14-17 The study was approved by the National Health Research Institutes, Taiwan. Using International Statistical Classification

of Diseases and Related Health Problems, 9th edition (ICD-9) codes to define the presence of diseases (Supporting Table 1), we first identified all hospitalized patients who were admitted with a primary diagnosis of PUB (ICD-9 codes 531.0, 531.2, 531.4, 531.6, 532.0, 532.2, 532.4, 532.6, 533.0, 533.2, 533.4, and 533.6) for the first time between January 1, 1997, and December 31, 2006. Those who were admitted again Paclitaxel cell line or transferred to another hospital within 3 days of discharge from the index hospitalization were considered in the same bleeding episode. Patients <20 years of age and those receiving gastric resection or vagotomy prior to discharge were excluded. The definition of liver cirrhosis required both the specific admission code (ICD-9 code 571) and certification in the Registry for Catastrophic Raf inhibitor Illness Patient Database, a subsection of the NHIRD.15, 16 Patients with cirrhosis had to have hepatic encephalopathy, gastroesophageal varices, or ascites

to certify the cirrhosis-related catastrophic illness. Presumably, these stringent criteria affirmed the diagnosis of cirrhosis but enrolled patients at advanced stages. We matched each patient with cirrhosis with four controls selected from the same PUB population according to age (±2 years), sex, and the frequency of taking antisecretory drugs (defined as proton pump inhibitor or histamine type 2 receptor antagonist). Recurrent PUB was defined as rehospitalization with a primary diagnosis of PUB after the index bleeding episode during the study period. The length of follow-up was calculated according Etofibrate to the calendar dates between discharge from the index hospitalization and readmission for the recurrent bleeding, occurrence of death, or the end of study period (December 31, 2006). Because this research enrolled open cohorts,

study subjects entered at various time points with various periods of follow-up. We defined presence of comorbidities according to the diagnoses coded on admissions prior to the index hospitalization. We considered acute coronary syndrome, cerebral infarction, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, dyslipidemia, and end-stage renal disease as potentially important confounders. We adjusted the factor of ulcerogenic drugs, which included aspirin, nonsteroidal anti-inflammatory drugs, and other antiplatelets and anticoagulants (Table 1). We also took into account the influence of propranolol, a nonselective beta-blocker frequently prescribed in patients with cirrhosis to manage portal hypertension.18 Enrolled patients who received eradication therapy for Helicobacter pylori before or after the index hospitalization were defined as having H. pylori–associated peptic ulcers.14, 15 The definition of H.