LAUSR

The study by Drolz et al. is important given the need to reduce bleeding complications and mortality in decompensated cirrhosis. The novel finding of this study was that fibrinogen is independently predictive of new onset major bleeding in critically ill patients with cirrhosis. It is increasingly recognized that despite coagulopathy of liver disease, thrombin generation is considered “rebalanced.” The role of fibrinogen, however, is relatively poorly studied in cirrhosis despite being quantitatively and qualitatively reduced. Hypofibrinogenemia is a key determinant of bleeding severity in other types of hemorrhage. To further investigate the importance of fibrinogen in cirrhosis, we retrospectively analyzed data from 1313 hospitalized patients with cirrhosis from a nationwide United Kingdom study of blood use in cirrhosis. This study prospectively enrolled all unselected cirrhotic patients admitted to 85 hospitals in the United Kingdom over 4 weeks in 2013. Of these, 192 (15%) patients were admitted with gastrointestinal (GI) bleeding, and the 28-day mortality was 10%. Fibrinogen level data were available for 551 (42%) patients. We used logistic regression to (1) assess the association between fibrinogen and 28-day mortality; and (2) to determine whether the association between fibrinogen and mortality differed according to GI bleeding status. The model included 12 prespecified covariates, and we assumed a linear association between continuous covariates and mortality. We assessed whether the association between fibrinogen and mortality differed according to baseline bleeding status using an interaction test. Multiple imputation using chained equations with 50 imputed datasets was used to account for missing data. The imputation model included all baseline covariates and mortality. Continuous covariates were imputed using predictive mean matching. With the exception of sex and baseline bleeding status, all variables were predictive of 28-day mortality (Table 1). Fibrinogen was an independent predictor of mortality, with a 29% increase in the odds of mortality for every 1 g/L reduction in fibrinogen. There was no evidence that the association between fibrinogen and mortality differed according to baseline bleeding status (no bleeding: odds ratio, 0.68 [95% confidence interval, 0.52-0.90]; bleeding: odds ratio, 0.93 [95% confidence interval, 0.52-1.66]; P value for interaction 5 0.28). It is unclear whether hypofibrinogenemia reflects severity of underlying liver disease or is a modifiable risk factor after bleeding that could be treated with fibrinogen supplementation. We consider this a key area for an interventional study in decompensated cirrhosis with GI bleeding.