Impact of Fresh Frozen Plasma Transfusion on Upper Gastrointestinal Bleeding in Patients With Cirrhosis

Abstract

Introduction: Patients with cirrhosis and upper gastrointestinal bleeding (UGIB) commonly have an elevated INR, resulting in transfusion of fresh frozen plasma (FFP) in an effort to reduce active bleeding. Our aim is to determine the impact of FFP transfusions on presence of active bleeding or evidence of recent bleeding at endoscopy in cirrhotic patients with UGIB. Methods: A retrospective chart review was used to identify patients with cirrhosis admitted with UGIB at a university and county hospital from 2010 to 2015. Patients were excluded if no EGD was performed, a non-UGI source or no bleeding source was identified, or warfarin was taken prior to admission. The outcome of interest was presence of active bleeding or evidence of bleeding at endoscopy. Patients were divided into three groups: no FFP transfusions prior to endoscopy, 1-2 units transfused, and 3+ units transfused. To control for covariates associated with either FFP transfusion or increased likelihood of bleeding at endoscopy, a multinomial propensity score was developed for each group prior to logistic regression analysis. An independent multivariate logistic regression without propensity scores was used to evaluate FFP transfusion and outcomes in patients with variceal bleeding. Results: 169 patients were included in the study. The following covariates were included in the multinomial propensity score: demographics, admission vitals, comorbidities, antiplatelet agents or anticoagulants prior to admission, labs at admission and time of endoscopy, and administration of proton pump inhibitor or octreotide within 12 hours of presentation. 111 patients received no FFP, 34 received 1-2 units of FFP, and 24 received 3+ units of FFP. Compared to those who received no FFP, patients who received 1-2 units of FFP (OR 0.02, P=0.3) or 3+ units of FFP (0.01, P=0.16) were no more or less likely to have evidence of active or recent bleeding at endoscopy. In the classic multivariate logistic regression model in patients with variceal bleeding, each additional unit of FFP transfused did not have an impact on active or recent bleeding at endoscopy (OR 0.99, P=0.35). Conclusion: FFP is often given to patients with cirrhosis who present with an UGIB. However, in this study, FFP transfusions were not associated with the presence of active or recent bleeding at endoscopy. These data may be useful to decrease unnecessary FFP transfusions to minimize patient risk and improve resource utilization.