605 Independent Risk Factors for Severity and Mortality in Lower Gastrointestinal Bleeding and Proposal of New Prognosis Score

Abstract

INTRODUCTION: Risk stratification is key in the assessement and management of patients with lower gastrointestinal bleeding (LGIB). AIM: To identity risk factors for mortality and severity, namely: critical care requirement, prolonged hospitalization, rebleeding, blood transfusions, re-admissions or surgery in a cohort of patients with LGIB in a hospital in Peru. METHODS: Analytic prospective cohort study, performed between June 2016 and April 2018 in a tertiary care hospital. The sample size (80 Patients) was calculated considering an incidence of 0.3% with a confidence level of 95%. Data was collected during the hospitalization, including sex, age, comorbidities, medication, etiology, time to colonoscopy, vital signs, laboratory and the outcomes. The independent factors were determined using binomial logistic regression, these factors were used to developed a score. Thirty – day survival was calculated using a Kaplan-Meier curve. RESULTS: A total of 98 patients were included, of which 13 patients (13, 3%) died, and 56 (57, 1%) met severity criteria. Sixty-two patients (63, 3%) were male, the average age was 54.89 ± 19.37 years. The most frequent etiologies were hemorrhoids (20.4%), colorectal cancer (19.4%), and diverticulosis (13.3%). Thirteen patients died and the most frequent cause of death was pneumonia in twelve (92.3%). The independent risks factor for mortality was Glasgow scale under 15, and for severe bleeding were: Systolic blood pressure under 100 mm Hg and albumin lower than 2, 8 g/dL. With these factors the score SGA (S systolic, G Glasgow, and A albumin) was developed. An area under Receiving Operator Curve (AUC ROC) of 87.5% with a P < 0.000, the value of 2 or more predicted mortality with a sensitivity of 90.9% and a specificity of 69.7%, positive predictive value 33.7%, negative predictive value 97.9%. The value 2 or more predicted severity with a sensitivity of 59.6% and a specificity of 93.3%, positive predictive value 93.3%, negative predictive value 59.6%. The survival curve with the kaplan meier test is statistically significant, when comparing the value 0 or 1 versus the value 2 or 3. CONCLUSION: We found a high mortality rate and frequent severity parameters in our population with LGIB, the main risk factors being systolic blood pressure under 100 mm Hg, Glasgow score lower than 15, and albumin lower than 2.8 g/dL. Early identification of these associated factors would improve the management of LGB in the emergency room.