Maryam Alimirah, MD, Omar Sadiq, MD, Deepak Venkat, MD
Henry Ford Hospital, Detroit, MI
Introduction: Risk stratification scoring systems for upper gastrointestinal bleeds, including the Glasgow Blatchford and Rockall scores, don’t accurately predict outcomes in acute variceal bleeds (AVB). Similarly, the Child–Pugh score has poor prognostic value in AVB, likely due to its subjective parameters. The goal of this retrospective study was to identify factors that predict in-hospital mortality in patients presenting with AVB.
Methods: Data was obtained through the Medical Information Mart for Intensive Care III (MIMICIII), which is a large, freely available database comprised of patients admitted to Beth Israel Deaconess Medical Center. 236 patients were included in this study who were admitted to the ICU from 2001 to 2012 with an ICD code of cirrhosis and variceal bleed. After excluding 136 patients who did not undergo esophagogastroduodenoscopy (EGD) during admission, 100 patients remained and were analyzed. Factors such as transfusion requirements, number of EGDs, demographics, and laboratory values were collected using postgresql queries. Statistical analysis was done using R 3.5.1.
Results: 19 of the 100 patients died during admission (mortality rate of 19%). There were no significant differences between the deceased and survival groups with regards to age, sex, race and insurance type. There were significant differences between both groups regarding hemoglobin nadir (p=.0001), MELD-Na score (p< .0001), MELD score (p< .0001), INR (p< .0001) and bilirubin (p=.005). Other factors analyzed included number of EGDs during admission, length of admission and blood product requirements. Compared with the survival group, the deceased group received a larger number of red blood cell transfusions (p=.02) and fresh frozen plasma (p=.007). The number of cryoprecipitate and platelet transfusions between both groups was not statistically significant (p=0.14 and p=0.07) respectively. The number of EGDs in the deceased group was higher, compared with the survival group (p=0.008).
Discussion: We were able to identify several risk factors of mortality in variceal bleeders admitted to the intensive care unit. We found the MELD-Na score, INR as well as hemoglobin nadir to have the strongest association with mortality. Number of EGDs was also associated with increased mortality suggesting that in-hospital rebleeding is a strong predictor of outcome. The MELD score was as strongly associated with mortality as the MELD-Na score, contrary to previous studies of mortality in cirrhosis.
Citation: Maryam Alimirah, MD, Omar Sadiq, MD, Deepak Venkat, MD. P0395 - FACTORS ASSOCIATED WITH IN-HOSPITAL MORTALITY IN ACUTE VARICEAL BLEEDING. Program No. P0395. ACG 2019 Annual Scientific Meeting Abstracts. San Antonio, Texas: American College of Gastroenterology.