The threshold for transfusion in actively bleeding patients has been debated throughout the years. In 2013, a landmark trial demonstrated the utility of a restrictive transfusion protocol in patients with suspected upper gastrointestinal bleeding. A restrictive protocol (transfusion for hemoglobin <7.0) as compared to a liberal protocol (transfusion for hemoglobin <9.0) resulted in a lower 45-day mortality rate (5 % versus 9 %, p = 0.02), lower risk of further bleeding, lower likelihood of requiring rescue therapies, and a lower rate of complications including pulmonary edema and transfusion reactions. The authors postulated that the increase in mean hepatic venous pressure gradient in the liberal group likely contributed to these findings [16]. Of note, only around a quarter of the patients in this study had variceal bleeding. Furthermore, patients with poorly defined “massive exsanguinating bleeding” were excluded from this study. In general, a goal hemoglobin level of 7.0 is reasonable in patients with variceal bleeding for these reasons, although in the setting of massive exsanguination, red blood cell transfusion should not be directed at a hemoglobin level but should be based on the clinical situation.

There have been no randomized, controlled trials evaluating the administration of crystalloid compared to colloid fluids in patients with acute variceal bleeding. However, in other patient populations with hemorrhagic shock, colloid fluids such as albumin or hetastarch have no advantage compared to crystalloids [17]. In the cirrhotic patient population with variceal bleeding, it is reasonable to transfuse red blood cells if the hemoglobin is less than 7.0 and crystalloid fluids when the hemoglobin is 7.0 or greater and the patient has signs of hypoperfusion.

Cirrhosis often results in decreased levels of all clotting factors, dysfibrinogenemia, and thrombocytopenia, all of which can contribute to ongoing bleeding. In addition, multiple transfusions can result in a dilutional coagulopathy. Frequent monitoring of prothrombin time, platelet count and fibrinogen levels is warranted.

Unfortunately, prothrombin time has not been shown to correlate well with clinical bleeding in cirrhotic patients [18, 19]. No randomized trials of Vitamin K supplementation in patient with variceal bleeding have been performed, and routine use is neither recommended nor discouraged [20]. Providers should aim to avoid over-transfusion solely to correct numerical lab values as this will likely result in higher portal venous pressures and possibly more bleeding. Recommendations for a threshold at which to transfuse fresh frozen plasma, cryoprecipitate and platelets vary and become even more complicated in the setting of cirrhosis. While trauma literature supports the use of a “massive transfusion protocol,” which entails transfusing packed red blood cells, fresh frozen plasma and platelets in a 1:1:1 ratio [2], this strategy has not been tested in a randomized trial of cirrhotic patients with variceal bleeding. It is unknown if this is the correct resuscitation approach in our patient population, and it should not be universally applied.

Recombinant factor VIIa, a product first available for patients with hemophilia-related bleeding, has been considered as an agent that could halt gastrointestinal bleeding in cirrhotic patients by activating the extrinsic pathway and increasing thrombin production. While recombinant activated factor VIIa has been shown to normalize prothrombin time in patients with cirrhosis and variceal bleeding [21], a Cochrane review that included approximately 500 patients with various stages of cirrhosis and active upper gastrointestinal bleeding from two trials did not find a mortality benefit when this factor was administered in addition to usual care. Furthermore, there was no significant difference in packed red blood cell transfusion requirement [22]. An exploratory analysis from one of the two trials suggested that the subgroup of patients with variceal bleeding, Child-Pugh scores B or C, and more severe coagulopathy may benefit from recombinant factor VIIa [23]. Therapy with recombinant factor VIIa remains controversial in this setting.

Case reports have described success in achieving hemostasis in cirrhotic patients with the administration of Prothrombin Complex Concentrate (PCC), a mixture of coagulation factors II, VII, IX and X. This agent cannot be recommended on the basis of current evidence [24].

Elective intubation has not been proven to improve outcomes or prevent aspiration events or pneumonia in patients undergoing endoscopy in the intensive care unit. A retrospective cohort study compared patients with suspected variceal hemorrhage who underwent elective pre-endoscopy intubation and those who did not. All patients had active bleeding or stigmata suggestive of high risk of bleeding on endoscopy. Patients who underwent elective intubation had higher mortality and higher rates of aspiration pneumonia. However, patients with a known aspiration event prior to presentation, respiratory distress, intoxication, or hepatic encephalopathy greater than stage 1 were excluded. Patients who were alert at the time of endoscopy protected their own airways better [25].

In another retrospective study, pre-endoscopy intubation did not change the likelihood of witnessed aspiration during EGD, new radiographic infiltrates post-EGD, hospital length of stay, or mortality [26], although it may have prevented massive aspiration events. To this date, no prospective randomized trial has been performed to truly demonstrate the utility or lack thereof of pre-endoscopy elective intubation. Intubation prior to endoscopic therapy should not routinely be employed and should be carefully considered for each individual patient with hemodynamics, mental status and respiratory status in mind.

Proton pump inhibitor administration following esophageal variceal ligation has been demonstrated to speed the resolution of post-ligation ulcer healing [27], but their use has been challenged [28]. In a retrospective cohort study, prolonged infusion of proton pump inhibitors in addition to octreotide infusion was compared to short term infusion. No difference in transfusion requirements or mortality rate was observed [29]. However longer proton pump inhibitor infusions may have been selectively administered to patients who were more critically ill, biasing the results.