0.83]). Sensitivity for ileocolonic CD improved when K50.8 and K50.9 (CD, unspecified) codes are combined (0.85 [0.68-0.94]). 112 surgical resections were performed. The administrative data was accurate in identifying partial small (PPV 0.87 [0.75-0.94]) or large bowel resections (PPV 0.81 [0.64-0.91]), but less accurate for partial rectal excisions (PPV 0.57 [0.22-0.88]). It was also accurate for defining elective surgery (PPV 0.90 [0.79 - 0.96], and open (PPV 0.93 [0.84 - 0.97]) vs. laparoscopic (PPV 0.88 [0.72-0.96]) approach but was only moderately predictive of post-surgical anatomy (Table 1). Conclusions: In CD patients requiring bowel resection, administrative data accurately identifies large or small bowel CD, surgical urgency, location, and approach but is limited for defining ileocolonic CD and post-surgical anatomy. This may reflect the heterogeneous clinical phenotype and complex operations required in this cohort. S-788 AGA Abstracts Table 1 - Validation of Surgical Procedure Codes Mo1833 Clinical Characteristics of Crohn's Disease and Ulcerative Colitis Patients Who Switched From an anti-TNF-α Agent or Natalizumab to Vedolizumab or an Alternative anti-TNF-α Huifang Liang, Sudhakar Manne, Jesse Shick, Shawn Yu, Gregory Fusco, Paul Dolin Background: Vedolizumab (VEDO) has been available since June 2014 in the US for treatment of adults with ulcerative colitis (UC) or Crohn's disease (CD). This study presents a retrospective comparative analysis of VEDO use in its first year, in a large US health insurance claims database. Aim: To describe clinical characteristics of CD or UC patients who switched from anti-TNF-a (TNF) or natalizumab to VEDO or to an alternative TNF. Patients and methods: A non-interventional, retrospective analysis was done using the Optum insurance claims database covering the period 1 May 2000 through 30 June 2015. Patients meeting all of the following criteria were included in the analysis: patients had a diagnosis of UC or CD; patients had both medical and pharmacy insurance coverage; patients switched to VEDO or a TNF from a different TNF agent or natalizumab on or after 01 June 2014; and patients had continuous insurance enrolment for 6 months before that switch in biologic therapy. Descriptive statistics were used to describe and compare patients switching to VEDO or to a TNF. Results: 246 patients switched biologic therapy to VEDO and 921 switched to another TNF were identified and included in the study. The 2 groups of patients were similar with regards to mean age (VEDO 43 ± 15 years vs TNF 44 ± 15 years) and gender (VEDO 55% vs TNFa 57% female) and duration of IBD (5 years). Those switching to VEDO were more likely to have UC (VEDO 35%, vs TNF 24%), to be their first switch in biologic therapy (VEDO 59% vs TNF 47%) and had 8 months shorter time since first biologic therapy was started. All these differences were statistically significant (p<.05). In the 60 days immediately prior to the switch, the medication use of VEDO treated group included corticosteroids (26% of patients), immunosuppressants (24%) (AZA 15%), 5-ASAs (29%), and antibiotics (16%). Among those switching to another TNF, medication in 60 days prior to switch included corticosteroids (20%), 5-ASAs (17%), immunosuppressants (21%) (AZA 14%), and antibiotics (12%). Analysis of medical claims prior to switch showed those switching to VEDO were more likely to have a history of GI infections (VEDO 46% vs TNF 33%, p<.05), this was seen for both CD and UC patients. There were no difference between groups in upper respiratory tract infections (VEDO 38% vs TNF 38%), opportunistic infections (12% vs 11%), or lower respiratory tract infections (3% vs 3%). Conclusion: Several differences in clinical characteristics (such as history of GI infections) were found between patients on biologic therapy switching to VEDO as compared to those switching to an alternative TNF. This is likely to reflect VEDO being a new treatment option for patients with unmet medical needs. Mo1834 Clinical Characteristics of Crohn's Disease or Ulcerative Colitis Patients Who Initiated Vedolizumab or an anti-TNF-α As First Biologic Therapy Huifang Liang, Sudhakar Manne, Jesse Shick, Shawn Yu, Gregory Fusco, Paul Dolin Background: Vedolizumab (VEDO) has been available since June 2014 in the US for treatment of adults with ulcerative colitis (UC) or Crohn's disease (CD). This study presents a retrospective analysis of vedolizumab use in its first year, in a large US health insurance claims database. Aim: To describe clinical characteristics of CD or UC patients initiating vedolizumab or anti-TNF-a (TNF) without prior biologic use. Methods: A non-interventional, retrospective analysis was done using the Optum insurance claims database covering the period 1 May 2000 through 30 June 2015. Patients meeting all of the following criteria were included in the analysis: patients had a diagnosis of UC or CD; patients had both medical and pharmacy insurance coverage; patients initiated their first-ever biologic (VEDO or TNF) from 01 June 2014 to 30 June 2015; and patients had continuous insurance enrolment for 6 months before initiation of that biologic agent. Descriptive statistics were used to describe and compare VEDO and TNF initiators. Results: During the study period