A National Survey of the Prevalence and Demographics of Gastrointestinal Bleeding in Cirrhotic Patients

Abstract

Introduction: Gastrointestinal (GI) bleeding in cirrhotic patients continues to carry a high risk for morbidity and mortality. Few studies have examined all sources of GI bleeding within this population. Given the lack of comprehensive data within the literature, we aimed to evaluate the prevalence and characteristics of GI bleeds within subsets of cirrhotic patients utilizing a national population-based dataset. Methods: Using ICD-9 codes, we analyzed patients ≥ 18 years with a discharge diagnosis of alcoholic, nonalcoholic, or biliary cirrhosis from 2000-2012. Within this population, we identified all subjects with a principal or secondary diagnosis code representing a GI bleed. We further categorized these codes into Upper (UGI), Lower (LGI), or Other GI bleeding. In addition to patient prevalence, we analyzed demographic factors, GI bleed location, liver decompensation, and overall mortality. Results: The estimated prevalence of GI bleeding within the total cirrhotic population in the US from 2000-2012 was 11.1%. Males, whites, and UGI bleeds held the highest prevalence within all cirrhotic subgroups. Biliary cirrhosis accounted for only 2.2% of the population with women representing 87% of all GI bleeds. Adjusting for age, gender and using white race as reference, Native Americans and Hispanics showed higher likelihood for UGI bleeds [OR:1.29 95%CI:1.25-1.32 vs OR:1.07 95%CI:1.06-1.08, p<0.001]. Blacks ultimately showed the strongest likelihood for LGI bleeds [OR:1.31 95%CI:1.26-1.35, p<0.001]. Females were also 24% more likely to have LGI bleeding than males [p <0.001]. Alcoholic cirrhosis was a stronger predictive factor than nonalcoholic cirrhosis for both UGI bleeds and overall mortality [OR:1.81 vs 1.10, OR:2.15 vs 1.52, p <0.001]. No positive association was revealed when comparing location of bleed and overall mortality. In contrast, signs of decompensated liver disease were strong predictors of mortality [OR 2.36 95%CI: 2.34-2.37, p <0.001]. Conclusion: While a higher prevalence of Whites, males, and UGI bleeds exist within the cirrhotic population, Hispanics and Native American patients may have been more likely to develop UGI bleeds. Furthermore, both Black and female demographics may show higher likelihood of LGI bleeding. Alcoholic cirrhosis showed the highest likelihood of mortality. Unlike the location of bleed, signs of decompensated liver disease showed a positive association with mortality. To expand on these conclusions, future studies are warranted.